遥感影像空间分辨率对居民地信息提取的影响

Influence of the Varied Spatial Resolution of Remote Sensing Images on Urban and Rural Residential Information Extraction

利用遥感手段,以低廉的成本有效获取包括居民地在内的地表信息是促使卫星遥感数据转化为现实生产力的根本所在,也是遥感应用领域中亟待解决的重要课题。目前虽然已有"空间分辨率越高,遥感分类精度也越高"的共性认识,但是遥感影像空间分辨率的不同对于城镇和乡村这两种不同类型的居民地信息提取产生怎样的影响等,目前尚无明确答案。本文首先以同一区域不同空间分辨率的4种卫星遥感影像数据为信息源,定性探讨居民地在不同分辨率遥感影像上的表现特征;然后以两种不同分辨率的影像ALOS和TM为实验对象,对这两种影像上城镇和乡村居民地信息提取精度进行比较分析。结果显示:虽然基于10m-ALOS影像的居民地信息提取精度较30m-TM影像约提高6.09%,并且无论是依据哪一种空间分辨率的遥感影像,城镇居民地的信息提取精度都明显高于乡村居民地,但是,对于分散式分布的乡村居民地,遥感影像空间分辨率的下降令其遥感识别更容易受到周边环境的影响,其信息提取精度会随着遥感影像空间分辨率的增加而显著提高;相比之下,城镇居民地的信息提取精度随遥感影像空间分辨率增加的趋势是有限的。造成这一现象的原因主要在于虽然居民地的轮廓清晰性以及像元的纯净度都与遥感影像的空间分辨率基本上呈正相关,但是空间分辨率的提高会放大居民地内部结构的细节,从而对居民地的遥感识别与信息提取工作增加一些干扰性的噪声信息。

This paper discusses the characteristics of residential areas on four types of remote sensing images with different spatial resolution in the same area based on the data from them,and analyzes the accuracy of information extraction of urban and rural residential areas from ALOS and TM images with different resolution.The results show that the accuracy of information extraction of residential areas from 10m-ALOS image has increased by 6.09%compared with that from 30m-TM image and the accuracy of urban residential areas is obviously higher than that of rural residential areas based on either remote sensing image.Nevertheless,the accuracy of dispersed rural residential areas will notably improve with the increase of spatial resolution and the remote sensing identification is more likely to be influenced by the neighboring environment due to the reduced resolution,and the situation for urban residential areas is different.The reason is that the structural detail of residential areas will be magnified by the increased spatial resolution and that will interfere in the remote sensing identification and information extraction of the residential areas,although the contour clarity of residential areas and the pixel purity are positively correlated with the spatial resolution of remote sensing images.