Abstract Person re‐identification (Re‐ID) is one of the most remarkable research topics that widely applied in our daily lives. For person Re‐ID in bird's eye scenes, traditional computer vision‐based methods used multiple features, for example, texture and color, of a pedestrian's head and shoulders. Those methods are difficult to cope with environments of variety and the change the appearance of different people due to the instability of feature detection. On the other hand, although recent advanced deep learning‐based methods are powerful to extract discriminative features, the requirement of a large amount of annotated training data restricts the appliable tasks. To overcome this problem, in this article, we propose a novel method fusing multiple heterogeneous features through a multi‐feature subspace representation network (MFSRN) to maximize the classification performance while keeping the disparity among features as small as possible, that is, common‐subspace constraints. We conducted comparative experiments with state‐of‐the‐art models on the bird's‐eye view person dataset, and extensive experimental results demonstrated that our proposed MFSRN could achieve better recognition performance. Furthermore, the validity and stability of the method are confirmed.

In this paper, we present an efficient framework for 3D indoor scene labeling based on a Conditional Random Field model. To make this framework invariant to camera rotation, a novel feature vector is developed, which is simple but discriminative. Meanwhile, we re-define the pairwise potential to improve the performance. A method for learning the labeling compatibility is proposed to exploit the strong contextual relations between class labels. We evaluate our approach on three datasets and the experimental results show that it achieves higher accuracy compared with several state of the art researches.

In this paper, we propose an innovative algorithm for detecting single or multiple ground planes and estimating the orientation of the ground plane with a RGB-D camera. In our algorithm, we use Kernel Density Estimator for selecting the ground plane reliably under a given tilt angle. Camera tilt angle is estimated with probability theory and then used as a feedback to improve the accuracy of detection. Our algorithm is evaluated on several challenging data set and is shown to hold great superiority and effectiveness.

This paper describes a novel descriptor Color Distinctiveness Feature (CDF). The CDF is calculated based on Bayes rule using colors in two areas: a main area and a support area. We identify persons in the images captured with a Bird's-eye views system by using the CDF. The identification is carried out by estimating Bhattacharyya distance or Hamming distance between the registered CDF and the one obtained from the input images. We confirmed that our method can identify persons at video-rate by only using the color information of shoulders and heads, and without using any face-related information. We also confirmed that CDF can be used to identify multicolored objects.

It is difficult to carry out visualization of the large-scale time-varying data directly, even with the supercomputers. Data compression and ROI (Region of Interest) detection are often used to improve efficiency of the visualization of numerical data. It is well known that the Run Length encoding is a good technique to compress the data where the same sequence appeared repeatedly, such as an image with little change, or a set of smooth fluid data. Another advantage of Run Length encoding is that it can be applied to every dimension of data separately. Therefore, the Run Length method can be implemented easily as a parallel processing algorithm. We proposed two different Run Length based methods. When using the Run Length method to compress a data set, its size may increase after the compression if the data does not contain many repeated parts. We only apply the compression for the case that the data can be compressed effectively. By checking the compression ratio, we can detect ROI. The effectiveness and efficiency of the proposed methods are demonstrated through comparing with several existing compression methods using different sets of fluid data.

This paper describes a sophisticated method to track irises in a monocular video sequence with a particle filter that uses a newly proposed Iris-Eyelid separability filter (IESF). In order to reduce the influence of eyelids and to cope with the variance of the appearance of eyes of different people, we propose an Iris-Eyelid separability filter (IESF), and use it to estimate the likelihood of hypotheses in the particle filter. We confirmed that our method has the ability of tracking various eyes for different people whose head motion including translation, rotation and zoom, even when people are wearing glasses. Through the comparative experiments of the IESF and the conventional circular separability filter (CCSF), we confirmed that our iris tracking algorithm is more robustly.

Abstract This paper presents a new clustering method based on the Affinity Propagation (AP) for gesture recognition. AP has been successfully applied to broad areas of computer science research because of its better clustering performance over traditional methods such as k-means. In order to obtain high quality sets of clusters, the original Affinity Propagation algorithm exchanges real-valued messages between all pairs of data points iteratively until convergence. Therefore, the original AP is not suitable for handling big data. In this paper, we add two improvements to the original AP. In order to increase the processing speed, we define the continuous optical flow in video sequence and use it to reduce the number of the data to be clustered. In order to guarantee the accuracy, we define a function for evaluating the preference according to the data distribution.

In this paper, we present a high-speed and high-accuracy method for estimating scene flow with a Kinect. The region where motion objects may exist is detected by calculating the difference between three frames. From this region, the bounding box of the moving object is then determined by segmenting the depth image. We estimate the scene flow by finding the corresponding key-point pairs between frames. In the process of establishing the correspondence between key-points, we use the SURF feature and the depth with variable cuboid search algorithm. We have compared our method with other existing methods by using a common test data set.

This paper describes a method for transforming local image features with relevant component analysis (RCA) for extracting a compact set of visual words from a large set of local image features. By applying this method to a set of local image features that distributed in chaos, the local image features are transformed to a proper number of well separated groups. By selecting a visual-word from each group, a compact set of visual words can be obtained that has much less ambiguity and redundancy than the un-transformed set of local features, thus the performance of object recognition can be improved much by using those visual words. We applied this method to the recognition of daily necessities and confirmed its effectiveness through several comparison experiments using common image database. The proposed method should be useful for supporting the daily life of disabled people.

This paper presents a novel camera calibration method for a system consists of a sensor-less mobile robot and an external camera. This method can estimate the focal length and the viewing direction of the camera relative to a plane where the sensor-less mobile robot moves from a single image of an arbitrary circle on that plane. Since it is easy to let a four-wheel mobile car to move along a circular locus, our method does not requires specially designed calibration objects for camera calibration. The method does not use the position of the center and the radius of the circle and the circle is allowed to be partially occluded. Through computer simulations and experiments with a real camera and a robot, we have confirmed that our method is more robust than the Homography based calibration using four point correspondences.

Viola et al. have introduced a fast object detection scheme based on a boosted cascade of haar-like features. In this paper, we introduce a novel ternary feature that enriches the diversity and the flexibility significantly over haar-like features. We also introduce a new genetic algorithm (GA) based method for training effective ternary features through iterations of feature generation and selection. Experimental results showed that the rejection rate can reach at 98.5% with only 16 features at the first layer of the constructed cascade detector. This indicates the high performance of our method for generating effective features. We confirmed that the training time can be significantly shortened compared with Violas's method while the performance of the resulted cascade detector is comparable to the previous methods.

Viola et al. have introduced a fast object detection scheme based on a boosted cascade of haar-like features. In this paper, we introduce a novel ternary feature that enriches the diversity and the flexibility significantly over haar-like features. We also introduce a new genetic algorithm based method for training effective ternary features. Experimental results showed that the rejection rate can reach at 98.5% with only 16 features at the first layer of the cascade detector. We confirmed that the training time can be significantly shortened while the performance of the resulted cascade detector is comparable to the previous methods.

Describes a method to estimate the three-dimensional pose of a human head in a single image. An efficient method based on a perceptually uniform color system and fuzzy theory is used to extract the skin region and the hair region from an image, which are then used to detect the face and to estimate the approximate pose of it. The viewable facial features and the positions of them are then predicted with the head pose. The facial features are detected within the predicted region, which are then used to estimate the 3D pose of the head.

This paper describes a new method to detect faces in color images based on the fuzzy theory. We make two fuzzy models to describe the skin color and hair color, respectively. In these models, we use a perceptually uniform color space to describe the color information to increase the accuracy and stableness. We use the two models to extract the skin color regions and the hair color regions, and then comparing them with the prebuilt head-shape models by using a fuzzy theory based pattern-matching method to detect face candidates.

Presents a method to estimate the 3D pose of a human head in an image. To increase the stableness of the pose estimation, we use continuous representations instead of binary images to describe the skin/hair regions in an image. We use them to calculate the geometrical properties such as area, center, and the axis of least inertia of the skin, the hair, and the head regions. Then we use them to estimate the 3D pose of the head.

This paper describes a fuzzy theory based face detection system. We use a perceptually uniform chromatic system to represent color information for increasing the robustness of our system. We build two models that describe the skin color and hair color, and use them to estimate the skin color likeness and the hair color likeness. We treat the appearance of faces in images as a combination of the skin part and the hair part, and model it with several two-dimensional patterns. Then we detect the "face like" regions by finding out patterns that are similar to one of the face models from input images using the fuzzy pattern matching method. Extensive experiments show the effectiveness of our method in the face detection from images with many people and complex background.

The outputs of many face detection systems are face candidates that may contain some false faces. This paper describes an algorithm to verify the face candidates. Once a face candidate is detected from an image, the positions and the sizes of the facial features on the face are predicted based on the knowledge about the arrangement of the facial features on a human face. Then the facial features are detected from the predicted positions with a coarse to fine approach. The face candidates are verified by considering whether facial features can be extracted and how well they match with a relational face model that describes the geometric relationship among the facial features of a generic human face.

This paper presents an approach to obtain optimal images for face recognition with an active camera. Once the face is detected from an input image, the 3-dimensional position and the pose of the face relative to the camera are estimated by making use of the active camera and both the skin and hair parts of the face extracted from input images. It is then used to guide the active camera system to change its view point and direction to obtain a face image where the face is in the desired size and pose.

The factorization method, first developed by Tomasi and Kanade (1992), recovers both shape and motion from a sequence of images, by tracking a large number of feature points and using singular value decomposition (SVD). However, in this approach, the nonlinear perspective projection is linearized by approximating it using orthographic projection, weak perspective projection, and para-perspective projection, which limits the range of motions that the approach can be accommodated. In this paper, we present a new approach based on a higher-order approximation of perspective projection to recover 3D shape and motion from image sequences. The accuracy of this approximation is higher than orthographic projection, weak perspective projection and para-perspective projection, so it can be used in wider circumstances in the real world. Experimental results with synthesized and real data show that this approach is promising.

The paper describes an approach to detect faces whose size and position are unknown in an image with a complex background. The candidates of faces are detected by finding out "face like" regions in the input image using the fuzzy pattern matching method. The perceptually uniform color space is used in our research in order to obtain reliable results. The skin color that is used to detect face like regions, is represented by a model developed by us called skin color distribution function. The skin color regions are then extracted by estimating a measure that describes how well the color of a pixel looks like the skin color for each pixel in the input image. The faces which appear in images are modeled as several 2 dimensional patterns. The face like regions are extracted by a fuzzy pattern matching approach using these face models. The face candidates are then verified by estimating how well the extracted facial features fit a face model which describes the geometrical relations among facial features.>

A method for object manipulation using an active camera is introduced. By actively moving the camera to proper positions with desired viewing angle from objects of interest, it is possible to acquire data to realize the robust visual feedback control of a robot. This approach avoids complicated 3-D modeling, and the image processing carried out is driven by tasks. Active camera control for robot guidance and image understanding is addressed.>

A scheme that uses an active camera to guide object manipulation in an environment that has not been modeled is introduced. By actively moving camera to desired position with proper distance and viewing angle from objects of interest, it is possible to acquire good data for robot visual feedback control when locating objects. The environment is described in an object-centered coordinates system and is measured relatively in consecutive 2-D image spaces. This approach is flexible and efficient in manipulation, since it avoids complicated 3-D modeling and image processing is driven by tasks. It is shown that robot operation and camera motion are guided by visual feedback. The strategies to control this active guidance are studied.>