- 學位論文
結合主動形狀模型與色彩資訊的人臉識別系統
A Face-Recognition System That Combines Active Shape Models and Color Information
摘要
身份驗證的需求在近年來越來越殷切，而相較於其他的生物辨識方式，如虹膜識別、指紋辨識、手掌靜脈辨識等，人臉識別具有自然且非接觸式的優點，並且可以同時進行多個使用者的身份識別。但是人臉識別技術也被視為生物辨識中最困難的研究課題，因為不同的人其人臉特徵相似度很高，這樣的特點雖有助於人臉偵測，但對於人臉識別來說卻是一大困擾，此外，人臉特徵會因為頭部姿勢、臉部表情、環境明亮度等多方面因素而有極大的變化。 我們認為除了五官特徵關係之外，顏色資訊也是一個辨識人臉的重要依據，而且顏色資訊也不需要透過複雜的運算求得，因此本論文使用主動形狀模型(Active Shape Model)擷取人臉特徵，除了特徵點上的顏色之外，我們還抓取臉頰區域的顏色特徵。經由實驗結果得知，本論文所提出的演算法，相較於原來的主動形狀模型，可提升約4~5%的辨識成功率。
並列摘要
In recent years, authentication demand becomes more and more eager. Face recognition has the advantage of natural and contactless than other biometric such as iris, fingerprint, or palm vein, etc. Furthermore, face recognition can identifies multiple users at the same time. But face recognition technology was regarded as the hardest research on biometrics. Because similarity degree of facial features is higher on different people, though it is great for face detection, but it is vexation a for face recognition. In addition, facial features have considerable variation because head pose change, expression change, and environment brightness change. We think color information also is important element of recognize face besides facial features. And color information does not complexity operating. Hence, this paper uses Active Shape Model extract facial features, we capture color features on cheek besides color on the positions of feature points. Experimental result shows that the proposed method promotes recognition success rate is about 4~5% than original ASM.