中文摘要：

彭州地处成都平原与龙门山的过渡地带,土地利用状况具有强度高、类型多、复杂度大的特点,土地类型变化非常快。同时,＂5·12＂汶川大地震严重破坏了该区域的生态环境,这场大灾难导致了地震灾区土地利用在类型和结构上都发生了巨大变化。因此,如何快速的获取最新的土地利用信息对于灾后重建及政府决策具有重要意义。利用无人机高空间分辨率影像作为数据源,首先引入SIFT算法完成了无人机影像的自动拼接,然后采用面向对象的分析技术获取了两期无人机影像的土地利用类型,着重研究了影像最优分割尺度的获取,最后建立了土地利用动态监测系统获取了两期土地利用信息的变化量,提出了一种适用于四川盆地多云雾山地区域土地信息动态监测方法。通过对实验区两期影像的分析研究,快速获取了各类型地类的变化量,并对变化原因进行了分析。研究结果表明：利用无人机影像进行土地利用动态监测能克服多云雾山地区域卫星影像难以实时获取问题,能快速、高效的获得土地变更信息。

英文摘要：

Pengzhou located in the Chengdu Plain and Longmen mountain transition zone.The land use of Pengzhou has the characters of high intensity,much of styles and complex.The land use changes very quickly.Meanwhile,the 5.12 Wenchuan earthquake led to a serious damage to the ecological environment in the quake-stricken areas,this exceptional natural disaster result in the type and structure of land use have changed in great degree in the quake-stricken areas.The land use change information can provide basis for development and use of regional resources and making sustainable development decision.On account of the weather of multi-cloud in Sichuan basin,it is not conducive to acquire satellite images and manned aircraft images in real-time,and the low-altitude UAV aerial photography can solve the data acquisition difficulty in multi-cloud mountainous areas.The UAV images automatic mosaic were completed based on SIFT algorithm,and the object oriented-analysis method was used for obtained two phase of the land use type of UAV images,The key problem of segmentation is focused on identifying whether the resulting object has good internal homogeneity,whether the neighbor object has a good heterogeneity,so focus on the how to obtain optimal segmentation scale,and a rapid inspection of land information system was established for the two phase information change of land use,and a landuse information dynamic monitoring method was proposed for the multi-cloud mountainous areas.Through the analysis of the experimental area of two phase images,change of each land types were obtained quickly,and the reasons for such changes were analysed.Experimental results show that the proposed land inspection method based on UAV images could acquire land change information rapidly and efficiently in the multi-cloud region.