Research On Image Recognition Algorithm And GPU Acceleration Of Large Vehicle Rapid Location And Orientation System

In recent years,with the development of industrial technology in China and the proposal made by the state 2025 China,intelligent manufacturing has become the trend of industry development.At present,a large number of sensors have been installed in the field of home cars,industrial large transporters and automation robots to help achieve more automation,which is heading into an intelligent and unmanned direction.Among them,the navigation and location technology represented by machine vision has been developing for a long time.In this paper,the problem of rapid positioning and orientation of large vehicles with machine vision is studied.The image matching algorithm,target discrimination method,digital recognition method and data processing method are studied to solve the problem.Finally,the GPU acceleration algorithm for large vehicle rapid location and orientation system is introduced.The main contents of the full text are as follows:(1)Introduced the problems encountered in the process of transportation and handling of large vehicles.Discussed the background of the research on the rapid orientation and orientation system of large vehicles.Sums up the research status at home and abroad.Based on the existing technology level and problem needs,Put forward the research train of thought in this paper.(2)The process and main components of image matching are summarized,and the advantages and disadvantages of the gray image matching algorithm,such as the mean absolute difference algorithm,the absolute error and the algorithm,the average error square sum algorithm and so on,are analyzed.The matlab program is used to write the image matching with the absolute error and the algorithm as an example.The image matching algorithm,such as invariant moment matching algorithm,distance matching algorithm and minimum mean square error matching algorithm,is analyzed.The moment invariant algorithm is used as an example,and the matlab calculation moment is written and the image matching degree is normalized.Several typical ways of image matching search are introduced,including Pyramid search,genetic algorithm search,and other fast search methods.(3)Image matching algorithm based on neighborhood gray coding is introduced,and the algorithm is verified by an example.Compared with the gray image matching algorithm in the second chapter,the algorithm has the advantages of good stability and high computing efficiency,and two methods of target recognition based on template character recognition and pattern recognition based on neural network are analyzed.(4)The related technologies of image processing are introduced,including camera calibration technology,image pre-processing technology and target digital recognition technology.(5)By identifying the identification line in the image,the algorithm of calculating the relative angle of the vehicle and finding the number of the identification location through the recognition line and identifying the number,thus the algorithm of the specific location of the vehicle is obtained.And an example is given to illustrate the algorithm.(6)introduced the immune clonal algorithm for the rapid positioning and directional system of large vehicles.In view of the problem of relatively low efficiency in the algorithm,a PICFS acceleration algorithm based on GPU is proposed,and the different sample size is used to compare with the serial algorithm,and the example is verified.