HSV变换和多尺度分割相结合的高分辨率遥感影像阴影检测

Shadow Detection from High Resolution Remote Sensing Imagery Based on HSV Transformation and Multilevel Segmentation

针对阴影在高分辨率遥感影像的特性,提出了一种色彩空间变换和多尺度分割相结合的阴影检测方法。该方法首先对原始影像进行连续两次HSV变换,并分别提取前后两次变换的亮度分量和色度分量;然后引入面向对象思想,进行多个尺度的影像分割并依次实现每一尺度下的阴影检测;最后将多个尺度的检测结果进行决策级融合获取最终检测结果。利用高分二号和Google Earth影像分别进行实验,实验结果表明,该方法有效结合了粗细尺度优势,阴影检测误检率和漏检率较低,同时对较亮阴影和较暗地物均具备较好的识别效果。

According to the shadow characteristics in high resolution remote sensing imagery,a method based on color space transformation and multilevel segmentation is proposed in this paper. Firstly,the value and hue of the original image are extracted respectively by two consecutive HSV transformation. Then,object-oriented analysis is used for multi-scale segmentation and shadow detection in single scale. Finally,the ultimate extraction is accomplished by decision level fusion. The presented method is evaluated with GF-2 and Google Earth imageries. The results show that the proposed shadow detection algorithm can combine the advantages of different scales and has a performance in terms of a low errors and leakages and great recognition to the partial shadow and dark object.