Deep network image enhancement method and system based on derived graph and Retinex

Abstract

The invention discloses a deep network image enhancement method and system based on a derived graph and Retinex. The image enhancement method comprises the following steps: decomposing an input imageinto a reflection image and a brightness image by using a depth decomposition network, and enhancing the brightness image of the input image by using a depth enhanced image; processing a reflection image of the input image through fast mean filtering, and removing noise in the input image; besides, for the problems of low contrast ratio, low overall brightness and fuzzy details of a dark area of an input image, generating a derived image for processing; and finally, fusing the derived image, the filtered reflection image and the enhanced brightness image by adopting a fusion strategy to obtainan enhanced image of the input low-brightness image. According to the method, a shallow image derived image and a deep enhanced image obtained by a deep learning network are fused to realize enhancement of a low-illumination image.

Description

technical field [0001] The invention belongs to the technical field of image processing, and in particular relates to a deep network image enhancement method based on a derived graph and Retinex, and an image enhancement system applying the method. Background technique [0002] In recent years, as image enhancement technology has become more and more widely used in fields related to national economy and people's livelihood, such as medical field, intelligent transportation, intelligent identity authentication and satellite remote sensing imaging, it has promoted the continuous progress of image enhancement technology and methods. To solve the problems of low contrast, low overall brightness and blurred details in dark areas of images collected under weak light conditions, we can use image enhancement methods based on deep learning. Using the shallow image enhancement method to obtain the image derived image under weak light conditions, using the deep convolutional neural net...

AI Technical Summary

Problems solved by technology

Although deep learning algorithms have obvious advantages over traditional methods in terms of feature extraction and mathematical modeling, there are still some problems in image enhancement using deep learning methods, mainly in the following: the lack of training data sets for deep learning methods for image enhancement, and currently no Contains a public dataset of low-light and corresponding normal-light images, resulting in inconsistent training datasets

In addition, some depth models are less robust to image enhancement in different scenes. How to effectively enhance images with uneven illumination collected in different scenes still has many difficulties and challenges.

Images