Image fusion method for brightness consistency learning

Abstract

The invention discloses an image fusion method for brightness consistency learning, which adopts an unsupervised depth learning method and adopts a generator-based image fusion method. A deep learningmodel based on a generative adversarial net (GAN) predicts the lightness channel of an image, the inherent semantics of the foreground part is maintained, and the background appearance is ensured toremain unchanged by adding a background covering layer, so that the image fusion result with the consistency of light and shade is obtained, and the problem of inherent semantics loss in the traditional image fusion technology is solved.

Description

technical field [0001] The present invention relates to the technical field of image synthesis, in particular to an image fusion method for brightness consistency learning, which uses a deep learning model based on a generator-adversarial network structure (GAN) to predict the brightness channel of an image, ensuring the accuracy of the fusion region Consistency of light and shade, to obtain a fusion effect with a sense of reality and consistency of light and shade. Background technique [0002] The problem of image fusion is to fuse the foreground image with the background image. At present, in the key technology of image fusion, the main difficulty is to make the fused image have illumination, texture and other aspects under the premise of maintaining the inherent characteristics of the foreground object. consistency. Existing image fusion techniques can be divided into image fusion methods based on gradient field and image fusion methods based on multi-resolution. [00...

AI Technical Summary

Problems solved by technology

At present, the problem with the Poisson fusion method is that it only retains the gradient field of the foreground image, that is, the relative change feature, while ignoring other original features such as the hue, light and shade of the

Images