摘要
圆检测Hough变换(CHT)在数字图像处理领域中有广泛的应用,为了满足圆检测准确和实时性的要求,本文首先对标准圆检测Hough变换在实际应用中存在的一些问题进行了分析,然后针对这些影响因素,提出了一种利用圆特性的Hough变换改进算法。该算法主要利用了圆边界上各对称点的梯度在水平和垂直方向上存在对称关系的几何特性,通过判断各点是否需要参与变换、筛选出候选像素点和减少参与变换各点的半径搜索范围等手段,既保证了计算速度的高效性,又对干扰噪声不敏感,降低了错误判断率,并用实验加以了证实。
The Circle Hough Transform (CHT) is one of the most popular algorithms for detecting circle. However, this technique has many drawbacks when applied in digital images. In this paper, we showed that significant wrong evidence could be generated when the CHT was used. The article presented a new method, based on CHT, which offered robust identification of circles by using geometrical features of circles. The geometric property that the midnormal to symmetry points of the circumference of a circle passed through the center of circle was used to filter non-circular feature points and determine the searching range of radius. The proposed technique is simple in implementation, accurate and efficient in computation and robust to noise. Experimental results confirmed the efficiency of the proposed method in both synthetic drawings and real images.
出处
《模式识别与人工智能》
EI
CSCD
北大核心
2004年第1期72-76,共5页
Pattern Recognition and Artificial Intelligence
