Multidimensional signal processing group
The main research area of the group is multimedia signal processing with emphasis on image and video processing, coding, compression, visual object segmentation, tracking, and recognition.
- Image and video compression. The group is developing advanced techniques and algorithms for wavelet-based image and video compression, MPEG-4 based coding, object-based coding, and wavelet-fractal based image compression and restoration. The algorithms are being used for efficient multimedia data storage and transmission, and personalized multimedia content delivery over heterogeneous networks.
- Image content analysis and scene understanding. The research is concentrated on low-level image feature extraction, as well as high-level recognition and content understanding. Image segmentation using multiresolution techniques, active contours and level-sets, and invariant object shape descriptors provide an input to high-level image content analysis using formal systems and fuzzy logic. Content analysis was applied to case studies of complex traffic scene analysis and medical image diagnostics.
- Video object detection and tracking. Real-time video analysis is extensively used in video surveillance, quality control in industry, and interactive multimedia communication systems. We develop algorithms for video object detection, active contours for object boundary tracking, layered video segmentation, and multiple object tracking in cluttered scenes. Applications include object-based video coding and video traffic control.