局部方向模式在非接触掌纹识别中的应用

提高非接触掌纹识别系统的性能是目前掌纹识别领域一个具有实际意义的热点。针对非接触掌纹识别系统的鲁棒性问题,以掌纹图像的纹理特征为基础,提出一种基于局部方向模式(LDP)的掌纹识别方法,设计并实现了符合应用环境的嵌入式系统。LDP方法主要利用了Kirsch八方向算子的边缘响应值,从而获取图像的纹理方向模式特征。首先给出LDP算法的基本模型和流程,然后将非接触掌纹图像分成大小均匀的区块,利用LDP算法获取不同区块的纹理特征直方图向量,并进行融合形成总的模式特征,最后使用Chi距离测度进行匹配识别。在香港科技大学(HKUST)和自建的非接触掌纹图库上进行了实验测试,结果表明,该方法正确识别率可达97.824 4%和96.754 7%,相比其他典型和流行方法,最高可提升6.452 9%和5.995 6%。同时在室内环境下,利用自行设计的嵌入式原型装置进行了初步实际测试,结果表明,该方法正确识别率可达96.193 3%,具有可行性和有效性,提高了非接触掌纹识别系统的性能。

Application of local directional pattern in non-contact palmprint recognition

Improving the performance of non-contact palmprint identification system is a hot spot that has practical significance in current palmprint recognition research field. Aiming at the robust problem of non-contact palmprint recognition system, a palmprint recognition method is proposed based on local directional pattern (LDP) on the basis of the rich texture features of palmprint image; and an embedded system is designed and realized according to the application environment. The Kirsch edge response values in eight directions are mainly used in the LDP method to obtain the texture direction pattern features of the image. Firstly, the basic model and procedures of the LDP algorithm are given, and then the non-contact palmprint image is divided into blocks with uniform size; Secondly, the LDP algorithm is used to obtain the texture feature histogram vectors for different blocks, and then these features are fused to form the overall pattern features; Finally, the measure of Chi distance is used to perform matching identification. The experiment tests were conducted on Hong Kong University of Science and Technology (HKUST) non-contact palmprint image database and the self-built non-contact palmprint image database. Experiment results show that the proposed method can achieve correct recognition rates (CRR) of 97.824 4% and 96.754 7%, the CRRs are increased by 6.452 9% and 5.995 6% at most, respectively for the two databases compared with other typical and popular methods. At the same time, the preliminary actual tests were carried out using a self-designed embedded prototype device in indoor environment, the results show that the CRR can reach 96.1933%, which demonstrates that the proposed method is feasible and effective and improves the performance of non-contact palmprint identification system.