Design of Fingerprint Verification System Based on STM32
|School||Hangzhou University of Electronic Science and Technology|
|Course||Circuits and Systems|
|Keywords||fingerprint recognition STM32 fingerprint sensor fingerprint acquisition fingerprint algorithm|
With the development of society, the safety requirements of lives of the people have graduallyimproved. Fingerprint recognition technology which has the status of security certificate has greatlydeveloped under the people’s attention and demand. Nowadays, a variety of fingerprintidentification systems have emerged in our lives. However, most fingerprint identification system isbased on the PC machine. These systems still have many problems in the demand of mobility andconvenience. People take more attention to research the embedded fingerprint identification, andmake some achievements at present. But there are still many deficiencies in real time, powerconsumption and price. So they need a further research and improvement to enhance the system’sapplication prospect.This article makes a brief analysis about research and development state of the currentfingerprint recognition system. Combined with my own knowledge of the embedded system and thefingerprint algorithm, I put forward a plan of fingerprint recognition design system based onSTM32 chip, and make the design implement. According to the practical demand of the system,fingerprint collection and processing algorithm have been improved. Final we implement theembedded fingerprint identification system. The design mainly includes two aspects of hardwareand software, specific as follows:Hardware:1. The system uses MBF200 fingerprint sensor to collect fingerprints, and uses ARM cortex-M3kernel STM32F-103ZET6 act as main controller, then achieves communication between themaster controller and the sensor to collect image through the SPI interface, thus saves the imagedata in the memory space. The main controller sends image data to the computer throughasynchronous serial communication, then displays fingerprint image on PC.2. As the fingerprint recognition algorithm requires processing a large amount of data, the systemtakes up 76.8KB space to store a complete fingerprint image collection. But the host controllerSTM32 internal RAM is only 64KB. The system makes expand external SRAM processing formicroprocessor. The intermediate data of collection fingerprint and of processing algorithms arestored in SRAM. Procedures and characteristic points of fingerprint algorithm extraction arestored in internal FLASH of chip.3．The system uses USB to supply power, and then provides the stable voltage 3.3V for the systemby the transform of the internal circuit, then takes simulation and debug with JTAG interface.There uses LCD1602 to display fingerprint recognition results. Software:1. Fingerprint collect and implement: the SPI communication is realized by the timing of MBF200fingerprint sensor data sheet based the research and analysis SPI communication. Then thedifferent control registers are read and write by the commands of the sensor to collectedfingerprint images2. Fingerprint algorithm improvement: Through the study of existing fingerprint algorithms, I takeprogramming and implement the algorithms in VC++, and improve fingerprint processingalgorithms. I use image enhancement of orientation information based the fingerprintenhancement algorithms to improve image quality; I choose the way of separated from thefingerprint image into some same small block, there calculating the threshold of every blockimplement fingerprint image binary. In order to enhance efficiency, I put forward the way ofinitial match and global similarity method about the matching algorithm. The experiments showthat the fingerprint recognition system has been improved in the recognition accuracy andefficiency by improving the algorithm,.3. PC communication design: System implements serial communication between the mastercontroller STM32 and PC host computer, and sends the collection of fingerprint date to the PC.Then realizes of receiving the serial programming from the control by MScomm of VC + +Software. At the same time make man-machine interface display grayscale fingerprint image.4. Fingerprint algorithm transplant: the algorithm which is developed in VC needs to betransplanted to the main controller for embedded fingerprint identification. In the transplantprocess, the store of system program, fingerprint data and the template data of feature points areassigned. The fingerprint image is cut and Sampled Zoom by system to reduce the amount ofdata and improve execution efficiency.