基于边缘检测与凹点匹配的接触珍珠分离方法

Method for Separating Contact Pearls Based on Pit Detection

针对机器视觉分选珍珠过程中,相互接触的珍珠分离精度低等问题,提出一种基于凹点检测的接触珍珠分离方法。采用背光成像获取多颗珍珠图像,通过预处理算法提高珍珠图像质量。通过计算连通域面积比提取接触珍珠区域,以边缘跟踪算法得到接触区域的轮廓特征,建立边缘轮廓点夹角的数学模型,将夹角小于60°判定为候选的凹点,再通过凹点匹配算法准确得到凹点,以欧式距离做了度量函数实现接触珍珠的分离。实验结果表明:论文方法较Harris算法具有较好的的分割效果,分离正确率达到为98.6%,面积损失率为1.443%,平均时间为0.280s。

Aiming at the problem of low separation precision of pearls in contact with each other in the process of machine vi⁃sion sorting pearls,a method for separating contact pearls based on pit detection is proposed.Backlight imaging is used to obtain multiple pearl images,and the quality of pearl images is improved through preprocessing algorithms.The contact pearl area is ex⁃tracted by calculating the area ratio of connected domains,the contour feature of the contact area is obtained by the edge tracking al⁃gorithm,the mathematical model of the angle between the edge contour points is established,and the angle is less than 60°as the candidate concave point,and then the concave point matching algorithm is used.The pits are accurately obtained,and the Euclide⁃an distance is used as a measurement function to realize the separation of contact pearls.The experimental results show that the method in this paper has better segmentation effect than Harris algorithm.The accuracy of separation is 98.6%,the area loss rate is 1.443%,and the average time is 0.280s.