Local and global parallel learning-based style transformation method and system for ten-million-level pixel digital image

Abstract

The invention discloses a local and global parallel learning-based style transformation method for a ten-million-level pixel digital image, and the method comprises the following steps: S1, constructing a stylized model training sample set, including an original image sample set, a corresponding image retouching sample set obtained by manual processing of a professional image retouching person, and a semantic segmentation image sample set corresponding to the original image sample set; S2, compressing the original image sample set and the corresponding retouching image sample set to obtain a small-size small image training sample set; S3, training to obtain a small graph stylized model; S4, based on the training sample set, cutting the original image sample set to obtain a corresponding slice pair, training and recording coordinate information, and obtaining a slice stylized model; S5, obtaining a fusion model; S6, jointly training the three networks in the steps S3 to S5. The invention further discloses a local and global parallel learning-based style transformation system for the ten-million-level pixel digital image. According to the method, local and global parallel learning is realized, the processing speed is higher, and the effect is better.

Description

technical field [0001] The invention belongs to the field of image processing, and in particular relates to a style transfer technology for imaging a digital single-lens reflex camera. A deep convolutional neural network obtained by training a stylized image dataset composed of a stylized image manually processed by a retoucher to obtain a stylized image, specifically relates to a local and global parallel based local and global parallel for digital images of tens of millions of pixels. Learning style transformation methods and systems. Background technique [0002] The problem to be solved at present is to stylize the photos taken by photographers in some specific layouts or scenes to obtain photos that are more visually beautiful and stylized than the original images, such as figure 1 The picture shows a common photo stylization processing of photos taken in the West Lake Scenic Area. The upper picture is the original photo, and the lower picture is the stylized photo. At...

AI Technical Summary

Problems solved by technology

The problem with this method is that the estimation of its affine bilateral grid is still an approximate estimate. The 16*16*8 affine bilateral grid obtained by the author based on the training picture size of 512 is indeed still accurate when inferring pictures with millions of pixels. Better results can be obtained, but when the size of the picture to be inferred reaches a higher level of tens of millions of pixels, the size of the training picture and the size of the affine bilateral grid need to be increased accordingly, and the training cost will also increase with the In addition, the structure of the model designed by this method is fixed, and the setting of this method forces the model to concentrate most of the parameter fitting pressure on the generation part of the affine bilateral grid. When the effect of the model is not good, it is difficult to carry out two sub-optimization

Images