Color correction based on point clouds alignment in the logarithmic RGB space

Abstract

Color distortion which is caused by different illuminations and different camera settings in imaging process, incurs many problems. To overcome these problems, color correction is required. In this paper, a novel method for point cloud alignment-based color correction is proposed to eliminate unwanted color variation between images of the same or similar scenes automatically. The method is based on the Lambertian reflection model. For the advantages inherent in the logarithmic representation of an image, we first project the RGB color space into the logarithmic color space; then, in the logarithmic color space, we prove that the point clouds of the same or similar objects in different images can be related by a 3D translation vector. Next, we seek to find an accurate translation vector to align corresponding point clouds; and finally, when the source image is transformed by the translation vector, it can exhibit a similar appearance to the reference image. Various experimental results verify the validity of the proposed algorithm.