摘要
为了提高煤矸石分选的识别率,研究煤与煤矸石在同一场景下不同特征的数据差异;将煤与煤矸石各分为2类,从各类中抽取样本,经预处理后分析灰度特征、纹理特征、灰度分布直方图,以及不同阈值时各类样本灰度级为255的像素点频率;基于灰度特征与纹理特征,采用支持向量机对样本进行训练。结果表明,各类煤与煤矸石在灰度特征的灰度能量与灰度熵、纹理特征的能量与熵、灰度分布直方图特征以及阈值为55~70且灰度级为255时的频率存在明显差异,基于支持向量机与多种特征的煤矸石识别率最高可达96.4%。
To improve identification rate of coal gangue,data difference of different features of coal and coal gangue in the same scene was researched.The coal and ther coal gangue were respectively divided into 2 categories.Sampling from each category and after pre-processing,gray features,texture features,gray distribution histograms,and pixel frequencies with gray level of 255 of each category at different thresholds were analyzed.Based on gray features and texture features,support vector machine was used to train the samples.The results show that gray energy and gray entropy of gray features,energy and entropy of texture features,gray distribution histogram features,and frequencies with threshold from 55 to 70 and gray level of 255 are obviously different.The highest identification rate of coal gangue based on support vector machine and multi-features reaches 96.4%.
作者
范振
陈乃建
黄玉林
张来伟
李映君
FAN Zhen;CHEN Naijian;HUANG Yulin;ZHANG Laiwei;LI Yingjun(School of Mechanical Engineering,University of Jinan,Jinan 250022,Shandong,China)
出处
《济南大学学报(自然科学版)》
CAS
北大核心
2021年第3期277-284,共8页
Journal of University of Jinan(Science and Technology)
基金
国家自然科学基金项目(51875250)
山东省自然科学基金项目(ZR2017MF023)
山东省高等学校“青创科技支持计划”项目(2019KJB018)。
关键词
图像处理
煤矸石识别
灰度特征
纹理特征
支持向量机
image processing
coal gangue identification
gray feature
texture feature
support vector machine