多模态视觉生物特征嵌入式识别装置设计与实现

发布时间：2018-04-27 09:58

本文选题：多模态生物特征识别 + 嵌入式Linux系统　；参考：《哈尔滨工程大学》2012年硕士论文

【摘要】：随着世界人口数量的日益增大和流动性的不断增强，用于身份认证的生物特征识别技术得到了越来越广泛的重视。多模态生物特征识别技术利用了人体所固有的生物特征进行身份识别，并克服了单模态生物特征识别技术的不足，因此具有相当的可靠性。随着嵌入式技术的迅速发展，人们对系统的便携性和实用性提出了更高的要求。嵌入式系统具有专用性强、实时性高、灵活性强等特点，能够适合各种工作环境，可以完成多模态生物特征识别系统的所有功能，因此，本文将设计一款多模态生物特征嵌入式识别装置。本文所设计的多模态生物特征嵌入式识别装置主要实现视觉生物特征（如指纹、手指静脉等）的采集，，得到有效的图像数据，为以后在嵌入式系统上实现多模态生物特征识别功能提供硬件平台与软件平台。本文针对多模态生物特征嵌入式识别装置的功能要求，选用TI公司的TMS320DM6446作为嵌入式系统的核心处理器，并选用Linux作为操作系统。本文主要完成了系统硬件平台的设计工作，并基于硬件平台进行了系统软件平台的设计与实现。本文主要研究内容包括系统硬件平台架构设计、嵌入式Linux操作系统设计以及视频/图像采集接口设计等，并重点对相应接口的Linux驱动程序进行了编写。为了实现多模态生物特征采集功能，视频/图像采集功能模块设计为双路视频/图像采集接口。论文详细阐述了双路视频/图像采集接口的设计方法，包括接口的硬件电路设计以及软件驱动设计。在视频/图像采集接口设计过程中，针对处理器TMS320DM6446结构特点，并结合系统的视频/图像采集功能要求和嵌入式Linux操作系统驱动架构，对基于V4L2架构的视频/图像驱动程序进行了深入研究，最终完成了包含V4L2驱动、VPFE驱动、以及复合视频采集接口驱动或数字图像采集接口驱动在内的完整的视频/图像采集驱动程序。本系统设计了多模态生物特征嵌入式识别装置的软硬件平台，并实现了生物特征采集功能，为后续的多模态生物特征识别功能算法实现提供了有效的硬件技术支持。
[Abstract]:With the increasing population and mobility of the world, biometric identification technology for identity authentication has been paid more and more attention. Multi-modal biometric recognition technology makes use of the inherent biological characteristics of human body to identify and overcome the shortcomings of single-mode biometric identification technology, so it has considerable reliability. With the rapid development of embedded technology, people put forward higher requirements for the portability and practicability of the system. The embedded system has the characteristics of high specificity, high real-time, strong flexibility and so on. It can adapt to all kinds of working environment, and can complete all the functions of the multi-modal biometric recognition system. In this paper, a multi-modal biometric embedded recognition device will be designed. The multi-modal biometric embedded recognition device designed in this paper mainly realizes the collection of visual biometric features (such as fingerprints, finger veins, etc.), and obtains effective image data. It provides hardware platform and software platform for the realization of multimodal biometric recognition in embedded system. In this paper, according to the functional requirements of multi-modal biometric embedded recognition device, TI's TMS320DM6446 is selected as the core processor of embedded system, and Linux is selected as operating system. This paper mainly completes the design of the system hardware platform, and the design and implementation of the system software platform based on the hardware platform. The main contents of this paper include the design of system hardware platform, embedded Linux operating system and video / image acquisition interface, etc. The Linux driver of the corresponding interface is mainly written in this paper. In order to realize the multi-modal biometric acquisition function, the video / image acquisition module is designed as a dual-channel video / image acquisition interface. The design method of dual-channel video / image acquisition interface is described in detail, including hardware circuit design and software driver design. In the process of designing video / image acquisition interface, according to the characteristics of processor TMS320DM6446 structure, combined with the system video / image acquisition function requirements and embedded Linux operating system driver architecture, The video / image driver based on V4L2 architecture is studied in detail. Finally, the VPFE driver including V4L2 driver is completed. And the composite video capture interface driver or digital image acquisition interface driver including the complete video / image acquisition driver. This system designed the hardware and software platform of the multi-modal biometric embedded recognition device, and realized the biometric collection function, which provided an effective hardware technical support for the subsequent implementation of the multi-modal biometric recognition algorithm.
【学位授予单位】：哈尔滨工程大学
【学位级别】：硕士
【学位授予年份】：2012
【分类号】：TP391.4;TP368.1

【参考文献】