Landsat-7 ETM+ 是Landsat 系列中最新的一顆衛星，它擁有30公尺的空間解析力及185公里的影像寬度，由於它的FOV為14.7°，因此相當適合用來監測大範圍的地表資訊，但由於掃描校正器故障(SLC, Scan Line Corrector)的原因，造成了所得到的影像會有空隙像元(Gap Pixels)，這些空隙像元的部份佔整張影像22%，間隔性分佈於整張影像，使Landsat-7 ETM+ SLC-off影像不易運用，因此，發展一填補空隙像元的方法，使Landsat-7 ETM+ SLC-off影像仍可以使用為重要課題。 在此次研究中，我們提出了兩階段式填補空隙像元的方法。在第一階段中，利用建立工作影像及參考影像的轉換函式來填補空隙像元，由於兩張影像可能因時間的不同，造成會有變異的產生，因此我們將有變異的像元加以移除，再利用第二階段進行填補，第二階段是以內插法來填補空隙像元，內插法是利用周圍非空隙的像元進行填補，使用內插法填補空隙像元可完整填補剩餘的空隙像元。無論在模擬案例或實際案例中，實驗成果都顯示填補影像具有影像的完整性與高影像品質。 Landsat-7 ETM Plus, one of latest satellite in Landsat series, is capable of acquiring images with spatial resolution in 30 meters and swath in 185 kilometers. The specifications make it possible to monitor the earth surface for a relatively large scale area. This performance is attributed to its whisk-broom sensor with 14.7 degrees FOV. However, the Scan Line Corrector (SLC) used to compensate the forward motion of whisk-broom sensor has malfunctioned on May 31, 2003. As a result, the images acquired from Landsat-7 have data gaps that occupy about 22% of the entire scene. Because of distribution of gapped pixels, Landsat-7 ETM Plus SLC-off images are difficult to use. Consequently, the methods used to fill the gaps on Landsat 7 images become an important issue in the study of image processing. In this study, a two-stage approach for filling the gapped pixels of Landsat-7 image is proposed. In the first stage, filling by mapping method is employed. The first stage is to fill the gapped pixels with other reference images through a mapping process. The method particularly makes an effort to erase changed pixels between the working and reference images. These changed pixels are filled by second stage. In the second stage, filling by interpolating method is used to fill the gapped areas that fail to fill in the first stage. The second stage is to fill the gapped pixels by interpolating the non-gapped pixels that neighbor to them. No matter simulated case or real case, experimental results indicate that gap-filled images are complete and accurate.
