Effect of checkerboard on the accuracy of camera calibration

Abstract

Camera calibration is an important step in many fields, and its accuracy is influenced by many factors. In this paper, the influence of different numbers and sizes of squares of the checkerboard on calibration is studied, and the errors caused by this are analyzed. We propose SBI algorithm to improve the calibration accuracy. In order to better judge the quality of calibration, we map the texture map onto the depth map by coordinate transformation to achieve the effect of registration. Besides, we use the Canny operator to extract the depth map edges and the mapped texture map edges, and utilize their matching results to evaluate the calibration accuracy. Experimental results show that our registration effect is the best and the number of mismatched pixels on the edges is the least, which indicates our calibration accuracy is the highest.