Research on Low Light Meteorological Observation Image Enhancement Algorithm Based on Deep Learning

This article combines brightness adjustment, sharpening, and multi-scale fusion to achieve image enhancement. Based on computer deep machine learning algorithms, the Retinex method is used to segment the original image and obtain the corresponding brightness values. Then, uniform brightness is obtained through gamma correction. For information with excessive exposure range caused by gamma correction, this article will use sharpening method to enhance the details of the original image. Use multi-scale fusion cone method to weight two images to obtain high-quality images. The subjective effect of improving contrast, brightness, and color saturation. Objective evaluation indicators such as PSNR and SSIM have also significantly improved.