髪型の認識と合成  [in Japanese] Recognition and representation of the hair style for the automatic facial caricaturing  [in Japanese]

Abstract

自然なヒューマン・マシン・インタフェースを目指して, われわれは視覚による顔の意味理解技術, および顔の表現技術の研究開発を進めている. この技術の応用例として, 似顔絵クリエーション技術を開発した. 開発したシステムではコンピュータビジョン技術を用いて様々な人種や顔・髪の種類に対してロバストに特徴抽出ができる. さらに, ファジィ推論による部品選択・部品配置・部品変形ルールから, 複数のイラストレータや漫画家などのタッチ(画風)で似顔絵を自動的に描画することができる. 本稿では似顔絵クリエーション技術の中で重要となる髪型認識の手法とイラストレータタッチの髪型を合成する手法について述べる. 具体的には, 髪領域抽出法, 髪型の特徴によるクラスタリングとテンプレートマッチングを併用した髪型部品の選択方法について述べる.

Aiming at a flexible human-machine interface, we are advancing the research and development about the understanding technology of the face and expression technology of the face. As an example of applying this technology, we have developed the facial caricaturing technology. The facial feature is extracted robustly by the computer vision technology for the kind of various races, faces and hairs, and it is possible to draw in the likeness automatically by the style of drawing such as two or more caricaturists, illustrators, and comic artists. In this paper, we explain the method which is important in our technology to recognize the hair style and to represent the hair style. Concretely, we will describe the method for the extraction of the hair region, and for the selection of the hair parts using classification with features of the hair style and template matching together. Owing to this method, it is possible to draw a hair style as if illustrators or comic artists draw.

Journal

Technical report of IEICE. PRMU   [List of Volumes]

Technical report of IEICE. PRMU 97(387), 25-32, 1997-11-21  [Table of Contents]

The Institute of Electronics, Information and Communication Engineers

References:  6

You must have a user ID to see the references.If you already have a user ID, please click "Login" to access the info.New users can click "Sign Up" to register for an user ID.

Preview

Preview

Codes

  • NII Article ID (NAID) :
    110003274269
  • NII NACSIS-CAT ID (NCID) :
    AN10541106
  • Text Lang :
    JPN
  • Article Type :
    ART
  • Databases :
    CJP  NII-ELS