摘要
变化检测方法作为遥感信息获取的主要技术手段之一,在灾害监测中已得到了一定应用。然而基于遥感图像灰度信息的变化检测方法,其阈值选取受到传感器和环境因素引起的"伪变化"信息影响。由于纹理特征具有移不变特性,可以利用地物空间结构信息开展变化检测,从而避免图像灰度"伪变化"信息的影响。该文在现有变化检测技术流程之上,通过纹理特征参数的分析,结合基于贝叶斯的图像阈值分割技术,建立了基于纹理特征的高分辨率遥感影像变化检测方法,并应用在汶川地震前后的福卫-2影像,与基于光谱影像的变化检测方法进行了分析对比。
Change detection using remote sensing images is one of the important methods on obtaining land feature information,and is widely used to monitoring the natural disaster.In the change detection methods based on image gray information,the threshold selection is affected by pseudo change information from the effect of sensors and environmental factors.Texture feature has a property of shift invariant,which can use the spatial structure information in change detection and avoid the effect of pseudo change from image gray.An algorithm was presented that using the texture characteristic and image segmentation techniques.The algorithm was validated and analyzed using the sample data.The results show that the algorithm based on the texture characteristic can effectively extract the land features with a better precision.
出处
《灾害学》
CSCD
北大核心
2014年第3期97-101,共5页
Journal of Catastrophology
基金
国家科技支撑项目(2011BAH23B03)
国家自然科学基金资助项目(41301501)
关键词
纹理特征
遥感图像
灾害监测
变化检测
texture characteristic
remote sensing images
hazard monitoring
change detection
