中文摘要：

圆形结构特征大量存在于工业CT图像中，有效地检测出圆并获得其参数，在图像测量及逆向工程应用中具有重要意义。针对现有存在概率图的圆检测方法计算时间长、占用内存大的缺点，通过事先确定潜在圆心选择范围的方法，将存在概率数值符合设定条件的圆的参数信息存储在链表中，并根据被检测圆的数量确定圆的参数，标识其位置。该方法大大降低了计算量，由于利用链表方式存储数据，降低了内存的开支。实验表明，该方法在计算效率上比现有方法提高了36～178倍。

英文摘要：

The circle structure widely exists in the Industrial CT images. It is significant to detect the circle, obtain the parameter and apply to the image measurement as well as the reverse engineering. To overcome the defect of long-time calculation and memory consuming in the current method of existence probability map for detecting circles, the present study first made sure the potential range of the center of circle, and stored the message which met the required numerical values in the chain and then decided the parameter of circle and marked the position according to the checked number. The research will reduce calculation task and save the internal stored capacity while using the form of chain. The result shows that this method has improved greatly the calculation rate36～178 times than that of the original one.