基于Curvelet变换和支持向量机的磁瓦表面缺陷识别方法

针对磁瓦表面缺陷对比度低、自动识别困难的问题,作者提出了一种对磁瓦图像应用快速离散Curvelet变换(FDCT)提取特征,并用支持向量机(SVM)分类器进行分类的磁瓦微小缺陷自动识别方法。该方法首先对磁瓦图像做分块处理,并对各分块图像应用FDCT,计算分解系数的l2范数,获得磁瓦不同方向的纹理频域特征;然后以归一化的分解系数l2范数作为支持向量机分类器的特征向量,对图像做出分类。对不同缺陷占比的图像进行实验测试,结果显示,当缺陷部分占分块图像的比例在1/64以上时正确识别率大于83%。

Defect detection on magnetic tile surfaces based on fast discrete curvelet transform and support vector machine

Difficulties exist in automatically inspecting surface defects because of the low intensity image contrast. To overcome these difficulties, this paper describes a textures analysis method for detecting defects on the magnetic tile surfaces. In this methodology the original image is divided into several equal sized squares, and decomposed based on a fast discrete curvelet transform (FDCT) at different scales and orientations. Then the l2 norms on the curvelet coefficients are calculated as the feature vector for support vector machine (SVM) classifier. The experimental results show that, the defects retrieval accuracy achieved 83% when defects accounted for more than 1/64 of magnetic tile image.