Color correction without color patterns for stereoscopic camera systems

Color consistency between image pairs from stereo camera systems is crucial for 3D display systems as well as depth estimation by stereo matching. Even though we use the same type of cameras with the same hardware settings, color response can be significantly different due to the variation of lighting conditions and radiometric characteristics. In this paper, we propose a color correction method for stereoscopic camera systems. Previously, color patterns were used to search out color matches between two or multiple images. In spite of accurate estimation of the previous work, using color patterns has the limitation for various applications. Without color patterns, we automatically find correspondences of a stereo pair using feature detection and RANSAC-based stereo matching. Then, by calculating color transform matrix, we reduce color discrepancy between the stereo pair. The accuracy of color correction is dependent on feature detectors, matching procedures and color transform methods. Experimental results show that appropriate types of them should be chosen according to target applications in the view of the processing time and accuracy.