块离散余弦变换与灰色聚类相结合的表面瑕疵识别
|
摘要
针对工件或产品表面局部细微瑕疵不易识别、不利于在线检测问题,提出了一种块离散余弦变换(Block-Discrete Cosine Transform,BDCT)与灰色聚类相结合的表面瑕疵识别方法。该方法首先将图像分解成等尺寸的非重叠图像块,利用BDCT算法进行图像变换,提取待测图像的能量谱特性;然后将能量特性作为质量特征进行T2统计分析,初步查明待测图像表面瑕疵形状和性质;最后采用灰色聚类方法处理表面疑似瑕疵,进一步确定瑕疵具体位置和形状。为验证算法的有效性,设计了LED灯镜头表面瑕疵识别的检测方案和试验步骤,试验结果表明:本文所提出的算法相较经典的Otsu算法和分水岭算法在表面瑕疵正确识别率、速度等方面具有明显优势,可用于表面瑕疵的在线检测和识别。
It was difficult to identify and detect surface local slight defect of products online,a surface defect method was provided by Block-Discrete Cosine Transform( BDCT) combined grey clustering. An image was divided into no-overlapping image blocks and the energy spectrum was extracted by using BDCT algorithm. Then taken energy characteristics as quality feature to analyze T2 statistics,and found approximate surface defects form and features of image. Lastly suspected surface defects and shapes were further located by means of grey clustering. Testing scheme and procedures of LED lens surface defects were designed in order to verifying the validity of algorithm. Experimental results shows that presented algorithm has obvious advantages in speed and correct rate compared with classic Otsu and watershed algorithm,can be used to identify and detect surface defect of products online.
It was difficult to identify and detect surface local slight defect of products online,a surface defect method was provided by Block-Discrete Cosine Transform( BDCT) combined grey clustering. An image was divided into no-overlapping image blocks and the energy spectrum was extracted by using BDCT algorithm. Then taken energy characteristics as quality feature to analyze T2 statistics,and found approximate surface defects form and features of image. Lastly suspected surface defects and shapes were further located by means of grey clustering. Testing scheme and procedures of LED lens surface defects were designed in order to verifying the validity of algorithm. Experimental results shows that presented algorithm has obvious advantages in speed and correct rate compared with classic Otsu and watershed algorithm,can be used to identify and detect surface defect of products online.
引文
[1] ROSTEN E,PORTER R,DRUMMOND T. Faster and better:a machine learning approach to corner detecting[J]. IEEE Transaction on Pattern Analysis and Machine Intelligence,2010,32(1):105-119.
[2] PISHKENARI H N,MEGHDARI A. Surface defects characterization with frequency and force modulation atomic force microscopy using molecular dynamics simulations[J]. Current Applied Physics,2010,10(2):583-591.
[3]刘国平,胡环星,胡瑢华.基于纹理特征的磁瓦表面缺陷提取[J].现代制造工程,2015(7):119-123.
[4] PERNG D B,WANG W C,CHEN S H. A novel quasi-contact lens auto inspection System[J]. Journal of the Chinese Institute of Industrial Engineers,2010,27(4):260-269.
[5] HUANG K Y. An auto-recognizing system for dice games using a modified unsupervised grey clustering algorithm[J].Sensors,2008(8):1212-1221.
[6] SHANMUGAMANI R,SADIQUE M,RAMAMOORTHY B.Detection and classification of surface defects of gun barrels using computer vision and machine learning[J]. Measurement,2014,60(1):222-230.
[7]白雪冰,郭景秋,陈凯,等. GVF Snake模型在木材表面缺陷分割的应用研究[J].中南林业科技大学学报,2015,35(2):90-95.
[8]王世通,杨甬英,赵丽敏.光学元件表面缺陷散射光成像数值模拟研究[J].中国激光,2015,42(7):0708005-1-0708005-10.
[9]李新科,高潮,郭永彩,等.利用改进的SIFT算法检测桥梁拉索表面缺陷[J].武汉大学学报(信息科学版),2015,40(1):71-76.
[10]张玉燕,吕坤坤,温银堂,等.基于小波变换的高精度表面损伤测量[J].中国激光,2014,41(12):1208007-1-1208007-5.
[11]查方龙,张定华,黄魁东,等.基于预分割轮廓的CT图像亚体素表面检测方法[J].仪器仪表学报,2012,33(6):1308-1314.
[12]王宏宇.一种检测DCT编码图像方块化的客观质量评价方法[J].华东理工大学学报(自然科学版),2011,37(3):377-381.
[13] CHIU Yuan-ShyiPeter,LIN Hong-Dar. An innovative blemish detection system for curved LED lenses[J]. Expert Systems with Applications,2013(40):471-479.
[14]刘思峰,蔡华,杨英杰,等.灰色关联分析模型研究进展[J].系统工程理论与实践,2013,33(8):2041-2046.
[15]朱颖,童纪新.基于灰色关联分析的旅游产业研究———以江苏省常州市为例[J].重庆理工大学学报(自然科学版),2015(7):124-128.
[16]钱丽丽,刘思峰,谢乃明.基于熵权和区间灰数信息的灰色聚类模型[J].系统工程与电子技术,2016,38(2):352-356.
[17]周同,邹丽新,尤金正,等.基于改进Canny边缘检测算子的电子稳像算法研究[J].计算机应用研究,2010,27(2):506-508.
[2] PISHKENARI H N,MEGHDARI A. Surface defects characterization with frequency and force modulation atomic force microscopy using molecular dynamics simulations[J]. Current Applied Physics,2010,10(2):583-591.
[3]刘国平,胡环星,胡瑢华.基于纹理特征的磁瓦表面缺陷提取[J].现代制造工程,2015(7):119-123.
[4] PERNG D B,WANG W C,CHEN S H. A novel quasi-contact lens auto inspection System[J]. Journal of the Chinese Institute of Industrial Engineers,2010,27(4):260-269.
[5] HUANG K Y. An auto-recognizing system for dice games using a modified unsupervised grey clustering algorithm[J].Sensors,2008(8):1212-1221.
[6] SHANMUGAMANI R,SADIQUE M,RAMAMOORTHY B.Detection and classification of surface defects of gun barrels using computer vision and machine learning[J]. Measurement,2014,60(1):222-230.
[7]白雪冰,郭景秋,陈凯,等. GVF Snake模型在木材表面缺陷分割的应用研究[J].中南林业科技大学学报,2015,35(2):90-95.
[8]王世通,杨甬英,赵丽敏.光学元件表面缺陷散射光成像数值模拟研究[J].中国激光,2015,42(7):0708005-1-0708005-10.
[9]李新科,高潮,郭永彩,等.利用改进的SIFT算法检测桥梁拉索表面缺陷[J].武汉大学学报(信息科学版),2015,40(1):71-76.
[10]张玉燕,吕坤坤,温银堂,等.基于小波变换的高精度表面损伤测量[J].中国激光,2014,41(12):1208007-1-1208007-5.
[11]查方龙,张定华,黄魁东,等.基于预分割轮廓的CT图像亚体素表面检测方法[J].仪器仪表学报,2012,33(6):1308-1314.
[12]王宏宇.一种检测DCT编码图像方块化的客观质量评价方法[J].华东理工大学学报(自然科学版),2011,37(3):377-381.
[13] CHIU Yuan-ShyiPeter,LIN Hong-Dar. An innovative blemish detection system for curved LED lenses[J]. Expert Systems with Applications,2013(40):471-479.
[14]刘思峰,蔡华,杨英杰,等.灰色关联分析模型研究进展[J].系统工程理论与实践,2013,33(8):2041-2046.
[15]朱颖,童纪新.基于灰色关联分析的旅游产业研究———以江苏省常州市为例[J].重庆理工大学学报(自然科学版),2015(7):124-128.
[16]钱丽丽,刘思峰,谢乃明.基于熵权和区间灰数信息的灰色聚类模型[J].系统工程与电子技术,2016,38(2):352-356.
[17]周同,邹丽新,尤金正,等.基于改进Canny边缘检测算子的电子稳像算法研究[J].计算机应用研究,2010,27(2):506-508.