通过Tetrolet变换将热轧钢板表面图像分解成不同尺度和方向的子带,提取子带的Tetrolet高通系数矩阵特征,得到一个高维的特征矢量。利用核保局投影算法对高维特征矢量进行降维,将降维后的低维特征矢量输入支持向量机,从而实现热轧钢板表面缺陷的分类识别。对现场采集到的热轧钢板表面图像样本进行试验,包括横向裂纹、纵向裂纹、横向划伤、纵向划伤、结疤、麻点、网纹、压痕等8类常见热轧钢板表面缺陷,以及氧化铁皮和无缺陷等样本。试验结果表明基于Tetrolet变换方法对样本图像的识别率可达97.38%,比基于Curvelet变换、Contourlet变换等方法得到的识别率提高1%左右。
Sample images of hot-rolled steel plates are decomposed into multiple subbands with different scales and directions by Tetrolet transform. The high-pass Tetrolet coefficients of subbands are combined into a high-dimensional feature vector. Kernel locality preserving projection(KLPP) is applied to the high-dimensional feature vector for dimension reduction, which results in a low-dimensional feature vector. The low-dimensional feature vector is fed into support vector machine(SVM) for surface defect recognition of hot-rolled steel plates. The method is tested with sample images from an industrial production line, including transversal cracks, longitudinal cracks, transversal scratches, longitudinal scratches, scars, pimples, net cracks, impressions, scales and no defect. The results show that the recognition rate with Tetrolet transform is 97.38%, which is about 1% higher than that with Curvelet transform and Contourlet transform.
[1] 毕昕, 丁汉. TFT-LCD Mura缺陷机器视觉检测方法[J]. 机械工程学报, 2010, 46(12)：13-19.
BI Xin, DING Han. Machine vision inspection method of Muradefect for TFT-LCD[J]. J. Mech. Eng., 2010, 46(12)：13-19.
[2] 徐科, 艾永好, 周鹏, 等. 基于Contourlet 变换的连铸坯表面缺陷识别[J]. 北京科技大学学报, 2013, 35(9)：1195-1200.
XU Ke, AI Yonghao, ZHOU Peng, et al. Recognition of surface defects in continuous casting slabs based on Contourlet transform[J]. J. Univ. Sci. Technol. Beijing, 2013, 35(9)：1195-1200.
[3] 徐科, 孙浩, 杨朝霖, 等. 形态滤波在中厚板表面裂纹在线检测中的应用[J]. 仪器仪表学报,2006, 9(9)：27-31.
XU Ke, SUN Hao, YANG Chaolin, et al. Application of morphological filtering to on-line detection of surface crack of medium and heavy steel plate[J]. Chinese Journal of Scientific Instrument, 2006, 9(9)：27-31.
[4] 宋克臣, 颜云辉, 李骏, 等. 基于邻域信息评估的热轧钢板表面缺陷图像识别方法：中国,201310210470.6[P]. 2013-05-30.
SONG Kechen, YAN Yunhui, LI Jun, et al. Recognition of surface defects on Hot-rolled Steel Plates based on neighborhood information evaluation：China,201310210470.6[P]. 2013-05-30.
[5] 宋强, 徐科, 徐金梧. 基于结构谱的中厚板表面缺陷识别方法[J]. 北京科技大学学报, 2007, 29(3)：342-345.
SONG Qiang, XU Ke, XU Jinwu. Recognition of surface defects on medium and heavy plates based on structure spectrum[J]. Journal of University of Science and Technology Beijing, 2007, 29(3)：342-345.
[6] DO M N, VETTERLI M. The contourlet transform：an efficient directional multiresolution image representation[J]. IEEE Trans. Image Process, 2005, 14(12)：2091-2106.
[7] 李财莲, 孙即祥, 康耀红. 基于Tetrolet变换的图像降噪新算法[J]. 计算机应用研究, 2011(11)：4373-4375.
LI Cailian, SUN Jixiang, KANG Yaohong. New image denoising approach based on Tetrolet transform[J]. Application Research of Computers, 2011(11)：4373-4375.
[8] 延翔, 秦翰林, 刘上乾, 等. 基于Tetrolet变换的图像融合[J]. 光电子•激光, 2013(8)：1629-1633.
YAN Xiang, QIN Hanlin, LIU Shangqian, et al. Image fusion based on Tetrolet transform[J]. Journal of Optoelectronics·Laser, 2013(8)：1629-1633.
[9] 陈原, 张荣, 尹东. 基于Tetrolet Packet变换的SAR图像稀疏表示[J]. 电子与信息学报, 2012(2)：261-267.
CHEN Yuan, ZHANG Rong, YIN Dong. SAR image sparse representation based on Tetrolet packet transform[J]. Journal of Electronics & Information Technology, 2012(2)：261-267.
[10] 吴秀永. 多尺度子空间分析及其在板带表面缺陷识别中的应用[D]. 北京：北京科技大学, 2009.
WU Xiuyong. Multiscale subspace analysis and its application to surface defect recognition[D]. Beijing：University of Science and Technology Beijing, 2009.
[11] JENS K．Tetrolet transform：A new adaptive Haar wavelet algorithm for sparse image representation[J]. VisCommun Image R, 2010, 21(4)：364-374.
[12] ROWEIS S T, SAUL L K. Nonlinear dimensionality reduction by loacally linear embedding[J]. Science, 2000, 290(5500)：2323-2326.
[13] BELKIN M, MIKHAIL P. Laplacian eigenmaps and spectral techniques for embedding and clustering[M]. Cambridge：MIT Press, 2002.
[14] HE X, CAI D, MIN W. Statistical and computational analysis of locality preserving projection[C]//Proceedings of the 22nd International Conference on Machine Learning, 2005：281-288.
