一种煤和煤矸石图像识别的新方法  

A New Method for Image Recognition of Coal and Coal Gangue

作　　者：余乐[1] 

机构地区：[1]工业智能化技术与系统福建省高校工程研究中心,华侨大学,泉州362021

出　　处：《现代计算机（中旬刊）》2017年第6期66-70,共5页Modern Computer

基　　金：华侨大学研究生科研创新能力培育计划资助项目(No.1511422005)

摘　　要：构造一种基于非线性灰度压缩扩阶共生矩阵的煤和煤矸石图像识别的新方法,用于对煤矿成分中的煤和煤矸石进行纹理数据分析并分选。采用对原有图像0-255级灰度信息前部分进行灰度级压缩,并在该部分进行灰度级扩阶,其剩余灰度部分保持原灰度级不变并且不作处理,并计算在此处理情况下4个不相关纹理特征值:能量、熵、惯性矩、相关性。实验结果表明,该方法比常规灰度共生矩阵、间隔灰度压缩扩阶共生矩阵具有更高的区分度,对煤和煤矸石分选的正确率能达到92.5%。Constructs a new method of coal and coal gangue image recognition based on non-linear gray-scale compression-expansion co-occurrence matrix, which is used to analyze and sort the coal and coal gangue in coal mine. The gray scale of the original image 0-255 gray level information is used to compress the gray scale of the original image, and the gray scale is extended in the part. The remaining gray scale is kept unchanged and not processed. In this case, four irrelevant texture eigenvalues： energy, entropy, moment of inertia, con＇elation. The experimental results show that the proposed method has a higher degree of discrimination than the conventional gray level co-occurrence matrix and the interval gray scale compression expansion covariance matrix, and the correct rate of coal and coal gangue sorting can reach 92.5%.

关 键 词：煤和煤矸石图像识别 特征提取 共生矩阵 非线性灰度压缩扩阶共生矩阵 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]