顾及水域的QuickBird影像阴影检测方法研究

Research on QuickBird image shadow detection method considering the waters

针对城市高分影像中阴影检测不准确以及水体影响阴影检测的问题,结合影像各波段灰度值特征,改进一种基于多波段灰度值差异的阴影自动检测方法。首先,分析阴影与非阴影在不同波段的灰度值差异,进行波段组合,通过建立的波段组合构造多波段阴影指数MSI以增强阴影信息得到疑似阴影区域;然后,在归一化水体指数NDWI的基础上根据比值法原理构造水体指数WI加大水体与其他地物的差异性,得到水体区域;最后,通过形态学滤波算法对检测结果进行后处理并结合布尔运算得到影像中较完整的阴影区域。实验结果表明,该方法能够准确区分影像中的阴影与水体,准确提取影像中的阴影区域。

Aiming at the problem of inaccurate shadow detection in urban high resolution images and the shadow of water body effects detection and combined with the band digital number characteristics of the images, a method of automatic shadow detection based on multi band digital number difference is improved. First, it analyzes the digital number difference between the shadow and non shadow, executes the bands change, and constructs a multi band shadow index by creating a band combination to enhance the shadow information to get a suspected shadow area. Then, on the basis of the normalized difference water index, it constructs the water index according to the principle of the ratio method to increase the difference between the water and other objects, and getts the water area. Finally, the morphological filtering algorithm is used to post process the detection result, and combines the Boolean operation to get the entire shaded area in the image. Experimental result shows that the detection method in this paper can distinguish the shadow and water in the image accurately, and extract the shaded area in the image precisely.