Moving Object Segmentation in Video Using Stationary Wavelet Transform

Various video processing applications, such as tracking, requires low complexity and reliable segmentation of objects. Global motion and background clutter often acts as key constraints to perform reliable segmentation. In this paper, we propose a video segmentation algorithm for tracking application that handles these constraints by operating on high and low frequency wavelet bands simultaneously. Furthermore, our method incorporates novel motion adaptation, clutter removal and region creation techniques. It successfully deals with various types of obstacles, such as large global motion and high background clutter. Simulation results demonstrate that the proposed algorithm achieves appropriate performance in segmentation, at a low complexity level.