Global feature based female facial beauty decision system

This paper presents an automated female facial beauty decision system based on Support Vector Machine (SVM). First, we constructed manually two classes of female faces with respect to their facial beauty, by requesting personal opinions of people. As the second step, Principal Components Analysis (PCA) and Kernel PCA(KPCA) were applied to each class for extracting principal features of beauty. Support Vector Machine (SVM) was used for judging whether a given face is beautiful or not. Since judging the beauty is subjective, the decision results of our system were evaluated by comparing the system generated decision results with the corresponding ones made by the persons. Based on this criteria, our results showed that KPCA with a success ratio of 89% outperformed PCA with a success ratio of 83%.