摘要： 将传感器技术用于大空间火灾探测时，存在距离短、误报率高、可靠性差等不足。为此，在TMS320DM642微处理器和TI的DSP/BIOS实时操作系统上，设计基于视频图像的远程火灾探测系统。建立RGB空间颜色模型，对连续数帧火灾图像做预处理，分析频闪特性并进行模糊聚类，提取疑似目标区域，以火焰相关性、面积变化率和圆形度3个特征作为火灾识别依据。实验结果表明，该系统提高了大空间图像型火灾探测的精度和速度，可满足远程火情监测需要。

关键词: 视频火灾探测, 颜色模型, 微处理器, 火焰特征, 模糊聚类, 嵌入式系统

Abstract: Remote fire detection system based on TMS320DM642 microprocessor and DSP/BIOS real-time operating system is implemented for resolving the probleme of detection shorter distance, higher misinformation and lower reliability of sensor technology in large space. Firstly a series of frames are pre-processed in RGB space. Then suspected target areas are extracted depending on the flickering feature and fuzzy clustering analysis. The features, such as fire correlation, rate of area growth and circularity are regarded recognization criterion of fire. DSP/BIOS and RF5 based-system is designed and tested to illuminate the scheme feasibility. Experimental result shows that the proposed system improves the accuracy and speed of image fire detection in a variety of large space, and it can meet the needs of remote fire detection.

Key words: video fire detection, color model, microprocessor, flame characteristic, fuzzy clustering, embedded system

中图分类号: 

• TP391