建立卡通化人臉模組在電腦圖學領域中是一個有趣並且重要的研究題材,然而大部份被提出來的卡通化的做法在於將照片轉換為具有特定風格的NPR (Non-Photorealistic)圖片,或是將較具像的擬真3D模型形變成具有肖像畫風格的3D模型。 在這篇論文中,我們提出一個不同的方法來將2D的臉部照片產生出3D的卡通化模型,這個模型將可以作為在社交網站或是遊戲的虛擬替身(Avatar) ,或是運用於製作客製化的Q版公仔模型。 為了達成上述的目標,我們使用了一個評分樹來分析與表現人臉的重要特徵,並且使用了DXUT(The DirectX Utility Library)架構來開發出使用者介面,然後以Open CV(Open Source Computer Vision Library)這套電腦視覺函式庫來自動擷取基本的臉部特徵點,再經由簡化與卡通化原本具象模型的方式來產生一個生動且另人印象深刻的臉部3D NPR模型。
Modeling 3D caricaturing face is an interesting and important issue in computer graphics. However, most of the method propose for caricaturing face focus on translating a photograph to a NPR(Non-Photorealistic Rendering) image or deforming of a realistic 3D Model to a caricaturing one. In this paper, we propose a different approach to build 3D cartoonized face model form 2D photograph(s). This 3D model can be used as an avatar in social web sites or games. It can also be used to make user-customized cute figurines. To achieve the above goal, we adopt an evaluation tree to analyze and characterize important features of face. In addition, we use the DXUT (The DirectX Utility Library) framework to develop a user interface and adopt the computer vision library – Open CV (Open Source Computer Vision Library) to detect the basic facial features automatically. Then we simplify and cartoonize the realistic model to create a vivid and expressive 3D NPR face model.