Abstract: In terms of the difficulties in identifying the coal image and calculating coal volume in non-uniform distributed coal particles, this paper proposes a binocular vision detection method for coal quantity in belt conveyor.The method consists of three modules, namely, identification module, three-dimensional information extraction module and quantity calculation module.First, the identification module uses wavelet transform algorithm and K-means clustering algorithm to enhance and segment the coal image, respectively; Then, the three-dimensional information extraction module is employed to obtain the depth information of each point of coal image using binocular vision method, so as to get the three-dimensional point cloud information of coal; Finally, the quantity calculation module combines Delaunay and T-S fuzzy reasoning algorithms to calculate the coal volume, and further adopts coal calculation formula to achieve the coal quantity detection.Experimental studies curried out using the actual images show the effectiveness of the proposed method.