领域统计火焰检测算法及其实验研究

针对目前火焰检测器中采用的火焰燃烧判别算法存在的不足，提出了一种新的邻域统计火焰检测算法。首先根据选定的火焰图像区得到火焰图像的特征灰度级，然后通过统计特征灰度级邻域范围内的像素数对火焰状况进行判断，得出了正确的检测结果。该方法可以在线修改参数，为正常燃烧状态、不稳定燃烧状态、低负荷状态和灭火状态设定不同的判别阈值，保证了在不同工况下准确地检测出火焰燃烧状况，提高了检测的可靠性。

Study on algorithm of neighborhood statistic flame detection in FSSS

In order to improve the algorithm of flame detection in furnace safeguard supervisory system, a new algorithm of neighborhood statistic was proposed in the paper. Characteristic threshold of flame gray was obtained according to chosen image. The flame state was judged by the pixeis number of flame in the threshold pixei neighborhood area. The advantages of the method are that parameters can be modified online and the flame states can be obtained by the different thresholds. The method satisfied the requirements of furnace safeguard supervisory system.