Research on video motion characteristics extraction and description based on human visual characteristics

With the rapid development of Internet and multimedia, there is new challenge on streaming media processing and transmission. On one hand, the demand for the quantity and quality of multimedia resources are getting higher, which requires the matching capabilities of data storage and transmission. On the other hand, restricted to the computer's processing power, heterogeneous network environment, network bandwidth, terminal processing capacity and so on, the limited computing resources can only be allocated to the important visual information for analysis and processing. In order to accurately describe the important visual information, this paper proposes a video motion description method which is in line with human visual characteristics. Combined with the human visual perception characteristics and eye movement analysis, three characteristics are introduced: motion intensity, motion complexity and motion density. The dynamic characteristics of the frame is described by weighting the object and background motion vector intensities. The objective motion value can accurately reflect the human visual perception.