Method and apparatus for aggressive compression, storage and verification of the dynamics of handwritten signature signals

• US 7,474,770 B2
• Filed: 06/28/2005
• Issued: 01/06/2009
• Est. Priority Date: 06/28/2005
• Status: Expired due to Fees

• Current Assignee: Homayoon S.M. Beigi
• Original Assignee: Homayoon S.M. Beigi
• Inventors: Beigi, Homayoon S. M.
• Primary Examiner(s): Mariam; Daniel G
• Application Number:
  US11/160,530
• Publication Number:
  US 20060291703A1
• Time in Patent Office: 1,288 Days
• Field of Search: 382/119, 382/173, 382/177, 382187-188, 382/232, 382/244, 382/246, 382/251, 341/51, 341/65, 358/426.01, 738/654, 178/18.01, 178/18.03, 348/161
• US Class Current: 382/119
• CPC Class Codes: G06Q 20/40145 Biometric identity checks
• CPC Class Codes: G06V 40/30 Writer recognition; Reading and verifying signatures
• CPC Class Codes: G07F 7/10 together with a coded signal , e.g. in the form of personal identification information, like personal identification number [PIN] or biometric data
• CPC Class Codes: G07F 7/1066 PIN data being compared to data on card

Abstract

The fields of biometrics and commerce can benefit from the aggressive compression techniques presented here. A novel approach to compressing the on-line handwriting signal without losing the dynamic information (i.e., velocity and order) is presented which provides the means to fit almost any signature sample within the 57-byte limit of one of the free magnetic strips on the back of conventional transaction cards such as credit or debit cards or other identity cards. The apparatus requires an on-line signature capture tablet capable of sampling the handwritten signature in a series of sampled X and Y coordinates. The compression method starts with segmenting the on-line handwriting signal into portions which may be approximated with a set of features from which the velocity and position may be reconstructed. These features are then compressed further using standard compression techniques. A Huffman coding technique is used to demonstrate this step. The size of the average signature including its dynamic information is stored in a space less than 57 bytes which is the magic limit imposed by conventional, magnetic transaction (credit or debit) card strips. This is by far less than about 1000 bytes required if only loss-less techniques were used. A loss-less compression technique is also briefly presented. Normally, out of the three strips which are provided on the back of a card, two strips are available and may be used for storing the template of the owner'"'"'s signature for future verification.

Claims

1. In a system that captures handwriting on a digitizer tablet, a method performed by a processor for generating and storing data representing the handwriting comprising:

said processor processing the captured handwriting to identify strokes of the handwriting;

said processor partitioning each stroke into a plurality of segments;

said processor generating a plurality of parameters for each segment;

said processor generating a compressed representation for each given segment by transforming the parameters for the given segment;

said processor generating data representing the handwriting based upon the compressed representation for each given segment; and

said processor storing the data representing the handwriting.

2. The method of claim 1, wherein:

the partitioning involves identifying points with zero velocity along a predetermined direction.

3. The method of claim 2, wherein:

the predetermined direction is selected from the group including an x direction and a y direction.

4. The method of claim 1, wherein:

the plurality of parameters for a given segment are dynamic features derived from a second order mechanical model of a hand and stylus movement on the digitizer tablet.

5. The method of claim 1, wherein:

the parameters for a given segment represent velocity along two orthogonal directions.

6. The method of claim 5, wherein:

the parameters for a given segment include

i) first components representing amplitude for each one of the two orthogonal directions;

ii) second components representing frequency for each one of the two orthogonal directions;

iii) third components representing phase for each one of the two orthogonal directions;

iv) fourth components representing mean velocity for each one of the two orthogonal directions; and

v) a fifth component representing initial position of the given segment.

7. The method of claim 6, wherein:

the transforming of the parameters of a given segment comprises reducing the number of parameters for the given segment.

8. The method of claim 7, wherein:

reducing the number of parameters for the given segment is accomplished by using one of the second components associated with a corresponding one of the two orthogonal directions to represent frequency for the segment along both of the two orthogonal directions.

9. The method of claim 7, wherein:

reducing the number of parameters for the given segment is accomplished by using the phase difference between the third components for the two orthogonal directions to represent phase for the segment.

10. The method of claim 5, wherein:

the transforming of the parameters of a given segment comprises reducing bit length quantization of at least one component of the parameters of the segment.

11. The method of claim 10, wherein:

reducing bit length quantization of at least one component of the parameters of the segment is accomplished by using 4 bits to represent each one of the first components for the two orthogonal directions.

12. The method of claim 10, wherein:

reducing bit length quantization of at least one component of the parameters of the segment is accomplished by using 5 bits to represent each one of the second components for the two orthogonal directions.

13. The method of claim 1, wherein:

the transforming of the parameters of a given segment comprises reducing bit length quantization of data derived from the parameters of the segment.

14. The method of claim 13, wherein:

comprises reducing bit length quantization of data derived from the parameters of the segment is accomplished by using 4 bits to represent a phase difference along two orthogonal directions for the segment.

15. The method of claim 1, further comprising:

generating a non-linear table associated with a given parameter to provide increased resolution.

16. The method of claim 1, further comprising:

applying a coding scheme to the data representing the handwriting.

17. The method of claim 16, wherein:

the coding scheme employs Huffman Coding.

18. The method of claim 1, wherein:

the handwriting is a signature.

19. The method of claim 1, wherein:

the data representing the handwriting is stored on the memory of a transaction card.

20. The method of claim 19, wherein:

the data representing the handwriting is stored on media in accordance with ISO-7811/2 standards.

21. In a system that captures a handwriting sample on a digitizer tablet, a method embodied by a processor for verification of the handwriting sample comprising:

said processor storing data representing a handwriting signal in accordance with the method of claim 10;

said processor accessing the stored data representing the handwriting signal and converting such data to parameters for the segments of the handwriting signal;

said processor processing the captured handwriting sample to generate parameters for segments of the handwriting sample; and

said processor comparing the parameters of the segments of the handwriting signal to the parameters of the segments of the handwriting sample to make a verification decision.

22. A method according to claim 21, wherein:

the comparing involves calculating a distortion measure between the parameters of the segments of the handwriting signal and the parameters of the segments of the handwriting sample.

23. A method according to claim 21, further comprising:

said processor processing the parameters for the segments of the handwriting signal to reconstruct the handwriting signal for display.