Scale Space Mining Algorithm and Application Analysis Based on Computer Image Processing

In the process of computer IP(CIP), computer vision is used as the basis of image processing(IP). Images contain a lot of information, which is the basis for people to analyze image data. However, because images contain noise that affects the analysis results, intelligent extraction of image information has become the key to computer vision analysis of image data. According to the SS theory, this paper proposes three CIP algorithms, such as the EM image segmentation algorithm based on one-dimensional and multi-dimensional feature scale space(SS) and the IP algorithm based on adaptive multi-scale wavelet transform(WT). Compared with the other two algorithms, the three algorithms obtained the MSE value, PSNR value and SSIM value of the Q1 image. With the increase of the denoising level, the WT IP algorithm has the best denoising effect on the image, and with the increase of the denoising level. With the increase of sampling rate, the algorithm also has the best effect on image reconstruction.