提出了一种新的TM图像中桥梁目标的识别方法。算法充分利用了TM图像的特点，在底层处理中运用形态学的方法提取出潜在桥梁目标；在中层处理中使用链码表示目标并提取其特征参数；最后在高层处理中进行桥梁识别和一些后处理。算法识别速度快、准确率高，整个算法可以进行自动的识别，也可以有少量人工干预使得算法更稳健。实验证明该算法对于TM图像中的桥梁识别是很有效的。

In this paper, a new technique to solve the problem of bridge recognition from TM images is presented The algorithm takes advantage of the characteristics of TM images It uses morphology operators to collect candidate bridge objects in the segmented images; then chain code is used to represent these candidates and computer their parameters; and in the last step we recognize those true bridges according to some knowledge and make the result more rational by post processing The algorithm works fast and can work automatically, even though some man intervention can make it more robust The result is satisfying considering the low resolution of the TM images Results on the TM images of three different areas are given to demonstrate the efficiency of our method