Advanced Retinex-Net image enhancement method based on value component processing

When capturing images under low-light lighting conditions, the resulting often suffer low visibility. Such low-visibility not only affect visual effect but also cause many difficulties in practical application. Therefore, image enhancement technology conditions has always been a challenging problem algorithms. Considering that most of existing methods are based on RGB color space technology, correlation among three primary colors is ignored, which makes distortion phenomenon easy to occur when enhanced. To solve problems poor visibility and deviation this paper an advanced Retinex network method proposed. In method, firstly transformed into HSV space, decomposition used decompose enhance value component separately, thus increasing resolution through up-sampling operation; then, for hue saturation component, nearest neighbor interpolation increase their resolutions, enhanced combined convert back obtain initial image; finally, wavelet transform fusion fuse with original eliminate over-enhanced part image. The analysis experimental results shows proposed obvious advantages brightness restoration images. Especially, comparing NIQE decreases by average 19.49%, standard increases 41.35%. algorithm expected be effectively fields security monitoring biomedicine.