摘要
低质量指纹图像通常会产生较多的伪特征点,或者当待识指纹图像与模板图像之间存在较大平移和旋转时,识别算法的性能会受到很大的影响。利用多匹配算法的融合技术可以提高系统的识别率,但是这种数据融合的方法既增大了计算量,同时又难以充分利用原始指纹图像的信息。针对上述问题,提出了一种评估特征点置信度的方法,并在匹配算法中根据特征点置信度来修正匹配分数;然后运用特征层数据融合技术,将一个指纹的多个模板的特征点融合成一个特征点的本征模板,再与待识指纹图像的特征点进行匹配。实验结果表明,该方法在不明显增加存储空间和在线计算量的同时,可以达到很好的识别性能。
Presently, most fingerprint identification algorithms can not effectively work on some occasions, for instance, in which there are many false characters in a low quality fingerprint image, or existing place large motivation and rotation between a template fingerprint image and a compared fingerprint image. Although the technology of multi algorithm fusion can improve the true rate of an identification system, it will occupy much more calculating time and chunk away some information of an original fingerprint. This paper, firstly, for false characters in a fingerprint image, proposes a method to evaluate the credibility of each character and continuously to modify their matching scores by the character’s credibility in the identification process and secondly, for large motivation and rotation, fuses a set of templates into a character’s eigen template before the matching process. The experimental results show that, by the method of character’s layer data fusion, the algorithm can achieve a better identification performance, while it does not apparently increase the memory occupancy and the online calculating time.
出处
《中国图象图形学报（A辑）》
CSCD
北大核心
2005年第2期170-174,共5页
Journal of Image and Graphics
基金
国家自然科学基金项目(90104030)
