Design And Realization Of The Fire Early Warning System Based On ARM And DSP

The traditional sensor-based fire detection techniques is difficult to monitor the fire of large space,bad environment for industrial sites and outdoor. With the . development of digital image processing technology,network technology and embedded technology, make a video-based intelligent fire detection and early warning methods be possible. Fire monitoring with the image sensor technology replacing the traditional detection techniques can meet the early fire warning requirements of modern large space buildings.This paper designed an embedded fire early warning systems based on ARM and DSP, ARM is the master controller, DSP is equivalent to be co-processor. In the ARM and Linux system, capturtion of video images, wireless transmission of images, alarm function were completed.In the DSP system, image recognition algorithm for the fire was completed. This fire image recognition algorithm based on flames model and smoke model.Two models operating simultaneously, as long as one model suspected the fire the system will alarm. In this way, the fire forecast can be more rapid.Firstly, the paper analyzed the background and significance of the research and then introduced the research and development of fire warning system which based on digital image processing. Then introduced the design principles,overall structure and block diagram of the embedded Fire Early Warning System based on ARM and DSP, described the system hardware design of each module. According to the system's hardware resources, builded an embedded Linux software platform and established a cross-compiler development environment. This paper focuses on the fire recognition algorithms.Improved fire recognition algorithms which based on wavelet analysis, gaussian mixture models, RGB and HIS model can be good to eliminate interference and determine fire accurately. Then described the system's overall software process, highlighted the the software implementation of image capture by V4L and wireless image transmission by GPRS.System testing aimed at monitoring system's performance and stability, and simulation of the fire recognition algorithm. Test results show that the system's various hardware modules are able to work security and stability, the operating system can run stability, the pictures captured by V4L are clear, the speed of the image transmission by GPRS meet the requirements. The algorithm was simulated, the results show that the algorithm can be good to identify fire. Finally, analysised system's problems and the solutions, summarized and prospected the trends of this topic, pointed out the video_fires recognition system's future direction.