Abstract: Extracting minutiae is difficult for high-resolution palmprint because of the strong influence from principal lines, creases and other noises. A novel high-resolution palmprint minutiae extraction method is proposed based on Gabor phase feature and image quality estimation. Firstly, the Gabor amplitude-phase model is introduced for palmprint representation. Based on the obtained Gabor phase, a multi-scale minutiae detection method is proposed. Then with the information from minutiae detection, the authors design Fourier response based quality estimation method and Gabor amplitude based quality estimation method. Finally, the two image quality estimation results are fused for minutiae screening to remove the unreliable minutiae. Experiment results show that the proposed method can extract minutiae effectively from high-resolution palmprint and remove the unreliable ones. Compared with other methods, the proposed method obtains better minutiae extraction results.

Key words: palmprint recognition, minutiae extraction, image quality estimation, Gabor convolution

摘要： 针对主线、褶皱线及噪声等造成高分辨率掌纹细节点提取困难的问题, 提出一种基于Gabor 相位和图像质量评价的高分辨率掌纹细节点提取算法。首先使用Gabor 振幅相位模型对掌纹图像进行描述, 并基于 Gabor 相位提出多尺度的细节点检测方法; 然后利用提取细节点过程中的信息, 设计基于傅里叶响应的质量评价方法和基于Gabor 振幅的质量评价方法; 最后融合两种质量评价结果, 并对检测到的细节点进行筛选。实验结果表明, 所提方法能够有效地提取高分辨率掌纹图像的细节点, 并去除不可信细节点。与其他方法相比, 具有更好的细节点提取结果。

关键词: 掌纹识别, 细节点提取, 图像质量评价, Gabor 卷积