In this paper, the authors present a novel Content-Based Video Retrieval (CBVR) system based on Bounded Coordinate of Motion Histogram (BCMH). The authors propose to characterize videos by using spatio-temporal features (e.g. motion direction, intensity and the residual information features). Bounded Coordinate of Motion Histogram is introduced to characterize and compare video subsequences in a reduced computation time. Moreover, the proposed approach is adaptive: a training procedure is presented. For instance, the cost — expressed in similarity measurement — has its precision increased by more than twice 38%) in comparison with extended fast dynamic time warping approach on a dataset of 1707 movie clips.