摘要：Harris corner detection algorithm called Harris corner detector is a very effective corner algorithm for gray-scale images. The corners extracted by Harris corner detector are stable, reliable, homogeneous and reasonable. However, it has own inevitable limitations. For the shape recognition of parts, an improved Harris corner detector is proposed in this paper. Based on the analysis of Harris corner detector, B-spline function is chosen as smooth filter instead of Gaussian function. And then the convolution template of B-spline function is provided. Some experiment results indicate that the improved Harris corner detector is effective and superior.
