Optimization of Corner Detection Algorithm for Video Stream Based on FAST
Yifan Wang, Jiaqi Wang, Haofeng Lv, Yang Li, Zeli Yang · 2021 International Conference on Electronic Information Engineering and Computer Science (EIECS) · 2021
For the FAST algorithm commonly used in corner detection of video stream, there are problems of image edge missing and corner redundancy. This paper proposes a constant filling method to fill the target image. Then judge the pixels points of 90 degrees Bresenham circle, reduce by a quarter the number of discrete contrast points for Bresenham circles in FAST algorithm. Finally, SAD non-maximum suppression is used to obtain corner points in video stream. The experimental results show that under the premise of detecting the same image and ensuring that the real corners of the original image are not lost, compared with the FAST algorithm, the number of fictitious corners detected by the proposed algorithm is reduced by 94.08% on average, and the detection time of corners is shortened by 23.33 %, and the missing detection of image edges is effectively avoided.