-
[show abstract]
[hide abstract]
ABSTRACT: In recent years stochastic context-free grammars have been shown to be effective in modeling human activities because of the hierarchical structures they represent. However, most of the research in this area has yet to address the issue of learning the activity grammars from a noisy input source, namely, video. In this paper, we present a framework for identifying noise and recovering the basic activity grammar from a noisy symbol string produced by video. We identify the noise symbols by finding the set of non-noise symbols that optimally compresses the training data, where the optimality of compression is measured using an MDL criterion. We show the robustness of our system to noise and its effectiveness in learning the basic structure of human activity, through an experiment with real video from a local convenience store.
Motion and Video Computing, 2007. WMVC '07. IEEE Workshop on; 03/2007
-
[show abstract]
[hide abstract]
ABSTRACT: From the viewpoint of an intelligent video surveillance system, the high-level recognition of human activity requires a priori hierarchical domain knowledge as well as a means of reasoning based on that knowledge. We approach the problem of human activity recognition based on the understanding that activities are hierarchical, temporally constrained and temporally overlapped. While stochastic grammars and graphical models have been widely used for the recognition of human activity, methods combining hierarchy and complex queries have been limited. We propose a new method of merging and implementing the advantages of both approaches to recognize activities in real-time. To address the hierarchical nature of human activity recognition, we implement a hierarchical Bayesian network (HBN) based on a stochastic context-free grammar (SCFG). The HBN is applied to digressive substrings of the current string of evidence via deleted interpolation (DI) to calculate the probability distribution of overlapped activities in the current string. Preliminary results from the analysis of activity sequences from a video surveillance camera show the validity of our approach.
Visual Surveillance and Performance Evaluation of Tracking and Surveillance, 2005. 2nd Joint IEEE International Workshop on; 11/2005
-
[show abstract]
[hide abstract]
ABSTRACT: Automatic range image registration without any knowledge of the viewpoint requires identification of common regions across different range images and then establishing point correspondences in these regions. We formulate this as a graph-based optimization problem. More specifically, we define a graph in which each vertex represents a putative match of two points, each edge represents binary consistency decision between two matches, and each edge orientation represents match quality from worse to better putative match. Then strict sub-kernel defined in the graph is maximized. The maximum strict sub-kernel algorithm enables us to uniquely determine the largest consistent matching of points. To evaluate the quality of a single match, we employ the histogram of triple products that are generated by all surface normals in a point neighborhood. Our experimental results show the effectiveness of our method for rough range image registration.
3-D Digital Imaging and Modeling, 2005. 3DIM 2005. Fifth International Conference on; 07/2005
-
[show abstract]
[hide abstract]
ABSTRACT: We propose an ICP-based registration method for range images that preserves fundamental features, i.e., local structures and color, of object surfaces. The method employs local surfaces as an attribute for establishing correspondences between range images where local surfaces are evaluated geometrically and photometrically. In estimating correspondences between range images, our method evaluates consistency of shape patterns and chromaticity of local surfaces together. In estimating transformation parameters relating the coordinates between different range images, on the other hand, our method evaluates skewness and chromaticity of correspondences. These two kinds of evaluation enhances accuracy of the estimation and results in preserving local structures and color of object surfaces.
3D Data Processing, Visualization and Transmission, 2004. 3DPVT 2004. Proceedings. 2nd International Symposium on; 10/2004
-
[show abstract]
[hide abstract]
ABSTRACT: We propose a registration method for range images that preserves local structures of object surfaces. The method introduces shape patterns and a skewness of correspondences, both of which are extracted from the local surface nearby a point of interest in each image. The shape patterns are used to eliminate false corresponding pairs of surfaces, while the skewness is used to estimate the transformation that relates the coordinates between different range images. These two features enable us to estimate the transformation that preserves local structures of object surfaces.
Pattern Recognition, 2004. ICPR 2004. Proceedings of the 17th International Conference on; 09/2004
-
[show abstract]
[hide abstract]
ABSTRACT: In this paper, we propose a method for recovering the reflectance properties of a moving Lambertian object from an image sequence of the object taken by a fixed camera under unknown, complex illumination. Our proposed method is based on the spherical-harmonic representation of Lambertian reflectance under arbitrary illumination. Then, by combining the geometry reconstructed by shape-from-motion (SFM), we recover the albedo of the object and the illumination distribution only from the image sequence of the moving object. The proposed method enables us to synthesize realistic images of the object in arbitrary poses under arbitrary lighting conditions by using reconstructed shape and albedo. We conducted a number of experiments by using both synthetic and real images to confirm the effectiveness of our proposed method.
Pattern Recognition, 2004. ICPR 2004. Proceedings of the 17th International Conference on; 09/2004
