摘要
由于虹膜自身具备独特性和随机性,因此需要对其进行特征提取操作,找出具有识别意义的虹膜纹理信息并将其编码,然后依据特定的判决标准来判断两幅图像是否源于同一人,从而实现虹膜识别的功能。该文对传统LBP算子进行改进,极大程度上缩减了描述码维数,并结合位移差分图像匹配方法,得到一种快速、有效的特征方法。通过对CASIA虹膜数据库进行实验比较,该方法能够获得理想的识别效果。
Because of the uniqueness and randomness of the iris image,feature extraction operation is im-portant,finding and coding the iris texture information to judge whether two iris images from the same person, so as to realize the function of iris recognition.This paper improve the traditional LBP operator to reduced di-mension description code,and combined with the Hamming Distance match method,get a fast and effective method of feature extraction.Experiment results on CASIA database demonstrate that this method is ideal.
出处
《电子质量》
2014年第3期76-82,共7页
Electronics Quality
