手話における非手指信号を含めた手話単語および文法分類の検討

DOI 機関リポジトリ (HANDLE)

書誌事項

タイトル別名
  • INVESTIGATION OF SIGN LANGUAGE WORDS AND GRAMMAR CLASSIFICATION INCLUDING NON-MANUAL SIGNALS IN SIGN LANGUAGE

この論文をさがす

説明

Sign language is composed of hand and finger signals representing the movements of fingers and arms, and non-manual signals representing the movements of the head. There have been many studies on sign language recognition using only manual signals, but few studies using non-manual signals. In this study, we tried to classify words and grammars using manual signals and non-manual signals. In order to acquire both signals, it is desirable to acquire not only skeleton points of the body but also coordinate points of fingers and faces. In this study, we used OpenPose, which can acquire both finger and non-finger signals. We obtained two-dimensional time-series data from color moving images expressing sign language words and tried to extract key frames from the moving images using the acceleration of the head and left and right wrists. For each extracted frame, we calculated the features expressing words and grammatical meanings, performed principal component analysis on them, and worked on evaluation experiments using the subspace method. We aimed to categorize the grammar for each of the five affirmative forms, negative forms, and yes / no question forms of the target Japanese sign language.

収録刊行物

キーワード

詳細情報 詳細情報について

  • CRID
    1390853649761506688
  • NII論文ID
    120006897019
  • NII書誌ID
    AA12677220
  • DOI
    10.15002/00022880
  • HANDLE
    10114/00022880
  • ISSN
    21879923
  • 本文言語コード
    ja
  • 資料種別
    departmental bulletin paper
  • データソース種別
    • JaLC
    • IRDB
    • CiNii Articles
  • 抄録ライセンスフラグ
    使用可

問題の指摘

ページトップへ