【摘要】 提出一种基于Gabor小波和改进的广义K-L变换的掌纹识别方法。该方法首先对测试样本的掌纹ROI灰度图像进行Gabor小波变换,得到其Gabor特征向量,然后利用改进的广义K-L变换方法将高维特征向量变换到低维空间,最后将得到的低维特征向量利用欧氏距离法与训练样本库中的特征向量作匹配识别。该方法首次将基于时频变换的特征提取算法与基于子空间的特征提取算法结合起来,既充分利用了Gabor函数优良的特征提取性能,又有效解决了高维特征的降维处理问题。通过使用自行采集的数据库对该方法作对比实验,获得了94%的识别率,验证了该方法的有效性和准确性。更多还原

【Abstract】 This paper presented a novel method of palmprint feature extraction using Gabor wavelet and improved generalized K-L transform.The new method made use of Gabor transform to obtain Gabor eigenvector of palmprint from palmprint ROI image in testing samples,then used the improved generalized K-L transform to extract the master components and reduced dimensions of Gabor features.Finally,matched the features with other features in training samples to realize recognition.The method integrated the feature extraction algorithm based on time-to-frequency transform with the algorithm based on subspace transform,and made full use of Gabor function’s fine capability of feature extraction and also effectively solved the problem of feature’s high dimensions,the 94% recognition rate shows the method’s effectiveness.更多还原