| 電子情報通信学会 研究会発表申込システム 講演論文 詳細 |
技報閲覧サービス [ログイン] 技報アーカイブ |
| トップに戻る | 前のページに戻る | [Japanese] / [English] |
| 講演抄録／キーワード | ||
| 講演名 | 2015-11-24 13:45 Adaptive Nearest Feature Space Method for Remote Sensing Images Classification Yang-Lang Chang（NTUT）・○Chihyuan Chu（G-AVE）・Hirokazu Kobayashi（OIT）・Tzu-Wei Tseng（NTUT）  SANE2015-62 |
|
| 抄録 | （和） | In this paper a novel technique based on nearest feature space (NFS), known as adaptive nearest feature space (ANFS), is proposed for supervised remote sensing image classification. The NFS has been proven to be efficient for remote sensing image classification in recent years. Although NFS can perform well for classification, in some instances, it decreases the efficiency when samples of different classes are not far apart or even overlapped. Due to the different neighborhood structures of overlapping training labels, the traditional NFS can't perform well for classification of remote sensing images. In response, ANFS is proposed to overcome this problem. It combines and adopts two methods, NFS and incenter-based nearest feature space (INFS) which makes use of the incircle of three labeled samples to form an INFS, to achieve the best classification accuracy. In ANFS, a fitting preprocessing of NFS is presented to determine what the best-fix model (NFS/INFS) is for the three nearest labeled samples of the same classes. Experimental results demonstrate the proposed ANFS approach is suitable for land cover classification in earth remote sensing. It can achieve the better performance than NFS classifier when the class samples distribution overlaps. |
| （英） | In this paper a novel technique based on nearest feature space (NFS), known as adaptive nearest feature space (ANFS), is proposed for supervised remote sensing image classification. The NFS has been proven to be efficient for remote sensing image classification in recent years. Although NFS can perform well for classification, in some instances, it decreases the efficiency when samples of different classes are not far apart or even overlapped. Due to the different neighborhood structures of overlapping training labels, the traditional NFS can't perform well for classification of remote sensing images. In response, ANFS is proposed to overcome this problem. It combines and adopts two methods, NFS and incenter-based nearest feature space (INFS) which makes use of the incircle of three labeled samples to form an INFS, to achieve the best classification accuracy. In ANFS, a fitting preprocessing of NFS is presented to determine what the best-fix model (NFS/INFS) is for the three nearest labeled samples of the same classes. Experimental results demonstrate the proposed ANFS approach is suitable for land cover classification in earth remote sensing. It can achieve the better performance than NFS classifier when the class samples distribution overlaps. | |
| キーワード | （和） | remote sensing images classification / nearest feature space / incenter-based nearest feature space / adaptive nearest feature space / / / / |
| （英） | remote sensing images classification / nearest feature space / incenter-based nearest feature space / adaptive nearest feature space / / / / | |
| 文献情報 | 信学技報, vol. 115, no. 320, SANE2015-62, pp. 71-74, 2015年11月. | |
| 資料番号 | SANE2015-62 | |
| 発行日 | 2015-11-16 (SANE) | |
| ISSN | Print edition: ISSN 0913-5685 Online edition: ISSN 2432-6380 | |
| 著作権に ついて |
技術研究報告に掲載された論文の著作権は電子情報通信学会に帰属します．(許諾番号：10GA0019/12GB0052/13GB0056/17GB0034/18GB0034) | |
| PDFダウンロード | SANE2015-62 |
|
| 研究会情報 | |
| 研究会 | SANE |
| 開催期間 | 2015-11-23 - 2015-11-24 |
| 開催地（和） | アジア工科大学，バンコク，タイ |
| 開催地（英） | AIT, Bangkok, Thailand |
| テーマ（和） | ICSANE 2015 |
| テーマ（英） | ICSANE 2015 |
| 講演論文情報の詳細 | |
| 申込み研究会 | SANE |
| 会議コード | 2015-11-SANE |
| 本文の言語 | 英語 |
| タイトル（和） | |
| サブタイトル（和） | |
| タイトル（英） | Adaptive Nearest Feature Space Method for Remote Sensing Images Classification |
| サブタイトル（英） | |
| キーワード(1)（和/英） | remote sensing images classification / remote sensing images classification |
| キーワード(2)（和/英） | nearest feature space / nearest feature space |
| キーワード(3)（和/英） | incenter-based nearest feature space / incenter-based nearest feature space |
| キーワード(4)（和/英） | adaptive nearest feature space / adaptive nearest feature space |
| キーワード(5)（和/英） | / |
| キーワード(6)（和/英） | / |
| キーワード(7)（和/英） | / |
| キーワード(8)（和/英） | / |
| 第1著者 氏名（和/英/ヨミ） | Yang-Lang Chang / Yang-Lang Chang / |
| 第1著者 所属（和/英） | National Taipei University of Technology (略称： NTUT) National Taipei University of Technology (略称： NTUT) |
| 第2著者 氏名（和/英/ヨミ） | Chihyuan Chu / Chihyuan Chu / |
| 第2著者 所属（和/英） | G-AVE Technology Company (略称： G-AVE) G-AVE Technology Company (略称： G-AVE) |
| 第3著者 氏名（和/英/ヨミ） | Hirokazu Kobayashi / Hirokazu Kobayashi / |
| 第3著者 所属（和/英） | Osaka Institute of Technology (略称： OIT) Osaka Institute of Technology (略称： OIT) |
| 第4著者 氏名（和/英/ヨミ） | Tzu-Wei Tseng / Tzu-Wei Tseng / |
| 第4著者 所属（和/英） | National Taipei University of Technology (略称： NTUT) National Taipei University of Technology (略称： NTUT) |
| 講演者 | 第2著者 |
| 発表日時 | 2015-11-24 13:45:00 |
| 発表時間 | 25分 |
| 申込先研究会 | SANE |
| 資料番号 | SANE2015-62 |
| 巻番号（vol） | vol.115 |
| 号番号（no） | no.320 |
| ページ範囲 | pp.71-74 |
| ページ数 | 4 |
| 発行日 | 2015-11-16 (SANE) |
[研究会発表申込システムのトップページに戻る]