A new approach for real-time segmenting moving objects under cluttered background
Ke Xiang, Xuanyin Wang, Songxiao Cao, Xiaojie Fu · 2012
Concerning traditional motion segmenting algorithms, such as TD and GMM, can not meet the actual requirement that not only be robust stability to minor disturbance but also response rapidly while facing abrupt background changes, we propose a novel method named IDGMM to solve this problem by modeling the Inter-frame Differencing image with Gaussian Mixture Models. By taking both the long-term statistical analysis and instantaneous change response into account, this method is able to gain better performances than traditional algorithms. Experimental results on standard video sequences extremely show the effectiveness and robustness of this method.