基于DCT鉴别分析的掌纹特征提取

离散余玄变换是一种经典的图像处理技术，而鉴别分析是一种常用的图像特征提取技术。本文将这两种技术有机地结合起来，提出了一种新的掌纹特征提取方法。该方法首先对于掌纹的离散余玄变换图像，提出了一个二维可分性判据来选择具有良好可分性的频段;然后提出了一种改进的费舍脸方法来提取鉴别特征。在掌纹图象公共数据库上的实验结果验证了本文所提出的方法的有效性。

Palmprint Feature Extraction based on DCT Discriminant Analysis

The discrete cosine transform is a classical image processing technique, and the discriminant analysis is a commonly used image feature extraction technique. These two techniques are effectively combined in the paper, and a new palmprint feature extraction approach is proposed. After performing discrete cosine transform on the palmprint images, the proposed approach first uses a two-dimensional separability judgment to select the frequency bands with favorable separability. Then from the selected bands, it extracts the discriminant features by using an improved Fisherface method. The experimental results on the public palmprint image database demonstrate the effectiveness of the proposed approach.