Research On True Color Restoration Of Visible And Near Infrared Image Under Low Illumination

There are some shortcomings in the application of long time exposure or supplemental light methods in obtaining true color images under low illumination conditions.Under low illumination,the color image has the real color of the scene,but the noise is large and the brightness is very low.Under low illumination,the color image with near infrared has high signal-to-noise ratio,but the color is red.In this paper,we build an image acquisition experimental platform for the same scene to simultaneously collect visible band true color images,color images including visible and near infrared bands,gray images including visible and near infrared bands,and true color images using complementary light.Under the condition of clear and full moon,the illumination is between 0.01-10 lux,and more than 60 experimental scenes including different scenes are taken.The polynomial nonlinear regression and chroma brightness separation algorithm under different image sources are studied to realize true color restoration,and verify the feasibility of the algorithm through actual shooting.Specific work includes:(1)A color restoration method based on visible light and near-infrared color images under low illumination is proposed.The correction matrix is solved by polynomial regression method through the collected color sample color images and near-infrared color images,and then the color images are solved by the color scene images and correction matrix containing near-infrared.Some of the color samples were selected and some higher-order terms were added to the polynomial regression method.The experimental results show that the improved algorithm has a better color restoration effect,and the peak signal-to-noise ratio(PSNR)is about 1-3d B higher than that of the collected color images.(2)A gray image color restoration method based on visible light and near infrared under low illumination is proposed,because the color restoration effect of the green area in the experimental results of the previous algorithm is not good.And gray image containing near-infrared high,the algorithm increases the gray image containing nearinfrared,also increased the higher order term in the polynomial regression method and select part of the sample in the color samples,the experimental results show that the improved algorithm has good effect of color restoration and PSNR value relative to the acquisition of color images probably increased 1-4 dB.(3)A research method of color restoration based on chromaticity separation of visible and near-infrared grayscale images under low illumination is proposed.This method first estimates an image according to the relative difference of local contrast between visible and near-infrared grayscale images.Then,the method is extracted from the gray image containing near-infrared image details of the outline,according to the estimate of the image to extract the details of the weighted and injected into the visible light image,the experimental results show that the improved algorithm has good color restoration effect and PSNR value relative to the acquisition of color images probably increased 1-4 dB.