Fingertips tracking algorithm for guitarist based on temporal grouping and pattern analysis

Zhao Wang, Jun Ohya

    研究成果: Conference contribution

    2 引用 (Scopus)

    抜粋

    This paper proposes a temporal Grouping and pattern analysis-based algorithm that could track the fingertips of guitarists during their guitar playing towards the actualization of the automatic guitar fingering recognition system. First a machine learning-based Bayesian Pixel Classifier is used to segment the hand area on the test data. Then, the probability map of fingertip is generated on the segmentation results by counting the voting numbers of the Template Matching and Reversed Hough Transform. Furthermore, a temporal Grouping algorithm, which is a geometry analysis for consecutive frames, is applied to removal noise and group the same fingertips (index finger, middle finger, ring finger, little finger). Then, a data association algorithm is utilized to associate 4 tracked fingers (index finger, middle finger, ring finger, little finger) with their correspondent tracked results frame by frame. Finally, particles are distributed only between the associated fingertip candidates to track the fingertips of guitarist effectively. The experimental result demonstrates that this fingertip tracking algorithm is robust enough for tracking fingertips (1) without any constrains such us color marker; (2) under the complex contexts, such us complicated background, different illumination conditions, (3) with the high tracking accuracy (mean error 3.36 pixels for four fingertips).

    元の言語English
    ホスト出版物のタイトルComputer Vision - ACCV 2016 Workshops, ACCV 2016 International Workshops, Revised Selected Papers
    出版者Springer Verlag
    ページ212-226
    ページ数15
    10118 LNCS
    ISBN(印刷物)9783319545257
    DOI
    出版物ステータスPublished - 2017
    イベント13th Asian Conference on Computer Vision, ACCV 2016 - Taipei, Taiwan, Province of China
    継続期間: 2016 11 202016 11 24

    出版物シリーズ

    名前Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
    10118 LNCS
    ISSN(印刷物)03029743
    ISSN(電子版)16113349

    Other

    Other13th Asian Conference on Computer Vision, ACCV 2016
    Taiwan, Province of China
    Taipei
    期間16/11/2016/11/24

    ASJC Scopus subject areas

    • Theoretical Computer Science
    • Computer Science(all)

    フィンガープリント Fingertips tracking algorithm for guitarist based on temporal grouping and pattern analysis' の研究トピックを掘り下げます。これらはともに一意のフィンガープリントを構成します。

  • これを引用

    Wang, Z., & Ohya, J. (2017). Fingertips tracking algorithm for guitarist based on temporal grouping and pattern analysis. : Computer Vision - ACCV 2016 Workshops, ACCV 2016 International Workshops, Revised Selected Papers (巻 10118 LNCS, pp. 212-226). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); 巻数 10118 LNCS). Springer Verlag. https://doi.org/10.1007/978-3-319-54526-4_16