Presentation is loading. Please wait.

Presentation is loading. Please wait.

To solve the problem of limited documentation and example code available on the subject of biometrics. Research that is done in this field can not directly.

Similar presentations


Presentation on theme: "To solve the problem of limited documentation and example code available on the subject of biometrics. Research that is done in this field can not directly."— Presentation transcript:

1

2 To solve the problem of limited documentation and example code available on the subject of biometrics. Research that is done in this field can not directly be used in an application; the programmer must develop the code themselves using the research as a guide.

3 Easy to useA programmer of any skill level should be able to use SmallFile size must be small FastMust not require loads of CPU time Cross platform compatible Must run on any platform with little to no change in code CustomizableOpen source and most changed items must be easily accessible

4 Fingerprints are matching by comparing minutia points Two basic types of minutia points Line ending Line branching Fingerprint verification vs fingerprint recognition: Verification systems need to have more accuracy Recognition system must be able to process many prints quickly *This project is a verification system CostAverage cost is around $1000.00 SizeLarge file size due to unneeded functions Resource Requirements Large amounts of memory and processor time Cross compatibilityDesigned for only certain operating systems, database server, or input devices Non-customizableNot open source, making it difficult to customize Hard to useRequires large amounts of documentation reading to learn how to use Problems associated with commercial SDKs (Software Development Kits):

5 C\C++ Compiler Basic Text editor or Development IDE Hex editor Image manipulation program

6 1. ResearchFind information on fingerprint matching and image manipulation 2. DesignDesign and layout library with flow charts 3. CodeCode using designs from step 2 4. Compile and Debug Compile and debug fixing any typographical errors 5. TestTest using sample fingerprints 6. AdjustAdjust for better accuracy 7. PublishMake final copy available

7 Edge Detection with Logarithm Algorithm

8 Thinning with Skeletierung’s Algorithm Breaks found Final rewritten thin

9 Match Part 1 – Shifting Move the verifying print vertically and horizontal to find the spot were the most pixels line up. A true match will have a certain percentage line up. Lines up Does not line up

10 Match Part 2 – Minutia Matching = Line Branching = Line Ending

11 My data has shown that this system is not 100% accurate, but no prints that were not suppose to pass did. With a little bit of tuning the accuracy of the system can be easily improved. Also most of the goals for the project have been met, with the exception of speed. As for speed, a revision of Thin() and Match_Part1() are required to optimize these functions. Unfortunately smudged prints still cannot be matched without further correction of the images. Overall the project was a success and continued work will only improve upon it.

12 Image manipulation – including scaling and rotation Faster Thinning Faster Matching Part1 Design Embedded System Correction of smudged and other imperfections in images

13 R. Haralick and L. Shapiro Computer and Robot Vision, Vol 1, Addison-Wesley Publishing Company, 1992. A. Jain and S. Pankanti Automated Fingerprint Indentification and Imageing Systems, Dept. of Comp. Sci. and Eng., Michigan State University, 1996. A. Jain, S. Prabhakar and J. Wang Minutia Verification and Classification for Fingerprint Matching, Dept. Of Comp. Sci. and Eng., Michigan State Unversity. D. Verna Machine Vision, Prentice-Hall, 1991.


Download ppt "To solve the problem of limited documentation and example code available on the subject of biometrics. Research that is done in this field can not directly."

Similar presentations


Ads by Google