A Low Illumination Images Enhancement Method Based on Conditional Generative Adversarial Networks with Self Attention Mechanism

Abstract

AbstractAiming at the problems of low signal-to-noise ratio, low resolution and low illumination in low illumination environment, this paper proposes a low illumination images enhancement method based on generative adversarial networks with self attention mechanism. Firstly, the DenseNet framework is used to build the generator network, and the self attention mechanism module and constraint conditions are introduced. At the same time, the average discriminator is used to improve the traditional binary discriminator. Secondly, the low illumination images is introduced into the generator network to generate the illumination enhancement images, and then the discriminator network is used to supervise the enhancement effect of the generator on the low illumination images. Through the game between the generator and the discriminator, the network weight is continuously optimized, and finally the generator has better enhancement effect on the low illumination images. Experimental results show that, compared with the existing mainstream methods, this method not only has obvious advantages in brightness enhancement and clarity restoration of low illumination images, but also has significant advantages in objective evaluation indexes of images quality such as peak signal-to-noise ratio and structural similarity.KeywordsLow illumination images enhancementConditional generative adversarial networksDenseNetSelf attention mechanismAverage discriminator