中文摘要：

针对基于SPOT5影像提取水体时易受地形阴影、居民地等因素影响提取精度的问题,本文构建了一种新的水体提取方法。该方法以水体的遥感图像本底值研究为基础,通过构建能够同向增强的水体指数及多指数集成计算模型,实现水体灰度值的极化,在此基础上利用改进的阈值自动选取算法、数学形态学滤波及细化等算法,实现了水体的高精度自动提取。经过试验比较表明,该方法能够有效地提取较细水体,且能够有效去除地形阴影、居民地及河流边滩等的影响。

英文摘要：

Water body feature extraction from remotely sensed images can be easily affected by several factors such as topographic shadowing and habitation settlement place. In order to solve the precision problem in water body feature extraction based on SPOT5 images, this paper proposes a novel water body extraction approach. The proposed approach is based on the study of a water body＇s remote sensing image background value. It is able to polarize water body gray values by constructing water body index that is able to enhance in the same direction and across multiple indices the integration cal- culation model. Based on these techniques, the proposed approach utilizes an automatic threshold se- lection algorithm, mathematic morphology filtering, and thinning algorithms to enable automatic wa- ter body featre extraction with high precision. The experimental results and corresponding compari- sons show that the proposed approach is able to effectively extract small water bodies and remove top- ographic shadowing, habitation settlement places, and river alternative bars etc.