Color Calibration with Isolated Distortion Pixels Removal Based on Human Visual Characteristics

摘要

Taking into account human visual characteristics, when isolated distortion pixels disperse sufficiently in an image, they have minimal effect on the visual image. This paper presents two color calibration methods based on excluding isolated pixels with brightness distortion under la? color space. The experimental results show that the methods can obtain better brightness and color uniformity, as well as make the histograms of corrected images in three channels get closer to normal distribution. Among them, the maximum isolated degree removal method can obtain higher richness of color, and the brightest - the darkest isolated pixels removal method can make the corrected image has higher similarity with the original image. Furthermore, the methods suggested in this paper can better reflect the transformation of surface features in the images.