A Multispectral Image Edge Detection Algorithm Based on Improved Canny Operator

The traditional canny operator performs edge detection, which needs to artificially intervene in the variance of Gaussian filtering, and the choice of variance will affect the edge retention and denoising effects. When filtering out the noise, many edge details are lost. Aiming at the shortcomings of the traditional canny algorithm, this paper proposes an improved canny algorithm for edge detection. After multispectral image Gaussian filtering, the mixed and enhanced operation is made on multispectral image. This operation filters out the noise and retains the important edge detail information. In addition, when the gradient amplitude image is obtained, more edge information are obtained by changing the size of the sobel operator. The edge details of the multispectral image processed by this operation are more abundant and the detection is more accurate. The multispectral image is then subjected to non-maximum suppression and double threshold processing. Experiments show that compared with the traditional canny edge detection effect, the algorithm proposed in this paper has greatly improved the effect of edge connection and pseudo edge removal, and the objective evaluation and visual effect have been greatly improved than before.