Infrared Image Enhancement Algorithm Based on Retinex and Multi-scale Filter

Aiming at the problems of low contrast, Gaussian white noise degradation of image quality and loss of detail information in infrared images collected by traditional infrared detectors, an infrared image target enhancement algorithm integrating multi-scale filtering and image detail enhancement is proposed. The algorithm solves the problem of loss of details and edge information in infrared images caused by Gaussian blur in the traditional Retinex algorithm by improving the blurring method in the traditional Retinex algorithm, i.e., replacing the Gaussian blur with bilateral blur, and at the same time realizes the enhancement of the contrast of infrared images. On this basis, the halo artifacts appearing due to excessive contrast enhancement of infrared images are eliminated by multi-scale filtering and denoising processes such as 3D block matched filtering and non-local mean filtering. The feasibility of this algorithm is verified through experiments and the comparison experiments show that the algorithm in this paper is more effective in removing the Gaussian white noise in infrared images. The experimental results show that the infrared target image obtained by this algorithm has enhanced detail information, good clarity, high contrast, and is better than other image enhancement algorithms.