デジタルデータあり（Japan Link Center）
すぐに読む
CiNii Research
全国の図書館の所蔵
国立国会図書館以外の全国の図書館の所蔵状況を表示します。
所蔵のある図書館から取寄せることが可能かなど、資料の利用方法は、ご自身が利用されるお近くの図書館へご相談ください
書誌情報
この資料の詳細や典拠（同じ主題の資料を指すキーワード、著者名）等を確認できます。
- 資料種別
- 記事
- 著者・編者
- 鈴木 貴士長沼 一輝辻 裕之木村 誠聡
- 並列タイトル等
- An Estimate the Standard Deviation of Gaussian Noise Based on the Additiveness of Gaussian Distribution
- タイトル（掲載誌）
- 神奈川工科大学研究報告. A・B, 人文社会科学編・理工学編 = Research reports of Kanagawa Institute of Technology. Pt. A, pt. B, Humanities and social science, science and technology / 神奈川工科大学 編
- 巻号年月日等（掲載誌）
- (44):2020
- 掲載号
- 44
- 掲載ページ
- 37-42
- 掲載年月日（W3CDTF）
- 2020
- ISSN（掲載誌）
- 2188-2878
- ISSN-L（掲載誌）
- 2188-2878
- 出版事項（掲載誌）
- 厚木 : 神奈川工科大学
- 出版地（国名コード）
- JP
- 本文の言語コード
- jpn
- NDLC
- 対象利用者
- 一般
- 所蔵機関
- 国立国会図書館
- 請求記号
- YH247-1383
- 連携機関・データベース
- 国立国会図書館 : 国立国会図書館雑誌記事索引
- 書誌ID（NDLBibID）
- 030328552
- 整理区分コード
- 632
- 要約等
- As a method of estimating Gaussian noise superimposed on the image, there is an estimation method based on MAD. The method based on MAD has good estimation accuracy for images with many flat area. However, the estimation accuracy is not good for images with many edges and detail signals. We proposed the method to extend the method based on MAD to correct the Gaussian noise estimate according to the type of image. As a result, it was possible to improve the estimation accuracy even in an image including many edges and detail signals. However, improvement in estimation accuracy is effective only when the Gaussian noise is large, and a very effective result cannot be obtained when the Gaussian noise is small. In this paper, we propose the method for improving estimation accuracy for images with small Gaussian noise and many edges and detail signals. In the proposed method, an estimation method that focuses on the additiveness of the Gaussian distribution is applied only to images that contain many edges and detail signals. The proposed method improved the noise estimation accuracy by about 27% compared to the conventional method.
- DOI
- 10.34411/00032028
- 記録形式（IMT）
- application/pdf
- 一次資料へのリンクURL
- kkb-044-006.pdf （fulltext）
- オンライン閲覧公開範囲
- インターネット公開
- 掲載誌（NCID）
- AA12669200
- 連携機関・データベース
- 国立情報学研究所 : 学術機関リポジトリデータベース（IRDB）（機関リポジトリ）
- 提供元機関・データベース
- 神奈川工科大学 : 神奈川工科大学学術情報リポジトリ
- 要約等
- As a method of estimating Gaussian noise superimposed on the image, there is an estimation method based on MAD. The method based on MAD has good estimation accuracy for images with many flat area. However, the estimation accuracy is not good for images with many edges and detail signals. We proposed the method to extend the method based on MAD to correct the Gaussian noise estimate according to the type of image. As a result, it was possible to improve the estimation accuracy even in an image including many edges and detail signals. However, improvement in estimation accuracy is effective only when the Gaussian noise is large, and a very effective result cannot be obtained when the Gaussian noise is small. In this paper, we propose the method for improving estimation accuracy for images with small Gaussian noise and many edges and detail signals. In the proposed method, an estimation method that focuses on the additiveness of the Gaussian distribution is applied only to images that contain many edges and detail signals. The proposed method improved the noise estimation accuracy by about 27% compared to the conventional method.
- DOI
- 10.34411/00032028
- 関連情報（URI）
- 連携機関・データベース
- 国立情報学研究所 : CiNii Research
- 提供元機関・データベース
- Japan Link Center学術機関リポジトリデータベース雑誌記事索引データベースCiNii Articles
- 書誌ID（NDLBibID）
- 030328552
- NII論文ID
- 120006875556