基于Harris的棋盘格角点检测改进算法

角点作为含有图像重要信息的一种特征点,运用角点计算可以有效减少图像运算量.正因为如此,角点检测在机器视觉中更是起着不可忽略的作用,并广泛应用于视觉测量、图像匹配和摄像机标定中.针对Harris的棋盘格角点检测的优劣性,综合尺度空间、Harris算子、Fonster算子和SVD算子的思想,提出一种由粗到精的多层策略,快速自动实现棋盘格角点检测与提取,既采用了Harris算子良好的定位性能和鲁棒性,又结合了其他算子良好的优越性,克服了单种算子的局限性.实验证明,该方法针对棋盘格角点有较高的检测精度、准确率和稳定性,可以满足实时性要求.

Improved checkerboard corner detection algorithm based on the Harris

Corner of the image containing important information as a feature point,the use of corner point calculations can effectively reduce image computation.Corner detection plays a non-negligible role in machine vision,it widely used in visual measurement,image matching and camera calibration.For the pros and cons of checkerboard corner detection with Harris,integrate scale space,Harris operator,Fonster operator and Singular Value Decomposition of thinking,proposes a coarse-to-fine multi-layer strategy to achieve rapid automatic checkerboard corners detection and extraction.It uses both good positioning performance and robustness of Harris operator,but also combines the advantages of other good operators to overcome the limitations of single species operator.Experimental results show that the method for the checkerboard corners have higher detection accuracy,the accuracy and stability,to meet real-time requirements.