Research on image enhancement algorithm based on deep neural network algorithm

Abstract

Image denoising and enhancement technology plays an important role in remote sensor image processing. Due to the low brightness, signal-to-noise ratio and contrast of image processing, the traditional remote sensor is limited in obtaining object information. Based on the principle of biological segmentation algorithm, neural network algorithm technology can carry the advantages of image evaluation algorithms such as PSNR, IFC and SSIM, through the end-to-end training of nonlinear infinite approximation, and carry out the research of image fitting low light level image enhancement and denoising. On this basis, the effectiveness of the research results is evaluated. The results show that compared with the traditional remote sensor imaging technology, the PSNR value of the image constructed by the depth neural network algorithm is significantly improved by about 15.2401, the IFC value is improved by about 0.4523, and the SSIM value is improved by about 0.4500, which improves the low light level imaging quality to a higher extent, and provides excellent technical guidance for the remote sensor to obtain the object information at night.