A Real-Time Eye Gaze Tracking Technique

[[abstract]]本論文主要目的在探討如何以簡易的設備完成人眼之視線追蹤，以改良傳統視線追蹤成本過高的問題，本系統不僅成本低廉，且不須要任何額外之配戴裝置，甚至僅需一臺含視訊裝置之筆記型電腦即可使用，可攜帶性極高！此套系統可估算出人眼凝視的方向及位置，可應用於醫護，軍事，娛樂等。 在傳統的視線追蹤方式，不外乎利用光線反射原理、配戴視線追蹤裝置或眼動圖資訊(electro-oculogram，EOG)等，不僅所需成本較高，且有些光線反射裝置更可能對人眼造成侵入式的傷害。本論文基於低成本、不具侵略性之前提下，使用一部web camera偵測裝置就能運作。 本論文提出的視線追蹤演算法概分為三大部份，首先，使用YCbCr色彩空間將人臉的色彩資訊切割出來，以便快速定位出人臉位置，接著運用樣本比對的方式找出人眼粗略部位，此時我們採用擴張方格的方式，以便能快速的搜尋到人眼位置，進而利用垂直投影與水平投影資訊推論得知人眼凝視的方向。[[abstract]]The objective of thesis is to present a set techniques integrated into a low-cost eye gaze tracking system. An eye gaze tracking system is a system that can estimate the direction and a position of the human eye gaze. It can be used in medical, military and entertainment. There are several ways of tracking the direction of the eye-gaze by using reflection of light, electro-oculogram (EOG), or contact lens, etc. Each of them has its own advantages and disadvantages such as high complexity, high cost, and invasive way. Price and convenience are two important considerations, and therefore we proposed the theory not just only easy to use but also install a web camera device it can be operated. In this thesis three-stage algorithm is proposed to estimate the eye gaze direction. At the first stage, an efficient face detection based on PCA Feature is employed to locate the user’s face. Then use template matching to detect the human eye. Last, we used horizontal projection and vertical projection to identify where the eye looking at.[[note]]碩