多人臉影像視覺辨識技術

The Technique of Image-Vision Recognition of Multiple Faces

吳明芳(Ming-Fang Wu) ； 李振興(Jen-Hsing Li) ； 王炳聰(Ping-Tsung Wang) ； 詹慧珊(Hui-Shan Chan) ； 黃建邦(Chien-Pang Huang)

《崑山科技大學學報》 8期 (2011/06) Pp. 75-95

https://doi.org/10.29922/JKSUT.201106.0007

摘要

隨著科技的發展，舉凡門禁管制系統、金融管理、犯罪偵查與電腦認證等等，都是需要借助一套強而有力的身份確認系統。目前有關人臉辨識或偵測的研究已有許多，但多為單人辨識技術或多人人臉偵測方法。至於多人人臉的辨識技術仍有許多發展的空間，所以本研究目的在發展結合單人人臉辨識技術及多人人臉偵測方法的多人人臉辨識系統，此類系統特別有利於犯罪的偵查。做法上，本研究實際應用臉部特徵點處理的技術於多人人臉場合的影像中，發展以臉部特徵為主的多人人臉身份確認系統。此系統利用各種不同的基本影像處理方法自動找出特徵點的座標後，計算出其所對應的特徵向量組合。在比對方面，本研究利用歐幾里德距離和漢明距離之差異值計算法來比對特徵向量間的相似程度，並考慮人臉偏轉情形進行修正，以求較高的辨識率。完成理論的推演後，本論文最後以實際單人及多人合照影像進行實驗，以驗證本研究所提之身份確認系統及改善辨識率方法的實際功效。

關鍵字

視覺技術 ； 特徵 ； 多人臉 ； 辨識

並列摘要

Along the development of modern technology, the entrance control system, financial management, criminal detection, and computer certification etc., which all require a set of strong identity verification system. Although many studies have been proposed for identity verification or detection of face, but most current study only concentrate on the identity verification technology of single face or the detecting method of multiple faces. There are still a lot of development-space of identity verification technology of multiple faces. So the purpose of this thesis is to develop a system for the recognition of multiple faces by combining both the single face recognizing technology and the multiple faces detection method. And this system is especially favorable to the investigation of the crime.On the method, this study use the processing technology of face features to deal with the images with multiple faces, and to develop a set of identity verification system that relying mainly on face features. In this system, the basic image processing techniques are used to find out the coordinates of feature points automatically, and then to calculate their combination of corresponding eigenvectors. On the comparison aspect, this study has made the use of difference value calculation methods including Euclidean distance and Hamming distance to compare and to check the degree of similarity among the eigenvector to achieve the goal of identification. This study also considers the correction of face's angle to upgrade the identification rate. After the completion of the theoretical inference, this study, in addition to use practical single and multiple people images for experiments to test and verify the practical efficiency of the multi-face identity-verification system proposed in this thesis.

並列關鍵字

vision technique ； feature ； multiple faces ； recognition