Moving target detection in image sequences

Abstract

A novel and effective approach to the global motion estimation and moving object extraction is proposed in this paper. Firstly, the translational motion model was used by the fact that complex motion can be decomposed as a sum of translational components. Then in this application, the edge gray horizontal and vertical projections were used as the block matching feature for the motion vector estimation. The proposed algorithm reduces the motion estimation computations by calculating the one-dimensional vector rather than the two-dimensional ones. Once the global motion is robustly estimated, relatively stationary background can be almost completely eliminated through inter-frame difference method. To achieve an accurate object extraction result, the higher-order statistics (HOS) algorithm was used to discriminate background and moving object. Experiments have shown a robust result for global motion estimation and object extraction.