Research And Implementation Of RaspberryPi Moving Object Detection Algorithm

With the development of digital circuits and microelectronic technology,moving target detection technology in many areas of industrial inspection,consumer electronics,intelligent traffic control and security,etc.it has been widely used.At present,the research on the algorithm of the moving target detection system has been mature,so that the room for improvement of algorithm is limited.But most are implemented by C language in the PC,which fails to meet the requirements for instantaneity and mobility.In this thesis,I design a moving target detection system based on RaspberryPi is designed,which satisfies the mobility detection requirements of moving targets.The main work of this thesis is as follows:(1)Development status and background significance of moving target detection technology and the principle of image pre-processing and image morphology are summarized.Three kinds of motion detection algorithm are introduced,that is frame difference method,background difference method,optical flow method,which focuse on the Horn-Schunck algorithm of optical flow method,Lucas-Kanade algorithm and simplified optical flow calculation method,comparatively analyze the complexity of Horn-Schunck algorithm and Lucas-Kanade algorithm in computation.The simplified optical flow method will reduce the amount of calculation greatly,and can be applied on the hardware platform.What's more,the effect and H-S optical flow algorithm is basically the same.In this thesis,the designed system is based on the simplified optical flow algorithm.(2)The simulation model of the optical flow algorithm is built in Simulink.The threshold judgment model,the median filter model,the Blob analysis model,the red box marker model and the detected model are built.Quantity Displays the model and displays the number of detected moving objects.The simulation results show that the simplified optical flow method can detect the moving target correctly,and the simulation result is better,which verifies the correctness of the system design.(3)The hardware platform of the moving target detection system is set up,and the moving target detection system based on RaspberryPi is designed and realized.The system through the USB network camera(UVC mode)image acquisition kit to obtain real-time video image sequence,image resolution is 640*480,using Raspbian operating system and socket library functions for image data transmission,the camera captured video source through the USB interface into the raspberrypi development board,using the optical flow method for motion detection the goal of video image sequence in the raspberrypi to complete the moving target detection module.Parameter information of moving target,the processed data through the Ethernet interface real-time transmission to the PC terminal,the real-time video processing results will be displayed on the PC.(4)The system of 640*480 image detection accuracy of more than 97%.The experimental results show that the moving object detection system designed in this thesis achieves the desired target detection function,and has good real-time performance and accuracy.