基于凹点方向线的粘连药品图像分割方法  

Based on concave point direction line adhesion drug image segmentation method

作　　者：李宏辉[1,2,3,4] 郝颖明[2,3,4] 吴清潇[2,3,4] 段红旭[2,3,4] 周静[2,3,4] 

机构地区：[1]中国科学院大学,北京100049 [2]中国科学院沈阳自动化研究所,沈阳110016 [3]中国科学院光电信息处理重点实验室,沈阳110016 [4]辽宁省图像理解与视觉计算重点实验室,沈阳110016

出　　处：《计算机应用研究》2013年第9期2852-2854,共3页Application Research of Computers

摘　　要：在利用视觉技术进行药品检测的过程中,药品相互粘连的图像会给后续的特征提取工作带来很大的困难,针对此问题提出了一种基于凹点方向线的粘连图像分割方法。首先依据颜色距离对药品图像进行预分割;然后对不同颜色的预分割图像进行4链码运算,寻找所有的凹点,并对所有的凹点进行标记;最后根据每一个凹点得到方向线,根据方向线进行配对,将配对凹点的连线作为分割线。实验结果表明,该算法具备良好的准确性和通用性。For using visual technology in the process of drug testing, drug mutual adhesion image will bring a lot of difficulties to subsequent feature extraction. For this problem, this paper presented a segmentation method based on the concave point di- rection line. It mainly made the color image segmentation in advance according to the color distance, then took 4 chain code to different color segmentation images to find all concave points and marked all the concave points. Then it got the direction line according to all concave points and matched the concave points, took the pairing concave points attachment as a parting line. The test results show that the algorithm, has a good accuracy and generality.

关 键 词：粘连分割 颜色距离 链码 凹点配对 凹点方向线 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]