Fingerprint Verification System Cheryl Texin Bashira Chowdhury 6.111 Final Project Spring 2006 Abstract This report details the design and implementation of a fingerprint verification system. The system consists of two parts: an image acquisition module and an image verification module. The image acquisition module involved capturing an image of an inked fingerprint via a web camera and transmitting that image through a video decoder onto a block RAM in the 6.111 lab kit. Using the stored image, the image verification module involves verifying the image via image processing filters to find a match in a pre-formed print database. The image verification module was completed while the image acquisition module was not completed due to wiring and timing errors.Table of Contents Introduction ……………………………………………………………………4 Design Overview………………………………………………………………..5 Image Verification Overview……………………………………………….…9 Match FSM……………………………………………………………………14 Matching……………………………………………………………………….19 Appendix (Code for Image Verification and Acquisition)………………….21 Image Acquisition Overview……..…………………………………………….5 Camera and Video Decoder………………………………………………… 6..,.Asynchronous FIFO Image Transfer………………………………………....7 Block RAM……………………………………………………………………..8 Sobel Edge Detection Filter…………………………………………………..11 Direction Filter………………………………………………………………..13 Control FSM…………………………………………………………………..16 Conclusion……………………………………………………………………..20 2Table of Figures Figure 1. Block Diagram of Fingerprint Verification System………….5 Figure 2. Block Diagram of Image Acquisition Module ……………….6 Figure 3. State transition diagram for control FSM …………………..7 Figure 4. State transition diagram for FIFO interface …………………8 Figure 5. Block diagram for image verification module ………………10 Figure 6. Sobel filter state transition diagram …………………………12 Figure 7. Direction Filter FSM ………………………………………….13 Figure 8. Major FSM state transition diagram ………………………..14 Table 1. Match-sum switches ……………………………………………16 Table 2. Memory mapping switches ……………………………………16 Figure 9. Match FSM state transition diagram ………………………..17 3I. Introduction Despite the multitude of personal identification methods currently in practice, fingerprint verification has traditionally been considered one of the most reliable methods available. Typically, fingerprint verification has been manually conducted by experts in print identification, which creates a significant backlog of work due to the tedious and time-consuming task of the manual identification. Consequently, autonomous fingerprint verification systems are in high demand for a wide range of applications that cannot use the manual verification process. These applications include access control, security verification, as well as the more commonly known criminal identification. An autonomous fingerprint verification system takes advantage of many features of the ridge topology of the fingerprint. In particular, verification systems focus on the the minutiae of the fingerprint. The most commonly examined minutiae are endpoints and bifurications. Fingerprints are obtained by a sensor or a camera and preprocessed to obtain the minutiae of interest. The following matching stage uses the minutiae of interest to establish a correlation between the sample print and a print in a database. For our final project in 6.111, we focused on implementing a robust fingerprint verification system with a small database. As a part of this design project, we researched, planned, and implemented our fingerprint verification system using the 6.111 lab kit and a web camera obtained through resources available in 6.111. This report details our design and implementation of our system. 4II. Design Overview This fingerprint verification system is composed of two modules: an image acquisition and image verification, both of which are detailed below. Control 8Decoder 8 Display Result Controller8 Image Static RAM Camera FSM Video VGA interface Processing FSM Matching FSM Figure 1. Block Diagram of Fingerprint Verification System III. Image Acquisition Modules To acquire an image of the fingerprint, a web camera provided by the 6.111 laboratory was used to capture an image of an inked print. Through the video decoder file provided by the 6.111 staff, the inked image was stored in the ZBT. However, only one frame of the print image was needed. In order to store just one frame of the image in the ZBT, a button on the lab kit was used to trigger a freeze mode on the video decoder. Via button zero on the lab kit, the signal triggered causes the ZBT to hold the last frame 5captured from the camera. This mode is shown in the state transition diagram on the following page. After the last frame of the print image is stored on the ZBT, the image is transferred to an asynchronous FIFO. This step was needed in order to synchronize the clocks of the image acquisition and image verification modules. The camera required the image acquisition module to run at a 65 Mhz clock while the image verification module ran at the 31.5 Mhz clock for its image processing procedures. The asynchronous FIFO allows the image to be stored at 65 Mhz into memory and then later read at the 31.5 Mhz clock. Finally, the image was to be transferred to a block RAM in order to allow the image verification module to access the print sample. In addition, several block RAMs were to be created in order to create the small database needed for meaningful functioning. However, this step could not be achieved. In the following sections on debugging the image acquisition module, the difficulties in achieving this task will be detailed. The following block diagram gives a quick visual
View Full Document
Unlocking...