A robust homography technique for color correction in high dimensional environment

Abstract

One of the crucial steps in the processing of images is color correction, which is used to improve newlinethe color quality of the image during collection or pre-processing before it is being used in newlinesubsequent steps. Over the years a significant number of techniques have been developed for newlinemapping colors in target images from reference images. In the present work, an effective and newlineefficient color correcting method is proposed that is based on Alternate Least Square (ALS) and newlineRoot Polynomial (RP) techniques. In the proposed work, the two techniques are hybridized to newlineform a hybrid ALS-RP color correction method. The primary goal of the suggested model is to newlineminimize discrepancies between two photos, improving overall visual quality. newlineTo deal with this task, Amsterdam Library of Object Images (ALOI) dataset is used which newlinecontains thousands of pictures of different objects. These images are obtained under various newlinelightening conditions and angles. Moreover, a hybrid ALS-RP color correction model is applied newlineto the target picture in which colors are fixed as per the reference image. The images are newlineconverted into three color models of LAB, LUV and RGB into XYZ format to match color newlinecoordinates effectively. Furthermore, the difference in colour between the reference image and newlinethe target image is calculated using values for characteristics like Mean, Median, 95% Quantile, newlineand Maximum Error. newline