基于二维Gabor的掌纹图像预处理研究

Research on palmprint image preprocessing based on 2D Gabor

掌纹识别属于相对较新的一种生物特征识别技术,是利用人手掌上丰富的纹理特征来进行身份识别。掌纹图像的质量是影响掌纹识别性能的关键,因此,由掌纹的特点入手,对掌纹图像采用基于形态学方法进行感兴趣区域(ROI)的分割,为了防止由于采集时手放置位置的旋转或偏移导致的掌纹图像的差异,通过中值滤波、二值化、膨胀腐蚀等操作确定了特殊角点,再利用角点连线确定旋转角度,来旋转掌纹图像。然后对掌纹图像感兴趣的区域采取小波阈值法来去除噪声。最后结合Gabor滤波器的方向性,采用基于二维Gabor滤波器对掌纹纹线的特征进行提取。为了验证所提出的掌纹图像预处理方法的有效性,在PolyU掌纹图像库上进行实验并取得了较好的实验效果。

Palmprint recognition is a biometric technology , which makes use of rich texture features on palm for i-dentification .The quality of a palmprint image is the key to the performance of palmprint recognition , therefore , this article starts with the palmprint features .The palmprint image is based on morphological methods for region of interest （ROI） segmentation.In order to prevent the differences of the palmprint image caused by rotation or offset of hand position in image acquisition , the palmprint image was rotated using corner connections to determine the ro-tation angle by means of median filtering , binary technology , the expansion of corrosion and other operations to de-termine the special corner .Then, the wavelet thresholding method was taken to remove the noise of the ROI region . Finally, taking advantage of the directionality of Gabor filter , the palmprint ridge characteristics were abstracted based on 2D Gabor filter.In order to verify the efficiency of proposed method for palmprint image preprocessing , some experiments were made in the PolyU palmprint image database , achieving good results .