Abstract: Existing pixel level edge detection algorithm is dependent on image gray value, and the calculating neighborhood is not clear. Aiming at this problem, an improved method of sub-pixel centre of the circle mark in photogrammetry is proposed. The sub-pixel edge based on digital coarse boundaries using a dual threshold method is found by rotating image coordinate system according to the local gradient direction of the coarse boundaries. An adaptive parameter S in rotated coordinate system is used to search local sub-pixel edge points, and the centre of ellipses is detected on the basis of these points. The proposed algorithm is realized using Matlab, and the experimental results demonstrate its excellent performance and high precision in good agreement with that of the Steger edge algorithm. It is concluded that the new method can satisfy the demands in the opto-electronic measurement.

Key words: coding mark, sub-pixel, edge detection, ellipse fitting, gradient, adaptive selection

摘要： 针对现有亚像素级边缘检测算法过于依赖图像灰度值，且计算邻域不明确等不足，提出一种改进的圆形标志中心亚像素级检测方法。利用双阈值法得到边缘的粗定位点，将坐标轴按粗定位点的梯度方向进行旋转，在新坐标系下通过邻域控制参数S的自适应选择，选取合适的邻域进行边界点的亚像素级定位，利用椭圆拟合确定目标边缘。通过Matlab编程实现算法，并与原算法进行比较，结果显示改进方法的精度较高，可满足摄影测量等光电测量的要求。

关键词: 编码标志, 亚像素, 边缘检测, 椭圆拟合, 梯度, 自适应选择

CLC Number: 

• TN911.73