基于区域生长的雪糕棒毛刺检测算法
|
摘要
为提高雪糕棒产品质量,避免雪糕棒对人体造成伤害,根据雪糕棒缺陷类型对其进行分类处理是一种有效手段。但雪糕棒各类缺陷复杂多样且具有一定的相似性使得区分缺陷类别成为一个难点。针对以上问题,文章采用了一种基于区域生长的缺陷检测方法对毛刺与矿物线缺陷进行准确分类。首先对图像进行滤波处理来排除噪声干扰,然后获取缺陷区域骨架,并通过分析骨架端点8邻域内像素灰度值特征来获取区域生长的起点;在此基础上将骨架的方向以及骨架端点的灰度值作为相似性度量准则进行下一步生长,最后根据生长后的区域与雪糕棒边缘的连通性区分缺陷类别。利用该方法对建立的ICS-F缺陷图库进行测试,毛刺与矿物线的准确识别率分别达到99.6%、98.4%。该方法为雪糕棒工业现场实时缺陷分类提供可行性,具有一定的实用价值。
In order to improve the quality of the ice cream bar products and avoid the harm to the human body, it is an effective means to classify it according to the type of ice cream bar defect. However, the complex and diverse defects of the ice cream bar have made it a difficult problem to distinguish the defect categories. In view of the above problems, a defect detection method based on regional growth is used to accurately classify the defects of burr and mineral lines. Firstly,image filtering to eliminate noise, and then obtain the defect region framework, and through the analysis of the skeleton endpoints in the 8 neighborhood pixel gray value feature to obtain the starting point of regional growth; on the basis of the direction of backbone and gray skeleton endpoints value as the similarity metrics for the next step of growth, according to the regional and ice cream bar edge after growth of connectivity between defects. The method is used to test the established ICS-F defect library, and the accurate recognition rate of the burr and the mineral line is 99.6% and 98.4%respectively. This method provides the feasibility for the real-time defect classification of the ice cream bar industry, and has certain practical value.
In order to improve the quality of the ice cream bar products and avoid the harm to the human body, it is an effective means to classify it according to the type of ice cream bar defect. However, the complex and diverse defects of the ice cream bar have made it a difficult problem to distinguish the defect categories. In view of the above problems, a defect detection method based on regional growth is used to accurately classify the defects of burr and mineral lines. Firstly,image filtering to eliminate noise, and then obtain the defect region framework, and through the analysis of the skeleton endpoints in the 8 neighborhood pixel gray value feature to obtain the starting point of regional growth; on the basis of the direction of backbone and gray skeleton endpoints value as the similarity metrics for the next step of growth, according to the regional and ice cream bar edge after growth of connectivity between defects. The method is used to test the established ICS-F defect library, and the accurate recognition rate of the burr and the mineral line is 99.6% and 98.4%respectively. This method provides the feasibility for the real-time defect classification of the ice cream bar industry, and has certain practical value.
引文
[1]王玉春.雪糕棒生产工艺简介[J].林业科技开发,1993(3):22.
[2]关胜晓.机器视觉及其应用发展[J].自动化博览,2005(9):80-82.
[3]周赟,李久贤,夏良正.基于区域生长的红外图像分割[J].南京理工大学学报,2002(S1):81-84.
[4]陈宏希.基于边缘保持平滑滤波的Canny算子边缘检测[J].兰州交通大学学报,2006(01):86-90.
[5]Mehnert,Andrew,Jackway,et al.An improved seeded region growing algorithm[J].Pattern Recognition Letters,1997,18(10):1065-1071.
[6]刘宝生,闫莉萍.几种经典相似性度量的比较研究[J].计算机应用研究,2006,23(11):1-3.
[7]Zhang X F,Zhang Y.Image edge detection method of combining wavelet lift with canny operator[J].Procedia Engineering,2011(15):1335-1339.
[8]孙秋成,谭庆昌.一种亚像素精度的边缘检测方法[J].北京工业大学学报,2009,35(10):42-47.
[2]关胜晓.机器视觉及其应用发展[J].自动化博览,2005(9):80-82.
[3]周赟,李久贤,夏良正.基于区域生长的红外图像分割[J].南京理工大学学报,2002(S1):81-84.
[4]陈宏希.基于边缘保持平滑滤波的Canny算子边缘检测[J].兰州交通大学学报,2006(01):86-90.
[5]Mehnert,Andrew,Jackway,et al.An improved seeded region growing algorithm[J].Pattern Recognition Letters,1997,18(10):1065-1071.
[6]刘宝生,闫莉萍.几种经典相似性度量的比较研究[J].计算机应用研究,2006,23(11):1-3.
[7]Zhang X F,Zhang Y.Image edge detection method of combining wavelet lift with canny operator[J].Procedia Engineering,2011(15):1335-1339.
[8]孙秋成,谭庆昌.一种亚像素精度的边缘检测方法[J].北京工业大学学报,2009,35(10):42-47.