In this paper,the author analyzed the similarities and differences of material volume measurement methods of belt conveyor.As for the traditional surface subdivision volume measurement method,there are shortcomings such as high precision,high maintenance cost and difficult popularization.Binocular vision technology was used to obtain the image of materials,and Census transformation was combined to carry out stereo matching,so as to obtain the three-dimensional coordinate information of belt conveyor under no-load and full-load scenes.After visualizing the three-dimensional coordinate information,the author obtained the point cloud model after three-dimensional reconstruction.Based on the point cloud model,a method to improve the calculation accuracy of material volume based on pixel coordinates was proposed,and the calculation formula of material volume was optimized and deduced.The experimental results show that the improved calculation accuracy of material volume has reached more than 95%,which has certain practical value.