摘要： 应用独立分量分析（ICA）方法从高阶统计相关性角度出发提取人耳图像的特征变量，并采用基于欧氏距离测度的最近距离分类器进行人耳图像的识别。与传统的主成分分析（PCA）方法相比具有更好的鉴别能力。通过与PCA的对比实验结果表明，该方法具有更高的识别率，对姿态和光照的变化也具有较好的鲁棒性。

关键词: 人耳识别, 独立分量分析, 主成分分析, 最近距离分类器

Abstract: This paper puts forward a new method of ear recognition. In this method, independent component analysis(ICA) which considers the high-order statistics and obtains the inner character of images, is applied to extract ear feature, and then adopts the nearest distance classification based on Euclidean distance to recognize the ear image. Compared with principal component analysis(PCA), the ICA has a better identify ability. The result of experiments shows that the ICA algorithm is more efficient and has better robust to illumination and pose variation.

Key words: Ear recognition, Independent component analysis(ICA), Principal component analysis(PCA), Nearest distance classification