一种实用的X型靶标亚像素角点提取方法

为了提取亚像素角点和实现高精度的标定,提出了一种基于Harris算子和空间矩的亚像素角点提取方法。利用Harris算子,在优化后的范围内提取像素级角点;运用改进后的梯度模板提取像素级角点周围部分边界点,并利用空间矩的方法得到边界点的亚像素级坐标;将亚像素边界点进行直线拟合,并将交点的平均值作为该角点的亚像素坐标。实际测试证明:利用该方法提取到的角点精度可以达到0.1pixel,可满足实际的公差要求,为X型靶标的角点提取提供了一种新的思路,目前已经将该方法应用到了嵌入式机器视觉工业现场。

A practical method for detection of sub-pixel corners for X-target

In order to extract sub-pixel corners to achieve high-precision calibration,a method for extracting sub-pixel corners based on Harris operator and spatial moment is proposed. The Harris operator is made to extract pixel-level corners within the optimized range and extract part of the boundary points around the corners by improved template of the gradient. It uses spatial moment approach to obtain the sub-pixel coordinates of the boundary points,and fits the sub-pixel boundary points to a straight line,and saves the average intersection points as the sub-pixel coordinates of corners. The actual test shows that the precision of this method can reach 0.1 pixels,it can meet the actual tolerance requirements,and provide a new idea for the X-type corner target object. At present the method has been used in embedded machine vision industrial field.