Abstract—Harris algorithm is widely used in image feature point extraction, but the algorithm only can reach the accuracy at pixel level. Moreover, the stability of the algorithm is significantly influenced by the choice of K value. In this paper, according to mathematics principles, threshold calculation method for the algorithm has been improved and the influence of k selected by users on the stability of the algorithm is eliminated. On the basis of extracting feature points via Harris algorithm, a method for solving the extreme points of the intersecting line of curved surface and gradient plane has been proposed by using the fitting surface geometry model of gradation value at the near neighborhood of corner, and the positioning of feature point at sub-pixel level and application to camera calibration based on checkerboard template have been realized. It is shown with experimental results that this method is effective in improving the precision and accuracy of feature point positioning.

Index Terms—Harris algorithm, threshold, surface fitting method, contour line, extreme point, sub-pixel.

Z. J. Zhang, H. L. Lu, W. Q. Li, and W. Q. Yuan are with Technology Computer Vision Group, Shenyang University of Technology, Shenyang, China. X. Li was with Liaoning Broadcasting & TV Transmission Station, Shenyang, China.

 

Cite:Zhijia Zhang, Hongliang Lu, Xin Li, Wenqiang Li, and Weiqi Yuan, "Application of Improved Harris Algorithm in Sub-Pixel Feature Point Extraction," International Journal of Computer and Electrical Engineering vol. 6, no.2, pp. 101-104, 2014.