基于四邻域对角线上灰度约束的亚像素角点检测

Sub-pixel corner detection method based on gray constrain on diagonal line of four-neighborhood

针对微棋盘格亚像素角点检测过程中,存在漏检、多检及检测精度较低等问题,在研究现有算法的基础上,提出一种基于四邻域对角线上灰度约束的角点检测方法。通过角点的四邻域中心几何特征及其对角线方向上最大灰度特征的分析,采用SINC函数灰度分布约束角点位置,实现亚像素级精度的角点检测。开展对比实验及误差分析,其结果表明,与现有检测方法相比,该方法的多检率降低了约18%,漏检率降低了约2%,均达0;其检测精度提高了约50%,达到0.3Pix。该方法显著提高了亚像素角点检测的准确率、精度和重复率,在显微标定中具有一定的应用价值。

To solve the problems of high missing and false detection rate and low detection accuracy in the process of sub-pixel corner detection on micro-checker,a corner detection method based on gray scale constraint on four-neighborhood diagonal line was proposed.By analyzing the geometric features of the four-neighborhood center and the grayscale distribution in its diagonal direction,the detection of the sub-pixel level was realized by the corner position constraint.A SINC function grayscale distribution was used to constrain the four-neighborhood diagonal line.The contrast experiments and error analyses were carried out.Comparing with the existing methods,the missing and false detection rates decrease by about 2%and 18%,respectively,both of them reach zero.Meanwhile,its detection accuracy increases by about 50%,reaches±0.3 Pix.Experimental results show that the proposed method can dramatically improve the detection rate,accuracy and repeatability,and it has certainly applicable value in microscopic calibration.