彩色視覺技術於臉型判別及膚質分析上的應用

The application of color vision technique on face-shape decision and skin-character analysis

林祈帆(Chi-fan Lin)

指導教授 ： 吳明芳 ； 張永昌

崑山科技大學/工程學院/電機工程研究所/博士(2005年)

摘要

隨著科技的進步，影像擷取設備及周邊產品在弁鄐W日新月異及價格下降；而電腦運算能力提昇以及影像處理技術的快速進展，使得影像處理這門領域有了很廣泛的研究及使用。
本論文目的在應用彩色視覺技術於臉型判別及膚質分析。由於美容造型師在判斷人臉臉型以及皮膚膚質通常都是靠感官直覺及經驗累積，較無效率，所以本論文以能提供美容造型師精準分辨臉型及膚質為目的，發展出一套判別臉型以及膚質分析的視覺處理系統。此系統在臉型判別方面，利用臨界值法、影像外形處理與影像投影等基本影像處理自動找出特徵點的座標後，計算相對應的特徵向量組合，並依臉型長與寬之比例關係找出其對應的權值。在膚質分析方面，則利用小波係數將皮膚影像三角化的技術，計算出三角化後的三角形個數以及重要節點數。在比對方面，依系統所找出的特徵值將可以有效代表臉部特徵或是膚質特徵，接著利用歐基里德距離及漢明距離來完成比對工作，也就可以完成臉型判別及膚質分析的結果。本論文最後以實際人臉及皮膚影像進行實驗，其結果顯示不論在臉型判別或膚質分析上皆有不錯的辨識率。

關鍵字

膚質 ； 特徵 ； 影像處理 ； 臉型

並列摘要

With the progress of science and technology, the function of image grabbed equipments and peripheral products is changed with each passing day, so their prices dropped. And the promotion of computer operating ability and the fast progress of image processing technique make the field of image has very extensive research and application.
This thesis purpose is to use the application of color vision technique on face-shape decision and skin-character analysis. Because the judgment of face shape and skin quality by cosmetic modeling teacher is usually depended on organ intuition and experience, which are relatively inefficient. So in order to provide the accurate recognition of face shape and skin quality, this thesis develops a vision processing system for face-shape decision and skin-character analysis. In the face-shape decision of this system, the basic image processing techniques including thresholding method, image form treatment and image projection etc., are first used to find out the coordinates of feature points automatically, and then their combination of corresponding eigenvectors can be calculated as features, and its corresponding weights can be found out by using the ratio of length and breadth of face shape. In the aspect of skin-character analysis, the number of triangles and the number of major nodes are found as features by the triangulating technique of skin image using wavelet coefficient. In the aspect of comparison between features, both Euclidean distance and Hamming distance are used to compare face features or skin features, and then the results of face-shape decision and skin-character analysis can be obtained. Finally, this thesis carries out the experiment with real face and skin images, its result shows that there is good distinguishing rate no matter on the face-shape decision or the skin-character analysis.

並列關鍵字

feature ； face shape ； image processing ； skin quality

參考文獻

[5]R. C. Gonzalez and R. E. Woods, “Digital image processing,” 2nd Edition, Prentice Hall, 2003.

[6]R. Jain, R. Kasturi and B. G. Schunck, “Machine Vision,” 1995, pp.35-36.

[7]林咸仁，“改良現性鑑別式分析在少量訓練樣本下之人臉辨識研究”，碩士論文，國立成功大學資訊工程研究所，2002。

[10]C. Liu, and H. Wechsler, “Gabor Feature Based Classification Using the Enhanced Fisher Linear Discriminant Model for face Recogition,” Proc. of IEEE Trans. on Image Processoing, Vol.11, pp467-476, 2002.

[11]C. Garcia and G. Tziritas, “Face detection using quantized skin color regions merging and wavelet packet analysis,” IEEE Transactions on Multimedia, Vol.1, No.3, Sept.1999, pp.264-277.