【摘要】 煤流实时监测是实现煤炭开采数字化过程的重要技术，传统立体匹配方法 RAFT计算效率低下，为提升煤流监测方法的效率，提出了一种基于双目实时匹配的输送带煤流监测方法，首先采用基于Mask约束的煤流图像快速立体匹配算法计算煤流视差、点云，然后利用PUGeo-Net网络对煤流点云进行加密，最后通过网格分割法计算煤流体积和质量。实验结果表明，该方法计算煤流质量平均耗时为0.6 s/帧，煤流质量误差3.5%以内，相较传统方法效率提升4.16倍，验证了其针对输送带实时煤流监测效率、精度的有效性。更多还原

【Abstract】 Real time monitoring of coal flow is an important technology to realize the digital process of coal mining. The traditional three-dimensional matching method RAFT calculation efficiency is inefficient, so as to improve the efficiency of coal flow monitoring method, proposes a coal flow detection method for conveyor belt based on binocular real-time matching. First, a fast stereo matching algorithm for coal flow images based on Mask constraints is used to calculate the parallax and point cloud of coal piles, then, the PUGeo-Net network is used to encrypt the coal flow point cloud, and finally, the volume and quality of the coal flow are calculated using grid segmentation method. The experimental results show that the average time of this method to calculate coal flow quality is 0.6 s/frame,and the coal flow quality error is less than 3.5%, which is 4.16 times higher than that of the traditional method, which verifies the effectiveness of real-time coal flow monitoring efficiency and accuracy for conveyor belts.更多还原