An Image Enhancement Method in Coal Mine Underground Based on Deep Retinex Network and Fusion Strategy

Due to the high dust concentration and low illuminance in coal mines, the collected images have shortcomings such as low contrast and overall darkness, which brings severe challenges to subsequent image registration and target detection. This paper proposes a mine image enhancement method based on deep learning and fusion strategies. First, CNN is used to decompose the image into illumination components and reflection components, and then based on the reflectivity consistency constraint and the illumination mapping smoothness constraint, the enhancement network is used to enhance the image And adjust to obtain the initial enhanced image, and then introduce the camera response model to simulate the multiple exposure process, find the best exposure rate, generate a virtual over-exposure image, and finally combine the original low-illuminance image, use image block decomposition for reconstruction and fusion to obtain underground low-illuminance Enhance the image. The experimental results show that compared with mainstream enhancement algorithms, the enhancement strategy in this article is better than other algorithms in subjective and objective evaluation. It can improve the contrast and brightness of the image while avoiding color distortion and over-enhancement, making it a follow-up image The processing work laid the foundation.