摘要
为降低厚度效应的影响,提高识别的环境适应能力以及准确度,提出一种基于R值与改进多阈值Otsu分割的煤矸识别方法。首先利用二值图像结合Canny算子边缘检测的方法完成煤矸目标与输送带背景的分割;然后采用边缘检测优化的多阈值Otsu算法自适应地计算分割R值,根据实际生产需求设定分选阈值含煤量的大小;最后以不同地区的焦煤、肥煤、气煤以及矸石作为研究对象,开展不同品质、不同比例下的煤矸识别综合试验。试验结果表明:分割R值受不同煤质与样本整体混矸率的共同影响,煤矸整体识别准确率较高,稳定性较好。
In order to reduce the influence of thickness effect,improve the environmental adaptability and accuracy of recognition of coal and gangue,a method based on R-value and improved multi-threshold segmentation algorithm is proposed.Firstly,the segmentation of coal and gangue target area and conveyor belt background was completed by using binary image combined with Canny operator edge detection method.Then,the multi-threshold Otsu algorithm optimized by edge detection is used to adaptively calculate the segmentation R-value.Then set the value of coal content for the separation threshold according to the actual production requirements.Finally,taking coking coal,fat coal,gas coal and gangue from different areas as the research object,then the comprehensive test of coal and gangue identification under different quality and proportion was carried out.The experimental results show that the segmentation R-value is jointly affected by different coal qualities and the overall gangue mixing rate of the sample.The coal and gangue achieved a higher recognition accuracy and better stability.
作者
于中山
YU Zhongshan(China Coal Technology and Engineering Group Shanghai Co.,Ltd.,Shanghai 200030,China)
出处
《煤炭技术》
CAS
北大核心
2022年第4期130-134,共5页
Coal Technology
基金
安徽省科技重大专项资助项目(18030901049)。
关键词
煤矸识别
双能X射线探测
物质属性R值
Otsu多阈值分割
边缘检测
图像处理
recognition of coal and gangue
dual-energy X-ray detection
material attribute R-value
multi-threshold Otsu segmentation algorithm
edge detecting
image processing
