【作者】 周鹏；

【导师】 王凤文；

【作者基本信息】 东北大学 ， 通信与信息系统， 2009， 硕士

【摘要】 煤炭产量、运量控制系统对于解决煤炭生产行业架构的特殊性造成的难以控管的局面,实现煤炭产业的现代化管理具有重要意义。本文主要研究产运煤源点(生产矿)、结算站和储煤场的煤炭运量控制系统中的车载物(煤和非煤物)识别系统,即利用数字图像处理技术实现煤与非煤物自动识别,在此过程中所应用的技术和方法为今后的深入研究和将其应用于实际生产奠定了坚实的基础。本文所研究的煤与非煤物的识别是基于产运煤源点(生产矿)、结算站和储煤场这些特定的应用环境中,研究的是散煤与非煤物的识别。通过研究、分析煤与非煤物的图像,和大量的基于灰度共生矩阵的煤与非煤物的纹理特征数据,总结出了描述煤的纹理特征向量,通过支持向量机对图像样本进行训练和分类,并得到了良好的识别效果。每一种物质都具有自己特有的纹理特征,利用计算机图像处理技术完成煤与非煤物纹理特征的计算,可以获得我们需要的纹理特征向量,通过支持向量机对所获特征向量进行分类就可以判断所识别图像是煤或者非煤。本文提出了一种满足煤炭运量系统要求的煤与非煤物的图像识别算法：基于灰度共生矩阵的纹理分析识别算法的改进算法,在纹理特征中加入了灰度特征,提高了识别率。并且通过对灰度共生矩阵的可调参数的讨论与选择,克服了灰度共生矩阵对图像尺度变化敏感的缺点。更多还原

【Abstract】 Coal output and transport volume control system is great significance for solving the situation that the management of coal production industry is difficult and realizing the modernization management of the coal industry. In this paper, recognition of stowage(coal and non-coal) in coal transport volume control system of sources of coal production and transportation (ore production)、settlement station and coal storage yard was studied. Digital image processing technology was used to study and achieve recognition algorithm of coal and non-coal automatic identification. The techniques and methods in this paper lay a solid foundation for further researching and applicationIn this paper, coal and non-coal recognition was different from the previous coal and non-coal separation recognition system. It was based on the specific application environments of coal production and transportation (ore production)、settlement station and coal storage yard. Through the research and analysis of coal and non-coal image, and research and analysis the data of coal and non-coal textural property based on Gray-Level Co-Occurrence Matrix, summing up the coal grain property eigenvector. Through training and classification of the sample image by SVM, in this way can get a well recognize effect. Every material have its own textural property, make use of computer image processing technologies to finish the compute of coal and non-coal textural property, we can get the required grain property vector. We can judge the material if it is coal by classified the eigenvector got by SVM.The recognize method in this paper realized automatic recognition of coal and non-coal in coal yard. The experimental proof that we get the eigenvector based on image gray level co-occurrence matrix can describe the coal feature exactly and recognize the coal and non-coal.更多还原