電子情報通信学会ソサイエティ大会講演要旨
CS-1-1
教師なしPolSAR地表分類のためのQuaternion Auto-Encoder による偏波特徴抽出と自己組織化マッピング
◎△キム ヒョンス・廣瀬 明(東大)
We propose an unsupervised PolSAR land classification system based on combination of quaternion auto-encoder and quaternion self-organizing map (SOM). Most of the existing PolSAR land classification systems extract a set of feature information that human beings designed beforehand. Such methods will face limitations in the near future when we expect classification into a large number of land categories recognizable to humans. In our proposed system, quaternion auto-encoder extracts feature information based on the natural distribution of PolSAR features, and quaternion SOM classifies the extracted features. As a result, we succeeded to classify PolSAR data into new and more detailed land categories such as factory sites and furrowed farmlands.