摘要

A new graph-cuts algorithm for video moving object detection is presented. Using the area, perimeter and shape of the moving objects, the energy function from the low-order Markov random field (LMRF) model is extended to the HMRF model, and the approximate graph-cuts algorithm based on this HMRF model is derived. Experimental results show that the proposed method is more effective than the traditional graph-cuts algorithm for video moving object detection.