Studies And Designs Of Fire Detection System Based On Image-type

Fires has brought great harm to people’s production and life,when it happens inspacious buildings,it is difficult for the traditional smoke、temperature sensors to detectfires.Eventually,the fire continually spread and leads to more damage.Video-imagebased fire detection technology has a lot of advantages,such as it could cover a widerange,has a high sensitivity and low cost.Image-type based fire detection technologycan slove this problem effectively.Generally,image-type fire detection system canroughly divided into two types,based on PC and based on DSP.Because of the powerconsumption、limited used and not conducive to the development and production,it isnot a wise choice to use PC in fire detection.In order to slove this problem,animage-based fire detection system based on the TMS320DM642platform is studied inthe paper which uses TI’s C6000DSP as the core processor has a real-time、highstability、low power consumption and easy to upgrade.In this paper,according to the actual requirement.After the study of video capturemodule、external memory expansion module、GSM fire alarming module and Ethernetcommunication module;A hardware module is designed which is based on DM642thatas a core processor. Analysis of TI’s DSP/BIOS real-time operating system and RF5program development reference framework,based on this,developing the video capture,double thread of detection algorithm processing and alarm,network transmissiontask;Analysis of the optical flow method、the color model method and backgroundsubtraction those three flame segmentation methods,through the experiment verifiedthe shortage of the algorithm and figured out the reasons,proposing the algorithm ofsingle-Gaussian-dynamic characteristics.Excluding the reflected light、red objects、alcohol light,improve the accuracy of the algorithm;the experiment of Flash burn wasdid,implement the system to work offline;Users can check the result of videoprocessing by B/S module; The development of GSM SMS alarm through EMIFinterface has been done,which could improve the system’s response speed androbustness;The experiment results shows that the system can detect the fire accurately and has a good real-time performance.