Kmeans聚类与多波谱阈值相结合的烟检测算法研究

Study On Smoke Detection Using Kmeans Clustering and Multithreshold Approaches

森林火灾破坏人类赖以生存的宝贵而有限的自然资源,造成环境污染,引发生态失衡,并且危害林区周边的城镇安全。卫星遥感技术可以有效地监测森林火灾。传统的烟检测算法往往利用遥感卫星某些通道的反射率、亮温或者波段数据的组合,设置绝对阈值来判定。阈值方法具有一定的主观性,对先验知识要求较高,并且在不同季节和不同地区阈值适用性不同,容易产生误判或者漏判。采用聚类分析和多波谱阈值相结合的方法,提高了检测算法的适用性和准确性。

Forest fires destroy the natural resources and the environment which human survive in.Satellite remote sensing is an effective approach for detecting the fire.smoke detection is an important step for monitoring forest fires.Traditional methods compare brightness temperature or reflectivity of different bands based on the analysis of spectral characteristics of different cover types, such as vegetation, water and so on.In this process, specific thresholds are utilized.However, threshold method is subjective and cannot adapt to all seasons and regions.In this study, new remote sensing methods for detecting smoke were developed.Kmeans clustering and multithreshold techniques were explored for application to improve the applicability and accuracy of detection algorithm with MODIS imagery.