融合Gabor小波能量与方向特征的掌纹识别算法

Palmprint identification algorithm based on energy and direction feature fusion of Gabor wavelet

提出了一种融合Gabor小波能量与方向两级特征的掌纹识别算法,有效地解决了纹线的方向特征难以描述的困难。首先对预处理后的掌纹进行不同方向不同尺度的Gabor小波多通道滤波,再对得到的各个方向子带进行模糊分块,计算每一小块区域的能量,在同一尺度不同方向上找出能量最大的区域,并标记此区域对应的Gabor方向。采用格雷码对方向特征编码,再分别计算能量特征的欧氏距离和方向特征的汉明距离,最后将2种距离按权相加进行掌纹识别。与利用单一能量或方向特征的算法相比,本文提出的算法具有较高的检索识别率。

A palmprint identification algorithm based on energy and direction feature fusion of Gabor wavelet is proposed, which solves the difficulty of describing the direction feature efficiently. First, a set of Gabor wavelet is applied to filter the preprocessed palmprint in different scales and different directions, and the obtained sub-bands are divided into fuzzy blocks. Then the energy of each small block is calculated, the small block with maximum energy in the same scale and different directions is found, and the Gabor filter direction of the block with maximum energy is marked. Gray code is used to encode the direction feature, and the Eculidean distance of energy feature and the Hamming distance of direction feature are calculated respectively. Finally these two distances are fused according to their weights to identify the palmprint. Compared with the methods of using only single energy or direction feature, the algorithm proposed in this paper has higher retrieval identification rate.