基于计算机视觉的中药饮片检测与分级研究  

Study on Detection and Grading of Chinese Herbal Pieces Based on Computer Vision

作　　者：钱丹丹 周金海[1] QIAN Dan-dan;ZHOU Jin-hai(Nanjing University of Chinese Medicine,Nanjing 210023,China)

机构地区：[1]南京中医药大学,江苏南京210023

出　　处：《时珍国医国药》2019年第1期203-205,共3页Lishizhen Medicine and Materia Medica Research

摘　　要：为了适应大规模商业化中药饮片生产同时弥补人工质检方法不足,基于计算机视觉技术和朴素贝叶斯分类模型构建中药饮片生产线快速检测与分级系统。该系统工作时,生产线上的传送带将中药饮片送至图像采集区进行图像采集,由计算机视觉系统对采集到的图像进行面积、颜色H分量值和缺陷面积百分比的特征值提取,然后将提取到的特征值输入到朴素贝叶斯分类器中,并以中药饮片的3个等级分类为输出,最后用训练数据和试验数据对模型进行检测。测试结果表明:中药饮片检测与分级系统可以高效、较精确的完成分类工作,为中药饮片的分级提供了一种新方法。In order to adapt to the large-scale commercialization of traditional Chinese medicine decoction pieces and to make up for the lack of manual quality inspection methods,a rapid detection and grading system for the traditional Chinese medicine decoction pieces production line was constructed based on computer vision technology and naive Bayes classification model.When the system is in operation,the conveyor belt on the production line sends the Chinese herbal pieces to the image collection area for image acquisition.The computer vision system extracts the characteristic values of the area,color H component value and defect area percentage of the collected image and then extracts the extracted image.The eigenvalues were input into the Naive Bayesian classifier,and the three grades of the TCM decoction pieces were classified as the output.Finally,the model was tested with the training data and the test data.The test results show that the Chinese medicine decoction piece detection and grading system can efficiently and accurately complete the classification work and provide a new method for the grading of the decoction pieces of traditional Chinese medicine.

关 键 词：中药饮片 分级 计算机视觉 朴素贝叶斯 

分 类 号：R282.5[医药卫生—中药学]