基于SVM和纹理的煤和煤矸石自动识别

Recognition of coal and stone based on SVM and texture

为提高煤矸石的自动识别和分选效率,提出了基于支持向量机(SVM)和纹理识别煤矸石的方法。选取两种煤和一种煤矸石的图像作为样本,经过图像预处理及图像灰度和纹理特征分析后,发现灰度均值、灰度共生矩阵最大值、二阶矩、对比度、相关、熵为有效特征。在此基础上,采用了支持向量机来完成图像的自动识别过程,选取上述6个参数作为支持向量机的训练特征,实验结果表明,该支持向量机识别煤和煤矸石的成功率较高。

To increase the efficiency of the separation of gangue from coal, a new method based on support vector machine （SVM） and texture analysis is proposed. First, the images of two kinds Of coal and one kind of gangue as samples are selected. After a series of samples analysis as digital processing, intensity analysis and texture analysis, the six parameters including the features of average gray, the maximum of gray-level be effective parameters for the separation. At last, matrix, moment, contrast, relevant, and entropy are found to SVM classification algorithm is adopted in image classification, which uses the six parameters mentioned above in procedure of training. Experimental result prove that the accuracy of the differentiation between coal and gangue through SVM is high.