Greenhouse mapping has attracted much attention recently, especially in China where greenhouse practice has been growing dramatically. Remote sensing based greenhouse extraction methods can generate the geographical locations and spatial distribution of greenhouses efficiently. Most of the existing greenhouse extraction algorithms rely on high-resolution remote sensing images or aerial images, which are often expensive to obtain and require complicated algorithms to process. To solve this problem, this paper proposes a fast algorithm for greenhouse extraction based on Landsat images that are freely available. First, an enhanced water index was introduced to characterize winter greenhouse, based on an observed natural phenomenon that water vapor inside a greenhouse is usually condensed to form a layer of dew on the inner surface of the greenhouse plastic or glass. On one hand, the dew layer makes a greenhouse has a high water index value, which makes it to be distinguished easily from bare land. On the other hand, the dew layer increases a greenhouse’s reflectivity, which makes it different from natural water bodies. In order to extract greenhouses, a simple and efficient decision tree classifier was designed. Da’ao town of Jiangmen in Guangdong Province was chosen as an example, and the experiments were based on Landsat images taken in different years. The results show that the proposed method is effective in extracting greenhouses, with the advantages of high efficiency, low cost, and strong robustness.