中文摘要：

为了从Gabor滤波后的图像中提取简单有效、区分力强的人脸特征,提出了一种基于可变长起主导作用特征（VLDF）的人脸识别算法.即首先人脸图像与不同尺度、不同方向的Gabor滤波器进行卷积运算,然后利用局部二元模式（LBP）算子提取滤波输出的纹理特征,并根据纹理特征的统计分布规律,采用数量可变的起主导作用的纹理模式作为人脸的VLDF特征.最后构造了VLDF人脸特征之间距离的计算方法.该算法具有较小的特征向量维数和高的rank-1识别率.在FERET人脸数据库上的仿真结果验证了算法的高效性.

英文摘要：

To extract compact and distinctive information from images filtered by Gabor kernels, we propose a variable length dominant feature （VLDF） based algorithm for face recognition. Specifically, the face image is first convolved with Gabor fil- ters of different scales and different orientations, and then the texture information is extracted from the Gabor responses using local binary pattem（LBP）. Based on the statistical characteristics of the pattern distribution, the variable length dominant feature （VLDF） is derived. Finally, the distance between two VLDFs is constructed. VLDF has a lower feature vector dirnensionality and a higher rank-1 recognition rate. The experiments on FEREr database verify the high efficiency of the proposed algorithm.