基于视觉检测的仓储物料计数方法研究  

Research on warehouse material automatic counting method based on visual inspection

作　　者：王昱棠[1,2] 

机构地区：[1]中国科学院长春光学精密机械与物理研究所中国科学院航空光学成像与测量重点实验室,吉林长春130033 [2]中国科学院大学,北京100049

出　　处：《计算机工程与设计》2014年第7期2598-2601,共4页Computer Engineering and Design

基　　金：国家自然科学基金项目(61304032;61308099);长春光机所领域前沿创新基金项目(Y2CX1SS123)

摘　　要：针对仓储物料的计数问题,提出了一种基于视觉检测的自动计数方法,通过对圆形物料端面图像的识别实现物料的自动计数。以OpenCV库函数为基础,采用目标颜色全局阈值化及去孔洞等处理实现物料与背景的分割,使用边缘轮廓提取、Hough圆检测、圆心聚类分析等方法实现对物料的计数;根据物料堆放规律等先验知识对结果进行修正。该方法也适用于摆放不整齐的仓储物料的计数,可同时处理直径不同的仓储物料计数问题。实验结果表明了该方法的精度、鲁棒性和实时性。A warehouse material automatic counting method based on visual inspection was investigated and designed,which was able to count warehouse materials automatically by recognizing the round material ends.Based on OpenCV functions,it applied threshold segmentation and void filling to segment materials from background.Edge detection,circle Hough detection and clus-ter analysis were introduced to count the segmented material.Prior knowledge was used to revise the counting results.Experi-ment results showed that the proposed method was of satisfactory precision,robustness and speed,it could also handle conditions with materials in irregular arrangement and different diameters.

关 键 词：图像处理 视觉检测 仓储管理 自动计数 OPENCV 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]