用于指纹识别的基于离散余弦变换的二值化联合变换相关器

Discrete Cosine Transform Based Binary Joint Transform Correlator for Fingerprint Identification

提出了一种利用离散余弦变换对联合功率谱进行处理的二值化联合变换相关算法。利用这种二值化联合变换相关器 ,进行了一个待识别指纹同四个参考指纹的比较。这种相关器同采用局部中值阈值技术的二值化联合变换相关器相比 ,能提高输出峰边比 ,提高相关峰的强度。光电混合处理实验的结果表明 ,这种方法能大大提高多指纹识别系统的识别性能。

A new binary joint transform correlator algorithm is presented which uses the discrete cosine transform technique to process the joint transform power spectrum. This binary joint transform correlator can compare a target fingerprint with four reference fingerprints. Compared with other joint power spectrum processing method such as slide window local median threshold method, the correlator using this technique has better performance in terms of output peak to noise ratio, peak intensity and discriminate rate. The result of optical electrical hybrid experiment has shown that this technique can greatly improve the discrimination ability of multiple fingerprint identification system.