Perception-Based Gradient Domain Enhancement of Images

Common contrast enhancement methods compress the dynamic range of images while increasing contrast and enhancing fine details in the darker regions.In the mean time,details in the brighter regions are usually not enhanced,or even attenuated in some cases.The novelty of our method is to take the human visual perception sensitivity into consideration.We introduce the idea of relative gradients and propose a method that tends to magnify gradient more in brighter regions,and an enhanced image is reconstructed from the gradient field in least mean square sense.By transforming the RGB images to HSV space,the method is extended to the color image enhancement.Experimental results demonstrate that,by our method,details in both brighter and darker regions of the original images are enhanced or preserved.