土地覆盖图斑多时相遥感影像自动配准

摘要

针对点、面方法仅利用了遥感影像几何特征和部分波段的灰度特征进行多时相遥感影像配准的不足，提出一种稳定土地覆盖图斑的多时相遥感影像自动配准方法，充分利用遥感影像多光谱信息和大量存在的稳定土地覆盖图斑信息进行图像配准，并且选取了土地覆盖年际变化最为强烈的农业种植区作为试验区，分别利用同一传感器和不同传感器不同时相的遥感数据开展了试验研究.两次试验中，配准精度分别达到了0．57个像元、0．65个像元.试验结果表明，本文提出的方法能够有效地筛选出满足图像配准的同名图斑，具有较高的配准精度和适用性，提高了遥感影像的配准效率.%An automatic registration method for multi-temporal remote sensing images using stable land cover patches was proposed in this paper,taking full advantage of multi-spectral features as well as stable and widespread land cover patches in remote sensing images.Rural regions,where significant land cover changes happened within-and between-year,were chosen as study area,and multi-temporal remote sensing images from same or different sensors were used.Experimental results showed the effectiveness and adaptability on extraction of stable land cover patches of this method, and registration accuracies of 0.57 and 0.65 pixels were achieved.