基于监督双限制连接Isomap算法的带钢表面缺陷图像分类方法

根据带钢表面缺陷图像具有复杂纹理结构、包含大量干扰信息、具备高维非线性几何结构等特点，本文提出基于监督双限制连接Isomap方法的带钢表面缺陷图像降维方法（dls-Isomap）.该方法以Isomap降维方法为基础，对其邻域图的连接方式进行K邻域（K-nearest neighbor，KNN）和ε-半径两个方面的限制性连接，并使用数据类别作为监督对类间邻域点进行扩展连接.针对多类Roll-swiss数据实验表明，dls-Isomap降维方法不仅能够在低维空间中完整嵌入所有数据点，而且能保持数据各类内和类间的几何结构，以及解决Isomap算法存在的“短路边”问题；针对带钢表面缺陷图像分类实验表明，基于dls-Isomap的新分类方法适合含水、油渍等干扰较多的带钢表面缺陷的分类任务，其中冷轧带钢5类缺陷识别率可以达78%.含水渍的热轧带钢缺陷识别率可以达到93%，其中水渍干扰图像的识别率达到97.6%.

Strip Surface Defect Image Classification Based on Double-limited and Supervised-connect Isomap Algorithm

A double-limited and supervised-connect Isomap dimensionality reduction and classification method (dls-Isomap) is proposed in this paper to classify more accurately the stripe surface defect images with the typical characteristics of complex texture, much noise, and high-dimension non-linear geometry. Based on the dimensionality reduction technique from Isomap, the connection of neighborhood graph is limited by key parameters K-nearest neighbor (KNN) and ε-radius, and inter-class neighborhood points are connected extensionally with the supervision of class labels. According to multi-classes roll-swiss data experiments, all the points can be embedded in lower dimensions with the complete inter-class and intra-class geometric structure, and the "short circuit" in the Isomap can be solved by the dls-Isomap method. In addition, stripe surface defect images data experiments show that the proposed classification method is suitable for the classification of stripe surface defects including more water and oil, with a recognition rate of 78% for cold-roll strip images, and 93% for hot-roll strip images with water, among which the recognition rata of water defects is 97.6%.