Region Merging Based on Robust Statistical Testing

This paper addresses the problem of nding moving objects present in an image sequence. More speciically, a method is proposed to merge regions based on a coherent motion criterion. A Modiied Kolmogorov-Smirnov test is proposed which exploits both the motion information present in the residual distribution and the motion information of the motion parameter space. Therefore, all the available motion information is used. Moreover, the proposed test is consistent with robust motion estimation. Using the Modiied Kolmogorov-Smirnov test, the graph of the relationships between the diierent regions is built. The graph also integrates spatial informationas only adjacent regions are allowed to merge. Two graph clustering rules are proposed which enable to robustly deene the moving objects. The proposed method does not require any user input. Simulation results demonstrate the eeciency of the proposed method.