【摘要】 传统的火灾检测方法一般采用感烟、感温、感光探测器以及红外对射探测。本文提出了一种基于图像视觉特征的火灾检测方法,根据火灾火焰处于近红外波段的特征,采集近红外图像,并利用火灾初期火焰变化的各种特征信息,用图像处理方法提取这些特征值,并把其作为输入,利用遗传神经网络对其进一步识别,从而进行火灾判别,并进一步设计了基于以上思想的火灾识别系统。实验结果表明,该系统比传统系统更进一步减少了误报率且具有响应快、监控范围广等优点。更多还原

【Abstract】 Traditional methods for fire detection are smoke detection, temperature detection, light detection & infrared detection. A fire detection method based on the visual characters of images is proposed. According to the characters of fire images near to near infrared, near infrared images are collected. And making use of the changing information of the fire at the earlier period, the characteristic value is picked up by the image process as the input signal and identified by the genetic neural network to distinguish the fire. This is the idea of the fire identification system. The result is indicated that the system can decrease the error rate further, respond more quickly and monitor more widely than traditional system.更多还原