高分辨率遥感影像阴影的自动检测与自动补偿

Automatic Shadow Detection and Automatic Compensation in High Resolution Remote Sensing Images

提出了高分辨率遥感影像中阴影的检测与补偿的全自动处理方法.首先,利用简单阴影特征构造复合特征的方法联合构建多种光谱阈值检测条件,结合相应自动阈值获取方案实现阴影自动初检测,并设计自动形态学处理方法对初检测结果进行后处理;然后,以补偿模型为基础,通过统计阴影与非阴影区域的亮度均值与均方差、采用特征点提取与匹配的方法自动获取模型参数,再综合区域级整体补偿和窗口级局部补偿两种结果实现阴影的全自动补偿.结果表明,多条件联合检测与自动形态学处理可提升检测结果的总体精度和区域完整性,补偿模型参数自动获取方法可靠有效,综合补偿法充分结合整体与局部补偿的优势,提升整体亮度,突显纹理细节,使得被阴影遮挡的信息得以真实再现.

A method to automatically detect and compensate shadows in high resolution remote sensing images is proposed. This method consists of initial shadow detection and post-procession. Through utilizing simple shadow spectrum signatures and complicated signatures shadows are detected automatically with relative automatic thresholds. Then the initial detected result is improved by a post-procession scheme based on the principles of mathematical morphology. Meanwhile an automatic compensating method which is essentially a mathematic model is explored, whose parameters are automatically acquired by calculating the mean and variance of illumination in shadow and non-shadow areas and using feature points extraction and matching strategy. Additionally, an integrated method using the compensation information from the whole shadow area and the small pixel window respectively is used to promote the final result. Experimental results indicate that both the multifarious conditions and post-processing can improve the accuracy of shadow detection results. The post-processing can also strengthen the completeness of shadow area. Moreover, the method of automatically acquiring the parameters in the compensation model is feasible and effectual. The integrated compensation strategy makes the brightness of shadow area heightened and the details and texture enhanced. In consequence, the real information shaded by shadow can emerge again.