The Motion Estimation Research Based on the Histogram of Canny Edge Feature Points

S. Zhao Xi'an Municipal Facilities Administration Bureau Xi’an , China X.R. Mao Xi’an University of Science and Technology, School of Communication and Information Engineering Xi’an, China Abstract—The criteria of traditional block matching motion estimation algorithm is based on the gray value offset between macro block, which was sensitive to the noise and the treatment under uneven illumination environment. For above problem, the paper proposed that using canny algorithm extracted the edge profile of current frame, and then calculate the feature points distribution histogram of block estimated, finally, using the the distance of feature points histogram between two frames as matching criteria for all frames. The experiment showed that, the motion estimation accuracy of improved algorithm is high, and it has small difference between target and matching box. Improved algorithm has better robustness on the scene affected by the noise and poor illumination, and it can apply to intelligent image processing work of coal mine and other special environments.