Research On Key Technologies Of Intelligent Grab Ship Unloader Based On Binocular Vision

With the continuous growth of world trade,bulk terminals have developed rapidly.In the face of increasingly fierce competition,the intelligentization and the informationization have become the new trend in the development of the bulk terminal.The grab ship unloader is the core machinery of the bulk terminal.Its handling efficiency determines the competitiveness of the entire terminal.Therefore,in this thesis,the grab ship unloader is taken as the research object.The shortcomings of the traditional loading and unloading mode are first analyzed.Aiming to achieve intelligent ship unloading,the key technologies such as the three-dimensional reconstruction of bulk cargoes based on binocular stereo vision and the control strategies of ship unloading are then investigated.Firstly,the principle of binocular vision system applicable to the bridge grab ship unloader is researched and analyzed,including the characteristics of system matrix and processing object,the mechanism of the binocular stereo vision imaging,the layout,processing flow and parameter configuration of binocular vision system.According to the experimental requirements and material characteristics,appropriate hardware and software are selected to build a binocular vision experimental platform.A laser grid is projected on the surface of materials,the purpose of which is to artificially increase feature points,and images are taken by a binocular camera.As for the camera calibration,based on the step-by-step calibration idea,the Zhengyou Zhang’s plane calibration method is adopted to determine the camera main point and its pose information,and the camera focal length calibration method based on double vanishing points is used to calibrate the camera focal length in real time.By taking multiple checkerboard images and actual pile images,and using MATLAB to obtain internal and external parameters of the camera.Secondly,according to the requirements of the image matching,the material image is operated by median filter and histogram equalization in advance.In addition,the left and right views are stereo corrected to be strictly aligned based on the Bouguet algorithm.Through the analysis of the matching algorithm and the characteristics of the experimental object,this matching first uses the rectangle detection to obtain the quadrilateral feature,and performs stereo matching based on the feature.The original algorithm is optimized according to the experimental requirements.And the optimization focuses on two aspects: one is to adjust the threshold value of the quadrilateral detection parameter and set it to 0.5;the other is to reduce the number of matching key points,reduce the matching time and enhance the reconstruction effect.A matching experiment is performed on the material image,the results of which verify the effectiveness of the optimization algorithm.Then,combining the results of the camera calibration and stereo matching,a three-dimensional point cloud image corresponding to the image feature points is obtained.The Biharmonic spline interpolation is selected to fit the discrete point cloud to obtain a continuous and smooth reconstruction surface,thereby realizing the3 D visualization of bulk cargoes.Through the binocular vision experimental platform,three-dimensional reconstruction experiments are performed on four different shape bulk piles.The experimental results show that the reconstruction errors of the four cases are within the error tolerance range and the accuracy is high,that verify the accuracy and feasibility of the method.Finally,on the premise of realizing the visualization of bulk piles,the control strategy of unloading and reclaiming is designed.A slicing principle of bulk cargoes is proposed,and the rules of reclaiming strategy are studied and analyzed.And on this basis,the current layer unloading operation efficiency evaluation model is established,comprehensively consider the three evaluation indexes of material height,material position and cart position,the index weight is determined through the Analytic Hierarchy Process,and the index evaluation value is combined to obtain the unloading operation efficiency comprehensive evaluation results.After that,the simulation experiments of three different shape bulk piles are performed by using MATLAB.The simulation results are consistent with the actual situations,which verify the reliability of the strategy in this thesis.