This paper presents a new motion segmentation method using iterative maximum likelihood framework. The method consists of two steps. In the first step, motion regions are detected and motion vectors are computed for these detected regions. In the second step, a similarity matrix is computed from the motion vectors and motion segmentation is done using maximum likelihood method. The method has been tested using real world motion sequences and is found to give very low error rate and it also detects the slow moving objects. © 2010, DESIDOC.
cited By (since 1996)0
Ra Vrinthavani and Kaimal, M. Rb, “Motion segmentation Algorithm using Spectral Framework”, Defence Science Journal, vol. 60, pp. 39-47, 2010.