Abstract: This paper proposes a scalable video stream sorting approach based on content-aware in order to improve the transmission efficiency and the video broadcast quality. According to the human subjective perception of moving scenes, the model is established to describe the dynamic characteristics of the video. Object and background motion vectors are weighted adaptively. The video packets on the frame level are tagged and sorted by the approach mentioned in this paper, which is combined with the video coding structure and the rate-distortion model. Experimental results show that the method can accurately describe the video dynamic characteristic and effectively improve the video broadcast quality and human subjective perception. This method is more significant for the complex video sports scene.