Due to the wide variety of copy videos, the existing video copy detection methods using single feature face great challenges, especially for video content matching, which are difficult to deal with various copy video transformations. To overcome this problem, a video copy detection method based on sparse representation of MPEG-2 spatial and temporal features is proposed in this paper. Firstly, the key frames are extracted based on visual saliency model, Then the global feature (HSV color histograms) and local feature (ORB features) are extracted from the key frames, Meanwhile, the key frames are represented compactly by sparse coding which exploits ORB features, and motion vectors (MV) extracted from the video bitstreams are exploited to build MV angle histograms. Finally, spatial feature and temporal feature are compared respectively, and matching results are fused to generate the final copy detection judgement. The experimental results on dataset TRECVID 2009 show that the proposed method presents better robustness and higher time efficiency.