摘要:
在将遥感影像应用于土地覆盖制图的过程中,混合像元会产生负面影响,尤其是在空间分辨率较低的情况下.软分类和超分辨率(亚像元/子像元)制图技术可以解决上述问题.子像元交换算法是一种简单而有效的亚像元制图技术,但也存在计算效率不够高和在超分辨率制图因子(scale factor,S)较大时制图精度较差的问题,其原因可能是亚像元制图只是从软分类结果中获得信息.因此,将数字高程模型(digital elevation model,DEM)及其衍生的数据作为子像元交换算法的辅助信息,研究提高其制图精度的有效性.研究结果表明:①在DEM信息的辅助下,亚像元制图的精度得到了改善,即使在S较大时也是有效的;②同时使用高程和坡度信息时的效果要好于单独应用高程或坡度信息;③在S较大的情况下,制图精度对邻域范围大小的敏感性要低于S较小时;④在DEM信息的辅助下,计算效率得到提高.实验结果表明,将DEM作为辅助信息进行亚像元制图是有效和可行的.
Abstract:
When remote sensing images are used to provide information for land cover mapping, it is negatively affected by the occurrence of mixed pixels in the remote sensing images, particularly in the case of coarse spatial resolutions. Soft classification and super - resolution ( sub - pixel ) mapping techniques can solve this kind of problems. The pixel - swapping ( PS ) algorithm is a simple and efficient technique for sub - pixel mapping. However, its computation is inefficient and yields poor mapping accuracy when the super-resolution scale factor ( S) is large. A possible reason for this is that it only relies on the information from the fraction images. In this study, the digital elevation model( DEM) and their derivative data are employed as supplementary information for the PS algorithm so as to improve super-resolution mapping accuracy. Some conclusions have been reached: ①The sub-pixel mapping accuracy could be improved with the assistance of the DEM even if the scale factor is large;② The mapping accuracy by incorporating both elevation and slope information is better than that of using elevation or slope data alone;③Mapping accuracy is less sensitive to the number of neighbors when scale factor is large;④ The computing efficiency is improved when incorporating DEM in pixel-swapping. Thus, it is feasible to use DEM as supplemental information for sub-pixel mapping.
下载: