Research on Brightness Enhancement of Street View Data Set Using DLN

Abstract

Low-brightness image enhancement is a challenging and difficult task. Photos taken under dark conditions often have poor visual quality. In order to solve the problems of low contrast and high noise in low illumination images, this paper uses deep illumination network technology to analyze nighttime street scene images taken under low illumination conditions. Different from the traditional method, this method regards weak light enhancement as a residual learning problem on the basis of deep learning, that is, the residual between estimates. The experimental results show that the PSNR of the algorithm we use is 20.63567452, and the SSIM is 0.2153426. The algorithm not only improves the brightness of the low-light image, but also improves the color depth and contrast. In the objective evaluation index, it has better low-light enhancement effect.