摘要
传统基于中分辨率成像光谱仪(MODIS)数据的山火监测方法火点误报率较高、判别低温火点的能力较弱、无法识别工业火点。为此,提出了一种基于概率统计的输电线路山火监测算法。该算法利用概率统计的方式提取潜在火点,对传统算法进行了标准差系数和背景阈值的修正。利用改进算法从已有的潜在火点中识别火点,将火点中的噪声点滤除,通过归一化植被指数(NDVI)的大小判断该火点是否为山火点。结果表明,该算法可以降低火点的误报率,具有较强的工业火点识别能力以及一定的低温火点识别能力。该算法成功辨识出了2013–10–25在安徽境内特高压线路锦苏线附近的山火,及时进行了预警,避免了山火跳闸事故的发生。
The traditional forest-fire detecting algorithm based on moderate-resolution imaging spectroradiometer（MODIS） data presents a relatively high rate of false alarm, and is also unable to detect low-temperature fire spot and industrial fire spot efficiently. Therefore, we put forward an improved algorithm based on probability statistics to detect potential forest-fire spot. Compared with the traditional algorithm, the improved algorithm uses probability statistics to extract potential fire spots, and improves the coefficient of standard deviation and threshold value of background. Using the improved algorithm can recognize the fire spot from all potential fire spots, get rid of noises, and identify forest-fire spots according to the normalized difference vegetation index（NDVI）. The results show that, the improved algorithm can reduce the rate of false alarm, and has stronger ability for recognizing low-temperature fire spot and industrial fire spot. Using the new algorithm, we have successfully identified the forest fire spot near UHV transmission Jinsu-iine in Anhui Province in 25th of October, 2013, and have given early warning in time to avoid tripping accidents due to forest-fire.
出处
《高电压技术》
EI
CAS
CSCD
北大核心
2015年第7期2302-2307,共6页
High Voltage Engineering
基金
国家自然科学基金(51137003)~~
关键词
中分辨率成像光谱仪
山火监测
概率统计
归一化植被指数
火点
亮温
moderate-resolution imaging spectroradiometer
forest fire detecting
probability statistic
normalized difference vegetation index
fire spot
brightness
