本論文主要描述於使用電腦硬體上,以被動方式透過網路攝影機擷取影像,根據影像提供的資訊加以處理及分析,針對擷取影像後的膚色偵測求得臉部的所在位置,並且透過眼球偵測及雜訊過濾後的估算,以提供電腦使用者偵測眼睛獲得的資訊,達到預期的結果。 膚色偵測方面採用YCbCr色彩空間為基礎針對膚色做擷取,然後使用二值化加強影像,最後將保留下來的膚色使用水平值方圖及垂直值方圖的統計計算後,來描繪出人臉影像的所在位置。 眼睛偵測則是採用色塊分割的方式,先取得人臉上半部臉部份影像分析處理後進行兩次色塊分割再將所判別顏色為最深的色塊保留下來,進一步對保留下來的色塊進行雜訊處理及面積過濾,以獲得最後眼睛瞳孔的所在位置。
This thesis mainly describes a passively external network camera capture scheme. According to the information provided using image processing and analysis, we utilize Skin Color Detection to obtain the face position of the participants from scheme. Moreover, to achieve the predicted results, we comput the eyes from the imformation via Eyes Detection and Noise Filtering which can provide us to acquire the imformation. We utilize the Skin Color Detection based on YCbCr Color Space. Furthermore, we use Binary to enhance skin color image maintainence, then compute the vertical and horizontal pixels of histgram to obtain the description of human face position. Eyes Detection utilizes the method of Region Segmentation. To achieve the final position of eyes, first of all, we analyze upside face image acquired, maintain the darkest region after twice operations of Four Level Segmentations.