Research on infrared image denoising and enhancement algorithm

Infrared image is a special image that can reflect the information of infrared radiation energy in the environment. It contains special information that is not available in visible light images and has unique value in military and civilian applications. However, at the same time, infrared images often show the characteristics of high dynamic range, low contrast and low signal-to-noise ratio, which seriously reduces the quality of infrared images. Moreover, due to the high dynamic range, infrared images cannot be directly displayed on traditional display devices. In order to improve the visual effect of infrared images and facilitate subsequent image processing, it is necessary to enhance the details of infrared images. At present, there are two main methods to improve the imaging quality of infrared images. One is to improve the imaging quality of infrared images. The research on infrared image enhancement technology mainly focuses on two directions: high dynamic range image detail enhancement technology and multi-source image fusion enhancement technology. It is found that the processing results of the existing detail enhancement algorithms have the problems of gradient reversal, noise amplification and low overall contrast. The fusion enhancement algorithm has the problems of low contrast, block effect, small target loss and so on. In order to solve these problems, this paper makes an in-depth study of the above two technologies, and summarizes the mutual application of the two. Firstly, the infrared image is filtered by the improved non-local mean filter to obtain the basic noise-free base layer image. Then, the filtered base layer image is subtracted from the original input image to obtain the detail layer image and the random noise image. Image enhancement is performed on the noise-free base layer image to obtain the image after detail enhancement. CLAHE(Contrast Limited Adaptive Histogram Equalization) is used to reduce the excessive enhancement of flat areas such as the sky and reduce the block effect. The octagonal window filtering is used for the detail layer image and the random noise image, and the threshold discrimination is set to 0 background noise for filtering. Finally, the image is weighted to fuse the enhanced detail layer and the base layer to obtain the processed image.