Font Size: a A A

Forest Fire Vision Detection Based On FPGA And Co-design Of Software And Hardware

Posted on:2021-04-16Degree:MasterType:Thesis
Country:ChinaCandidate:Y GaoFull Text:PDF
GTID:2393330611469216Subject:Mechanical Manufacturing and Automation
Abstract/Summary:PDF Full Text Request
Forest fire is a kind of natural disaster with suddenness,strong destructiveness.Early warning of forest fire is an important research topics in the field of forest fire prevention.Smoke is the most significant visual signal in the early days of forest fires.Therefore,forest fire smoke detection based on video has important research significance.This paper introduces the front-end detection strategy into the near-field forest fire smoke video surveillance scheme.First,an FPGA based high-performance computing platform,which is designed independently is adopted as the hardware.Then,the forest fire smoke detection algorithm and the dedicated hardware acceleration methods proposed independently are used as the software.Based on the above two parts,a collaborative design solution from software algorithm to hardware device is realized.The proposal of this scheme can effectively reduce the large-scale network video transmission and power supply requirements.The research contents of this paper include:(1)The concept of “smoke root” is proposed,and a forest fire smoke detection method based on smokeroot and physical diffusion model is designed.The experiments have proved that the method has goodimmune performance against interferences which are similar to smoke and color(such as clouds,fog).(2)A smoke root detection method in full scale is proposed.The method takes into account the extractionof smoke roots both in long and close range scenes.Besides,the method also introduces a discretedatabase to replace the calculation process of the likelihood function,and update it in real time.Theindex structure based on continuous memory makes the algorithm friendly to implement in hardware.(3)An ARM-FPGA heterogeneous computing platform with the ZYNQ7Z020 chip and the binocularcamera PCB board with the OV7850 chip are designed independently.The successful implementationof joint debugging verified the design integrity of the platform.(4)The hardware acceleration design scheme of MSER based on FPGA is proposed,a part of dedicatedhardware acceleration circuit module is designed for the algorithms.Some detailed design scheme,parts of open source code,and timing simulation results are also given in the paper.All of the above work laid the foundation for the further deployment of software algorithms on heterogeneous platforms.
Keywords/Search Tags:Smoke root, Full scale detection, FPGA, Software and hardware collaboration
PDF Full Text Request
Related items