Bus Passenger Counting Based On ARM+FPGA Hardware System Design And Implementation

At present,the population of urban population in China has expanded rapidly,and traffic congestion in large and medium-sized cities has become increasingly serious.Encouraging citizens to take public transportation can effectively alleviate traffic congestion.Through the existing bus electronic stop sign,we can only know that there are still a few stops in the next car,and we can't know the degree of congestion in the car,and it can't provide a basis for people to choose whether to ride this car.Therefore,research on automatic passenger counting systems is very important.The traditional method of human traffic statistics is infrared detection,pressure detection,etc.,but the bus getting on and off is complicated,and such methods are not well qualified for passenger detection and counting.The method based on video image processing is of great significance to improve the low accuracy of the traditional counting method.The rapid development of embedded systems and processors is to perform video acquisition by installing a camera on a bus and data processing with an embedded system.It is possible to complete the statistical work of the number of people on board.This paper designs and implements an embedded bus passenger counting hardware system based on ARM+FPGA for the human head detection counting algorithm developed by the laboratory.The main contents are as follows:Analyze the human head detection and counting algorithm developed by the former laboratory.According to the actual situation that the image preprocessing part of the algorithm has a large amount of computation and the processing time is long,the hardware architecture of this paper is determined as ARM+FPGA,and the FPGA module performs video.Image acquisition,and give full play to its fast computing ability,hardware parallel acceleration of image preprocessing part,ARM module for subsequent processing of FPGA accelerated processing data.For the existing ARM core board interface,an ARM backplane has been designed and developed independently.This backplane only has the necessary interface to run the human head detection and counting algorithm.The experimental results show that the ARM backplane developed in this paper,combined with the existing ARM core board,can meet the requirements of running the human head detection and counting algorithm and achieve the expected results.At the same time,the development board with complicated functions and high price on the market is tailored.Save on production costs.Data transmission between the FPGA module and the ARM module is completed using a Gigabit Ethernet port.In the FPGA module,an Ethernet frame header,an IP header,and a UDP header are added to the preprocessed image data to implement data encapsulation and transmission.In the ARM module,the Socket API and OpenCV are used to implement image data reception and restoration,and finally The entire system is connected to achieve the function of detecting and counting the human head.The experimental results show that the FPGA module can encapsulate and send the image preprocessed data according to the Ethernet data format,and the ARM module can successfully receive the data from the FPGA module and restore the data to a frame image,and the ARM can restore the image.The image is processed for subsequent processing,and the function of counting and counting the human head is realized.The average counting accuracy is 78.2%,which can roughly reflect the congestion degree of the bus,and has certain practical value.