汶川地震区航空影像地质灾害自动识别技术

An Auto-identify Method of Geological Hazards in Wenchuan Earthquake Area by Airphoto

汶川地震引发的大量地质灾害在高精度航片上表现出明显的纹理特征。对航片进行正射校正后,采用数学形态学方法,选取包括全部方向的结构元素,设计单波段全方位结构元素边缘增强和多波段加权彩色合成相结合的边缘检测算法,自动提取汶川地震引发的灾害体信息,既考虑了灾害体具有亮度突变信息、色调突变信息和边界发展具有多方向性的特点,又考虑了各种色调地物在各个波段上具有不同亮度值的特点,对粘连的地物信息进行分离,根据各个波段与实际灾害体信息的相关性做加权合成,充分挖掘了高精度遥感的色调信息和纹理信息。

Huge amount of geological hazards were triggered by Wenchuan earthquake, the obvious hazards texture can be identified on airphoto. The airphoto used in this paper was ortho--corrected in the method of mathematical morphology and processed based on multi--directional structuring elements. The method of mono--band multi--directional structuring elements edge en- hancement algorithm combined with multi--band weighted color synthesizing algorithm was designed to auto--extract hazard information from high spatial resolution airphoto. This method not only embodies the luminance mutation, tone mutation and the multi--direction of the edge of the hazards, but also reflects the band character of remote sensing, it separates the messed infor- mation. The weighted synthesizing is based on the relation between each band and the hazard, it takes advantage of the remote sensing information thoroughly. The result showes that it is a kind of algorithm that costs less computation, and can identify the hazards.