一种顾及空间关系的遥感影像阴影检测算法

以城市区域内高大建筑阴影为研究对象,针对现有的阴影检测算法在复杂地物环境下检测精度和可靠性不高的问题,提出了一种结合颜色空间特征和空间关系的遥感影像阴影检测方法。首先,采用SLIC超像素算法对影像进行分割;然后基于Lab和HSI颜色空间构建初步检测条件,将阴影划分为阴影主体区域和待检测区域;最后,借助Canny边缘检测信息合并待判别区域内的超像素块,并利用阴影区域与造成干扰区域间的空间位置关系构建的检测条件进行判别。实验结果表明,该方法可以有效提高复杂地物环境下遥感影像阴影的检测精度和算法可靠性。

A Shadow Detection Method of Remote Sensing Image Combing with Spatial Relationship

In order to solve the problem that current shadow detection algorithms are easily disturbed in complex terrain environment,a new shadow detection method for remote sensing image whose major shadow regions are caused by tall buildings in urban is proposed,which is also based on color space features and spatial relations.Firstly,a super-pixel algorithm named SLIC is used to realize the image segmentation.And then a number of detection conditions which are based on either Lab or HSI color space are built to divide image into two sections,which are called as the primary area of the shadow region and the region to be detected.The super-pixels in the regions to be detected are merged with information from edge detection by Canny operator.Finally,spatial relations between shadow regions and regions that make interference are used to built a new detection condition.The experimental results show that this method can effectively improve shadow detection results of remote sensing image in complex environment which can be quantized by correct detection ratio,leak detection ratio and error detection ratio.