| Color is the important information of images.Whether the imaging device can correctly perceive the image color is very important for target recognition and security monitoring.However,in the night or dark environment,the image captured by the imaging device often has problems such as color distortion,reduced clarity,and loss of detail information,which leads to a reduction in the overall quality of the image,thereby affecting human visual perception and many computer vision applications.Based on a large number of color samples of color images in low and normal illumination scenes,this paper innovatively proposes a color reproduction method for low-illumination scene imaging,referred to as low-illumination color reproduction method.The main research contents are as follows:(1)Study color fading mechanism of low-illumination images by constructing a low-illumination scene imaging simulation model with a single illumination variable.(2)Based on Munsell color system and image clustering,design and implement the minimum reproducible illumination estimation method,which provides theoretical support for the realization of low-illumination color reproduction method.(3)According to color fading mechanism and characteristics of low-illumination images,aiming at the problem of color distortion in low-illumination scene imaging,design and implement a low-illumination color reproduction method based on gamut mapping,optimize the color reproduction effect and accuracy of the fidelity matrix by smoothness matrix,matrix weight distribution and gamut boundary constraint.(4)Through ablation experiments,subjective and objective evaluation of image quality,verify the effectiveness of the proposed low-illumination color reproduction method and its modules,and the effect is better than other methods.Based on a large number of experiments,this paper designs and implements a low-illumination color reproduction method,which provides a new idea for future digital image color reproduction,and forms a technology that supports the night applicability of computer vision applications.It has good theoretical and practical value for low-illumination scene imaging. |