An Online Video Segmentation using Improved Particle Swarm Optimization Technique

Abstract

Video segmentation is the original needed stage in the video method wherever the whole video is splits into volumetric sections, each segment consists of spatial substantive with temporal domain information. We proposed a video segmentation algorithm using Short Term Hierarchical Fast Watershed Algorithm (STHFWA )and Fractional Order Darwinian Particle Swarm Optimization(FODPSO) technique. In STHFWA ,the k-means clustering is used to find out the initial segments based on the color intensity present in the frames and Fast watershed algorithm is applied to construct the rigid lines based on the catchment basin coordinates to avoid over segmentation. Finally, FODPSO optimization is used to reduce the computational complexity. Also, the experimental results are analyzed on a VSB100 dataset, shows that the proposed algorithm outperforms modern online video segmentation techniques significantly.