Research on Low-Light Image Enhancement Algorithm Based on Generative Adversarial Networks

Low-light image enhancement has long been a hot research topic in the field of image processing. Under low-light conditions, issues such as poor image quality and loss of details are prominent. To improve the quality of low-light images, an improved weak-light image enhancement method based on CycleGAN is proposed. In this method, the EMA attention mechanism is integrated into the residual network to optimize the generator, enhancing the model's feature extraction capability and thus improving the enhancement effect. Additionally, the ASPP module is employed to strengthen the network's multi-scale feature extraction ability, expand the receptive field, and improve the model's generalization performance. Experimental results show that the proposed method achieves good performance on low-light image datasets.