基于概率神经网络的矿井运输胶带图像监测  

Method of probabilistic neural network in recognizing belt monitoring images

作　　者：刘晓阳[1] 王瑛[1] 王帅[1] 

机构地区：[1]中国矿业大学(北京)机电学院信息工程研究所,北京100083

出　　处：《辽宁工程技术大学学报（自然科学版）》2010年第1期124-127,共4页Journal of Liaoning Technical University (Natural Science)

基　　金：教育部博士点基金资助项目(20050290010)

摘　　要：针对煤矿井下输煤胶带图像形状和边缘特征难以提取的特点,为实现煤矿井下输煤系统胶带状态的监控,提出了胶带红外图像识别方法提出了用灰度共生矩阵提取胶带图像纹理特征,将提取的纹理统计量经正规化处理后表现出较好的可分性,采用概率神经网络对空载(或少煤)的胶带图像和有煤的胶带图像进行了识别。实验结果表明,提出的算法可以较好地识别胶带的两种状态,为煤矿井下胶带运输系统监控提供了新的途径。In order to monitor conveyer belt in coal mines, a recognition method of infrared image of conveyer belt is presented in this paper. As it is difficult to extract the features of belt shape and edge, the method presented uses gray level co-occurrence matrix to extract the texture features of the infrared image of conveyer belt. After a normalization process, the texture statistical variables show that they are fairly divisible. Therefore, a Probabilistic neural network is used to recognize between unloaded conveyer belt and loaded conveyer belt. Experimental results show that the algorithm proposed can distinguish the two states of a conveyer belt.

关 键 词：煤矿 概率神经网络 运输胶带 灰度共生矩阵 

分 类 号：TD6[矿业工程—矿山机电] TP229[自动化与计算机技术—检测技术与自动化装置]