Researching On Embedded Imaging System For Synchronous Multimodal Finger Feature Acquisition

Unimodal biometric identification systems are always ineffective to some extend in real applications. Therefore, multimodal biometrics has been a hot topic in biometric community. Conventional multimodal biometrics usually fuses the traits from different parts of a human body. This is actually inconvenient in practice, and many real multimodal biometrics systems are less practical since their working mechanism is supported by desktop computer systems.In this dissertation, the used multimodal traits all lie in a finger, which contains finger-vein, finger-knuckle-print and fingerprint. And the images of the three modalities are conveniently acquired by imaging them synchronously. Aiming at using the three modalities for person identification, we develop an embedded imaging system for synchronous multi finger feature acquisition.For programming image acquisition system, the embedded linux operation system is adopted, and Qt4.5 is transplanted as the user interferes. For hardware platform construction, an E8 card-like development board with S5PV210 CPU is selected.For acquiring the images of the three modalities, USB camera, OV3640 camera and USB fingerprint module are respectively used for the imaging finger-vein,finger-knuckle-print and fingerprint. The two cameras are all programmed using the Video for Linux and Frame Buffer API interface, so their structures are similar. The developed embedded image acquisition program uses several multi-threads to deal with video loop, image acquisition and information transmission, and achieve imaging synchronization by setting mutexes and condition variables among threads.The proposed embedded system is equipped with LCD touch screen for implementing human-computer interaction. To ensure data security, the system is able to automatically upload the captured multimodal images via FTP to a server for a backup.The experiments demonstrate that the proposed system is stable and reliable as a impendent embedded device for implementing multimodal finger image acquisition.This lays a good foundation for embedding recognition algorithm in the next step.