Sun Ji-xiang

A fast ship detection algorithm in SAR imagery for wide area ocean surveillance

Ship detection is a basic problem to be solved in SAR application of ocean surveillance. To deal with the rapid increasing SAR data, fast algorithm has become a hot topic in the research of ship detection. This paper proposes a fast ship detection algorithm in SAR imagery for wide area ocean surveillance. The algorithm adopts global two-parameter CFAR and adaptive CFAR based K distribution in the coarse and fine detection phase separately. Performances on detection accuracy and efficiency have been analyzed theoretically. Validation results on several space-born SAR images illustrate the fast method can preserve detection accuracy and improve the calculation efficiency.

Moving objects extraction method in H.264/advanced video coding bit stream of a complex scene

For the purpose of extracting moving objects from H.264/advanced video coding (AVC) bit stream of a complex scene, an algorithm based on maximum a posteriori Markov random field (MRF) framework to extract moving objects directly from H.264 compressed video is proposed in this paper. It mainly involves encoding information of motion vectors (MVs) and block partition modes in H.264/AVC bit stream and utilizes temporal continuity and spatial consistency of moving object’s pieces. First, it retrieves MVs and block partition modes of identical 4×4 pixel blocks in P frames and establishes Gaussian mixture model (GMM) of the phase of MVs as a reference background, and then creates MRF model based on MVs, block partition modes, the GMM of the background, spatial, and temporal consistency. The moving objects are retrieved by solving the MRF model. The experimental results show that it can perform robustly in a complex environment and the precision and recall have been improved over the existing algorithm.

Improved tensor voting for illusory contour

In this paper we improve the performance of tensor voting over large contour gaps by extending the naive tensor voting techniques. We propose an iterative optimization procedure together with three extensions such as orientation estimation modification, saliency measure revision and edge thinning. As the experimental result shows, the extended tensor voting is more efficient to complete large contour gaps.