Sparse LiDAR point clouds cause severe loss of detail of static structures and reduce the density of static points available for navigation. Reduced density can be detrimental to navigation under several scenarios. We observe that despite high sparsity, in most cases, the global topology of LiDAR outlining the static structures can be inferred. We utilize this property to obtain a backbone skeleton of a LiDAR scan in the form of a single connected component that is a proxy to its global topology. We utilize the backbone to augment new points along static structures to overcome sparsity. Newly introduced points could correspond to existing static structures or to static points that were earlier obstructed by dynamic objects. To the best of our knowledge, we are the first to use such a strategy for sparse LiDAR point clouds. Existing solutions close to our approach fail to identify and preserve the global static Li-DAR topology and generate sub-optimal points. We propose GLiDR, a Graph Generative network that is topologically regularized using 0-dimensional Persistent Homology (PH) constraints. This enables GLiDR to introduce newer static points along a topologically consistent global static LiDAR backbone. GLiDR generates precise static points using 32 × sparser dynamic scans and performs better than the baselines across three datasets. GLiDR generates a valuable byproduct - an accurate binary segmentation mask of static and dynamic objects that are helpful for navigation planning and safety in constrained environments. The newly introduced static points allow GLiDR to outperform LiDAR-based navigation using SLAM in several settings. Source code is available at https://github.com/GLiDR-CVPR2024/GLiDR.

Automated skill evaluation of a trainee is key to the utility of the surgical training system. The focus of this paper is to develop an automated tool for the assessment of trainees for micro-suturing task. The real-life training datasets for the micro-suturing task are often small, with long-tailed distribution, making it difficult to develop machine-learning-based tools for automated assessment. Further, micro-suturing is often performed at various magnifications and suture sizes, which makes the automated assessment more challenging compared to macro-suturing. Hence, currently, assessment is done manually by an expert using the final outcome image. In this paper, we propose a multi-task learning-based convolutional-neural-network regression model to score the effectualness of the micro-suturing task from the final outcome image. We propose a novel equivalent of the logit-adjustment (used in classification) applicable to regression formulation which effectively handles the problems associated with the long-tail distribution of the data. Additionally, we contribute the largest open-access dataset for suturing images and the first dataset pertaining to the micro-suturing task. We also demonstrate that the performance of the proposed algorithm surpasses the performance of human experts and also other state-of-the-art (SOTA) algorithms. The dataset and the code are available at: https: //aineurosurgery.github.io/microsuturing

The effectiveness of fingerprint-based authentication systems on good quality fingerprints is established long back. However, the performance of standard fingerprint matching systems on noisy and poor quality fingerprints is far from satisfactory. Towards this, we propose a data uncertainty-based framework which enables the state-of-the-art fingerprint pre-processing models to quantify noise present in the input image and identify fingerprint regions with background noise and poor ridge clarity. Quantification of noise helps the model two folds: firstly, it makes the objective function adaptive to the noise in a particular input fingerprint and consequently, helps to achieve robust performance on noisy and distorted fingerprint regions. Secondly, it provides a noise variance map which indicates noisy pixels in the input fingerprint image. The predicted noise variance map enables the end-users to understand erroneous predictions due to noise present in the input image. Extensive experimental evaluation on 13 publicly available fingerprint databases, across different architectural choices and two fingerprint processing tasks demonstrate effectiveness of the proposed framework.

A fingerprint region of interest (roi) segmentation algorithm is designed to separate the foreground fingerprint from the background noise. All the learning based state-of-the-art fingerprint roi segmentation algorithms proposed in the literature are benchmarked on scenarios when both training and testing databases consist of fingerprint images acquired from the same sensors. However, when testing is conducted on a different sensor, the segmentation performance obtained is often unsatisfactory. As a result, every time a new fingerprint sensor is used for testing, the fingerprint roi segmentation model needs to be re-trained with the fingerprint image acquired from the new sensor and its corresponding manually marked ROI. Manually marking fingerprint ROI is expensive because firstly, it is time consuming and more importantly, requires domain expertise. In order to save the human effort in generating annotations required by state-of-the-art, we propose a fingerprint roi segmentation model which aligns the features of fingerprint images derived from the unseen sensor such that they are similar to the ones obtained from the fingerprints whose ground truth roi masks are available for training. Specifically, we propose a recurrent adversarial learning based feature alignment network that helps the fingerprint roi segmentation model to learn sensor-invariant features. Consequently, sensor-invariant features learnt by the proposed roi segmentation model help it to achieve improved segmentation performance on fingerprints acquired from the new sensor. Experiments on publicly available FVC databases demonstrate the efficacy of the proposed work.

Latent fingerprints recognition is very useful in law enforcement and forensics applications. However, automated matching of latent fingerprints with a gallery of live scan images is very challenging due to several compounding factors such as noisy background, poor ridge structure, and overlapping unstructured noise. In order to efficiently match latent fingerprints, an effective enhancement module is a necessity so that it can facilitate correct minutiae extraction. In this research, we propose a Generative Adversarial Network based latent fingerprint enhancement algorithm to enhance the poor quality ridges and predict the ridge information. Experiments on two publicly available datasets, IIITD-MOLF and IIITD-MSLFD show that the proposed enhancement algorithm improves the fingerprints quality while preserving the ridge structure. It helps the standard feature extraction and matching algorithms to boost latent fingerprints matching performance.

Neuro-endoscopy is a challenging minimally invasive neurosurgery that requires surgical skills to be acquired using training methods different from the existing apprenticeship model.There are various training systems developed for imparting fundamental technical skills in laparoscopy where as limited systems for neuro-endoscopy.Neuro-Endo-Trainer was a box-trainer developed for endo-nasal transsphenoidal surgical skills training with video based offline evaluation system.The objective of the current study was to develop a modified version (Neuro-Endo-Trainer-Online Assessment System (NET-OAS)) by providing a stand-alone system with online evaluation and realtime feedback.The validation study on a group of 15 novice participants shows the improvement in the technical skills for handling the neuro-endoscope and the tool while performing pick and place activity.

Use of higher order clique potentials in MRF-MAP problems has been limited primarily because of the inefficiencies of the existing algorithmic schemes. We propose a new combinatorial algorithm for computing optimal solutions to 2 label MRF-MAP problems with higher order clique potentials. The algorithm runs in time O(2(k)n(3)) in the worst case (k is size of clique and n is the number of pixels). A special gadget is introduced to model flows in a higher order clique and a technique for building a flow graph is specified. Based on the primal dual structure of the optimization problem, the notions of the capacity of an edge and a cut are generalized to define a flow problem. We show that in this flow graph, when the clique potentials are submodular, the max flow is equal to the min cut, which also is the optimal solution to the problem. We show experimentally that our algorithm provides significantly better solutions in practice and is hundreds of times faster than solution schemes like Dual Decomposition [1], TRWS [2] and Reduction [3], [4], [5]. The framework represents a significant advance in handling higher order problems making optimal inference practical for medium sized cliques.

Use of higher order clique potentials for modeling inference problems has exploded in last few years. The algorithmic schemes proposed so far do not scale well with increasing clique size, thus limiting their use to cliques of size at most 4 in practice. Generic Cuts (GC) of Arora et al. [9] shows that when potentials are submodular, inference problems can be solved optimally in polynomial time for fixed size cliques. In this paper we report an algorithm called Approximate Cuts (AC) which uses a generalization of the gadget of GC and provides an approximate solution to inference in 2-label MRF-MAP problems with cliques of size k ≥ 2. The algorithm gives optimal solution for submodular potentials. When potentials are non-submodular, we show that important properties such as weak persistency hold for solution inferred by AC. AC is a polynomial time primal dual approximation algorithm for fixed clique size. We show experimentally that AC not only provides significantly better solutions in practice, it is an order of magnitude faster than message passing schemes like Dual Decomposition [19] and GTRWS [17] or Reduction based techniques like [10, 13, 14].

Writer recognition based on peculiarity of hand-writing is an important aspect of any forensic analysis. We present an approach for selecting best discriminative primitives for writer recognition. After selecting the primitives we also propose a hybrid system by combining both writer recognition and handwriting recognition for improved accuracy. We have also validated the performance of selected primitives on publically available dataset. We have performed this study on the Devanagri script. Experimental results verified the effectiveness of the proposed franework.

This paper addresses the problem of discovering activities and their temporal significance in surveillance videos in an unsupervised manner. We propose a generative model that can jointly capture the activities and their behaviour over time. We use multinomial distribution over local motion features to model activities and a mixture distribution over their time stamps to capture the multi-modal temporal distribution of these activities. We give a Gibbs sampling algorithm to infer the parameters of the model. We demonstrate the effectiveness of our approach on real life surveillance feed of outdoor scenes.

We address the problem of super-resolution—obtaining high-resolution images and videos from multiple low-resolution inputs. The increased resolution can be in spatial or temporal dimensions, or even in both. We present a unified framework which uses a generative model of the imaging process and can address spatial super-resolution, space-time super-resolution, image deconvolution, single-image expansion, removal of noise, and image restoration. We model a high-resolution image or video as a Markov random field and use maximum a posteriori estimate as the final solution using graph-cut optimization technique. We derive insights into what super-resolution magnification factors are possible and the conditions necessary for super-resolution. We demonstrate spatial super-resolution reconstruction results with magnifications higher than predicted limits of magnification. We also formulate a scheme for selective super-resolution reconstruction of videos to obtain simultaneous increase of resolutions in both spatial and temporal directions. We show that it is possible to achieve space-time magnification factors beyond what has been suggested in the literature by selectively applying super-resolution constraints. We present results on both synthetic and real input sequences.

This paper describes an artificial neuron structure and an efficient learning procedure in the complex domain. This artificial neuron aims at incorporating an improved aggregation operation on the complex-valued signals. The aggregation operation is based on the idea underlying the weighted root-power mean of input signals. This aggregation operation allows modeling the degree of compensation in a natural manner and includes various aggregation operations as its special cases. The complex resilient propagation algorithm ([Formula: see text]-RPROP) with error-dependent weight backtracking step accelerates the training speed significantly and provides better approximation accuracy. Finally, performance evaluation of the proposed complex root-power mean neuron with the [Formula: see text]-RPROP learning algorithm on various typical examples is given to understand the motivation.

The basic ideas to develop artificial neural network (ANN) were originated with the investigation of brain's micro-structure. It has been a steady endeavor in the research that followed to develop it further and integrate additional discoveries about the human brain with a view to evolve the artificial neuron model closer to the actual brain functioning. The pursuit has ever been on to replicate the typical characteristic of the neuron. The neuron response to the input signals impinged onto it, is defined how they are aggregated with in the unit. A substantial body of evidence has grown to support the presence of non-linear integration of synaptic inputs in the neuron cells. Superior functionality of ANN in complex domain has been observed in recent researches, which presented the second generation of development in ANN. In this paper, we explore the functional capabilities of a compensatory neuron model with complex-valued high order non-linear aggregation function. The strength and effectiveness of considered neuron is evaluated with an efficient learning algorithm in a complex domain. The performance analysis is carried out through a solid set of simulations.

In this paper we address the problem of time based activity inference in unsupervised manner for an area under surveillance. We use a Latent Dirichlet Allocation based model that captures the activities and how they change over time. We use agglomerative cluster-ing on optical flow vectors to code direction and spatial information. In this model each activity is associated with not only a mixture distribution over these cluster occurrences but also on the distribution over timestamps of their occurrences. Our method thus helps in determining the prominence and the correlation of activities over a period of time. 1

The paper proposes new neuron architecture for Neural Network models with an aggregation function based on geometric mean of all inputs. This new neuron model gives better accuracy compared to Multilayer Perceptron model (MLP) without increasing the number of parameters. Various error measures have been used with this model. The effectiveness of this model with different error measures have been illustrated on various data sets pertaining to classification, prediction and approximations problems.

This paper presents a practical and effective image compression system based on wavelet decomposition and contrast sensitive-SVR (support vector regression) for compressing still images. The kernel function in an SVR plays the central role of implicitly mapping the input vector (through an inner product) into a high-dimensional feature space. We study the different wavelet kernel for image compression application. Image quality is measured objectively, using peak signal-to-noise ratio, and subjectively, using perceived image quality. The effects of different wavelet kernels, image contents and compression ratios are assessed. A comparison with JPEG, SPIHT compression system is given. Our results provide a good reference to choose a suitable kernel for image compression application.

The counterpropagation network is a combination of competitive network (Kohonen layer) and Grossberg outstar structure. In this paper we have proposed a complex valued representation on conventional forward only counterpropagation network. Many researchers have investigated the computational capabilities of neuron models for real values only. The novel part of the paper is, while considering the complex values equal weightage is given to both the real and imaginary parts. A vectored approach is taken to compute the complex numbers while implementing it with complex valued counterpropagation network (CVCPN). The proposed network is tested on benchmark problem (two spiral problem), Julia's set, rotational transformations and color image compression. The complex valued counterpropagation network (CVCPN) exhibits less percentage of misclassification and error rate is considerably smaller when compared to the equivalent model in backpropagation network. The learning of intermediate forms of vector classes, manipulation with complex numbers, criterion for winning neuron, and the results of the proposed network with various benchmark and classification problems are discussed

Reduction of the image colors, which is also called color quantization (CQ), has been the focus of recent research interest. It is as an integral part of various digital image related areas such as compression, segmentation etc.. Neural networks play a significant role in either assisting conventional color quantization techniques or providing standalone solutions for color quantization. In the present work three new algorithms have been proposed using modified forward only counterpropogation network (MFOCPN). These algorithms introduce, 1) sub-clustering in Kohonen layer for enhancing the clustering process, 2) a new entropy metric based initialization of the Kohonen layer for efficient color-map design and faster convergence of network. Further, the two approaches are merged, to yield third algorithm to achieve better results. The proposed algorithms have been tested on standard test image.

Abstract In this paper we present a framework for designing quality walkthroughs. Our framework aids the content creator in designing tours where one sees interesting objects for the most part of the tour. Further, the speed of the tour is controlled by what is shown. In particular we consider the designing of walkthroughs of scenes constructed by image‐based techniques where such a design framework helps to optimize on the effort needed to create the content. Copyright © 2007 John Wiley & Sons, Ltd.

In this paper, learning algorithm for a single multiplicative spiking neuron (MSN) is proposed and tested for various applications where a multilayer perceptron (MLP) neural network is conventionally used. It is found that a single MSN is sufficient for the applications that require a number of neurons in different hidden layers of a conventional neural network. Several benchmark and real-life problems of classification and function-approximation are illustrated. It has been observed that the inclusion of few more biological phenomenon in artificial neural networks can make them more prevailing.

In this paper, a learning algorithm for a single integrate-and-fire neuron (IFN) is proposed and tested for various applications in which a multilayer perceptron based neural network is conventionally used. It is found that a single IFN is sufficient for the applications that require a number of neurons in different hidden layers of a conventional neural network. Several benchmark and real-life problems of classification and function-approximation have been illustrated. It is observed that the inclusion of some more biological phenomenon in an artificial neural network can make it more powerful.

We give an efficient, scalable, and simple algorithm for computation of a central path for navigation in closed virtual environments. The algorithm requires less preprocessing and produces paths of high visual fidelity. The algorithm enables computing paths at multiple resolutions. The algorithm is based on a distance from boundary field computed on a hierarchical subdivision of the free space inside the closed 3D object. We also present a progressive version of our algorithm based on a local search strategy thus giving navigable paths in a localized region of interest.

Abstract In this paper, we present a novel system for facilitating the creation of stylized view‐dependent 3D animation. Our system harnesses the skill and intuition of a traditionally trained animator by providing a convivial sketch based 2D to 3D interface. A base mesh model of the character can be modified to match closely to an input sketch, with minimal user interaction. To do this, we recover the best camera from the intended view direction in the sketch using robust computer vision techniques. This aligns the mesh model with the sketch. We then deform the 3D character in two stages ‐ first we reconstruct the best matching skeletal pose from the sketch and then we deform the mesh geometry. We introduce techniques to incorporate deformations in the view‐dependent setting. This allows us to set up view‐dependent models for animation. Categories and Subject Descriptors (according to ACM CCS): I.3.7 [Computer Graphics]: Three‐Dimensional Graphics and Realism ‐ Animation Our system takes as input a sketch (a), and a base mesh model (b), then recovers a camera to orient the base mesh (c), then reconstructs the skeleton pose (d), and finally deforms the mesh to find the best possible match with the sketch (e). image

Abstract We present a novel measure for compression of time‐variant geometry. Compression of time‐variant geometry has become increasingly relevant as transmission of high quality geometry streams is severely limited by network bandwidth. Some work has been done on such compression schemes, but none of them give a measure for prioritizing the loss of information from the geometry stream while doing a lossy compression. In this paper we introduce a cost function which assigns a cost to the removal of particular geometric primitives during compression, based upon their importance in preserving the complete animation. We demonstrate that the use of this measure visibly enhances the performance of existing compression schemes. Copyright © 2004 John Wiley & Sons, Ltd.

B-splines have today become the industry standard for CAD data representation. Freeform shape synthesis from point cloud data is an emerging technique. This predominantly involves B-spline curve/surface fitting to the point cloud data to obtain the CAD definitions. Accurate curve and surface fitting from point clouds needs a good parameterization model, i.e. the determination of parameter values of the digitized points in order to perform least squares (LSQ) fitting. Numerous works have been on the selection of such parameters. Nevertheless, it is difficult with the present approaches to estimate better parameters particularly when the points are irregularly spaced and lie on a complex base curve or surface. There is a need to evolve from all the available parameterization solutions an optimum set of parameters which in turn will generate curves/surface interpolating the given data closely. An approach based on genetic algorithms for parameter optimization is presented here. A novel population initialization scheme is proposed that ensures that the optimization procedure is both global in nature with less expensive convergence. The present study of parameterization is for non uniform B-spline curve fitting.

This paper presents a computational model for studying the mechanical properties of skin with aging. In particular, attention is given to the folding capacity of skin, which may be manifested as wrinkles. The simulation provides visual results demonstrating the form and density of folds under the various conditions. This can help in the consideration of proper measures for a cosmetic product for the skin.

The first virtual humans appeared in the early 1980s in such films as Dreamflight (1982) and The Juggler (1982). Pioneering work in the ensuing period focused on realistic appearance in the simulation of virtual humans. In the 1990s, the emphasis has shifted to real-time animation and interaction in virtual worlds. Virtual humans have begun to inhabit virtual worlds and so have we. To prepare our place in the virtual world we first develop techniques for the automatic representation of a human face capable of being animated in real time using both video and audio input. The objective is for one's representative to look, talk, and behave like oneself in the virtual world. Furthermore, the virtual inhabitants of this world should be able to see our avatars and to react to what we say and to the emotions we convey. We sketch an overview of the problems related to the analysis and synthesis of face-to-virtual-face communication in a virtual world. We describe different components of our system for real-time interaction and communication between a cloned face representing a real person and an autonomous virtual face. It provides an insight into the various problems and gives particular solutions adopted in reconstructing a virtual clone capable of reproducing the shape and movements of the real person's face. It includes the analysis of the facial expression and speech of the cloned face, which can be used to elicit a response from the autonomous virtual human with both verbal and nonverbal facial movements synchronized with the audio voice.

Wrinkles are an extremely important contribution for enhancing the realism of human figure models. We present an approach to generate static and dynamic wrinkles on human skin. For the static model, we consider micro and macro structures of the skin surface geometry. For the wrinkle dynamics, an approach using a biomechanical skin model is employed. The tile texture patterns in the micro structure of skin surface are created using planar Delaunay triangulation. Functions of barycentric coordinates are applied to simulate the curved ridges. The visible (macro) flexure lines which may form wrinkles are predefined edges on the micro structure. These lines act as constraints for the hierarchical triangulation process. Furthermore, the dynamics of expressive wrinkles-controlling their depth and fold-is modeled according to the principal strain of the deformed skin surface. Bump texture mapping is used for skin rendering.

The research activity in natural/synthetic image processing and representation reported in this paper, initiated under the Esprit project HUMANOID and currently continued under the ACTS project VIDAS, concerns the application of virtual reality methodologies to interpersonal audio/video communication. The 3D videophone scene is modeled in video (the talker's face) and in audio (the talker's speech) so that natural data can be efficiently mixed with synthetic data and adapted onto deformable parameterized structures. Robust image analysis/synthesis tools are necessary to extract the visual primitives associated to the talker's face and to adapt them onto suitable modeling structures (wire-frames). Image/speech analysis performed at the transmitter provides suitable audio/video parameters which are encoded and used at the receiver to synthesize the corresponding facial expressions together with synchronized lip movements.

Abstract Availability of Visible Human Dataset (VHD)has provided numerous possibilities for its exploitation in both medical applications and 3D animation. In this paper, we present our interactive tools which enable extraction of surfaces for different organs, including bones, muscles, fascia, and skin, from the VHD. The reconstructed surfaces then are used for defining the inter‐relationship of organs, a process we refer to as topological modeling. A data base is constructed, which encapsulates structural, topological, mechanical and other relevant information about organs. A 3D interactive tool enables the building and editing of this data base. Such a data base can later be used for different applications in fields such as medicine, sports, education, and entertainment.

Medical imaging can provide data for useful views of the interior details of human anatomy. In addition to visualization, which in general has been the primary reason for obtaining these data, many other uses are possible. These include modeling of different elements and their inter-relationships-topological modeling, simulation of physical processes, analysis of movements and validation of models. Here, we describe some of the modeling issues from medical imaging. The issues are particularly related to topological modeling of different anatomical elements: bones, muscles, articulations, etc. A 3D topological modeler is presented with which anatomists and other users can build a topological data base containing structural, topological, and mechanical information of anatomical elements.>

Abstract We describe the HUMANOID environment dedicated to human modeling and animation for general multimedia, VR, and CAD applications integrating virtual humans. We present the design of the system and the integration of the various features: generic modeling of a large class of entities with the BODY data structure, realistic skin deformation for body and hands, facial animation, collision detection, integrated motion control and parallelization of computation intensive tasks.

Most of the earlier and existing computational models for facial animation consider only muscular expressions. We address and emphasize issues related to modeling of vascular expression. The proposed model enables visual characteristics such as skin color to change with time and provide visual clues for emotions like paleness and blushing. An emotion is defined as a function of two signals in time, one for spatial changes (muscular effects) and the other for the color (vascular effects). For different regions of the face, the atomic vascular action is modeled as an image mask with its shape and the shade defined by Bezier functions to manipulate the texture image.>

Abstract This paper describes interactive facilities for simulating abstract muscle actions using Rational Free Form Deformations (RFFD). The particular muscle action is simulated as the displacement of the control points of the control‐unit for an RFFD defined on a region of interest. One or several simulated muscle actions constitute a Minimum Perceptible Action (MPA), which is defined as the atomic action unit, similar to Action Unit (AU) of the Facial Action Coding System (FACS), to build an expression.