水体指数可以抑制背景噪声和提高地表水体的可分性，已经广泛用于地表水体提取。传统FCM聚类算法考虑了地物的不确定性，但没有顾及地物的邻域空间信息，对背景异质性比较敏感。针对传统FCM聚类算法的不足，提出一种可变邻域的区域FCM聚类算法。由于复杂环境下高分二号（GF-2）遥感影像的城市地表水体具有复杂异质背景和不确定性的特点，本文利用水体指数和区域FCM聚类算法的优点，提出一种整合水体指数和区域FCM的城市地表水体自动提取算法，该算法主要步骤包括：（1）去除影像阴影后计算归一化差分水体指数NDWI（Normalized Difference Water Index）；（2）区域FCM聚类算法；（3）整合水体指数和区域FCM聚类的城市地表水体自动提取算法。最后采用两景GF-2高分辨率遥感影像（广州和武汉）进行实验，验证了该算法的有效性，并与经典地表水体提取算法进行对比分析。实验结果表明：该算法具有较高的水体提取精度，城市地表水体边界既具有较好的区域完整性又保持了局部细节，同时对城市地表水体复杂背景噪声具有较好的抑制作用，有效减少传统FCM聚类算法的“胡椒盐”现象。

The water index can suppress background noise and increase the separability of surface water. Thus, it has been widely used for surface water extraction. Traditional FCM clustering algorithm considers the uncertainty of ground objects without neighborhood spatial information, which is sensitive to background heterogeneity. On the basis of the shortcomings of traditional FCM clustering algorithms, this study proposed a regional FCM clustering algorithm and applied it to extract city surface water in complex environment regions using GF-2 remote sensing imagery. The main steps of the method include (1)Calculating the normalized difference water index after the removal of shadows; (2) Presenting a regional FCM clustering algorithm;(3)Proposing the urban surface water automatic extraction algorithm by combining the water body index and the regional FCM clustering algorithm. Finally, the proposed method was carried out on two GF-2 high-resolution remote sensing image data located in Guangzhou and Wuhan. The experimental results showed that the proposed method has better accuracy and water boundary than state-of-the-art methods. The proposed method also retains regional integrity and local details of surface water objects while effectively inhibiting noise from urban surface water in the complex background, thereby reducing the "salt and pepper" phenomenon found in traditional FCM clustering algorithm.