A Study On The Denoising And Enhancement Of Low Light Image

With the rapid development of computer technology and digital image processing technology in recent years,various new image enhancement algorithms and new theories have emerged.Video image devices are widely used in various neighborhoods such as intelligent transportation,medical research and real time monitoring.However,in some special occasions,such as nighttime,rainy days,uneven illumination,or other low-illumination environments,images obtained tend to have lower image quality,suffer greater noise pollution,and have lower contrast.On the one hand,it affects people's normal viewing and makes people have poor visual experience;on the other hand,it also affects the further processing of images,such as image segmentation,information extraction and image recognition,and so on.Therefore,in this case,the application and research of low illumination image processing technology is particularly important.This article first focuses on two characteristics of low-illumination images.First,low light image suffers from large noise pollution,which is not conducive to further processing.Second,low-illumination images have lower average than images under normal illumination.For brightness and contrast,this paper proposes a low-brightness image denoising and enhancement algorithm based on illuminance map and de-fog model.The proposed method is divided into two steps:(1)A combined median filtering and guided filtering synthesis denoising method is proposed to obtain initial denoising images to avoid noise amplification in subsequent enhancement processing.(2)Based on the visually very similar characteristics of reversal of low-light images and daytime outdoor fog images,this paper uses a defogging model to enhance denoised low-light images.By applying the denoised image to the Lab color space,the luminance map L is extracted to obtain a transmission map with adaptive weights.Then,the refined quadratic tree method is used to obtain the refined atmospheric light values.Finally,the final adaptive low-light enhanced image is obtained through the defogging model inversion.In the end,a large number of simulation experiments are used to compare the current advanced algorithm with the processing effect of this algorithm.The low-light image denoising and enhancement algorithm based on the defogging model and the luminance component proposed in this paper has a better denoising effect compared with the typical algorithm,and can retain more detailed information while being enhanced,which is inline with human eyes.The visual characteristics are more authentic and have wider adaptability to low-light scenarios in different situations.