本文中我們提出了一種基於Elliptical Gabor Filter的眼睛偵測方法。用Elliptical Gabor Filter的特性，調整適合的眼睛大小、建立人眼模型，來偵測眼睛。眼睛是人臉上最重要的特徵點，如何能夠快速而且精確的提出眼睛特徵是一個重要的課題，為了提高即時性以及準確性，這邊提出了一個使用Elliptical Gabor Filter的定位方法：首先統計人臉資料庫中，眼睛的比例，用來建立一個合適的Elliptical Gabor Filter使其能夠在眼睛的區域具有高反應值，再區分出左右區塊後，來定位出眼睛的位置。經過實驗顯示，此方法可以準確的找到眼睛，與其他的方法比較，結果顯示所提出的方法可以使用較少的步驟定位眼睛。
In this paper, we propose an elliptical Gabor filter-based eye detection method. Based on the size of the eyes, the parameters of elliptical Gabor filters are obtained and are applied to detect eyes. Eyes are the most important features of human face. Therefore, a fast and accurate eye location method is a very important step. To improve the execution time and accuracy, we propose an elliptical Gabor filter-based approach to detect eyes. After analyzing the proportion of eyes from the face database, we design a suitable elliptical Gabor filter to enable the center of eye with high response. After testing with face database, we show that the proposed method can accurately locate the eyes. To compare with other approaches, the proposed method needs fewer steps and the results are more accurate.
