人臉影像轉正應用於性別辨識與年齡預測之研究

Face Pose Frontalization apply to Gender Classification and Age Estimation

黃俊翔(Chun-Hsiang Huang)

指導教授 ： 王才沛

國立交通大學/資訊學院/資訊科學與工程研究所/碩士(2017年)

摘要

在人臉識別領域中，性別與年齡是很常見的議題，目前也已經有非常多的相關技術與理論可以參考，但大部分的實驗通常會使用大頭照影像這種外在環境與姿勢影響較小的影像資料集，本文將利用貼近現實生活的影像進行性別與年齡預測，希望能透過前處理將外在環境與姿勢的影響縮小已達到好的辨識效果。
本文將利用網路視訊串流擷取出較符合真實狀態下的人臉影像，此些擷取之人臉可能會有不同的角度或遮蔽，本論文將利用轉正方法將每個擷取之人臉姿勢轉正化至正臉，並且比較轉正後與轉正前之特徵應用在性別辨識與年齡值或年齡族群預測之效果與影響。本文使用的特徵包含像素值、梯度直方圖等邊界特徵，並利用特徵相似度搭配KNN與SVM和SVR三種方式比較轉正前後特徵的分類效果。最終我們可以發現在年齡部分轉正步驟在各個特徵下都有正面效果，而性別部分則並非全部特徵有效。

關鍵字

人臉轉正 ； 性別辨識 ； 年齡預測

並列摘要

In the research of human face recognition, gender and age recognition are very common issue. There are many related technology and acknowledge about gender and age recognition for reference. However, those experiment usually use bust shot or head shot as their dataset to reduce external influences of light or pose. In this paper, a face dataset which is more realistic will be used. Our dataset’s image was extracted from web stream, which containing light and pose difference problem.
In our experiment, we apply a preprocessing step to our face image before feature extraction. This step will change the pose angle of face and make it looks like front face. Next step, we extract features from frontal face and original face(without preprocessing). We will use feature similarity with KNN, SVM and SVR in these two kinds of feature to get their gender and age result. At last, we compare the frontal face result with original face result and we found that the frontal step have positive influence in age and gender recognition.

並列關鍵字

face gender recognition ； age estimation ； frontal face

參考文獻

[1]Lu, L. and P. Shi. A novel fusion-based method for expression-invariant gender classification. in Acoustics, Speech and Signal Processing, 2009. ICASSP 2009. IEEE International Conference on. 2009. IEEE.

[2]Li, B., X.-C. Lian, and B.-L. Lu, Gender classification by combining clothing, hair and facial component classifiers. Neurocomputing, 2012. 76(1): p. 18-27.

[3]Phillips, P.J., et al., The FERET database and evaluation procedure for face-recognition algorithms. Image and vision computing, 1998. 16(5): p. 295-306.

[4]Ojala, T., M. Pietikainen, and T. Maenpaa, Multiresolution gray-scale and rotation invariant texture classification with local binary patterns. IEEE Transactions on pattern analysis and machine intelligence, 2002. 24(7): p. 971-987.

[5]Zheng, J. and B.-L. Lu, A support vector machine classifier with automatic confidence and its application to gender classification. Neurocomputing, 2011. 74(11): p. 1926-1935.