Enhancement Algorithms for Images in Coal mines base on Dark Light Denoising Networks

Abstract

Facing the low contrast and heavy noises in images taken from coal mine tunnels, this paper designs a new depth neural network for low-contrast denoising, denoted as the LCDNN, based on the powerful denoising ability of the SSDA and the feature leaning ability of deep neural network. The proposed network consists of a contrast enhancement module and a denoising module, each of which is a separate SSDA. The enhanced image is obtained by minimizing the noise and improving the contrast. The proposed network was applied to denoise and enhance low-contrast images taken in a coal mine tunnel. The enhancement effect was compared with that of several popular image enhancement methods. The results show that the LCDNN outperformed the contrastive methods in all areas, leading to high contrast, low noise, rich details and good visual effect. The proposed network offers an effective tool to automatically learn basic signal features and noise structure from low-contrast images.