Abstract: Human biometrics possess uniqueness and stability, leading to the development of various biometric identification technologies such as fingerprint, facial recognition, voice recognition, iris scanning, and DNA structure analysis. Among these, fingerprints exhibit superior performance, and fingerprint recognition technology is maturing rapidly. Therefore, this paper designs an embedded fingerprint recognition system based on the OMAP architecture to meet the market demands of the automotive security field. This system boasts high recognition rate, fast processing speed, and strong scalability, making it a promising candidate for widespread application and research value. OMAP (Open Multimedia Applications Platform) is an open multimedia application platform based on DSP from TI. It employs a dual-core architecture, combining a high-performance, low-power DSP core with a powerful ARM microprocessor, offering advantages such as high integration, strong hardware reliability and stability, high speed, strong data processing capabilities, low power consumption, and good openness. The OMAP platform opens up not only the ARM but also the DSP through an advanced operating system platform. Through the DSP/BIOS bridge, DSP resources are accessed via the operating system's API, just like ARM peripherals. The DSP/BIOS bridge achieves seamless dual-core connectivity on the OMAP platform. The OMAP platform supports multiple operating systems, including WinCE, EPOC, Nucleus, VxWorks, and Linux, providing an easy-to-use open programming environment. The dual-core technology of the OMAP platform optimizes operating system efficiency and multimedia code execution. Real-time tasks are handled by the DSP, while non-real-time tasks and system control are handled by the ARM, minimizing system power consumption and successfully solving the problem of optimal performance-power balance. The system hardware design, based on the OMAP architecture, is an embedded fingerprint recognition system (taking automotive applications as an example). It mainly consists of an OMAP5912 embedded microprocessor, an FPS200 fingerprint sensor, a human-machine interface (HMI), a CAN bus interface, a USB communication interface, and power management modules. The OMAP5912 is the core of the entire system. On one hand, it controls the FPS200 fingerprint sensor to acquire fingerprint samples, establish a fingerprint sample library, and manage and maintain the library. On the other hand, it controls the DSP core to process complex fingerprint recognition algorithms, performing fingerprint image processing, fingerprint feature extraction, and matching. A user-friendly HMI is crucial during system operation. Under the control of the ADS7846 touchscreen controller and the OMAP5912, users can conveniently and quickly engage in human-machine interaction via the touchscreen and LCD screen, such as PIN authentication and fingerprint library management and maintenance. An open system inherently requires information resource sharing; the CAN bus interface enables the system to communicate with multiple controllers (i.e., electronic control units, ECUs) in the vehicle. The USB communication interface design makes the embedded system application more flexible and convenient, enabling communication with various types of peripherals. The powerful power management chip TPS65010 provides the necessary operating voltages (1.6V, 1.8V, 3V, 3.3V) for the core system and allocates power consumption appropriately. Due to the large amount of fingerprint image data and the complexity of the fingerprint algorithm, the design requires significant storage space; therefore, the system expands with one NOR flash memory and one DDR SDRAM. The system can be debugged online via the JTAG interface. The OMAP5912 embedded microprocessor is a dual-core application processor consisting of a 192MHz TMS320C55x DSP core and a 192MHz low-power, enhanced ARM926 microprocessor. The Traffic Controller (TC) controls access to external memory, with a maximum operating frequency of 75MHz. The TC provides an External Memory Fast Interface (EMIFF), an External Memory Slow Interface (EMIFS), and an Internal Memory Interface (IMIF). EMIFF can connect to SDRAM, while EMIFS can only connect to flash memory and slower ROM. The OMAP5912 also contains 192KB of internal memory, shared by the ARM and DSP. The TC (True Core) can be seen as a guardian of storage resources, with its internal arbitrator determining which core has the right to access certain resources and whether two cores can access them simultaneously. The OMAP5912 has a high-performance 9-channel system direct memory access (DMA) at its center. This 9-channel DMA allows data transfer between different ports without ARM interference. DMA-operable ports include EMIFF, EMIFS, IMIF, and peripheral components. The FPS200 fingerprint sensor is a fingerprint sensor from Veridicom based on standard CMOS technology. Utilizing the semiconductor silicon capacitance effect, the silicon sensor becomes one plate of a capacitor, and the finger is the other plate. The capacitance difference between the ridges and valleys of the fingerprint line and the smooth silicon sensor forms an 8-bit grayscale image of the fingerprint. The FPS200 fingerprint sensor operates between 3.3V and 5V, offering low power consumption and high efficiency. It can achieve image quality comparable to or even better than optical technology (higher resolution of 500 DPI) on a relatively small surface (1.28cm × 1.50cm). In terms of hardware design, a crucial point to note is the need to add a 74LV245 transceiver between the OMAP5912 and FPS200 to resolve timing conflicts; details will not be elaborated here. For the human-machine interface (HMI) design, due to the thin and lightweight nature of the touchscreen and its convenient and quick input, this design uses a four-wire resistive touchscreen as the input device. The HMI system consists of four parts: an OMAP5912 microprocessor, a touchscreen controller, a four-wire resistive touchscreen, and an LCD display. The touchscreen controller uses the Burn-Brown ADS7846 chip, a typical 12-bit ADC with a continuous approximation register, an internal 2.5V reference voltage, and a standard SPI data interface connected to the microprocessor. The ADS7846 opens the corresponding switch channel based on different command words received at the serial data input terminal DIN, receives the returned analog voltage, converts it to a digital value through A/D conversion, and then transmits it back to the microprocessor through the serial data output terminal DOUT. For the CAN bus interface design, after the embedded fingerprint recognition system completes the vehicle owner authentication, it needs to immediately send an ignition control signal to the engine electronic control system (EEC) to start the vehicle engine. Therefore, this system designs a CAN bus interface to achieve communication with the EEC. The interface circuit between the CAN bus and the microprocessor typically includes a CAN controller and a CAN transceiver. This design uses Microchip's MCP2510 CAN protocol controller and Philips' PCA82C250 CAN transceiver. For the USB communication interface design, to enable the embedded system to communicate with various types of peripherals, a USB communication interface is extended in the design. The OMAP5912 microprocessor has a built-in USB host controller corresponding to USB 1.1, and this design supports dual master-slave USB communication modes. A single USB interface adapter is all that's needed for easy master-slave mode switching. It should be noted that when selecting Client mode, FUNC_MUX_CTRL_D[5:3] = 000 needs to be set. Software Design Flow OMAP5912 Software Architecture The OMAP5912 software architecture is built on two operating systems: an ARM-based Linux operating system and a DSP/BIOS based DSP. The core technology connecting the two operating systems is the DSP/BIOS bridge. The DSP/BIOS bridge provides a seamless interface for using the DSP, allowing developers to access and control the DSP's operating environment using standard application programming interfaces on the GPP (General Purpose Processor). Using TI's CCS (Code Composer Studio) integrated development environment, from the developer's perspective, OMAP appears to complete all processing functions using only the GPP processor. In this way, developers do not need to program the two processors separately, thus greatly simplifying the programming work. Under the OMAP architecture, developers can program the OMAP dual-processor platform as if it were a single GPP. System Software Flow In order to prevent someone from stealing fingerprints and using fingerprint films to launch deceptive attacks on the fingerprint recognition system, it is necessary to take the dual authentication measures of "Personal Identification Code (PIN) + Fingerprint Recognition". The user inputs their PIN information, guiding the system to find a corresponding template in the fingerprint database. This template is then matched one-to-one with the collected user fingerprint for a "best" match. This prevents spoofing attacks and avoids the need for repeated one-to-N matching, allowing for efficient and rapid fingerprint recognition. The fingerprint recognition process includes: 1) Image preprocessing: This involves three steps: image segmentation, image enhancement, and binarization. The segmenter reads and cuts the input fingerprint image, reducing the amount of data to be processed in subsequent steps while preserving useful fingerprint information. Image enhancement smooths, sharpens, and filters the segmented fingerprint image to improve image quality. Binarization converts the 8-bit grayscale fingerprint image into a binary image of 0s and 255s, using a local thresholding method. 2) Thinning: To further compress the data, the binarized image needs to be thinned. Thinning should maintain the connectivity and directionality of the ridges, and the center of the ridges should remain essentially unchanged. After thinning, bridging and gaps may appear in the ridges, so denoising is necessary after thinning. 3) Feature Extraction. Fingerprint feature point information is extracted from the thinned binarized image. Bifurcation points or termination points are found in the thinned binarized image, and then the ridge trajectory is traced from these points. The shape of the ridges is calculated using these points. This shape data, point type, and point position are stored as feature points of the fingerprint image. 4) Feature Matching. After extracting fingerprint feature points, a set of vector points is obtained. Similarly, the fingerprint template stored in the fingerprint database is also a set of vector points. Determining whether two fingerprints match is transformed into determining whether two sets of vector points match. Fingerprint image matching is determined based on the maximum number of matching points supported by the feature points of the two images. If the maximum number of matching points supported is greater than a specified value, the two fingerprint images are considered to match. During the initialization process, control words are written to the corresponding registers of the FPS200 fingerprint sensor to set the parameters for fingerprint acquisition. The most crucial aspect is the parameter setting of the three registers: DCR, DTR, and PGC. This paper achieved optimal register parameter settings through repeated experiments. Conclusion This paper designs a complete, independently operating embedded fingerprint recognition system based on the OMAP architecture. The system has successfully passed laboratory testing and is currently applying for project approval. With the rapid development of biometric authentication technology, the system's expansion potential will become increasingly broad, and the system design will become more and more perfect. The development of in-vehicle multimedia networks has become an inevitable trend, and this system can be further developed into a GPS in-vehicle navigation system. The OMAP open multimedia application platform also reserves sufficient space for future system expansion. It is believed that this system will have enormous market potential.