摘要
现有基于火灾图像识别的报警系统一般采用前端摄像头与视频采集卡的结构,然后由传输网络把图像传输到后台计算机进行处理识别、报警等工作。文中设计了一种DSP+FPGA的硬件系统结构,利用火灾图像独有的特征,用图像处理方法提取这些特征值,并利用神经网络对火灾进行判别的软件系统。这些处理都在终端进行,减少了监控台的负担。实验结果表明,该系统与传统系统相比,更进一步减少了误报率且具有价格低、响应快、监控范围广、不用更换网络等优点。
The architecture of the existing fire images are recognition system is generally composed of stylus and video acquisition card, and then the images are transferred to backstage computer to be deal and so on. So design a kind of new architecture based on DSP ＋ FPGA, and extract the characteristic of fire images using image processing, and then the characteristics are used to recognize the fire using neural network, These processings are done on the terminal, so the burden of backstage computer is decreased. The result is indicated that the system can decrease the error rate further, respond more quickly, monitor more widely and does not need change the existing network than traditional system.
出处
《计算机技术与发展》
2006年第6期95-97,100,共4页
Computer Technology and Development
