融合DCT系数和生物特征的年龄估计方法.pdfVIP

融合DCT系数和生物特征的年龄估计方法 1 1* 1 庞秀梅 ，马丙鹏 ，王天江 1. 华中科技大学计算机科学与技术学院，湖北省武汉市洪山区珞喻路 1037 号，430074 E-mail: {pangxm,bpma,tjwang}@ *：通讯作者 摘 要: 本文提出了一种能够有效提高年龄估计准确率的新方法。该方法以“HMAX ”生物特征模型为基础，并 融合了DCT信息。该方法首先利用Gabor滤波器组提取人脸图像中的生物基本特征，并对图像中的不同区域进行DCT 变换，取DCT系数作为特征。本文方法的理论基础在于，DCT系数包含了空间频率信息，而人眼对不同频率的信息 敏感程度也不同，因此选择能量比率较大的DCT系数作为特征，可以有效提高年龄估计的识别率。为验证该方法的 有效性，我们在公开的FG-NET年龄数据库上进行实验，并和其他方法做了对比。实验结果表明，我们的方法能够 有效的提高年龄估计的准确率。 关 键 词: 年龄估计；DCT系数；生物特征；“HMAX ”模型 Age Estimation Using Bio-inspired Features and DCT Coefficients 1 1* Xiumei Pang , Bingpeng Ma , Tianjiang Wang 1. School of Computer Science and Technology, Huzzhong University of Science and Technology, 1037 Luoyu Road, Wuhan, China, 430074 E-mail: {pangxm,bpma,tjwang}@ *：Corresponding author Abstract: This paper proposed a novel method which can improve the accuracy of human age estimation greatly. The proposed method was based on “HMAX” model and merges the information of DCT. First, Gabor filters were applied to extract the biological inspiration features. Second, zero mean-variance was applied on the picescs to obtain the responsed results. Finally, DCT transform was applied on the responsed results and the DCT coefficients was the output of the method. There were two reasons for us to propose the method: first, DCT coefficients contained the frequency information; Second, humen were sensitive to the information of different frequencies. To show the effectiveness of the proposed method, we done a lot of experiements on the public FG-NET age database and compared the results with other methods. The results shown that the proposed method can