Implementation Of Embedded Image Fire Detection System

In order to solve problems that the traditional temperature sensing or smoke sensing fire detector is difficult to adapt to large space and easily affected by environmental. Image fire alarm has become a research hotspot. This paper introduces a kind of image fire alarm system based on embedded. It uses dynamic characteristics of early fire image to detect the fire. The system with high integration level and is fit for outdoor and multi-point network.This embedded system is mainly composed of FPGA and ARM, the FPGA mainly complete image sensor driver, and cache the collection of image data to the SRAM. To achieve image acquisition and processing at the same time, this paper uses two pieces of SRAM for rotating storage. ARM as the main part of the whole system, coordinating with FPGA, obtains image data from the memory SRAM, completes the image preprocessing and fire detection algorithm, sends the image of the data and the testing results to the computer terminals via Ethernet, allows people to get test results and also can realize monitoring on site at the same time.The targets of detection are smoke and flame. The paper selects dynamic characteristics of them as a basis for the early fire detection, analyzes differences in whole mobile rule and the position of the origination point and similarity between frames with other distracters, and describes each dynamic characteristic with geometric method. The process of image processing and recognition is as follows. First of all, ARM preprocesses the image, filters out the isolated noise with median filter and zoomed out to reduce calculated amount. Then the system uses background difference method and binary AND between adjacent frames to extract and segment the moving object. The system extracts dynamic characteristics of the early fire to detect the fire and analyses those characteristics above some successive frames over a sliding time window. The results of extracting are weighted judgment. If the weighted result reaches the threshold, the system gives the message of fire.Test results show that the system can work steady. The image collected by sensor is clear and intact, so it meets the image quality requirement. The system can detect the frame and smoke to implement alarm in the early fire. The system has a certain practical value with small volume, low consumption and low cost advantages.