New multi-view image color correction method with karhunen- loeve transform

Abstract

Inconsistent color between different viewpoints is a serious problem in multi-view image processing. A multi-view image color correction method is proposed. We first estimate whether global correction satisfied for the target image and source image. If no global correction is satisfied, image segmentation and Karhunen-Loeve (K-L) transform are performed. Then, local mapping relations between target and source images will be established. Finally, preferred region matching method is used to correct source image. Experimental results show that, compared with other correction algorithms, corrected multi-images, obtained by the proposed method, have higher similarity with target image and the proposed method can achieve better correction result