漫畫以藝術形式的發展在世界各地出現已經有上千年的歷史,而現代漫畫從正式出現到現在有百餘年的時間,在近50年間,不論是在電影、電玩等娛樂產業中,佔有相當重要的地位。而在近年,已經有不少以漫畫風格為形式的非寫實算圖(NPR)的研究,但直至目前的研究,許多僅只於預先算圖的研究,大都是先建好的面具模型,將面具模型與由臉部追蹤結果所取得的參數加以整合,並將控制點的資料輸入已建好的面具模型,予以合成並繪製出目標對象的面具表情,皆無法即時的繪製出依所追蹤的人臉的特性合成的漫畫風格肖像畫。而相關於臉部偵測(Facial Detection)與追蹤(Facial Tracking)的技術,在現今已經相當成熟,已經可以達到即時處裡的效果,在近年來皆已被廣泛應用在如安全監控系統及數位相機等裝置上。本文研究的目的,便是結合臉部偵測以及臉部追蹤技術,配合深度影像(Depth Map)的輔助,實踐即時的漫畫風格NPR效果。
Comic has been developed thousands years by the art history in the whole world, it is very important whatever in the movie or game industry recent fifty years. There are many research about the comic styling NPR now, but a lot of these are pre-Rendering with the modules which are prepared in advance, combine the modules and the parameters which extract from the result of facial tracking, generate a NPR face from destination target, most of all couldn't generate a real-time comic styling video. But the technology about Facial Detection and Facial Tracking has developed completely now, both of them can operate immediately, the technology had applied to the Security System, Digital Camera and others. The goal of the research is, using the Facial Detection and Facial Tracking technology, coordinate with Depth Map to build a real-time comic styling NPR.