中文摘要：

从高分辨率卫星图像中高效、准确地自动定位描述桥梁目标具有重要的应用价值。本文针对以往检测精度不高,尤其当河水处于枯水期时检测结果依赖检测窗口的选择等问题,从水域提取出发,在桥梁知识模型指导下,经过目标检测和目标定位过程精确定位描述目标。首次引入桥梁阴影特征,通过对它的提取辅助桥梁目标的检测。提取与目标密切相关的环境要素,提出基于环境要素提取的高分辨遥感图像中水上桥梁目标的定位方法,经过环境要素的联合特征提取和知识规则验证,实验验证本文方法提高了定位检测精度。尤其当河水处于枯水期,本文的检测方法优势尤为明显。以实际高分辨率遥感图像为例,实验结果证实了本文方法的有效性。

英文摘要：

Automatic detection and location of bridges in high-resolution satellite image has an important implication value. In this paper, of the exiting methods are briefly described and analyzed. In consideration of their drawback, a method for automatic recognition of bridges over water is presented, based on bridge knowledge models, composed of two steps： detection and location. Shadow features are extracted for the first time to assist location of targets. Extraction of environment elements （ water, bridge and shadow ） is closely related with targets and combined with verification of knowledge and rules to improve the positioning accuracy. Experiments are executed on the high-resolution remote image, and the results confirm the validity of the prooosed method.