基於賈伯小波之自動人臉年齡偵測

Abstract

近年來，年齡偵測在人臉辨識技術中，漸漸成為一項受重視的領域，加上科技的發達，年齡偵測被認為在多媒體通訊、人機介面、居家照護及安全監控等應用上有相當大的發展潛力。在本篇論文中，我們提出一個創新與可靠的自動化人臉年齡偵測架構。本篇論文提出藉由結合賈伯小波與正交區域保留投影(Orthogonal Locality Preserving Projections)，產生適用於年齡辨識的新人臉特徵。接下來，在考量到同一個體各年齡層的影像資料蒐集的難易度以及現行公開之年齡資料庫的資料稀少，為了獲得更佳的一般性，我們採用基於支持向量機(Support Vector Machines)的分類器結合新人臉特徵。此系統可以全自動即時提取人臉特徵，相較於大多數文獻所提出的半自動人臉特徵提取，更具有實際應用的潛力。 本篇論文的目標是建構一個全自動且即時分析的年齡偵測系統，由此系統所得到的實驗結果將可以提供年齡偵測領域上更深入的了解，並且有助於實際應用的開發。

In recent years, age estimation has become an important research topic in face recognition technology. Furthermore, age estimation is considered as a potential research which has lots of real-world potential applications such as multimedia communication, human computer interaction, and security. In this thesis, we present a novel and reliable framework for automatic age estimation. It exploits the whole new face feature based on the combination of Gabor wavelets and Orthogonal Locality Preserving Projections. In order to obtain more proper generalization ability with respect to sparse training samples, we use a support vector machine based classifier. Since this system can extract face aging features automatically in real-time, it has more potential in applications than other semi-automatic systems. The objective of this paper is to build a full automatic and real time age estimation system. The results obtained from this novel approach would provide better insight to operators in the field of age estimation to develop the real-world applications