食品包装数粒机图像检测系统设计  

Design of Image Detection System for Food Packaging Granulation Counter

作　　者：张州[1] 白克[1] ZHANG Zhou;BAI Ke(Henan Institute of Economics and Trade,Zhengzhou 450000)

机构地区：[1]河南经贸职业学院,郑州450000

出　　处：《食品工业》2020年第7期179-181,共3页The Food Industry

摘　　要：为提高食品包装数粒机速度、精度以及缺陷检测能力,采用图像处理技术设计一种包装数粒机检测系统。在原有数粒机基础上搭建一套视觉检测系统并详细论述系统结构。以圆形片剂为例,利用灰度补偿算法实现片剂图像预处理。通过双边阈值获取图像目标区域。根据面积、周长和圆形度等特征判断片剂的完整性。经过具体试验验证,所述检测系统对正常药粒的识别准确率可达99.4%,对残损药粒的识别准确率可达99.7%以上,准确度比较高;单幅图像的处理时间仅需45 ms,速度比较快。该包装数粒机检测系统能够满足高实时性及高识别率等要求,降低检测成本,在实际生产中应用成效显著。In order to improve the speed,accuracy and defect detection ability of food packaging granulator,a packaging granulator detection system was designed using image processing technology.A visual detection system was built on the basis of the original granulator and its structure was discussed in detail.Taking the circular tablet as an example,the image preprocessing of the tablet was realized by using the grayscale compensation algorithm.The target region of the image was obtained by bilateral threshold.According to the area,circumference and roundness of the tablets were used to determine the integrity of the tablets.Through specific experiments,the detection system could identify normal drug granules with an accuracy of 99.4%,and damage drug granules with an accuracy of more than 99.7%,with a relatively high accuracy.The processing time of a single image was only 45 ms,which was relatively fast.The detection system of packaging granulator could meet the requirements of high real-time and high recognition rate and reduce the detection cost.

关 键 词：数粒机 图像处理 缺陷检测 包装 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术] TS206.5[自动化与计算机技术—计算机科学与技术]