Research And Implementation Of Moving Object Detection Method Based On ZYNQ In Complex Scene

Moving object detection is a key component of intelligent video surveillance,and various complex interference factors in actual scenes pose challenges to moving object detection.Designing more robust moving object detection algorithms still has important research significance.With the development of science and technology,video surveillance is increasingly moving towards high resolution and frame rate.The method of transmitting the surge of video data to remote cloud servers for processing and providing feedback cannot meet the growing demand for video services.Sinking complex computing from the cloud to the edge,real-time processing and analysis of video data can reduce the network load,cloud data storage pressure,and computational load.At present,many moving object detection algorithms have been able to achieve good detection results,but often the algorithm complexity is high,making it difficult to meet realtime processing requirements on edge devices.Therefore,based on the characteristic of FPGA being able to achieve high-speed parallel computing,it has become a trend to design hardware circuits for moving object detection algorithms by combining FPGA at the edge.Based on the above background,this article proposes an improved algorithm for integrating texture features,EA-Vi Be,and designs the hardware circuit of this algorithm on the ZYNQ platform.In addition,this article also built a moving object detection system based on ZYNQ(Zynq7020),and used this system to verify the hardware circuit design of the EA-Vi Be algorithm at the board level.The main research contents of this paper are as follows:(1)Research on moving object detection algorithms.After studying various motion object detection algorithms and texture features,this article proposes a new texture feature called Enhanced Adjacent Scale Invariant Local Binary Similarity Pattern(EASILBSP).Then,combined with the Vi Be algorithm framework,an improved algorithm for integrating texture features,EA-Vi Be,is proposed.The algorithm was simulated using C++and Open CV,and the experimental results showed that the algorithm in this paper has a certain inhibitory effect on various dynamic interference factors,and the overall detection effect has been improved to a certain extent.(2)The hardware circuit design of the EA-Vi Be algorithm on ZYNQ.This article optimizes the EA-Vi Be algorithm and designs circuits based on FPGA through the method of software and hardware collaborative design.According to the algorithm process,the entire design scheme is divided into image preprocessing section,EA Vi Be algorithm section,and image post-processing section.Finally,the hardware circuit design method for each part was described in detail.(3)Implementation of a moving object detection system based on ZYNQ.This article uses ZYNQ as the hardware platform for moving object detection,and uses this system to perform board level validation on the algorithm proposed in this article.This article not only completes the construction of the hardware circuit for PL(Programmable Logic),but also completes the design of the software driver for PS(Processing System).After testing,the system has stable plot,can detect moving objects in real time,and has good detection effect in complex background.It is small in size and low in power consumption,which meets the requirements of Edge device,and has certain practical significance.