基於人工蜂群演算法和K-means之人臉辨識

An Algorithm of Face Recognition Based on Artificial Bee Colony Algorithm and K-means

葉宣伯(Xuan-Bo Ye)

指導教授 ： 涂世雄

中原大學/工學院/電機工程研究所/碩士(2014年)

https://doi.org/10.6840/cycu201400666

摘要

在本論文中，我們提出基於人工蜂群演算法和K-means法的人臉辨識方法，主要目的是希望透過我們所提出的系統，能快速的偵測並框選出人臉。在人臉辨識中使用此方法可以有效提升辨識率。

在本篇論文中人臉辨識系統中我們提出系統的架構、方法和實驗結果。首先我們介紹系統架構，它包含了影像前處理、人臉偵測、特徵擷取和人臉辨識。在本研究的方法中我們提出四個步驟。(1)臉部偵測中，我們先將影像作灰階轉換及直方圖等化，可以簡化影像的資料量及光源的影響。(2)利用AdaBoost法可以有效的過濾非人臉的影像並找出人臉區域。我們將框出的人臉做圖像正規化。(3)在特徵擷取方面是利用主成分分析保留變化大的影像特徵。(4)而在人臉辨識的部分，我們提出人工蜂群演算法結合K-means法來辨識人臉身分。利用人工蜂群演算法可以改善K-means易陷入局部極小值和穩定性差的缺點，以提高K-means的強健性和人臉辨識率。最後，我們使用Matlab來進行實驗的模擬。本實驗與其他方法比較，我們提出的人工蜂群演算法和K-means法的人臉辨識方法可以有效並能提高辨識率。實驗的結果證實本論文所提出的人臉辨識系統可行性。

本論文中我們主要的貢獻有下列幾點：
1. 改善K-means陷入局部極小值和穩定性差的缺點
2. 比較其他方法，本論文提出的方法可以提升人臉辨識率
3. 本人臉辨識系統能夠用於門禁系統、犯罪偵查、3C用品等用途。

關鍵字

人臉辨識 ； K均值 ； 人工蜂群演算法

並列摘要

In this thesis, we propose a face recognition method based on artificial bee colony algorithm and K-means. The main purpose is to hope that through this face recognition system, the face can be detected rapidly and framed up. We use this system to promote effectively the recognition rate of face recognition.

In this thesis, we will present the architecture, methods and experimental results in our system. At first, we introduce our system architecture including image preprocessing, face detection, feature extraction and face recognition. Then we present the methods used in our research. They can be given into the following four steps. (1) We convert the color image into grayscale image and use histogram equalization method to adjust contrast. It can simplify the amount of image data and reduce the effect of light. (2) We can availably filtrate non-face images and find out the face area by AdaBoost method and use the image normalization to normalize the face part. (3) We can effectively reduce dimensions of image and retain the large variation of image features by using Principal Component Analysis (PCA) in feature extraction. (4) We combine Artificial Bee Colony algorithm combine with K-means to do face recognition. The ABC-K-means method can improve the weakness of local minimum and poor stability. This method can raise the robustness of K-means and face recognition rate. Finally, we use Matlab to simulate the proposed face recognition system and take ours method to compare with other methods mutually. The experiments prove that the face recognition system in this thesis can effectively increase recognition rate and the feasibility of the face recognition system in this thesis.

In this thesis, the contributions of our research are as follows:
1. Artificial Bee Colony algorithm improves the weakness of local minimum and poor stability in K-means
2. The ABC-K-means method improves the face recognition rate to be better than other methods.
3. Our face recognition system can be used for access control systems, criminal investigation, 3C products, etc.

並列關鍵字

K-means ； Artificial Bee Colony Algorithm ； face recognition

參考文獻

[53] 林泳輝。基於螞蟻最佳化和K-means方法之人臉辨識。中原大學電機工程研究所碩士論文，2013。

[1] A.F. Abate, M. Nappi, D. Riccio and G. Sabatino. 2D and 3D face recognition: A survey. Pattern Recognition Letters, vol. 28, on.14, pp.1885-1906, Jan. 2007.

[2] T.Y. Jea and V. Govindaraju. Aminutia-based partial fingerprint recognition system. Pattern Recognition, vol. 38, on.10, pp.1672-1684, 2005.

[3] S. Attarchi, K. Faez and A. Asghagi. A fast and accurate iris recognition method using the complex inversion map and 2DPCA. Proc. of the IEEE Int. Conf. on 2008 Computer and Information Science, pp.179-184, Apr. 2008.

[6] Yang, G. and T. S. Huang. Human face detection in complex background. Pattern Recognition, 27(1), 53-63, 1994.