Semantic segmentation in real-world applications often requires not only accurate masks but also strict adherence to textual labeling guidelines. These guidelines are typically complex and long, and both human and automated labeling often fail to follow them faithfully. Traditional approaches depend on expensive task-specific retraining that must be repeated as the guidelines evolve. Although recent open-vocabulary segmentation methods excel with simple prompts, they often fail when confronted with sets of paragraph-length guidelines that specify intricate segmentation rules. To address this, we introduce a multi-agent, training-free framework that coordinates general-purpose vision-language models within an iterative Worker-Supervisor refinement architecture. The Worker performs the segmentation, the Supervisor critiques it against the retrieved guidelines, and a lightweight reinforcement learning stop policy decides when to terminate the loop, ensuring guideline-consistent masks while balancing resource use. Evaluated on the Waymo and ReasonSeg datasets, our method notably outperforms state-of-the-art baselines, demonstrating strong generalization and instruction adherence.

Artificial intelligence (AI) models for computer vision trained with supervised machine learning are assumed to solve classification tasks by imitating human behavior learned from training labels. Most efforts in recent vision research focus on measuring the model task performance using standardized benchmarks such as accuracy. However limited work has sought to understand the perceptual difference between humans and machines. To fill this gap, this study first analyzes the statistical distributions of mistakes from the two sources, and then explores how task difficulty level affects these distributions. We find that even when AI learns an excellent model from the training data, one that outperforms humans in overall accuracy, these AI models have significant and consistent differences from human perception. We demonstrate the importance of studying these differences with a simple human-AI teaming algorithm that outperforms humans alone, AI alone, or AI-AI teaming.

Medical image segmentation is crucial for diagnosis and treatment planning. Traditional CNN-based models, like U-Net, have shown promising results but struggle to capture long-range dependencies and global context. To address these limitations, we propose a transformer based architecture that jointly applies Channel Attention and Pyramid Attention mechanisms to improve multi-scale feature extraction and enhance segmentation performance for medical images. We evaluate our approach on the Synapse multi-organ segmentation dataset and report significant improvements in Mean Dice score and Hausdorff Distance (HD95) compared to TransUNet, a widely adopted baseline. Our model also outperforms other state-of-the-art methods, demonstrating improved segmentation accuracy for complex anatomical structures.

Vehicle object detection benefits from both LiDAR and camera data, with LiDAR offering superior performance in many scenarios. Fusion of these modalities further enhances accuracy, but existing methods often introduce complexity or dataset-specific dependencies. In our study, we propose a model-adaptive late-fusion method, VaLID, which validates whether each predicted bounding box is acceptable or not. Our method verifies the higher-performing, yet overly optimistic LiDAR model detections using camera detections that are obtained from either specially trained, general, or open-vocabulary models. VaLID uses a lightweight neural verification network trained with a high recall bias to reduce the false predictions made by the LiDAR detector, while still preserving the true ones. Evaluating with multiple combinations of LiDAR and camera detectors on the KITTI dataset, we reduce false positives by an average of 63.9%, thus outperforming the individual detectors on 3D average precision (3DAP). Our approach is model-adaptive and demonstrates state-of-the-art competitive performance even when using generic camera detectors that were not trained specifically for this dataset.

Avatar creation from human images presents challenges for direct neural approaches, which suffer from inconsistent predictions and poor interpretability due to the large parameter space with hundreds of ambiguous options. We propose a neurosymbolic tag-based annotation method that combines neural perceptual learning with symbolic semantic reasoning. Instead of directly predicting avatar parameters, our approach uses a neural network to predict semantic tags (hair length, curliness, direction) as an intermediate symbolic representation, then applies symbolic search algorithms to match optimal avatar assets. This neurosymbolic design produces higher annotator agreements (96.7% vs 31.0% for direct annotation), enables more consistent model predictions, and provides interpretable avatar selection with ranked alternatives. The tag-based system generalizes easily across rendering systems, requiring only new asset annotation while reusing human image tags. Experimental results demonstrate superior convergence, consistency, and visual quality compared to direct prediction methods, showing how neurosymbolic approaches can improve trustworthiness and interpretability in creative AI applications.

Vision-language models (VLMs) have shown strong performance on text-to-image retrieval benchmarks. However, bridging this success to real-world applications remains a challenge. In practice, human search behavior is rarely a one-shot action. Instead, it is often a multi-round process guided by clues in mind. That is, a mental image ranging from vague recollections to vivid mental representations of the target image. Motivated by this gap, we study the task of Mental Image Retrieval (MIR), which targets the realistic yet underexplored setting where users refine their search for a mentally envisioned image through multi-round interactions with an image search engine. Central to successful interactive retrieval is the capability of machines to provide users with clear, actionable feedback; however, existing methods rely on indirect or abstract verbal feedback, which can be ambiguous, misleading, or ineffective for users to refine the query. To overcome this, we propose GenIR, a generative multi-round retrieval paradigm leveraging diffusion-based image generation to explicitly reify the AI system's understanding at each round. These synthetic visual representations provide clear, interpretable feedback, enabling users to refine their queries intuitively and effectively. We further introduce a fully automated pipeline to generate a high-quality multi-round MIR dataset. Experimental results demonstrate that GenIR significantly outperforms existing interactive methods in the MIR scenario. This work establishes a new task with a dataset and an effective generative retrieval method, providing a foundation for future research in this direction

We present SplatFace, a novel Gaussian splatting framework designed for 3D human face reconstruction without reliance on accurate pre-determined geometry. Our method is designed to simultaneously deliver both high-quality novel view rendering and accurate 3D mesh reconstructions. We incorporate a generic 3D Morphable Model (3DMM) to provide a surface geometric structure, making it possible to reconstruct faces with a limited set of input images. We introduce a joint optimization strategy that refines both the Gaussians and the morphable surface through a synergistic non-rigid alignment process. A novel distance metric, splat-to-surface, is proposed to improve alignment by considering both the Gaussian position and covariance. The surface information is also utilized to incorporate a world-space densification process, resulting in superior reconstruction quality. Our experimental analysis demonstrates that the proposed method is competitive with both other Gaussian splatting techniques in novel view synthesis and other 3D reconstruction methods in producing 3D face meshes with high geometric precision.

We present a novel generative model that synthesizes photorealistic, biophysically plausible faces by capturing the intricate relationships between facial geometry and biophysical attributes. Our approach models facial appearance in a biophysically grounded manner, allowing for the editing of both high‐level attributes such as age and gender, as well as low‐level biophysical properties such as melanin level and blood content. This enables continuous modeling of physical skin properties that correlate changes in skin properties with shape changes. We showcase the capabilities of our framework beyond its role as a generative model through two practical applications: editing the texture maps of 3D faces that have already been captured, and serving as a strong prior for face reconstruction when combined with differentiable rendering. Our model allows for the creation of physically‐based relightable, editable faces with consistent topology and uv layout that can be integrated into traditional computer graphics pipelines.

This research-to-practice WIP paper describes an approach to engage high school students in research through the utilization of citizen science tools embedded with Machine Learning (ML) models. In the context of fostering early engagement in scientific research among high school students, this paper explores the integration of citizen science and ML using SmartCS, an existing platform for creating citizen science smartphone applications. The process requires no prior programming knowledge, making it accessible to a broad range of students. For our approach, a group of high school students participated in a two-month-long summer research program, where they were introduced to the principles of citizen science as a method for data collection across diverse scientific projects from different research domains. The program's initial task involved students in the conceptualization of a citizen science project, adopted based on a thorough literature review, followed by the practical task of developing a smartphone application for data collection and educational purposes. Students either created new datasets or curated existing ones to train lightweight ML models for computer vision tasks, specifically focused on providing visual guidance within these mobile apps. The final task involved deploying these applications for public use and collecting user feedback. Our experience suggests that this approach not only enabled students to learn aspects of computer science and engineering, particularly in the area of ML model training and mobile application software development, but also allowed them to experience firsthand the significant role citizen science can play in collecting and analyzing scientific data.

We present a hybrid machine learning and flow analysis feature detection method, RipViz, to extract rip currents from stationary videos. Rip currents are dangerous strong currents that can drag beachgoers out to sea. Most people are either unaware of them or do not know what they look like. In some instances, even trained personnel such as lifeguards have difficulty identifying them. RipViz produces a simple, easy to understand visualization of rip location overlaid on the source video. With RipViz, we first obtain an unsteady 2D vector field from the stationary video using optical flow. Movement at each pixel is analyzed over time. At each seed point, sequences of short pathlines, rather a single long pathline, are traced across the frames of the video to better capture the quasi-periodic flow behavior of wave activity. Because of the motion on the beach, the surf zone, and the surrounding areas, these pathlines may still appear very cluttered and incomprehensible. Furthermore, lay audiences are not familiar with pathlines and may not know how to interpret them. To address this, we treat rip currents as a flow anomaly in an otherwise normal flow. To learn about the normal flow behavior, we train an LSTM autoencoder with pathline sequences from normal ocean, foreground, and background movements. During test time, we use the trained LSTM autoencoder to detect anomalous pathlines (i.e., those in the rip zone). The origination points of such anomalous pathlines, over the course of the video, are then presented as points within the rip zone. RipViz is fully automated and does not require user input. Feedback from domain expert suggests that RipViz has the potential for wider use.

We designed the LilyTiny sewable microcontroller over ten years ago in an effort to make electronic textiles more accessible to students, educators, and novices; it was meant to be affordable, easy to get started with, and well-supported by curriculum. We also designed the LilyTiny with hopes of helping to bridge the gap between e-textile activities using only lights and batteries – and those requiring knowledge of Arduino programming. Following its pilot, the LilyTiny was released as a commercial product through SparkFun Electronics, costing about $5 (USD) and shipping pre-programmed to control various LED behaviors. Free curriculum was released alongside it, detailing six low-cost activities that could be taught without any prior electronics experience.

Face models are widely used in image processing and other domains. The input data to create a 3D face model ranges from accurate laser scans to simple 2D RGB photographs. These input data types are typically deficient either due to missing regions, or because they are underconstrained. As a result, reconstruction methods include embedded priors encoding the valid domain of faces. System designers must choose a source of input data and then choose a reconstruction method to obtain a usable 3D face. If a particular application domain requires accuracy X, which kinds of input data are suitable? Does the input data need to be 3D, or will 2D data suffice? This paper takes a step toward answering these questions using synthetic data. A ground truth dataset is used to analyze accuracy obtainable from 2D landmarks, 3D landmarks, low quality 3D, high quality 3D, texture color, normals, dense 2D image data, and when regions of the face are missing. Since the data is synthetic it can be analyzed both with and without measurement error. This idealized synthetic analysis is then compared to real results from several methods for constructing 3D faces from 2D photographs. The experimental results suggest that accuracy is severely limited when only 2D raw input data exists.

The highest quality 3D face reconstructions are produced using multi-view stereo methods, reporting errors below 0.5mm. Unfortunately, these methods typically employ dozens of high-resolution cameras in a large laboratory capture gantry. In contrast, monocular 3D face reconstruction using sophisticated deep learning models are suited for casual mobile phone imaging outside the lab and report a mean error of 1-2mm.This paper investigates whether classic stereo methods can be used in scenarios with only a few low-resolution images available. We expect to find that it cannot since multi-view stereo performs well only when many high-resolution images are provided. When only two low-resolution images are available, stereo produces very noisy results which are not directly usable. Surprisingly, however, our analysis shows that this visually noisy data has lower error than comparison state-of-the-art methods. We find that the visual artifacts from stereo can be removed using a morphable face model to constrain face shape.

The capability to generate clear and bright images in low light situations is crucial for photographers, engineers, and researchers. When it is not possible to modify the imaging conditions, an algorithm to enhance images is needed. Traditional methods require manually adjusting parameters to tune the image. Supervised learning methods need to collect a large amount of paired data for training. In this paper, we demonstrate an semi-supervised method for low light image enhancement, using a chain of cycle consistent generators. We show the effectiveness of our method by comparing it to existing image enhancement methods, both using standard image quality metrics and by using human perceptual judgements. We include an ablation study for features in our model. Our proposed method is computationally efficient and does not require paired training data.

Stylized 3D avatars have become increasingly prominent in our modern life. Creating these avatars manually usually involves laborious selection and adjustment of continuous and discrete parameters and is time-consuming for average users. Self-supervised approaches to automatically create 3D avatars from user selfies promise high quality with little annotation cost but fall short in application to stylized avatars due to a large style domain gap. We propose a novel self-supervised learning framework to create high-quality stylized 3D avatars with a mix of continuous and discrete parameters. Our cascaded domain bridging framework first leverages a modified portrait stylization approach to translate input selfies into stylized avatar renderings as the targets for desired 3D avatars. Next, we find the best parameters of the avatars to match the stylized avatar renderings through a differentiable imitator we train to mimic the avatar graphics engine. To ensure we can effectively optimize the discrete parameters, we adopt a cascaded relaxation-and-search pipeline. We use a human preference study to evaluate how well our method preserves user identity compared to previous work as well as manual creation. Our results achieve much higher preference scores than previous work and close to those of manual creation. We also provide an ablation study to justify the design choices in our pipeline.

This Innovative Practice Work in Progress paper reports on our experience scaffolding student success in the uncertain landscape of open source. Following participation in a faculty workshop on the subject, the first author spent two consecutive terms developing, teaching, and revising an upper-division open source software course. The difference between the two course offerings was astounding; students enrolled in the second iteration made more successful project contributions, spent more of their own time working outside of class, and felt a greater connection to both the project and the developer community of which they were a part. We detail our experiences here, with particular focus on the importance of project selection - as well as the revisions we believe to be most responsible for improvement: additional mentorship, supplemental in-class tutorials, more dedicated class time for teamwork, intentional team groupings, and access to large screens for collaboration.

In this Innovative Practice Full Paper, we report on a CS0-level computational craft course added to our departmental offerings in hopes of further broadening participation. We summarize the course design and structure, which emphasize algorithmic design (using Processing), handcraft, and digital fabrication. We share examples of creative computational work and feedback from students, as well as reflections on the course's efficacy within our funnel-style curriculum. Early evidence suggests that the course offers a highly personal and creative entry point to computing – and one that is effective at engaging a diversity of students while ensuring a smooth transition to CS1.

Face models are widely used in image processing and other domains. The input data to create a 3D face model ranges from accurate laser scans to simple 2D RGB photographs. System designers must choose a source of input data and then choose a reconstruction method to obtain a usable 3D face. If a particular application domain requires accuracy X, which kinds of input data are suitable? This paper takes a step toward answering this question. A variety of common input data types such as 2D landmarks and 3D scans are constructed from an existing high quality dataset. A morphable face model is then used to reconstruct 3D faces. By comparing to ground truth, an analysis of the relative error between different data types is obtained.

Research experiences today are limited to a privileged few at select universities. Providing open access to research experiences would enable global upward mobility and increased diversity in the scientific workforce. How can we coordinate a crowd of diverse volunteers on open-ended research? How could a PI have enough visibility into each person's contributions to recommend them for further study? We present Crowd Research, a crowdsourcing technique that coordinates open-ended research through an iterative cycle of open contribution, synchronous collaboration, and peer assessment. To aid upward mobility and recognize contributions in publications, we introduce a decentralized credit system: participants allocate credits to each other, which a graph centrality algorithm translates into a collectively-created author order. Over 1,500 people from 62 countries have participated, 74% from institutions with low access to research. Over two years and three projects, this crowd has produced articles at top-tier Computer Science venues, and participants have gone on to leading graduate programs.

We present a novel phase unwrapping framework for the Time-of-Flight sensor that can match the performance of systems using two modulation frequencies, within a single shot. Our framework is based on an interleaved pixel arrangement, where a pixel measures phase at a different modulation frequency from its neighboring pixels. We demonstrate that: (1) it is practical to capture ToF images that contain phases from two frequencies in a single shot, with no loss in signal fidelity, (2) phase unwrapping can be effectively performed on such an interleaved phase image, and (3) our method preserves the original spatial resolution. We find that the output of our framework is comparable to results using two shots under separate modulation frequencies, and is significantly better than using a single modulation frequency.

Convolutional Neural Networks (CNN) are state-of-the-art models for many image classification tasks. However, to recognize cancer subtypes automatically, training a CNN on gigapixel resolution Whole Slide Tissue Images (WSI) is currently computationally impossible. The differentiation of cancer subtypes is based on cellular-level visual features observed on image patch scale. Therefore, we argue that in this situation, training a patch-level classifier on image patches will perform better than or similar to an image-level classifier. The challenge becomes how to intelligently combine patch-level classification results and model the fact that not all patches will be discriminative. We propose to train a decision fusion model to aggregate patch-level predictions given by patch-level CNNs, which to the best of our knowledge has not been shown before. Furthermore, we formulate a novel Expectation-Maximization (EM) based method that automatically locates discriminative patches robustly by utilizing the spatial relationships of patches. We apply our method to the classification of glioma and non-small-cell lung carcinoma cases into subtypes. The classification accuracy of our method is similar to the inter-observer agreement between pathologists. Although it is impossible to train CNNs on WSIs, we experimentally demonstrate using a comparable non-cancer dataset of smaller images that a patch-based CNN can outperform an image-based CNN.

We present a novel system which incorporates photometric stereo with the Time-of-Flight depth sensor. Adding to the classic ToF, the system utilizes multiple point light sources that enable the capturing of a normal field whilst taking depth images. Two calibration methods are proposed to determine the light sources' positions given the ToF sensor's relatively low resolution. An iterative refinement algorithm is formulated to account for the extra phase delays caused by the positions of the light sources. We find in experiments that the system is comparable to the classic ToF in depth accuracy, and it is able to recover finer details that are lost due to the noise level of the ToF sensor.

Given the growth of Internet photo collections, we now have a visual index of all major cities and tourist sites in the world. However, it is still a difficult task to capture that perfect shot with your own camera when visiting these places, especially when your camera itself has limitations, such as a limited field of view. In this paper, we propose a framework to overcome the imperfections of personal photographs of tourist sites using the rich information provided by large-scale Internet photo collections. Our method deploys state-of-the-art techniques for constructing initial 3D models from photo collections. The same techniques are then used to register personal photographs to these models, allowing us to augment personal 2D images with 3D information. This strong available scene prior allows us to address a number of traditionally challenging image enhancement techniques and achieve high-quality results using simple and robust algorithms. Specifically, we demonstrate automatic foreground segmentation, mono-to-stereo conversion, field-of-view expansion, photometric enhancement, and additionally automatic annotation with geolocation and tags. Our method clearly demonstrates some possible benefits of employing the rich information contained in online photo databases to efficiently enhance and augment one's own personal photographs.

We present a new upsampling method to enhance the spatial resolution of depth images. Given a low-resolution depth image from an active depth sensor and a potentially high-resolution color image from a passive RGB camera, we formulate it as an adaptive cost aggregation problem and solve it using the bilateral filter. The formulation synergistically combines the median and bilateral filters thus it better preserves the depth edges and is more robust to noise. Numerical and visual evaluations on a total of 37 Middlebury data sets demonstrate the effectiveness of our method. A real-time high-resolution depth capturing system is also developed using commercial active depth sensor based on the proposed upsampling method.

3D displays are increasingly popular in consumer and commercial applications. Many such displays show 3D images to viewers wearing special glasses, while showing an incomprehensible double image to viewers without glasses. We demonstrate a simple method that provides those with glasses a 3D experience, while viewers without glasses see a 2D image without artifacts. In addition to separate left and right images in each frame, we add a third image, invisible to those with glasses. In the combined view seen by those without glasses, this cancels the right image, leaving only the left. If the left and right images are of equal brightness, this approach results in low contrast for viewers without glasses. Allowing differential brightness between the left and right images improves 2D contrast. We observe experimentally that: (1) viewers without glasses prefer our 3D+2DTV to a standard 3DTV, (2) viewers with glasses maintain a strong 3D percept, even when one eye is significantly darker than the other, and (3) sequential-stereo display viewers with glasses experience a depth illusion caused by the Pulfrich effect, but it is small and innocuous. Our technique is applicable to displays using either active shutter glasses or passive glasses. Our prototype uses active shutter glasses and a polarizer.

This paper uses GPS loggers and interviews to measure the time taken to collect water in two Kenyan informal settlements. The time devoted to water collection is widely believed to prevent women and girls, who do most of this work, from undertaking more creative tasks, including income generation and education. We studied collection times in two settlements to compare Nyalenda in Kisumu, where the utility has introduced a new piped water system, with Kibera in Nairobi, where no such improvement has been made. In addition to the primary results of quantitative collections times, we discuss the use of GPS in this context and our findings that the two methods of measurement provide insights which neither would have provided alone.

The reflectance function of a scene point captures the appearance of that point as a function of lighting direction. We present an approach to printing the reflectance functions of an object or scene so that its appearance is modified correctly as a function of the lighting conditions when viewing the print. For example, such a “photograph” of a statue printed with our approach appears to cast shadows to the right when the “photograph” is illuminated from the left. Viewing the same print with lighting from the right will cause the statue's shadows to be cast to the left. Beyond shadows, all effects due to the lighting variation, such as Lambertian shading, specularity, and inter-reflection can be reproduced. We achieve this ability by geometrically and photometrically controlling specular highlights on the surface of the print. For a particular viewpoint, arbitrary reflectance functions can be built up at each pixel by controlling only the specular highlights and avoiding significant diffuse reflections. Our initial binary prototype uses halftoning to approximate continuous grayscale reflectance functions.

This poster describes an alternative approach to handling the best document selection problem. Best document selection is a common problem with many real world applications, but is not a well studied by itself; a simple solution would be to treat it as a ranking problem and to use existing ranking algorithms to rank all documents. We could then select only the first element of the sorted list. However, because ranking models optimize for all ranks, the model may sacrifice accuracy of the top rank for the sake of overall accuracy. This is an unnecessary trade-off. We begin by first defining an appropriate objective function for the domain, then create a boosting algorithm that explicitly targets this function. Based on experiments on a benchmark retrieval data set and Digg.com news commenting data set, we find that even a simple algorithm built for this specific problem gives better results than baseline algorithms that were designed for the more complicated ranking tasks. 1.

Time-of-flight range sensors have error characteristics, which are complementary to passive stereo. They provide real-time depth estimates in conditions where passive stereo does not work well, such as on white walls. In contrast, these sensors are noisy and often perform poorly on the textured scenes where stereo excels. We explore their complementary characteristics and introduce a method for combining the results from both methods that achieve better accuracy than either alone. In our fusion framework, the depth probability distribution functions from each of these sensor modalities are formulated and optimized. Robust and adaptive fusion is built on a pixel-wise reliability weighting function calculated for each method. In addition, since time-of-flight devices have primarily been used as individual sensors, they are typically poorly calibrated. We introduce a method that substantially improves upon the manufacturer's calibration. We demonstrate that our proposed techniques lead to improved accuracy and robustness on an extensive set of experimental results.

Desktop video conferencing often suffers from bad lighting, which may be caused by harsh shadowing, saturated regions, etc. The primary reason for this is the lack of control over lighting in the user's environment. A hardware-based solution to this problem would be to place lights near the video camera, but these would be distracting to the user. We use a set of infrared lights placed around the computer monitor to gather a sequence of frames which is used to infer surface normals of the scene. These are used in combination with a visible spectrum image to create an improved relighting result.

Designing a robot controller that can optimally manage limited resources in a deterministic, real-time manner can be challenging. Behavior-based architectures, which split autonomy into levels, are very popular but neither have real-time features that enforce timing constraints nor support determinism. Even though real-time features are not included, it seems like a natural fit to make each level in the behavior-based architecture its own task or process. The only that thing that it lacks are the timing features. This has already been implemented using Suns Java Real-Time System. It has also been shown that timing constraints effect performance. This brings us to the question; why not use the more traditional language of C or C++ to implement this behavior-based real-time architecture? Are we not taught that Java is useful but slow compared to C and C++? If so why not use C++ and the features of Open Robot Control Software (OROCOS) to implement the architecture. This paper answers the question of does it really matter what language is used in a behavior-based real-time architecture. We implemented the architecture using OROCOS/C++ running on UBUNTU. Then compared our implementation to two other implementations of the architecture: Java/Player on Fedora; and Suns Java Real-Time System (RTS) on Solaris. Results, from experiments on a robot, show that our OROCOS/C++ implementation performed similarly to the Java RTS implementation. Both the OROCOS and Java RTS implementations performed better than the Player/Java implementation. This suggests that Java is in fact feasible for a behavior-based real-time robot architecture but it needs to be run using Java RTS not the regular version.

Depth maps captured with time-of-flight cameras have very low data quality: the image resolution is rather limited and the level of random noise contained in the depth maps is very high. Therefore, such flash lidars cannot be used out of the box for high-quality 3D object scanning. To solve this problem, we present LidarBoost, a 3D depth superresolution method that combines several low resolution noisy depth images of a static scene from slightly displaced viewpoints, and merges them into a high-resolution depth image. We have developed an optimization framework that uses a data fidelity term and a geometry prior term that is tailored to the specific characteristics of flash lidars. We demonstrate both visually and quantitatively that LidarBoost produces better results than previous methods from the literature.

Segmenting images into distinct material types is a very useful capability. Most work in image segmentation addresses the case where only a single image is available. Some methods improve on this by collecting HDR or multispectral images. However, it is also possible to use the reflectance properties of the materials to obtain better results. By acquiring many images of an object under different lighting conditions we have more samples of the surfaces bidirectional reflectance distribution function (BRDF). We show that this additional information enlarges the class of material types that can be well separated by segmentation, and that properly treating the information as samples of the BRDF further increases accuracy without requiring an explicit estimation of the material BRDF.

We show that multi-path analysis using images from a timeof-flight (ToF) camera provides a tantalizing opportunity to infer about 3D geometry of not only visible but hidden parts of a scene. We provide a novel framework for reconstructing scene geometry from a single viewpoint using a camera that captures a 3D time-image I(x, y, t) for each pixel. We propose a framework that uses the time-image and transient reasoning to expose scene properties that may be beyond the reach of traditional computer vision. We corroborate our theory with free space hardware experiments using a femtosecond laser and an ultrafast photo detector array. The ability to compute the geometry of hidden elements, unobservable by both the camera and illumination source, will create a range of new computer vision opportunities.

There is a wealth of information collected about national level socio-economic indicators across all countries each year. These indicators are important in recognizing the level of development in certain aspects of a particular country, and are also essential in international policy making. However with past data spanning several decades and many hundreds of indicators evaluated, trying to get an intuitive sense of this data has in a way become more difficult. This is because simple indicator-wise visualization of data such as line/bar graphs or scatter plots does not do a very good job of analyzing the underlying associations or behavior. Therefore most of the socio-economic analysis regarding development tends to be focused on few main economic indicators. However, we believe that there are valuable insights to be gained from understanding how the multitude of social, economic, educational and health indicators relate to each other. The focus of our work is to provide an integration of statistical analysis with visualization to gain new socio-economic insights and knowledge. We compute correlation and linear regression between indicators using time-series data. We cluster countries based on indicator trends and analyze the results of the clustering to identify similarities and anomalies. The results are shown on a correlation or regression grid and can be visualized on a world map using a flexible interactive visualization system. This work provides a pathway to exploring deeper relationships between socio-economic indicators and countries in the hands of the user, and carries the potential for identifying important underpinnings of policy changes.

Time-of-flight range sensors have error characteristics which are complementary to passive stereo. They provide real time depth estimates in conditions where passive stereo does not work well, such as on white walls. In contrast, these sensors are noisy and often perform poorly on the textured scenes for which stereo excels. We introduce a method for combining the results from both methods that performs better than either alone. A depth probability distribution function from each method is calculated and then merged. In addition, stereo methods have long used global methods such as belief propagation and graph cuts to improve results, and we apply these methods to this sensor. Since time-of-flight devices have primarily been used as individual sensors, they are typically poorly calibrated. We introduce a method that substantially improves upon the manufacturerpsilas calibration. We show that these techniques lead to improved accuracy and robustness.

As the applications delivered by cellular phones are becoming increasingly sophisticated, the importance of choosing an input strategy is also growing. Touch-screens can simplify navigation by far but the vast majority of phones on the market are not equipped with them. Cameras, on the other hand, are widespread even amongst low-end phones: In this paper we propose a vision-based pointing system that allows the user to control the pointer's position by just waving a hand, with no need for additional hardware.

In this paper, we present a keyframe-based human motion capture data retrieval system which uses a wooden doll as the input device. A user inputs a keyframe by posing an artist’s doll with painted joints in front of a stereo camera rig. The system interactively gives real-time feedback on the results of the joint detection and 3D pose reconstruction as the user is positioning and rotating the doll. The robust 3D joint reconstruction is achieved by integrating 3D joint positions from multiple views of the same pose. After the user has finished inputting all the keyframes, the motion sequences are retrieved from the database and ranked based on their similarities to the keyframes. Experiments show that the presented system is simple to use and has high quality of retrieval results.

The same game is often played in real and virtual worlds. We integrate in-person and on-line playing of board games such as Go, bringing the real world into the virtual world. A player may record an in-person game by placing their camera on the table next to the game board, taking photos of the game. After automatically detecting the board and playing pieces, we perform inference on the time series of detections to eliminate errors and accurately estimate long sequences of moves. The game transcript may be studied afterwards, shared with friends and teachers, or added to online compilations, bringing the attendant benefits of online game play to an in-person game. We present a method to transcribe the moves of a board game in the real world, automatically and unobtrusively. A player simply places a camera on the table next to the board, recording a series of photos; our algorithm automatically finds the board and detects moves, creating a complete transcription of the game.

Abstract Inappropriate lighting is often responsible for poor quality video. In most offices and homes, lighting is not designed for video conferencing. This can result in unevenly lit faces, distracting shadows, and unnatural colors. We present a method for relighting faces that reduces the effects of uneven lighting and color. Our setup consists of a compact lighting rig and a camera that is both inexpensive and inconspicuous to the user. We use unperceivable infrared (IR) lights to obtain an illumination bases of the scene. Our algorithm computes an optimally weighted combination of IR bases to minimize lighting inconsistencies in foreground areas and reduce the effects of colored monitor light. However, IR relighting alone results in images with an unnatural ghostly appearance, thus a retargeting technique is presented which removes the unnatural IR effects and produces videos that have substantially more balanced intensity and color than the original video.

Much of the research on video-based human motion capture assumes the body shape is known a priori and is represented coarsely (e.g. using cylinders or superquadrics to model limbs). These body models stand in sharp contrast to the richly detailed 3D body models used by the graphics community. Here we propose a method for recovering such models directly from images. Specifically, we represent the body using a recently proposed triangulated mesh model called SCAPE which employs a low-dimensional, but detailed, parametric model of shape and pose-dependent deformations that is learned from a database of range scans of human bodies. Previous work showed that the parameters of the SCAPE model could be estimated from marker-based motion capture data. Here we go further to estimate the parameters directly from image data. We define a cost function between image observations and a hypothesized mesh and formulate the problem as optimization over the body shape and pose parameters using stochastic search. Our results show that such rich generative models enable the automatic recovery of detailed human shape and pose from images.

We present a new post-processing step to enhance the resolution of range images. Using one or two registered and potentially high-resolution color images as reference, we iteratively refine the input low-resolution range image, in terms of both its spatial resolution and depth precision. Evaluation using the Middlebury benchmark shows across-the-board improvement for sub-pixel accuracy. We also demonstrated its effectiveness for spatial resolution enhancement up to 100 times with a single reference image.

We introduce a theoretical framework and practical algorithms for replacing time-coded structured light patterns with viewpoint codes, in the form of additional camera locations. Current structured light methods typically use log(N) light patterns, encoded over time, to unambiguously reconstruct N unique depths. We demonstrate that each additional camera location may replace one frame in a temporal binary code. Our theoretical viewpoint coding analysis shows that, by using a high frequency stripe pattern and placing cameras in carefully selected locations, the epipolar projection in each camera can be made to mimic the binary encoding patterns normally projected over time. Results from our practical implementation demonstrate reliable depth reconstruction that makes neither temporal nor spatial continuity assumptions about the scene being captured.

Video matting is the process of taking a sequence of frames, isolating the foreground, and replacing the background in each frame. We look at existing single-frame matting techniques and present a method that improves upon them by adding depth information acquired by a time-offlight range scanner. We use the depth information to automate the process so it can be practically used for video sequences. In addition, we show that we can improve the results from natural matting algorithms by adding a depth channel. The additional depth information allows us to reduce the artifacts that arise from ambiguities that occur when an object is a similar color to its background.

Abstract Hand held long exposures often result in blurred photographs due to camera shake. Long exposures are desirable both for artistic effect and in low‐light situations. We propose a novel method for digitally reducing imaging artifacts, which does not require additional hardware such as tripods or optical image stabilization lenses. A series of photographs is acquired with short shutter times, stabilized using image alignment, and then composited. Our method is capable of reducing noise and blurring due to camera shake, while simultaneously preserving the desirable effects of motion blur. The resulting images are very similar to those obtained using a tripod and a true extended exposure.

Nearly all existing methods for stereo reconstruction assume that scene reflectance is Lambertian and make use of brightness constancy as a matching invariant. We introduce a new invariant for stereo reconstruction called light transport constancy (LTC), which allows completely arbitrary scene reflectance (bidirectional reflectance distribution functions (BRDFs)). This invariant can be used to formulate a rank constraint on multiview stereo matching when the scene is observed by several lighting configurations in which only the lighting intensity varies. In addition, we show that this multiview constraint can be used with as few as two cameras and two lighting configurations. Unlike previous methods for BRDF invariant stereo, LTC does not require precisely configured or calibrated light sources or calibration objects in the scene. Importantly, the new constraint can be used to provide BRDF invariance to any existing stereo method whenever appropriate lighting variation is available.

Nearly all existing methods for stereo reconstruction assume that scene reflectance is Lambertian, and make use of color constancy as a matching invariant. We introduce a new invariant for stereo reconstruction called light transport constancy, which allows completely arbitrary scene reflectance (BRDFs). This invariant can be used to formulate a rank constraint on multiview stereo matching when the scene is observed in several lighting configurations. In addition, we show that this multiview constraint can be used with as few as two cameras and two lighting configurations. Unlikely previous methods for BRDF invariant stereo, light transport constancy does not require precisely configured or calibrated light sources, nor calibration objects in the scene. Importantly, the new constraint can be used to provide BRDF invariance to any existing stereo method, whenever appropriate lighting variation is available.

Most dense stereo correspondence algorithms start by establishing discrete pixel matches and later refine these matches to sub-pixel precision. Traditional sub-pixel refinement methods attempt to determine the precise location of points, in the secondary image, that correspond to discrete positions in the reference image. We show that this strategy can lead to a systematic bias associated with the violation of the general symmetry of matching cost functions. This bias produces random or coherent noise in the final reconstruction, but can be avoided by refining both image coordinates simultaneously, in a symmetric way. We demonstrate that the symmetric sub-pixel refinement strategy results in more accurate correspondences by avoiding bias while preserving detail.

Depth from triangulation has traditionally been investigated in a number of independent threads of research, with methods such as stereo, laser scanning, and coded structured light considered separately. In this paper, we propose a common framework called spacetime stereo that unifies and generalizes many of these previous methods. To show the practical utility of the framework, we develop two new algorithms for depth estimation: depth from unstructured illumination change and depth estimation in dynamic scenes. Based on our analysis, we show that methods derived from the spacetime stereo framework can be used to recover depth in situations in which existing methods perform poorly.

We introduce the SCAPE method (Shape Completion and Animation for PEople)---a data-driven method for building a human shape model that spans variation in both subject shape and pose. The method is based on a representation that incorporates both articulated and non-rigid deformations. We learn a pose deformation model that derives the non-rigid surface deformation as a function of the pose of the articulated skeleton. We also learn a separate model of variation based on body shape. Our two models can be combined to produce 3D surface models with realistic muscle deformation for different people in different poses, when neither appear in the training set. We show how the model can be used for shape completion --- generating a complete surface mesh given a limited set of markers specifying the target shape. We present applications of shape completion to partial view completion and motion capture animation. In particular, our method is capable of constructing a high-quality animated surface model of a moving person, with realistic muscle deformation, using just a single static scan and a marker motion capture sequence of the person.

Range scanning, manual 3D editing, and other modeling approaches can provide information about the geometry of surfaces in the form of either 3D positions (e.g., triangle meshes or range images) or orientations (normal maps or bump maps). We present an algorithm that combines these two kinds of estimates to produce a new surface that approximates both. Our formulation is linear, allowing it to operate efficiently on complex meshes commonly used in graphics. It also treats high-and low-frequency components separately, allowing it to optimally combine outputs from data sources such as stereo triangulation and photometric stereo, which have different error-vs.-frequency characteristics. We demonstrate the ability of our technique to both recover high-frequency details and avoid low-frequency bias, producing surfaces that are more widely applicable than position or orientation data alone.

We present an algorithm for creating globally consistent three-dimensional maps from depth fields produced by camera-based range measurement systems. Our approach is specifically suited to dealing with the high noise levels and the large number of outliers often produced by such systems. Range data is filtered to reject outliers within each scan. The point-to-plane variant of ICP is used for local alignment, including weightings that favor nearby points and a novel outlier rejection strategy that increases the robustness for this class of data while eliminating the burden of user-specified thresholds. Global consistency is imposed on cycles by optimally distributing the cyclic discrepancy according to the local fit correlation matrices. The algorithm is demonstrated on a dataset collected by an active unstructured-light space-time stereo vision system.

We present an unsupervised algorithm for registering 3D surface scans of an object undergoing significant deformations. Our algorithm does not need markers, nor does it assume prior knowledge about object shape, the dynamics of its deformation, or scan alignment. The algorithm registers two meshes by optimizing a joint probabilistic model over all point-to- point correspondences between them. This model enforces preservation of local mesh geometry, as well as more global constraints that capture the preservation of geodesic distance between corresponding point pairs. The algorithm applies even when one of the meshes is an incomplete range scan; thus, it can be used to automatically fill in the remaining sur- faces for this partial scan, even if those surfaces were previously only seen in a different configuration. We evaluate the algorithm on several real-world datasets, where we demonstrate good results in the presence of significant movement of articulated parts and non-rigid surface defor- mation. Finally, we show that the output of the algorithm can be used for compelling computer graphics tasks such as interpolation between two scans of a non-rigid object and automatic recovery of articulated object models. 1 Introduction The construction of 3D object models is a key task for many graphics applications. It is becoming increasingly common to acquire these models from a range scan of a physical object. This paper deals with an important subproblem of this acquisition task -- the problem of registering two deforming surfaces corresponding to different configurations of the same non-rigid object. The main difficulty in the 3D registration problem is determining the correspondences of points on one surface to points on the other. Local regions on the surface are rarely distinc- tive enough to determine the correct correspondence, whether because of noise in the scans, or because of symmetries in the object shape. Thus, the set of candidate correspondences to a given point is usually large. Determining the correspondence for all object points results in a combinatorially large search problem. The existing algorithms for deformable surface A results video is available at http://robotics.stanford.edu/drago/cc/video.mp4 Figure 1: A) Registration results for two meshes. Nonrigid ICP and its variant augmented with spin images get stuck in local maxima. Our CC algorithm produces a largely correct registration, although with an artifact in the right shoulder (inset). B) Illustration of the link deformation process C) The CC algorithm which uses only deformation potentials can violate mesh geometry. Near regions can map to far ones (segment AB) and far regions can map to near ones (points C,D). registration make the problem tractable by assuming significant prior knowledge about the objects being registered. Some rely on the presence of markers on the object [1, 20], while others assume prior knowledge about the object dynamics [16], or about the space of non- rigid deformations [15, 5]. Algorithms that make neither restriction [18, 12] simplify the problem by decorrelating the choice of correspondences for the different points in the scan. However, this approximation is only good in the case when the object deformation is small; otherwise, it results in poor local maxima as nearby points in one scan are allowed to map to far-away points in the other. Our algorithm defines a joint probabilistic model over all correspondences, which ex- plicitly model the correlations between them -- specifically, that nearby points in one mesh should map to nearby points in the other. Importantly, the notion of "nearby" used in our model is defined in terms of geodesic distance over the mesh. We define a probabilistic model over the set of correspondences, that encodes these geodesic distance constraints as well as penalties for link twisting and stretching, and high-level local surface features [14]. We then apply loopy belief propagation [21] to this model, in order to solve for the entire set of correspondences simultaneously. The result is a registration that respects the surface geometry. To the best of our knowledge, the algorithm we present in this paper is the first algorithm which allows the registration of 3D surfaces of an object where the object config- urations can vary significantly, there is no prior knowledge about object shape or dynamics of deformation, and nothing whatsoever is known about the object alignment. Moreover, unlike many methods, our algorithm can be used to register a partial scan to a complete model, greatly increasing its applicability. We apply our approach to three datasets containing 3D scans of a wooden puppet, a human arm and entire human bodies in different configurations. We demonstrate good registration results for scan pairs exhibiting articulated motion, non-rigid deformations, or both. We also describe three applications of our method. In our first application, we show how a partial scan of an object can be registered onto a fully specified model in a dif- ferent configuration. The resulting registration allows us to use the model to "complete" the partial scan in a way that preserves the local surface geometry. In the second, we use the correspondences found by our algorithm to smoothly interpolate between two different poses of an object. In our final application, we use a set of registered scans of the same object in different positions to recover a decomposition of the object into approximately rigid parts, and recover an articulated skeleton linking the parts. All of these applications are done in an unsupervised way, using only the output of our Correlated Correspondence algorithm applied to pairs of poses with widely varying deformations, and unknown initial alignments. These results demonstrate the value of a high-quality solution to the registra- tion problem to a range of graphics tasks. 2 Previous Work Surface registration is a fundamental building block in computer graphics. The classical so- lution for registering rigid surfaces is the Iterative Closest Point algorithm (ICP) [4, 6, 17]. Recently, there has been work extending ICP to non-rigid surfaces [18, 8, 12, 1]. These algorithms treat one of the scans (usually a complete model of the surface) as a deformable template. The links between adjacent points on the surface can be thought of as springs, which are allowed to deform at a cost. Similarly to ICP, these algorithms iterate between two subproblems -- estimating the non-rigid transformation and estimating the set of correspondences C between the scans. The step estimating the correspondences assumes that a good estimate of the nonrigid transformation is available. Under this assumption, the assignments to the correspondence variables become decorrelated: each point in the second scan is associated with the nearest point (in the Euclidean distance sense) in the deformed template scan. However, the decomposition also induces the algorithm's main limitation. By assigning points in the second scan to points on the deformed model inde- pendently, nearby points in the scan can get associated to remote points in the model if the estimate of is poor (Fig. 1A). While several approaches have been proposed to address this problem of incorrect correspondences, their applicability is largely limited to problems where the deformation is local, and the initial alignment is approximately correct. Another line of related work is the work on deformable template matching in the com- puter vision community. In the 3D case, this framework is used for detection of articulated object models in images [13, 22, 19]. The algorithms assume the decomposition of the object into a relatively small number of parts is known, and that a detector for each object part is available. Template matching approaches have also been applied to deformable 2D objects, where very efficient solutions exist [9, 11]. However, these methods do not extend easily to the case of 3D surfaces. 3 The Correlated Correspondence Algorithm The input to the algorithm is a set of two meshes (surfaces tessellated into polygons). The model mesh X = (V X , EX ) is a complete model of the object, in a particular pose. V X = (x1, . . . , xN ) denotes the mesh points, while EX is the set of links between adjacent points on the mesh surface. The data mesh Z = (V Z , EZ ) is either a complete model or a partial view of the object in a different configuration. Each data mesh point zk is associated with a correspondence variable ck, specifying the corresponding model mesh point. The task of registration is one of estimating the set of all correspondences C and a non-rigid transformation which aligns the corresponding points. 3.1 Probabilistic Model We formulate the registration problem as one of finding an embedding of the data mesh Z into the model mesh X, which is encoded as an assignment to all correspondence vari- ables C = (c1, . . . , cK ). The main idea behind our approach is to preserve the consis- tency of the embedding by explicitly correlating the assignments to the correspondence variables. We define a joint distribution over the correspondence variables c1, . . . , cK , rep- resented as a Markov network. For each pair of adjacent data mesh points zk, zl, we want to define a probabilistic potential (ck, cl) that constrains this pair of correspondences to reasonable and consistent. This gives rise to a joint probability distribution of the form p(C) = 1 (c (c Z k k ) k,l k , cl) which contains only single and pairwise potentials. Performing probabilistic inference to find the most likely joint assignment to the entire set of correspondence variables C should yield a good and consistent registration. Deformation Potentials. We want our model to encode a preference for embeddings of mesh Z into mesh X, which minimize the amount of deformation induced by the embedding. In order to quantify the amount of deformation , applied to the model, we will follow the ideas of Hahnel et al. [12] and treat the links in the set EX as springs, which resist stretching and twisting at their endpoints. Stretching is easily quantified by looking at changes in the link length induced by the transformation . Link twisting, however, is ill- specified by looking only at the Cartesian coordinates of the points alone. Following [12], we attach an imaginary local coordinate system to each point on the model. This local coordinate system allows us to quantify the "twist" of a point xj relative to a neighbor xi. A non-rigid transformation defines, for each point xi, a translation of its coordinates and a rotation of its local coordinate system. To evaluate the deformation penalty, we parameterize each link in the model in terms of its length and its direction relative to its endpoints (see Fig. 1B). Specifically, we define li,j to be the distance between xi and xj; dij is a unit vector denoting the direction of the point xj in the coordinate system of xi (and vice versa). We use ei,j to denote the set of edge parameters (li,j, dij, dji). It is now straightforward to specify the penalty for model deformations. Let be a transformation, and let ~ ei,j denote the triple of parameters associated with the link between xi and xj after applying . Our model penalizes twisting and stretching, using a separate zero-mean Gaussian noise model for each: P (~ ei,j | ei,j) = P (~li,j | li,j) P ( ~ dij | dij) P ( ~ dji | dji) (1) In the absence of prior information, we assume that all links are equally likely to deform. In order to quantify the deformation induced by an embedding C, we need to include a potential d(ck, cl) for each link eZ EZ . Every probability k,l d(ck = i, cl = j) corresponds to the deformation penalty incurred by deforming model link ei,j to generate link eZ and is defined in (1). We do not restrict ourselves to the set of links in EX , since k,l the original mesh tessellation is sparse and local. Any two points in X are allowed to implicitly define a link. Unfortunately, we cannot directly estimate the quantity P (eZ | e k,l i,j ), since the link pa- rameters eZ depend on knowing the nonrigid transformation, which is not given as part k,l of the input. The key issue is estimating the (unknown) relative rotation of the link end- points. In effect, this rotation is an additional latent variable, which must also be part of the probabilistic model. To remain within the realm of discrete Markov networks, allowing the application of standard probabilistic inference algorithms, we discretize the space of the possible rotations, and fold it into the domains of the correspondence variables. For each possible value of the correspondence variable ck = i we select a small set of candidate rotations, consistent with local geometry. We do this by aligning local patches around the points xi and zk using rigid ICP. We extend the domain of each correspondence variables ck, where each value encodes a matching point and a particular rotation from the precom- puted set for that point. Now the edge parameters eZ are fully determined and so is the k,l probabilistic potential. Geodesic Distances. Our proposed approach raises the question as to what constitutes the best constraint between neighboring correspondence variables. The literature on scan registration -- for rigid and non-rigid models alike -- relies on the preserving Euclidean distance. While Euclidean distance is meaningful for rigid objects, it is very sensitive to de- formations, especially those induced by moving parts. For example, in Fig. 1C, we see that the two legs in one configuration of our puppet are fairly close together, allowing the algo- rithm to map two adjacent points in the data mesh to the two separate legs, with minimal deformation penalty. In the complementary situation, especially when object symmetries are present, two distant yet similar points in one scan might get mapped to the same region in the other. For example, in the same figure, we see that points in both an arm and a leg in the data mesh get mapped to a single leg in the model mesh. We therefore want to enforce constraints preserving distance along the mesh surface (geodesic distance). Our probabilistic framework easily incorporate such constraints as correlations between pairs of correspondence variables. We encode a nearness preservation Figure 2: A) Automatic interpolation between two scans of an arm and a wooden puppet. B) Regis- tration results on two scans of the same man sitting and standing up (select points were displayed) C) Registration results on scans of a larger man and a smaller woman. The algorithm is robust to small changes in object scale. constraint which prevents adjacent points in mesh Z to be mapped to distant points in X in the geodesic distance sense. For adjacent points zk, zl in the data mesh, we define the following potential: 0 dist Geodesic (xi, xj ) > n(ck = i, cl = j) = (2) 1 otherwise where is the data mesh resolution and is some constant, chosen to be 3.5. The farness preservation potentials encode the complementary constraint. For every pair of points zk, zl whose geodesic distance is more than 5 on the data mesh, we have a potential: 0 dist Geodesic(xi, xj ) < f (ck = i, cl = j) = (3) 1 otherwise where is also a constant, chosen to be 2 in our implementation. The intuition behind this constraint is fairly clear: if zk, zl are far apart on the data mesh, then their corresponding points must be far apart on the model mesh. Local Surface Signatures. Finally, we encode a set of potentials that correspond to the preservation of local surface properties between the model mesh and data mesh. The use of local surface signatures is important, because it helps to guide the optimization in the exponential space of assignments. We use spin images [14] compressed with prin- cipal component analysis to produce a low-dimensional signature sx of the local surface geometry around a point x. When data and model points correspond, we expect their lo- cal signatures to be similar. We introduce a potential whose values s(ck) = i enforce a zero-mean Gaussian penalty for discrepancies between sx and s . i zk 3.2 Optimization In the previous section, we defined a Markov network, which encodes a joint probability distribution over the correspondence variables as a product of single and pairwise poten- tials. Our goal is to find a joint assignment to these variables that maximizes this proba- bility. This problem is one of standard probabilistic inference over the Markov network. However, the Markov network is quite large, and contains a large number of loops, so that exact inference is computationally infeasible. We therefore apply an approximate inference method known as loopy belief propagation (LBP)[21], which has been shown to work in a wide variety of applications. Running LBP until convergence results in a set of probabilis- tic assignments to the different correspondence variables, which are locally consistent. We then simply extract the most likely assignment for each variable to obtain a correspondence. One remaining complication arises from the form of our farness preservation constraints. In general, most pairs of points in the mesh are not close, so that the total number of such potentials grows as O(M 2), where M is the number of points in the data mesh. However, rather than introducing all these potentials into the Markov net from the start, we introduce them as needed. First, we run LBP without any farness preservation potentials. If the solution violates a set of farness preservation constraints, we add it and rerun BP. In practice, this approach adds a very small number of such constraints. 4 Experimental Results Basic Registration. We applied our registration algorithm to three different datasets, containing meshes of a human arm, wooden puppet and the CAESAR dataset of whole human bodies [1], all acquired by a 3D range scanner. The meshes were not complete surfaces, but several techniques exist for filling the holes (e.g., [10]). We ran the Correlated Correspondence algorithm using the same probabilistic model and the same parameters on all data sets. We use a coarse-to-fine strategy, using the result of a coarse sub-sampling of the mesh surface to constrain the correspondences at a finer-grained level. The resulting set of correspondences were used as markers to initialize the non-rigid ICP algorithm of Hahnel et al. [12]. The Correlated Correspondence algorithm successfully aligned all mesh pairs in our hu- man arm data set containing 7 arms. In the puppet data set we registered one of the meshes to the remaining 6 puppets. The algorithm correctly registered 4 out of 6 data meshes to the model mesh. In the two remaining cases, the algorithm produced a registration where the torso was flipped, so that the front was mapped to the back. This problem arises from am- biguities induced by the puppet symmetry, whose front and back are almost identical. Im- portantly, our probabilistic model assigns a higher likelihood score to the correct solution, so that the incorrect registration is a consequence of local maxima in the LBP algorithm. This fact allows us to address the issue in an unsupervised way simply by running loopy BP several times, with different initialization. For details on the unsupervised initialization scheme we used, please refer to our technical report [2]. We ran the modified algorithm to register one puppet mesh to the remaining 6 meshes in the dataset, obtaining the correct registration in all cases. In particular, as shown in Fig. 1A, we successfully deal with the case on which the straightforward nonrigid ICP algorithm failed. The modified algorithm was applied to the CAESAR dataset and produced very good registration for challenging cases exhibiting both articulated motion and deformation (Fig. 2B), or exhibiting deforma- tion and a (small) change in object scale (Fig. 2C). Overall, the algorithm performed robustly, producing a close-to-optimal registrations even for pairs of meshes that involve large deformations, articulated motion or both. The registration is accomplished in an unsupervised way, without any prior knowledge about object shape, dynamics, or alignment. Partial view completion. The Correlated Correspondence algorithm allows us to register a data mesh containing only a partial scan of an object to a known complete surface model of the object, which serves as a template. We can then transform the template mesh to the partial scan, a process which leaves undisturbed the links that are not involved in the partial mesh. The result is a mesh that matches the data on the observed points, while completing the unknown portion of the surface using the template. We take a partial mesh, which is missing the entire back part of the puppet in a particular pose. The resulting partial model is displayed in Fig. 3B-1; for comparison, the correct complete model in this configuration (which was not available to the algorithm), is shown in Fig. 3B-2. We register the partial mesh to models of the object in a different pose (Fig. 3B- 3), and compare the completions we obtain (Fig. 3B-4), to the ground truth represented in Fig. 3B-2. The result demonstrates a largely correct reconstruction of the complete surface geometry from the partial scan and the deformed template. We report additional shape completion results in [2]. Interpolation. Current research [20] shows that if a nonrigid transformation between the poses is available, believable animation can be produced by linear interpolation be- Figure 3: A) The results produced by the CC algorithm were used for unsupervised recovery of articulated models. 15 puppet parts and 4 arm parts, as well as the articulated object skeletons, were recovered. B) Partial view completion results. The missing parts of the surface were estimated by registering the partial view to a complete model of the object in a different configuration. tween the model mesh and the transformed model mesh. The interpolation is performed in the space of local link parameters (li,j, dij, dji), We demonstrate that transforma- tion estimates produced by our algorithm can be used to automatically generate believable animation sequences between fairly different poses, as shown in Fig. 2A. Recovering Articulated Models. Articulated object models have a number of appli- cations in animation and motion capture, and there has been work on recovering them automatically from 3D data [7, 3]. We show that our unsupervised registration capability can greatly assist articulated model recovery. In particular, the algorithm in [3] requires an estimate of the correspondences between a template mesh and the remaining meshes in the dataset. We supplied it with registration computed with the Correlated Correspondence algorithm. As a result we managed to recover in a completely unsupervised way all 15 rigid parts of the puppet, as well as the joints between them (Fig. 3A). We demonstrate successful articulation recovery even for objects which are not purely rigid, as is the case with the human arm (see Fig. 3A).

Depth from triangulation has traditionally been treated in a number of separate threads in the computer vision literature, with methods like stereo, laser scanning, and coded structured light considered separately. In this paper, we propose a common framework, spacetime stereo, which unifies many of these previous methods. Viewing specific techniques as special cases of this general framework leads to insights regarding the solutions to many of the traditional problems of individual techniques. Specifically, we discuss a number of innovative possible applications such as improved recovery of static scenes under variable illumination, spacetime stereo for moving objects, structured light and laser scanning with multiple simultaneous stripes or patterns, and laser scanning of shiny objects. To suggest the practical utility of the framework, we use it to analyze one of these applications: recovery of static scenes under variable, but uncontrolled, illumination. Based on our analysis, we show that methods derived from the spacetime stereo framework can be used to recover depth in situations in which existing methods perform poorly.

Pan-tilt cameras are often used as components of wide-area surveillance systems. It is necessary to calibrate these cameras in relation to one another in order to obtain a consistent representation of the entire space. Existing methods for calibrating pan-tilt cameras have assumed an idealized model of camera mechanics. In addition, most methods have been calibrated using only a small range of camera motion. We present a method for calibrating pan-tilt cameras that introduces a more complete model of camera motion. Pan and tilt rotations are modeled as occurring around arbitrary axes in space. In addition, the wide area surveillance system itself is used to build a large virtual calibration object, resulting in better calibration than would be possible with a single small calibration target. Finally, the proposed enhancements are validated experimentally, with comparisons showing the improvement provided over more traditional methods.

Robotic navigation and interaction frequently require that the shape and motion of external objects and events be observed. Many interesting events occur at mixed scales. Subtle localized shape and motion often occurs together with long-range movements. One of the chief challenges in recovering these events is to obtain high resolution imagery suitable for resolving small details, while simultaneously increasing the working volume in which recovery is possible. This paper proposes architecture for mixed scale motion recovery. The robust coverage of a large working volume is provided by a wide area of tracking system. This system localizes interesting motions, and guides a separate foveated system of pan tilt cameras to observe the detailed event at high resolution. We demonstrate two applications, foveated structured light scanning and the capture of muscle deformation while walking. Both applications allow subtle detailed recovery that would not be possible using existing single scale systems.

The paper introduces a method to calibrate a wide area system of unsynchronized cameras with respect to a single global coordinate system. The method is simple and does not require the physical construction of a large calibration object. The user need only wave an identifiable point in front of all cameras. The method generates a rough estimate of camera pose by first performing pair-wise structure-from-motion on observed points, and then combining the pair-wise registrations into a single coordinate frame. Using the initial camera pose, the moving point can be tracked in world space. The path of the point defines a "virtual calibration object" which can be used to improve the initial estimates of camera pose. Iterating the above process yields a more precise estimate of both camera pose and the point path. Experimental results show that it performs as well as calibration from a physical target, in cases where all cameras share some common working volume. We then demonstrate its effectiveness in wide area settings by calibrating a system of cameras in a configuration where traditional methods cannot be applied directly.

Image mosaics are useful for a variety of tasks in vision and computer graphics. A particularly convenient way to generate mosaics is by 'stitching' together many ordinary photographs. Existing algorithms focus on capturing static scenes. This paper presents a complete system for creating visually pleasing mosaics in the presence of moving objects. There are three primary contributions. The first component of our system is a registration method that remains unbiased by movement-the Mellin transform is extended to register images related by a projective transform. Second an efficient method for finding a globally consistent registration of all images is developed. By solving a linear system of equations, derived from many pairwise registration matrices, we find an optimal global registration. Lastly, a new method of compositing images is presented. Blurred areas due to moving objects are avoided by segmenting the mosaic into disjoint regions and sampling pixels in each region from a single source image.

In a digital library system, documents are available in digital form and therefore are more easily copied and their copyrights are more easily violated. This is a very serious problem, as it discourages owners of valuable information from sharing it with authorized users. There are two main philosophies for addressing this problem: prevention and detection. The former actually makes unauthorized use of documents difficult or impossible while the latter makes it easier to discover such activity.In this paper we propose a system for registering documents and then detecting copies, either complete copies or partial copies. We describe algorithms for such detection, and metrics required for evaluating detection mechanisms (covering accuracy, efficiency, and security). We also describe a working prototype, called COPS, describe implementation issues, and present experimental results that suggest the proper settings for copy detection parameters.