Research On The Residual Coal Volume Measurement System Based On Binocular Vision For Coal Train Bottom

Looking at the current state of China’s resource and energy strategic planning,coal will continue to dominate China’s major energy consumption for quite some time to come,and ensuring the security of coal supply demand is vital to our economic,social and energy development.Therefore,in addition to regulating coal production,We need to focus on the delivery and safety of coal.The train carries coal supplies is the most economical and efficient way to transport coal.After the coal train has unloaded the coal,different amounts of coal piles will remain in the carriage,for the amount of coal left in the carriage most rely on the human eye to observe by experience-based feedback on whether the remaining coal in the carriage needs to be removed is inefficient.Laser measurement method can only be single point measurement,measurement point discontinuous.The use of binocular vision to measure the volume of residual briquettes in carriages is a good solution to the above problems,but the poor matching of areas with weak texture and the randomness of matching points do not meet the requirements of future industrial automation.For regional coal block images with weak texture,where feature point extraction and stereo matching are difficult,this study combines and improves the traditional ORB algorithm with o FAST feature point detection and r BRIEF feature descriptors,and gray prime center method was used to calculate the main direction of each feature point,thus improving the accuracy of the traditional ORB matching algorithm in the region with weak texture.For feature matching can only match the full range of image feature description points and cannot freely select the matching points in the matching region,the image segmentation and matching point automatic generation algorithm is proposed,using the mouse in the limit of the detection area of the captured image,limiting only the extraction of feature points inside the segmentation area to describe,calculates the average distribution coordinates of feature point positions by the coordinates of the effective rectangular area and the input measurement points.The number of feature points to be described is reduced,shortening the matching time and improving the matching efficiency.The current binocular vision volume measurement methods mainly rely on three-dimensional reconstruction to extract the overall outline of the object and the information in three-dimensional space coordinates,such methods need to consume a lot of time and have high requirements on the level of experimental equipment processing system.In this thesis,we use stereo matching of limited areas and left-right consistency detection to finalize the matching points and calculate the depth information of the coal pile based on the parallax of the left and right images.In this thesis,we use stereo matching of limited areas and left-right consistency detection to finalize the matching points and calculate the depth information of the coal pile based on the parallax of the left and right images,as well as combining the pixel size of the difference between the lower right and upper left points of the captured image in the carriage with the actual size of the carriage width to calculate the residual briquettes volume.Finally,a test platform is established in the laboratory by binocular camera,and the coal pile model is tested by volume calculation.The test results show that the proportional error of the volume measurement of the model with regular shape is about 5%,and that of the model with irregular shape is about 8%.To sum up,the binocular vision based method for measuring the volume of residual irregular coal piles in coal train carriages proposed in this thesis has certain reference value for actual production practice and future intelligent coal train residual volume measurement.