Font Size: a A A

The Improved CT Agorithm And Its Implementation On The Zedboard Hardware Platform

Posted on:2019-04-18Degree:MasterType:Thesis
Country:ChinaCandidate:W F ChenFull Text:PDF
GTID:2428330548973457Subject:Computer technology
Abstract/Summary:PDF Full Text Request
Motion target tracking is a very important research direction in the field of computer vision and is widely used in all aspects of social life.Compressed sensing tracking algorithm(CT)is a kind of movement target tracking field very classic algorithms,this algorithm applied compression perception theory,improved the way of feature extraction,raised the speed of the algorithm.With the development of embedded hardware development platforms such as Arm,FPGA and Zedboard,embedded software development has been deeply studied in the field of algorithms.Zedboard is a programmable SoC(on-chip system)chip of Xilinx Zynq series,consisting of dual-core ARM cortex-a9 processor and FPGA.It is the preferred hardware platform for embedded development in the current target tracking field.The work of this paper is mainly divided into three parts:In the first part,an in-depth analysis of existing CT algorithms reveals two deficiencies in the design of CT algorithms: the selection of candidate target samples still adopts the strategy of pixel by pixel;The measurement matrix cannot be adjusted according to the target.This paper introduces the construction of embedded Ubuntu operating system,serial port monitor and cross compiler on Zedboard platform,and expounds the basic principle and operation of embedded development on Zedboard platform.In the second part,in view of the shortcomings of CT algorithm,this paper proposes to improve CT algorithm by using adaptive measurement matrix and particle swarm optimization(PSO).Adaptive measurement matrix improves the algorithm timeliness by reducing the dimension of feature vector.Particle swarm optimization improves the algorithm timeliness by improving candidate target strategy.In this paper,APCT algorithm,an improved CT algorithm with higher timeliness,is obtained by combining two algorithms.In the third part,combined with the research field of movement target tracking,emphasis on Zedboard platform needed to run the algorithm library compilation of transplantation and algorithm design based on hardware platform are introduced,and the realization of the algorithm in Zedboard platform for the display.Through the improvement of CT algorithm,APCT algorithm with higher timeliness is obtained.Using Zedboard development board,the embedded development of APCT algorithm was realized.The algorithm research was extended to the hardware field,which laid a foundation for the further research on the combination of software and hardware.
Keywords/Search Tags:Compressed Sensing, Adaptive Measurement Matrix, Particle Swarm Optimization, Cross Compiling
PDF Full Text Request
Related items