TM数据经典水体提取方法存在对细小水体的漏判和对山体阴影的误判两方面缺陷。本文基于LBV变换提出了B分量阈值法和LBV关系法两种新的水体提取方法。实验结果表明:两种新方法的漏分精度分别为2.54%、2.73%,误分精度分别为15.71%、13.39%,整体提取效果优于经典方法。 There are two shortcomings of TM data classic water body extraction methods, such as the leakage judgment of small water body and the misjudgment of the mountain shadow. In this paper, we propose two new water body extraction methods based on LBV transform and B component threshold method. Experimental results show that the leakage accuracy of the two new methods are 2.54% and 2.73% respectively, and the accuracy of misclassification is 15.71% and 13.39% respectively. The overall extraction efficiency is better than the classical method.