Factors such as insufficient lighting, backlighting, and failed focusing can lead to low brightness, poor contrast, and noise in the collected images. These issues seriously affect the further analysis and processing of images, especially for color images with more information, which often result in color distortion after enhancement. In order to improve the quality of low light color images, this paper proposes a high-quality low light color image enhancement method. According to Retinex theory, the attention map and reflection map are initialized by constructing weight functions for three color channels, and used as inputs to the network. At the same time, using a dual discriminator structure and attention mechanism, an enhanced image is generated to achieve clearer image enhancement effects. Experimental results show that the method proposed in this paper can improve the quality of low light images with higher quality and effectiveness compared to mainstream enhancement algorithms.