• 作者： 黃俊雄
• 作者服務機構： 中央研究院資訊科學研究所
• 中文摘要： 一個數位化的二維形狀的重建是對物體辨認、字體自動產生、藝術工作等具有關鍵性的，雖然經過數位化後一些細節損失掉了。重建是需要對形狀的適當描述，而依不同的描述可分為概略重建及精細重建。本文在此給二維形狀之重建做一縱覽，此包含概略及精細重建的仔細討論。並對一些不對，不可靠或雜訊敏感的方法做些評論。對那些可靠實用的方法此文給予簡單及清楚的解釋。最後本文給一綜合性討論並指出改進目前方法的新研究方向。
• 英文摘要： The reconsturction of a digitized 2D shape, where some detail information is lost after digitization,is crucial to object recognition, automatic font generation, artist work, etc. The reconstruction requiresa description of the shape and can be classified into rough reconstruction and precise reconstruction,according to different descriptions such as the Fourier Descriptor, Circular Autoregressive Model, Para-metric Curve Fitting, etc. In this paper, a survey of 2D shape reconstruction methods is given, where bothrough and precise reconstruction methods are discussed in detail. Some comments are also given on thosemethods that are not correct, are unreliable, or are noise sensitive. For those practical methods, thissurvey gives brief and clear explanations. Finally, a discussion is given, and the new directions of researchto improve the current methods are indicated.
• 中文關鍵字： shape reconstruction; rough reconstruction; precise reconstruction; corner detection
• 英文關鍵字： --