Real-time monitoring of coal mines underground, collecting image information, and conducting safety alerts are important aspects of mine safety research. However, factors such as lighting and dust affect the clarity of images collected underground in coal mines. To address this issue, an image enhancement algorithm for coal mine underground images based on the HSV space Retinex has been proposed. This algorithm maps RGB images to HSV space, processes the corresponding components in HSV space, and then remaps them back to RGB space. In experiments, both images reflecting the overall information of the mine and images showing local details of objects in the mine were enhanced. The results show that compared to the traditional Retinex algorithm, the images processed by this algorithm have improved in terms of SSIM, PSRN, and information entropy, and are more effective in enhancing the edge details of the images, which is of great significance for the safety precautions in coal mines underground.