Research On Low-light Image Enhancement Algorithm Based On Multi-scale Fusio

With the vigorous development of social media such as microblog and Tiktok,it has become popular to record people’s daily lives with photos or videos and share them with others through the Internet.However,due to natural environmental factors and the lack of professional photography ability of most users,the quality of collected images is generally not high.For example,when in a dark or uneven light environment,the visual quality of collected images hardly meets human visual requirements,and it may also damage the performance of algorithms designed primarily for high visibility inputs.This type of image is generally characterized by low overall brightness,insufficient contrast,loss of detail,high noise,and color distortion.This type of image is called low-light image.For specific purposes and to reduce or eliminate degradation in low-light image,enhancement processing must be performed on low-light image.The main contents are as follows:Aiming at the problems of low-light image such as low overall brightness,insufficient contrast,and color distortion,a low-light image enhancement algorithm based on light reflection model was proposed.The algorithm first performs channel stretching on the original image to correct any type of color distortion.Secondly,the image is adaptively enhanced using Weber-Fechner law,in order to avoid artifacts while maintaining the brightening effect of chromaticity,adaptive chromaticity is introduced.Two source images are obtained by adjusting parameters and fused using principal component analysis.Then,the fusion image is processed using contrast limited adaptive histogram equalization to improve the overall contrast.Finally,subjective evaluation and objective analysis are conducted to compare the proposed algorithm with several existing image enhancement algorithms.Simulation results show that the proposed algorithm has significant advantages.Aiming at the problems of large noise and partial detail loss in low-light image,a low-light image enhancement algorithm based on multiscale fusion was proposed.First,a generalized unsharp mask algorithm and an improved gamma correction were used to process the image in parallel to enhance contrast,clarity,and brightness,and two images were obtained.And performs multiscale fusion with the image processed by a low-light image enhancement algorithm based on a light reflection model.Then,an improved guided filtering is used to process the fused image for denoising.Finally,a subjective and objective comparative analysis is conducted between this algorithm and several existing image enhancement algorithms.Simulation results show that the proposed algorithm has significant advantages.Based on MATLAB GUI,a low-light image enhancement system is built.Simulation tests show that the system can easily achieve human-computer interaction and visualization.