论文部分内容阅读
遥感影像时空融合技术可以从不同分辨率特征的传感器中,通过融合算法提取各自的高分辨率信息,生成双高甚至多高的融合影像。提出了一种基于多波段加权的时空数据融合方法,该方法利用近红、红、绿3个波段计算得到NDVI和NDWI值,用于提取MODIS影像的地物变化特征。结合单幅Landsat 7影像获得了不同变化区域重采样后的空间分布情况。并对STARFM原算法进行了改进,使之可适应不同区域的变化特征,即针对季节性变化区域,采用STARFM方法进行融合;对于地物类别变化区域,则采用距离加权法进行融合,最终得到具有高时间、高空间分辨率的融合影像。实验以深圳市为研究区,分别选取了2000年和2002年两年的遥感数据进行了实验。通过实验结果可知,基于多波段距离加权的融合结果同真实影像间的相关系数为0.86~0.89,优于STARFM算法(0.81~0.83)。此外,通过融合得到的多维高分辨率数据有利于丰富现有的遥感数据影像信息,可进一步应用于洪水监测方面的研究。
Remote sensing image spatio-temporal fusion technology can extract the high-resolution information from the sensors with different resolution features through the fusion algorithm to generate the double high or even high fusion image. A new spatio-temporal data fusion method based on multi-band weights is proposed. The method uses near-red, red and green bands to calculate NDVI and NDWI values, which are used to extract features of MODIS images. Combining with a single Landsat 7 image, the spatial distribution of resampled areas in different areas was obtained. The original algorithm of STARFM is improved to adapt to the changing features of different regions, that is, the STARFM method is used for the fusion of the seasonal variation regions, and the distance weighted method is used for the fusion of the STARFM regions. Finally, High time, high spatial resolution fusion image. The experiment took Shenzhen as the research area, and selected the remote sensing data of 2000 and 2002 respectively for experiments. The experimental results show that the correlation coefficient between fusion results and real images based on multi-band distance weights is 0.86-0.89, which is better than that of STARFM algorithm (0.81-0.83). In addition, the multi-dimensional high-resolution data obtained through the fusion is conducive to enriching the existing remote sensing data image information and can be further applied to flood monitoring research.