テクスチャ画像解析によるシロクローバとイネ科雑草群落の判別

• 田村 仁

  農業機械学会 東京理科大学経営学部

• 阿刀田 央一

  東京農工大学大学院生物システム応用科学研究科

• 酒井 憲司

  農業機械学会 東京農工大学農学部

書誌事項

タイトル別名

• Textural Image Analysis for Discriminating Colonies of Clover and Grass Weed.
• テクスチャ ガゾウ カイセキ ニ ヨル シロクローバ ト イネカ ザッソウ グンラク ノ ハンベツ

抄録

A method of computer-vision using a texture image analysis to discriminate the most predominant species in every small area of fields has been developed. The method treats each image of every small area as a random texture, and extracts micro-shapes as texture features by a “shape-pass” nonlinear filter bank proposed by the authors. A discriminant function is composed by statistically selected effective features to discriminate the predominant species in all extracted features. The method has been applied to a problem to discriminate white clovers (Trifolium repens L.) as a representative of leguminous cover crops from grass weeds, and an experiment to classify every small area of field into “White clovers predominant”, “Grass weeds predominant”, and “Neither” has been carried out. In the experiment, sample images prepared as the models have been discriminated the accuracy of 98% and the experimental results for actual field images have corresponded with the ones in 88% by human, the size of each area being set to 380×250mm.

収録刊行物

• 農業機械学会誌

  農業機械学会誌 64 (1), 43-51, 2002

  農業食料工学会

キーワード

• 植生分布図
• 画像処理
• テクスチャ解析

詳細情報

• CRID: 1390001204313456768
• NII論文ID: 10012318865
• NII書誌ID: AN00200470
• DOI: 10.11357/jsam1937.64.43
• ISSN: 18846025
• ISSN: 02852543
• NDL書誌ID: 6036456
• Web Site: http://id.ndl.go.jp/bib/6036456
• Web Site: https://ndlsearch.ndl.go.jp/books/R000000004-I6036456
• 本文言語コード: ja
• データソース種別:
  • JaLC
  • NDL
  • CiNii Articles
  • KAKEN
• 抄録ライセンスフラグ: 使用不可