階段式即時眼睛特徵擷取及其應用

Real-time Three-stage Eye Feature Extraction and Its Applications

周建男(Chien-Nan Chou)

指導教授 ： 洪一平

國立臺灣大學/電機資訊學院/資訊工程學研究所/碩士(2009年)

https://doi.org/10.6342/NTU.2009.01729

摘要

近年來人機互動模式的多元化發展，提供給使用者許多與以往不同的互動體驗。眼睛是人類接收與傳達資訊的重要感官之一，我們可以藉由觀察眼睛的變化來賦予使用者更貼近其需求的互動方式。本篇論文提出一個可即時偵測並追蹤使用者眼睛特徵點的方法，配合現有的低成本非紅外線視訊攝影機便可準確得到這些特徵點的位置，其中的關鍵技術在於眼睛位於人臉的位置變化差異性不大，因此能透過我們提出基於人臉結構的三階段式偵測法加快搜尋速度並提升穩定性。另一方面，由於系統使用一般的可見光視訊攝影機，不同於紅外光攝影機容易受到環境光的影響，室內、外皆可保有高度的準確性，進而確保此方法在現實生活中的簡易性與可行性。

同時，為了驗證系統的高效率與準確度，我們提出基於這些特徵點資訊的三種不同類型應用方式：一是計算虹膜中心點與眼角的距離變化判斷眼球的移動方式，配合臉部表情資訊可做為影片自動剪輯的參考依據。二是藉由計算使用者與視訊攝影機之間的相對位置來提供使用者類3D的顯示畫面。三是利用相同資訊藉由校正方法找出使用者注視在螢幕上的位置。

關鍵字

眼睛特徵擷取 ； 人機互動

並列摘要

The diversification of the development in human-computer interaction provides users more and more interactive experience which is totally different than usual. As one of the sense organs that serve as a proxy for human attention and intention, we can give users better interaction closer to their need by observing the change of the eyes. This paper proposes a low-cost system, using a non-infrared off-the-shelf webcam, which can extract and track the eye features on the face. For getting a real-time and robust result, we propose a three-stage method based on the low variety in the geometry of faces so that these eye features just need to be searched in the small eye rectangle derived from that. Also, our system works without IR lights which are easily affected by environmental ambient light, and it ensures getting high accuracy outdoors as well as indoors and being easy to set up in real life.

Meanwhile, we propose three different kinds of applications based on our eye feature tracking system in order to validate the efficiency and potentiality. First, we determine the eye movement by calculating the distance between the inner eye corner and the iris center, which will be one of the cues including facial expression for automatic video summarization. Second, we provide a view-dependent display to single user by tracking the relative location between his/her eyes and the camera. Finally, we estimate the point-of-gaze on the screen after calibrating the relation between these features and the sample dots shown on the screen.

並列關鍵字

Eye Feature Extraction ； Human-Computer Interaction

參考文獻

[2] Hansen, D. and Hansen, J. Review of Current Camera-based Eye Trackers. First Conference on Communication by Gaze Interaction, pp 7-9, 2005.

[3] A. T. Duchowski. Eye Tracking Methodology: Theory and Practice. Springer, 2007.

[4] A.Pérez, M.L.Córdoba, A.García, R.Méndez, M.L.Muñoz, J.L.Pedraza, and F.Sánchez. A precise eye-gaze detection and tracking system, in: Proc. of the 11th International Conference in Central Europe of Computer Graphics, Visualization and Computer Vision'2003, Plzen, Czech Republic, 2003.

[6] Zhiwei Zhu and Qiang Ji, Robust real-time eye detection and tracking under variable lighting conditions and various face orientations, Computer Vision and Image Understanding, vol. 98, no. 1, p.124-154, April 2005.

[7] Z. Zhu, Q. Ji, K. Fujimura, and K. Lee. Combining Kalman Filtering and Mean Shift for Real Time Eye Tracking Under Active IR Illumination," in International Conference on Pattern Recognition, Quebec, Canada, 2002.