针对大幅面遥感影像在分块边界特征不连续性和超像素分割地物边界精度不高等问题,提出了 一种面向地物特征的大幅面遥感影像分割方法.首先,通过降低辐射分辨率来压缩影像数据量,并通过 自适应紧密度参数方法获得具有相近尺寸的SLIC(simple linear iterative clustering)分割影像块;然后,对每个影像块分别构建影像的比值植被指数(relative vegetation index,RVI),通过最大类间方差法(OTSU)将像元标记为植被与非植被;最后,采用SLIC对标记像元精细分割,得到最终分割结果.以WorldView-2 卫星影像和高分二号卫星影像作为实验数据,进行面向地物特征的大幅面遥感影像分割.实验结果表明,改进算法效率得到提升,精度有所改善.
Feature-oriented Segmentation Method for Large Format Remote Sensing Images
Aiming at the problem that simple linear iterative clustering(SLIC)superpixel segmentation algorithm cannot process high-resolution remote sensing images with large size,a SLIC segmentation method for large high-resolution remote sensing images oriented to ground features is proposed.Firstly,the amount of image data is compressed by reducing the radiometric resolution,and the SLIC segmentation sub-image blocks with similar sizes are obtained by the adaptive tightness parameter method.Then,the relative vegetation index(RVI)of the image is constructed for each sub-image block,and the pixels are marked as vegetation and non-vegetation by OTSU method.Finally,SLIC is used for fine segmentation of labeled image elements to obtain the final segmentation results.Using WorldView-2 satellite imagery and Gaofen-2 satellite imagery as experimental data,large format remote sensing image segmentation oriented to ground features is carried out.The experimental results show that the efficiency and accuracy of the proposed method has been improved.
万方数据
维普
