Description

This paper presents an efficient color matching approach to address the photo metric in consistency problem that commonly exists in panoramic images. Color correction, as the first step, is to adjust the color and luminance of source images so that the differences between a djacent images can be minimized. Color blending is used after the color correction to further smooth the color transition between adjacent images. To establish the global error function, it is necessary to extract all the mean values of overlapping areas between every pair of adjacent images. In other words, the color correction process and panorama stitching process can only be started after the complete image sequence is captured, which causes unwanted delays. In this paper, an efficient image stitching approach is proposed to address the color/luminance inconsistency problem that existed in panoramic imaging. The proposed approach aims at panorama stitching on mobile devices and it improves the color correction and color blending processes proposed in to reduce the processing delay. Two major advantages of the proposed algorithm are: 1) the synthesis process of a panorama image can be started during image capturing instead of waiting until the whole input image sequence is captured as in 2) The proposed color correction and color blending methods are simpler than those in. The rest of the paper is organized as follows. The detailed color correction algorithm is introduced in Section II-A, followed by the proposed color blending approach in Section II-B. Section III shows the implementation and results. Section IV concludes the paper.