幾何特徵之多角度人臉辨識之研究

A Multi-Angle Face Recognition Based on Geometry Features

陳得恩(Te-En Chen) ； 卓裕榮(Yu-Jung Chuo)

《嶺東學報》 39期 (2016/06) Pp. 95-118

摘要

人臉辨識在現今的生活當中被廣泛的利用，例如手機的系統解鎖、大樓門禁、機場海關的快速通關、犯罪的破案依據，好的人臉辨識系統不但快速方便，準確又實用。本論文以幾何特徵為基礎，並於實驗系統中加入多角度的人臉來進行測試，偵測是否會影響辨識之成功率。本研究先利用色彩資訊結合形態學，將人臉從非單一色彩的背景中擷取出來，找出人臉的可能區域，之後再利用圓形檢測法找出眼睛於人臉區域中的位置，然後藉由生物特性找出嘴巴與鼻子的相對位置，再將嘴巴與鼻子的特徵點擷取出來，分別加權、正規化投影至特徵空間完成訓練與辨識的前置作業。結果證明，以幾何特徵為基礎的人臉辨識系統，在加入多角度的人臉辨識樣本後，不但不會影響系統在人臉辨識上的辨識成功率，反而有助於增加人臉辨識的成功率。

關鍵字

人臉偵測 ； 人臉辨識 ； 幾何特徵 ； 多角度

並列摘要

Nowadays, the face recognition is widely used in daily life such as the application on the unlock system of the cellphones, the access control system of the buildings, the E-Gate in the airport, and the basis to solve a case. A good face recognition system did not depend only upon the factor of fast and convenience, but also more important in precision. This study recognizes the face that based on the theory of the geometry features. Further, we also included the face in profile into the experiment to study the effect for the successful rate. In this investigation, we capture the face from complex background by using the methods of color information and morphology. After that, we positioned the eyes by using the circular detector and then to find out the specific points of nose and mouth in accordance with the biological character. Moreover, we used the weighted function and normalization method to project onto a feature vector space to recognize a particular face. However, the experimental result shows an outperformance by adding the face in profile to the recognized training system.

並列關鍵字

Face Detection ； Face Recognition ； Geometry Features ； Multi-angle

參考文獻

林詠章、黃明祥(2014)。資訊與網路安全概論：看見比特幣。台北市:麥格羅希爾。

楊煒達(2007)。簡易方法之少量人臉辨識系統。國立中央大學資訊工程研究所。

Poynton, C.A.(Ed.)(1996).A Technical Introduction to Digital Video.New York:John Wiley & Sons.

Kukharev, G.,Novosielski, A.(2004).Visitor Identification - Elaborating Real Time Face Recognition System.Proc. 12th Winter School on Computer Graphics (WSCG).(Proc. 12th Winter School on Computer Graphics (WSCG)).

Canny, J.(1986).A Computational Approach to Edge Detection.IEEE Transactions on Pattern Analysis and Machine Intelligence.8(6),679-698.