A novel zero-crossing edge detection method based on multi-scale space theory

This paper aims to explore the zero-crossing edge detection method based on the scale-space theory. After the one-dimensional signal and two-dimensional image are convolved with the second derivation of the Gaussian kernel respectively, the zero-crossing method is applied to And the zero-crossings of the second derivation. These points are located and then construct the edge of the signal or the image based on multi-scale space. In this paper, we proposed a new zero-crossing edge detection method based on the LOG (Laplacian of Gaussian) algorithm and multi-scale space. And then the result is compared with the gradient method used to detect border, and it shows that this method is more precise than the gradient method and other detection methods because of the one pixel width image border. This method can come close to or achieve the best level of detecting the edge.