GEOMETRICALLY DISTORTED IMAGE CORRECTION FOR SINGLE LENS USING DECIMAL NUMBER MAPPING

摘要

In the research field of image processing, many famous and classical algorithms aiming to correct geometrically distorted image have been documented. We also proposed a mathematically simple algorithm previously which is based on Seidel Aberration equation, but it suffered from pixel discontinuity because of its integer number mapping nature. The pixel discontinuity can be eliminated by carrying out an additional pixel interpolation step, which will greatly confine the processing speed. In this paper, we introduce an improved decimal number mapping (DNM) algorithm that needs no interpolation at all. Quantitative evaluation of the DNM method proved that DNM is superior in both smoothness of the corrected image and computational speed compared with our previously proposed algorithms.