低分辨率人耳图像识别方法研究*

针对人耳识别中存储量和计算速度的要求，同时考虑远距离拍摄时低分辨率人耳识别问题，探讨了低分辨率人耳图像识别性能，给出了分辨率与识别率的关系。首先采用高斯金字塔对人耳图像进行不同层的分解，然后对每一层图像应用广义判别分析方法（GDA）提取特征，最后计算样本间的余弦距离，通过阈值法分类识别。实验结果表明，当人耳图像分辨率降低为36×24时系统识别性能最好，满足实时生物识别系统的要求。

Study of low-resolution human ear image recognition

This paper discussed the low-resolution image recognition performance of the human ear and deduced the relationship between resolution and recognition rate in view of the requirements for memory capacity and computing speed in ear recognition and taking account of the low-resolution ear recognition problem at the long-range shooting. It first broke up the human ear images into different layers by using Gaussian pyramid, and then extracted the features of each layer on generalized discriminant analysis (GDA). Finally, calculated the cosine distance between the test samples and made classified recognition by the threshold value. The experimental results show that the system recognition performance is the best when the human ear image reduces to 36 × 24 and the requirements for a real-time biometric identification system are met.