改进的棋盘格自生长亚像素角点检测算法-《计算机技术与发展》

文章信息/Info

Title:: An Improved Checkerboard Self-growing Subpixel Corner Detection Algorithm

作者:: 昌海亮1; 2; 3 ; 席隆1; 2; 3; 1. 中国科学院空间应用工程与技术中心,北京 100094;
2. 中国科学院大学,北京 100049;
3. 中国科学院太空应用重点实验室,北京 100094

Author(s):: CHANG Hai-liang1; 2; 3 ; XI Long1; 2; 3; 1. Technology and Engineering Center for Space Utilization,Chinese Academy of Sciences,Beijing 100094,China;
2. University of Chinese Academy of Sciences,Beijing 100049,China;
3. Key Laboratory of Space Utilization,Chinese Academy of Sciences,Beijing 1000

Keywords:: ub-pixel corners; checkerboard; self growing; energy function; corner detection

摘要:: 针对当前棋盘格内角点检测算法需要提前指定棋盘格尺寸大小以及无法处理一张图片包含多张棋盘格情况的问题,提出了一种改进的棋盘格自生长亚像素角点检测算法。在分析现有算法的基础上,引入棋盘格的能量函数使棋盘格能够自适应生长。该算法的原理是首先定位棋盘格角点位置,接着亚像素化角点,提高精度。最后,优化引入的能量函数,使棋盘格生长到最大化。该算法鲁棒性好,因为棋盘格是自适应生长的,所以如果出现干扰,就会绕过干扰,生长出最大的棋盘格。因为该算法不需要提前指定棋盘格尺寸,可以满足很多自动化应用。相比较 Matlab 集成工具箱中的算法,该算法可以检测一个图片里包含多张棋盘格的情况。实验分析发现,该算法抗干扰强,能在复杂背景中识别出棋盘格角点。同时,在不同倾斜角度下对棋盘格的检测性能也比较良好。

Abstract:: The current checkerboard corner detection algorithm needs to specify the size of checkerboard in advance and cannot deal withthe situation that a picture contains multiple checkerboards. An improved checkerboard corner detection algorithm based on self-growingsub-pixel is proposed. On the basis of analyzing the existing algorithms,the energy function of checkerboard is introduced to make thecheckerboard grow adaptively. The principle of the algorithm is that Position the checkerboard corners,then subpixelate the corners toimprove the accuracy,and finally,optimize the introduced energy function to maximize the growth of the checkerboard. The algorithm isrobust because the checkerboard is adaptive,so if there is interference,it will bypass the interference and grow the largest checkerboard.Because this algorithm does not need to specify the checkerboard size in advance,it can satisfy many automation applications. Comparedwith the algorithm in Matlab integrated toolbox,this algorithm can detect the situation that a picture contains more than one checkerboard.Experimental analysis shows that the algorithm has strong anti - interference and can recognize the corner points of checkerboard incomplex background. At the same time,the detection performance of checkerboard under different tilt angles is also good.