手掌静脉识别技术在贵重物品物流中的应用研究

发布时间：2018-01-14 08:05

本文关键词：手掌静脉识别技术在贵重物品物流中的应用研究　出处：《沈阳大学》2015年硕士论文　论文类型：学位论文

【摘要】：贵重物品在物流中丢失、损坏、被冒领引起巨额财产损失和赔偿纠纷,传统的身份认证方式如身份证、短信、手机号码、工号、用户名、密码等存在容易被丢失、遗忘、复制及盗用的隐患,已经不能满足贵重物品物流的应用需求。由于手掌静脉特征具有稳定性、唯一性、含有丰富的信息量、难被复制和窃取、采集方式易于接受等独特的优势,本文采用新型的身份认证方式——手掌静脉识别技术对贵重物品物流中的客户、收派员、仓库管理员进行监督、管理,当贵重物品丢失、损坏、被冒领时,保证责任落实到个人。本文利用自建的掌脉图像数据库,重点围绕基于子空间的手掌静脉特征提取算法展开分析和讨论,提出基于主成分分析和FISHER线性判别的方法。FLD提取的是最佳分类特征,本文通过PCA降维克服了单独使用FLD方法时,出现的小样本问题。此外,提出上述方法的改进算法,识别阶段将降维过程提取的PCA特征与最终提取的FLD特征利用加法融合,取得了较好的识别效果。为了得到原始输入空间中非线性最佳分类特征,并解决小样本问题,本文提出基于核主成分分析和FISHER线性判别的掌脉特征提取方法,先用KPCA对图像降维,然后用FLD提取分类特征,最后采用欧式距离完成匹配。仿真结果表明,与传统的2DFLD、本文的PCA+FLD、改进的PCA+FLD相比,在不同的特征个数下,该方法均取得较高的正确识别率96%,识别时间较短,运行速度较快,满足贵重物品物流的应用需求。最后,利用MATLAB GUI编写了用户界面,完成从获取掌脉图像到匹配结果的软件调试,显示人机交互界面。仿真结果表明,此掌脉识别系统安全可靠,将其运用到贵重物品物流中,不仅具有重大的理论意义,而且具有广阔的应用前景。
[Abstract]:Valuables are lost and damaged in logistics, resulting in huge property losses and compensation disputes. Traditional authentication methods such as identity card, SMS, mobile phone number, worker number, user name. Password is easy to be lost, forgotten, copied and stolen hidden danger, can not meet the application needs of valuables logistics. Because of the palm vein features are stable, unique, rich in information. Difficult to be copied and stolen, easy to accept the acquisition of unique advantages, this paper uses a new type of identity authentication-palm vein identification technology to valuables logistics customers, receiving staff. The storekeeper supervises, manages, when the valuables are lost, damaged, is taken, guarantees the responsibility to carry out to the individual. This paper uses the self-built palm pulse image database. Focusing on the analysis and discussion of subspace-based palmar vein feature extraction algorithm, a method based on principal component analysis and FISHER linear discrimination is proposed to extract the best classification features. This paper overcomes the small sample problem when using the FLD method alone by using PCA dimension reduction. In addition, the improved algorithm of the above method is proposed. In the recognition stage, the PCA feature extracted from the dimensionality reduction process is fused with the FLD feature extracted finally, and good recognition effect is obtained. In order to obtain the best nonlinear classification features in the original input space. And to solve the problem of small samples, this paper proposes a palmar pulse feature extraction method based on kernel principal component analysis and FISHER linear discriminant. Firstly, KPCA is used to reduce the dimension of the image, then FLD is used to extract the classification features. Finally, the Euclidean distance is used to complete the matching. The simulation results show that compared with the traditional 2DF LDD, the PCA FLD of this paper, the improved PCA FLD, under different number of features. The method achieves high correct recognition rate 96, short recognition time, fast running speed, and meets the application requirements of valuables logistics. Finally, the user interface is written by MATLAB GUI. The simulation results show that the palmar pulse recognition system is safe and reliable, and it is used in valuables logistics. Not only has the great theoretical significance, but also has the broad application prospect.
【学位授予单位】：沈阳大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TP391.41

【参考文献】