Fast drawing method of circular patterns based on Gaussian circles for camera calibration

Abstract

Camera calibration is a fundamental technology for image measurement. Calibration using 2D planar patterns is the de facto standard, and is performed by displaying and detecting patterns such as a circle pattern or a checkerboard pattern, and calibrating based on the correspondence between the coordinates of the patterns. Recently, the patterns are often displayed on a liquid crystal display (LCD). While many previous papers have focused on the detection method, this paper focuses on the display method. This is because LCDs have coarser dots per inch than printing devices. In this paper, we propose a method for drawing circular patterns on LCD that accurately and fast detects feature points based on the Gauss circle problem. Experimental results show that the proposed method is 4 times more accurate and 2.5 times faster than previous drawing methods.