一种改进的基于Hough变换的精确圆检测方法

摘要

在空间航天飞行器的对接过程中,合作目标背景的复杂性使得对圆形对接靶标部位的准确快速识别能力显得至关重要.为了克服Hough变换的圆检测遍历边缘点的局限性,采用图像预处理方法,极大地减少了参与Hough变换的无效像素点,并对边缘进行抽样提高Hough变换圆检测效率;针对Hough变换圆定位精度低,对Hough变换检测出的像素点,采用最小二乘拟合圆心得到亚像素的圆心拟合精度.一方面利用Hough变换检测圆的鲁棒性,有效地解决了最小二乘拟合的平方项对离群点敏感的问题;另一方面利用了最小二乘法的精确性提高圆心的定位精度.仿真结果和实验结果均证明了方法可将圆定位精度提高到1/20像元.%Due to complex background,the fast and accurate position identification of cooperative circle targets is very important in the spacecraft docking.To overcome the limitations of the Hough transform for edge detection,the image processing method is used to greatly reduce the invalid pixels in Hough transform and improve the efficiency of Hough transform.By using least squares fitting method in the points detected by Hough transform,the subpixel accuracy of the center and radius is achieved.On the one hand,the Hough transform is so robust that it can effectively solve the problem that the least squares fitting is sensitive to the outliers;On the other hand,least squares method can significantly improve the localization accuracy of the center.The simulation results and experimental results show that the method can improve the circular positioning accuracy to 1/20 pixels.