ABSTRACT
Methods of motion segmentation aim at decomposing a video into moving objects and the background. This chapter presents a new motion segmentation method using iterative graph spectral framework. One popular method used for motion segmentation is background subtraction. In statistical approaches, motion segmentation is treated as a classification problem where each pixel has to be classified as background or foreground. The expectation maximization algorithm has proved to be cumbersome to use in practice, due to the problems of estimating the parameters of the motion mixture model and of controlling its structure. The computation of the motion vectors is done using single resolution block matching algorithm (BMA) using spatial/temporal correlation. Instead of performing BMA to the entire frame, first, the moving objects are detected by taking the difference of maximum intensity values and minimum intensity values for each pixel of a set of frames.
