摘要
根据图像中直线轮廓边缘点的灰度梯度方向一致的特性,提出了有向线段的概念并给出了限制性定义,将直线检测转化为有向线段的检测提高了检测精度和可靠性。将十字丝轮廓抽象成两组相交的有向线段,提出了一种基于随机Hough变换(RHT)的十字丝自动检测算法。实验结果表明该算法对十字丝目标的灰度、大小、形状变化具有鲁棒性,达到了十字丝目标的快速识别。
A concept of line-with-direction (LWD) is introduced. Basing on the property that all the points in a line contour have the same gray gradient direction the limited definition of LWD is given. It improves the precision and reliability of line detecting to use LWD detecting. A fast automatic cross detecting algorithm based on randomized hough transform (RHT) is proposed which regards the cross contour as two groups of parallel lines. Experiments with simulative images and real images show that the algorithm can recognize cross fast and it is robust to the changing of the cross gray, size and shape.
出处
《光学技术》
CAS
CSCD
2004年第3期351-353,356,共4页
Optical Technique
关键词
直线检测
有向线段
十字丝
梯度方向分类
随机HOUGH变换
line detecting
line-with-direction
cross
clustering by gradient direction
randomized hough transform
