Acquisition And Recognition System Based On Multispectral Palm

With the development of society, the security consciousness deeply into people’s lives,and there are more and more identification instrument. However, traditional personalidentification system simply extracts one kind of characteristic information, such asfingerprints, palm, etc. Single biometric information easily copied and forged, and bringshidden trouble to people’s life. Therefore, the traditional single modal identificationtechnique has been unable to meet the demand of people to safety. In the future, theidentification system with multi-modal, non-contact and network will become an inevitabletrend.In order to improve the traditional biometric system reliability and security, this paperdesigns a non-contact multi-modal hand image acquisition system. Under the conditions ofnatural light, the system simultaneous achieves collector’s palm print, palm-shape andpalm vein three features information, by hand without touching the case of acquisitiondevices.The core of the system is TMS320DM642, with the two Video Port connection ofdifferent image sensors, which can simultaneously capture image information, then sendsthe image information to the encoder chip and displays on the screen. To improve thequality of the images, focus on the study and analysis of the LED light source circuit, todesign a uniform illumination, programmable multi-spectral light source system. To makesystem’s recognition more accurate, the paper focus on the optical imaging part in thestudy of acquisition device,including filterselection, image sensor configuration, lens parameter selection andthe hardware connection betweendual image sensor and DSP Video Port. In order to improve acquisition speed, withoutDSP intervention, the system direct use of EDMA channel to move image data to externalmemory, reducing CPU workload. Meanwhile, the use of DM642EMAC interfaceconnecting the physical layer chip, that makes communication between DSP and PC,realizes the functions of online registration.When the device is used to identify the identity, the system will automatically extract the feature points which stored in FLASH, and matchthe information with the participants. If the match is successful, the system will broadcastcollector’s message by the voice module. The system has been able to separate off-lineoperation, and builds the ideal hardware platform for the subsequent hand three modalinformation fusion technologies.