【摘要】 本文提出了一种综合多个颜色空间分量的肤色特征提取方法,并通过SVM分类器进行肤色和非肤色的分类,从而实现肤色检测。特征提取先后采用了PFA和KPCA算法。肤色检测的实质是肤色和非肤色分类问题。针对先前提取的特征,采用基于SVM分类器进行分类。实验结果表明,基于PFA、KPCA特征提取和SVM分类的肤色检测正确率可以达到87.76%,误判率仅为14.62%。更多还原

【Abstract】 This paper proposes a new method for skin color detection with the combination of different components from many color spaces. First, features are successively extracted with the PFA and KPCA algorithms, and then SVM is employed to classify the skin and non-skin colors, as the skin detection is essentially the classification problem between the skin and non-skin colors. The experimental results show that the correct rate of skin color detection can reach 87.76%, while the false alarm value is only 14.62% based on the PFA and KPCA feature extraction methods and the SVM classification algorithm.更多还原