为了提高随机圆检测算法的检测效率和精密度, 提出了一种改进的随机圆检测算法。针对传统的随机算法无效累积多、执行效率低的问题, 研究了两步法无效圆模型筛选策略。首先根据图像梯度信息, 当迭代过程中取样点的梯度线与其构造圆的圆心距离小于设定阈值时, 判定圆模型为真; 然后取图像边缘点的子集, 当子集的局内点个数小于期望值时, 圆模型无效。针对随机圆检测算法精密度较差的问题, 提出了对边缘点聚类, 结合最小二乘法对结果进行优化。实验结果表明, 算法在执行效率上优于其他三种经典算法, 检测精密度与其他随机圆检测算法相比, 有显著提升。

In order to improve the detection efficiency and precision of the randomized circle detection algorithm, an improved method was proposed. Aiming at the problems of the invalid accumulation and low efficiency of traditional randomized algorithm, a two-step circle model screen strategy was studied. Firstly, according to the gradient direction of the image, only the circle whose center is approach to the gradient line is allowed to attend the next step. Then, a subsequence of the edge set will be extracted and the number points locating on the circle boundary will be counted. If the value is less than the expectation, the circle will be judged as false. Aiming at the problem of low detection precision of the randomized circle algorithm, an edge set clustering method was proposed, which combines the least square method to optimize the result. Experimental results illustrate that the proposed method shows a better performance than the other three classical algorithm, and its detection precision is also improved significantly compared with the other algorithms.