Research On Visual Recognition Technology For Double Lane License Plate And Loading Capacity Of Box Truck

Since the 21st century,the Chinese logistics transportation industry has developed rapidly.In 2021,the country’s road freight volume reached 39.14 billion tons,and the road freight turnover has reached 6,908.77 billion tons kilometers.The huge scale of logistics puts forward higher requirements for our logistics transportation industry.For the freight center,using machine vision technology to recognize the double-row license plate of the van and measure the load can effectively and quickly deploy the van,improve the transportation efficiency and reduce the operating cost,which is of great value to the development of logistics and transportation industry.Based on machine vision technology,this paper studies the double-row license plate recognition of van and the rapid measurement of loading capacity.The main research contents are as follows:(1)Aiming at problems such as the difficulty of double-line license plate recognition of vans,this paper uses text detection and text recognition technology based on deep learning to propose a method of dividing double-line license plate detection into two single-line license plates for recognition.Firstly,a large number of real vehicle license plate data was collected and the double-line license plate data set was marked by branch lines using PPOCRLabel labeling software.In view of the problem of fewer characters in other provinces and cities,partial license plates were synthesized by license plate template to supplement.Secondly,the DBNet detection algorithm and PPOCRv3 recognition model were used to train and fit the license plate recognition model in Paddle OCR platform.Finally,the test set is tested,and the license plate detection accuracy reaches 80.91% and recognition accuracy reaches 83.6%,which can basically meet the needs of double-row license plate recognition of van.(2)This paper explores a method of measuring the loading capacity of van based on monocular vision.Based on the analysis of the existing single visual distance model,an area fitting model is proposed,and on this basis,a measuring method of van load is proposed.In order to verify the feasibility of this method,a real vehicle measurement experiment was designed to collect images of different goods at different distances,and the image distortion was corrected by camera calibration.Then the corrected image is used to carry out measurement experiment.The error of experimental measurement results of van loading is within 6%,which can meet the requirements of van loading measurement under certain scenes.Intelligent construction of freight center is the future development direction of logistics transportation industry.Based on machine vision technology,this paper studies the problems of van double-row license plate recognition and rapid loading measurement,and has obtained certain research results.The verification experiment results show that the van double-row license plate recognition and loading measurement have high recognition accuracy and measurement accuracy.It provides some valuable data and reference for intelligent logistics applications such as rapid identification of trucks and real-time display of the remaining load.The research content of this paper is helpful to reduce the management cost of the freight center,improve the operation efficiency,and promote the development of the logistics industry.