The aim of this study is to construct time series images from multi-sensor images. This study compares different time series image fusion techniques for Formosat-2 (FS-2) and Landsat-8 (LS-8) images. Three different image fusion approaches are included to generate time series images from Formosat-2 and Landsat-8 images, including bi-cubic image resampling, high pass filter pan-sharpening and spatial and temporal adaptive reflectance fusion model (STARFM). The goal is to integrate satellite images from different sensors with different spatial and temporal characteristics. In order to assess the quality of simulated images, this study calculates the biases between the observed and the synthetic reflectance for Formosat-2 image. Also, this study demonstrates the simulation of Formosat-2 image from Landsat-8 image through different approaches. In qualitative analysis, the STARFM approach provides better results than other approaches. The quantitative results also indicated that the difference between simulated and real image via STARFM has the lowest bias. In summary, the time series satellite images can be constructed from multi-sensor satellite images successfully.
|Number of pages||4|
|State||Published - 1 Jan 2018|
|Event||39th Asian Conference on Remote Sensing: Remote Sensing Enabling Prosperity, ACRS 2018 - Kuala Lumpur, Malaysia|
Duration: 15 Oct 2018 → 19 Oct 2018
|Conference||39th Asian Conference on Remote Sensing: Remote Sensing Enabling Prosperity, ACRS 2018|
|Period||15/10/18 → 19/10/18|
- Image fusion