具有方向特性的X角点的亚像素检测定位

Sub-pixel X-corner detection and location with directional characteristics

X角点被广泛应用于相机标定与可见光视觉跟踪系统,不仅具有位置信息,还可以具有方向信息.针对X角点的精确位置检测和方向检测,提出了一种利用X角点对称性的检测与定位算法——对称性计算算子(SC,Symmetry Calculation),简称SC算子.该算法主要由两部分组成.先利用对称性计算值确定X角点像素位置,然后基于二次曲线拟合及直线相交求亚像素位置.通过角点定位实验,仿真分析高斯噪声和畸变对定位精度及鲁棒性的影响,并与Harris算法及Micron Tracker系统进行比较.实验表明,本文方法能有效检测X角点,排除其他类型的角点和干扰点,具有较高的定位精度和鲁棒性.

X-corner points are widely used in camera calibration and visible-spectrum visual tracking system as they can provide not only location information, but also direction information. To accurately detect the location and direction of X-corner points, a new algorithm based on the symmetry calculation of X-corner points was proposed. The algorithm mainly consists of two parts. First, the pixel position of the X-corner is determined by using the value of symmetry calculation. Then the sub-pixel location for the X-corner was obtained based on the quadratic curve fitting and inter-section of lines. Through a series of experiments and a comparison with Harris algorithm and Micron Tracker system, the impacts of Gauss noise and distortion on the positioning accuracy and robustness were analyzed. Experiments show that the method can effectively detect X-corners and exclude other types of corner points and disturbance, and thus has higher localization accuracy and robustness.