A Novel Approach for Moving Object Detection Based on Improved Particle Swarm Optimization Algorithm

Jin Yuyu,Xuan Zhou,Feng Qian
DOI: https://doi.org/10.1109/icca.2010.5524412
2010-01-01
Abstract:In order to extract from the video sequence in a complete and consistent moving target, a novel algorithm for video object segmentation based on improved particle swarm optimization (IPSO) is presented. The algorithm fuses brightness segmentation and color information at ‘region level’, as to make up for conventional ‘pixel level’ approaches. The IPSO is taken into account for spatial segmentation of the video frame, which combines the mixture Gaussian model of temporal framework in achieving better segmentation. Adapting to the real-time video surveillance, the proposed algorithm can speed up the process of image segmentation, and make background modeling accurately to update. Comparisons were performed with other method that the proposed algorithm can detect intact moving objects even when objects appear and disappear suddenly. The experiment across different types of video shows the efficiency and stability of video object segmentation by the novel approach.
What problem does this paper attempt to address?