Research on Image Edge Distortion Correction Algorithm Based on Improved Deep Learning Self-Coding

This paper proposes an algorithm for correcting image edge distortion, which is based on improved deep learning self-coding, which can solve the problem of large error of error correction of image edge distortion caused by the poor fitting of edge derived from linear projection. In this paper, the denoising algorithm will be used to transform the parameters at all levels to achieve the purpose of denoising. After obtaining the image denoising results, the deep learning results are improved by using Fisher vector coding, and then the corrected target optimization function is obtained based on the distortion morphology of the edge of the resulting image, so as to analyze the edge fracture in detail, and then fit the linear projection derived edge. Finally, the image edge distortion correction method will be determined to complete the image edge distortion correction, and the correlation evaluation function of the image edge error will be used in the process. The results show that the pincushion and barrel distortions of the images are corrected efficiently and with high precision by the algorithm proposed in this paper, and the number of edge bars in the corrected images of the two image distortions is consistent with the original image.