Abstract: Existing ceramic tile classification system has been mostly only using the color feature or simple texture edge information for classification, it only works for monochromatic tiles or a simple pattern of ceramic tiles, but it is not suitable for complex patterns of ceramic tile. To solve this problem, it combines with gray co-occurrence matrix and statistical geometrical features of tile image, and puts these features into SVM for hierarchical classification. Because texture features are nonlinear, it uses SVM classifier constructed by the radial basis kernel function and the optimal parameters obtained by [K] cross validation method. It uses a lot of ceramic tile images obtained from the production line to do experiment, the results show that the method in this paper could obtain high accuracy and good classification effect.

Key words: ceramic tile texture classification, Gray-Level Co-occurrence Matrix（GLCM）, statistical geometrical features, Support Vector Machine（SVM）, hierarchical classification

摘要： 在已有的瓷砖图像分类系统中，仅靠颜色特征和简单的纹理边缘信息只能对无花纹的单色砖或简单花纹的瓷砖进行有效分类，对复杂图案的瓷砖存在识别率低的问题。针对此种情况，结合瓷砖图像的灰度共生矩阵和统计几何特征，将这些特征输入支持向量机进行特征分层分类。采用基于径向基核函数和[K]交叉验证法所得到的最优参数构造支持向量机，解决瓷砖纹理特征具有非线性的分类问题。用瓷砖生产线上采集的大量图像进行实验表明，该方法准确率高，分类效果好。

关键词: 瓷砖纹理分类, 灰度共生矩阵, 统计几何特征, 支持向量机, 分层分类