An inter-frame collusion based video steganalysis scheme（ICS） was proposed by utilizing the temporal corre-lation among video frames to find out whether secret data was hidden in the frames or not.Motions existing in video se-quence decreased the detection accuracy of steganalysis algorithms

which were taken into account in ICS steganalysis scheme by modeling motions and messages in video frame as a bimodal noise.Suspicious video frames were recognized by decision module employing GRNN classifier with features extracted by the inter-frame correlation strategy.Experi-mental results show the satisfying performance of the proposed scheme.