基于纹理和改进径向基网络的煤监控图像识别

Identification of monitoring and control coal image base on texture features and improved radial basis function neural network

煤的红外监控图像的正确识别对矿井自动监控有重要的意义。计算煤监控图像的灰度相关矩阵各纹理统计量,分析其分布特征。在径向基函数神经网络(RBFNN)的输入层增加了正规化函数,用改进的RBFNN对图像进行了识别。结果表明,图像的纹理统计量在统计上有很好的分离性,改进的RBFNN能成功地识别出煤矿井下红外监控系统中面煤和块煤的图像。

A correct identification of an infrared monitoring coal image is an important to the mine auto monitoring and control. Calculated the matrix and texture statistic value related to the grey scale of the monitoring and control coal image and analyzed the distribution features. A normalization function was added to the input of the RBFNN. The images were identified with the improved RBFNN. The results showed that the texture statistics value of the images had a good separability in the statistics. The improved RBFNN could successfully identify the fine coal and lump coal images in the coal mine infrared monitoring and control system.