Seamless and Fast Panoramic Image Stitching

In this paper, a novel panoramic imaging method is presented which can compensate tone differences between images precisely by estimating ghost pixels in the overlap area. The method compensates the overlapped images bilaterally, then it prevents a panoramic image from getting darker or lighter as images are stitched. Therefore, the proposed method allows the overall panorama to keep the proper mid-tone, offering a high-speed panorama system with only software implementation in spite of its outstanding performance. Experimental results show that the proposed method can correct 95% serious corner cases taking less time than existing methods.