應用SPOT及空載全偏極合成孔徑雷達影像判識地表覆蓋

Landcover Classification Using Multi-sensor Images

江介倫(Jie-Lun Chiang) ； 鄭克聲(Ke-Sheng Cheng) ； 陳錕山(Kun-Shan Chen)

《農業工程學報》 51卷4期 (2005/12) Pp. 84-96

https://doi.org/10.29974/JTAE.200512.0007

摘要

遙測技術應用日漸廣泛、實務，故靠單一種感測器執行環境監測已漸漸無法滿足需求。由於不同感測器有其優缺點，故結合各種不同感測器將使我們得到更大效益。
本研究融合光學影像(SPOT)及空載主動式雷達影像(NASA/JPL PACRIM-Ⅱ)的特性，結合光譜特徵與組織結構粗糙度之資訊，以提高分類正確率。以嘉義熬鼓農場之SPOT及POLSAR影像為例，融合二者之分類總體正確率高達98.67%，高於單以SPOT的97.34%，更遠大於以AIRSAR C Band的HH、HV、VV等三特徵之分類正確率65.22%，尤其在二階段分類後可達99.30%。
結合兩種不同感測器的影像，除了正確率提高外，且雷達影像原有的雜訊也在分類過程中被忽略掉，而且融合後之影像又可呈現無法單從由光學影像看到的細節（如蚵架等），如此大幅提昇了遙測影像的應用價值。

關鍵字

SPOT ； 合成孔徑雷達影像 ； 影像分類 ； 融合

並列摘要

Remote-sensing techniques are getting more and more practical and pervasive. Since a single sensor has limited capability of monitoring the environment, combining multiple sensors achieves a higher coverage. Along with the development of satellite remote sensing techniques, the more sensors are used, the more images can be put in use. This paper studies the integration of SPOT and AIRSAR images in order to combine spectrum features and roughness information to improve classification accuracy.
As a case study, Au-Ku area in Taiwan was selected for the experimental evaluation. The overall accuracy from the integration of SPOT and AIRSAR images reaches to 98.67%, higher than the 97.34% of SPOT and much higher than the 65.22% of the SARC band. In addition to this accuracy, it reduces speckle effects of radar images, catches some details (the racks for breeding oyster in coastal waters for example) unavailable in optical images, and enhances the applicability of remote sensing images.

並列關鍵字

SPOT ； SAR ； Multi-sensor ； Classification ； Fusion