摘要
为实现基于机器视觉的绿色高效、高智能化的煤矸分选。探讨了煤和矸石共420张图像的2个灰度特征和4个纹理特征的分布情况,并分别模拟生产中的光照、淋水、粉尘环境对煤和矸石进行了图像采集,研究其对煤矸图像特征的影响;此外,针对光照强弱、湿度、煤粉沾染程度和样品种类4个试验因素,对影响因素进行了量化处理,应用Box-Benhnken Design(BBD)试验设计理论设计四因素三水平试验,以样本灰度均值为响应指标,研究各因素对煤矸图像灰度值影响的显著性及其交互作用,从而得到区分煤和矸石的最明显特征。特征分析表明,煤和矸石的灰度特征比纹理特征具有更好的区分度,从灰度均值和峰值来看,6~36 W的光照条件对灰度均值影响有限,却使灰度峰值波动严重;样本表面喷雾量的增加使灰度均值和峰值大幅下降,以0.08 g的喷雾量为转折点,灰度均值呈现出先急后缓的对数曲线下降趋势;煤粉量与灰度均值呈一次线性反比关系,灰矸的线性比例约为块煤和黑矸的4~5倍;单因素试验表明灰度峰值对环境变化较为敏感,而响应面试验表明煤和矸石的灰度均值在同一水平下区分度明显。研究结果有利于推进机器视觉煤矸分选技术的应用,实现井下煤矸分选,亦对煤岩界面识别技术具有参考意义。
In order to realize green, efficient and highly intelligent coal and gangue sorting based on machine visionIn this study, the distribution of two grayscale features and four texture features of a total of 420 images of coal and gangue were discussed, and the image collection of coal and gangue was carried out by simulating the illumination, water and dust environment in production to study its influence on the image features of coal and gangue.In addition, according to the four experimental factors of light intensity, humidity, degree of pulverized coal contamination and sample type, the influencing factors were quantified. The Box-Benhnken Design(BBD) experimental design theory was used to design four-factor three-level experiments. The mean value was the response index, and the significance and interaction of the influence of various factors on the gray value of the coal gangue image were studied, so as to obtain the most obvious features of distinguishing coal and gangue.The characteristic analysis shows that the grayscale characteristics of coal and gangue have a better degree of differentiation than the texture characteristics, from the point of view of the grayscale mean and peak values, 6-36 W light conditions have limited influence on the grayscale mean, but make the gray peak fluctuate seriously.With the increase of spraying amount on the surface of samples, both of them decreased significantly, taking 0.08 g spraying amount as the turning point, the gray mean showed a logarithmic curve descending trend from urgent to slow.There is a linear inverse ratio relationship between the gray mean and the amount of coal powder, and the linear ratio of ash gangue is about 4-5 times that of lump coal and black gangue.The single-factor experiment shows that the gray peak is sensitive to environmental change, while the response surface method shows that the gray mean of coal and gangue is distinguishable obviously at the same level.The results are helpful to promote the application of coal and gangue separation based
作者
李博
王学文
庞尚钟
高新宇
王璐瑶
丁恩发
暴庆保
LI Bo;WANG Xuewen;PANG Shangzhong;GAO Xinyu;WANG Luyao;DING Enfa;BAO Qingbao(College of Mechanical and Vehicle Engineering,Taiyuan University of Technology,Taiyuan 030024,China;Key Laboratory of Fully Mechanized Coal Mining Equipment,Taiyuan 030024,China;John Finlay Washing technology equipment Co.,Ltd.,Datong Coal Mine Group MEE Manufacturing Co.,Ltd,Datong 037300,China;State Key Laboratory of Mining Equipment and Intelligent Manufacturing,Taiyuan 030024,China)
出处
《煤炭科学技术》
CAS
CSCD
北大核心
2022年第8期236-246,共11页
Coal Science and Technology
基金
山西省重点研发计划资助项目(201903D121074)。
关键词
井下煤矸分选
煤矸识别
图像特征
灰度纹理
响应面法
underground coal gangue sorting
coal and gangue identification
image characteristics
grayscale texture
response surface method
