摘要
由于煤矿输送带视觉场景噪声高、地下环境复杂,输送带图像的异物识别效率较高。针对煤矿带式输送机存在异物造成输送机损坏的问题,提出了基于SSD技术的带式输送机表面异物视频检测方法。采用深度可分卷积方法减少SSD算法中的参数数量和提高速度;采用GIOU函数计算式代替原始SSD中的位置损失函数,以提高检测精度;对特征映射的提取位置和识别框的比例进行优化,以提高检测精度。实验结果表明,改进算法优于原SSD算法,平均准确率从87.1%提高到90.2%,检测帧率从32帧提高到41帧。
Due to the high noise of the visual scene of the coal mine conveyor belt and the complex underground environment,the foreign body recognition efficiency of the conveyor belt image is relatively high.Aiming at the problem of coal mine belt conveyor damage caused by foreign matter,a video detection method of foreign matter on the surface of belt conveyor based on SSD technology was proposed.The depth separable convolution method was used to reduce the number of parameters in the SSD algorithm and increase the speed;the GIOU function calculation formula was used to replace the position loss function in the original SSD to improve the detection accuracy;the extraction position of the feature map and the ratio of the recognition frame were optimized,In order to improve the detection accuracy.Experimental results showed that the improved algorithm was better than the original SSD algorithm,the average accuracy was increased from 87.1%to 90.2%,and the detection frame rate was increased from 32 frames to 41 frames.
作者
蒲志强
Pu Zhiqiang(Shaanxi Changwu Tingnan Coal Industry Co.,Ltd.,Changwu 713600,China)
出处
《能源与环保》
2021年第9期29-35,共7页
CHINA ENERGY AND ENVIRONMENTAL PROTECTION
基金
国家自然科学基金重点资助项目(41130419)。
关键词
带式输送机
异物
胶带表面
视频检测
智能系统
belt conveyor
foreign matter
belt surface
video detection
intelligent system
