Video Monitoring System Based on Amending Model of Motion Vector
Zujue Chen, Jia Wei, Zhixiong Zhang · 2008
In order to make the image in monitoring system more clear and stable, in paper, by making research about the process of motion estimation used in video coding, it analyses the block matching approach at large applied in video processing algorithm and illuminates the reasons of error existing in calculating of motion vector. According to the difference value of the vector of reference block and vector of reference frame attained in front video coding part, the error of parameters caused by moving objects has two instances, including big and small error areas. Amending model on motion vector is established, aiming at enhancing accuracy of vector by amending based on magnitude and direction of motion vector in small error area and primarily magnitude in big error area. Testing results show that the average PSNR of pixel is enhanced and the image becomes more clear.