Image processing method and device, electronic equipment and computer readable storage medium

Abstract

The invention discloses an image processing method and apparatus, an electronic device and a computer readable storage medium. The method comprises the steps of obtaining an original tripartite image and an original binary image corresponding to an original image; performing value correction on each pixel in the original binary image by adopting a conditional random field (CRF) algorithm to obtain a corrected binary image; and performing matting processing on the original image according to the corrected binary image and the original tripartite image to obtain an object to be matted. When the original image is matted, the CRF is adopted to carry out value correction on each pixel in the original binary image to obtain the corrected binary image, so that the precision of the corrected binary image is higher than that of the original binary image, and then, according to the corrected binary image and the original tripartite image, the accuracy of the original tripartite image is improved. The precision of the to-be-matted object obtained by matting the original image is higher than that of a traditional method, and the matting precision is improved.

Description

technical field [0001] The present application relates to the field of computer technologies, and in particular, to an image processing method, apparatus, electronic device, and computer-readable storage medium. Background technique [0002] Cutout technology refers to a technology that extracts foreground objects of any shape from the original image. It is an indispensable technical means in modern film and television production, and is widely used in magazine design, graphic art, advertising, film and television post-production and other fields. [0003] Traditional matting techniques mostly use semantic segmentation models to extract foreground objects in images. In some scenes, when the images contain some small holes, such as the sky through the holes between leaves and branches, the semantic segmentation model is used to extract foreground objects. When there are trees in , because the area of ​​the sky through which the holes pass is very small, it is difficult to ext...

AI Technical Summary

Problems solved by technology

[0003] Traditional matting technology mostly uses semantic segmentation model to extract the foreground objects in the image. In some scenes, when the picture contains some small holes, such as the sky through the holes between leaves and branches, the semantic segmentation model is used to extract the foreground object. When it comes to trees, since the area of ​​the sky through which the hole penetrates is very small, it is difficult to be extracted, resulting in low matting accuracy

Images