Spam fax filter

• US 20050030589A1
• Filed: 08/08/2003
• Published: 02/10/2005
• Est. Priority Date: 08/08/2003
• Status: Active Grant

• Current Assignee: j2 Cloud Services, LLC (Consensus Cloud Solutions, Inc.)
• Original Assignee: Venali Incorporated (Ziff Davis, Inc.)
• Inventors: Wolf, Hartmut, El-Gazzar, Amin, Nolasco, Alexander M., Puente, Adrian
• Granted Patent: US 7,184,160 B2
• Time in Patent Office: Days
• US Class Current: 358/402
• CPC Class Codes: H04N 1/32005 Automation of particular receiver jobs, e.g. rejecting unwanted calls requesting a communication from a transmitter H04N1/32771; with picture signal storage for forwarding messages H04N1/32358
• CPC Class Codes: H04N 1/3201 Rejecting unwanted calls
• CPC Class Codes: H04N 1/32021 according to the type of received information

Abstract

A spam fax filter which has been configured in accordance with the present invention can include a fax transformation processor configured to transform rasterized fax images into non-rasterized representations of fax images. The spam fax filter further can include a knowledge base of known spam characteristics. Finally, the spam fax filter can include a comparator configured to compare non-rasterized representations of fax images produced by the fax transformation processor to the characteristics in the knowledge base to detect spam fax in the spam fax filter.

Claims

1. A spam fax filter comprising:

a fax transformation processor configured to transform rasterized fax images into non-rasterized representations of fax images;

a knowledge base of fax characteristics comprising at least one fax characteristic selected from the group consisting of spam characteristics and non-spam characteristics; and

,

a comparator configured to compare non-rasterized representations of fax images produced by said fax transformation processor with at least one of said fax characteristics to detect spam fax in the spam fax filter.

2. The spam fax filter of claim 1, wherein said knowledge base comprises fax characteristics selected from the group consisting of a number of fax transmission pages, fax image intensity, and fax image patterns.

3. The spam fax filter of claim 1, further comprising an optical character recognition processor coupled to said fax transformation processor and configured to convert said rasterized fax images into textual representations of said rasterized fax images.

4. The spam fax filter of claim 1, further comprising:

an edge sharpening filter coupled to said fax transformation processor and configured to produce intermediate imagery for said rasterized fax images comprising an image having accentuated edges defining said rasterized fax image; and

,

means for computing a set of vectors for said accentuated edges in said intermediate imagery.

5. The spam fax filter of claim 1, further comprising:

a blurring filter coupled to said fax transformation processor; and

,

a robust hash processor configured to produce a robust hash value for a blurred form of said rasterized fax image.

6. The spam fax filter of claim 1, further comprising:

means for detecting a color value for a selection of pixels in a region of said rasterized fax images; and

,

means for numerically encoding each of said rasterized fax images with at least one numerical value accounting for a number of black pixels for a single rasterized fax image and a number of white pixels in said single rasterized fax images.

7. A method for filtering spam fax, the method comprising the steps of:

transforming a rasterized form of a candidate fax image into a non-rasterized form;

testing said non-rasterized form of said candidate fax image against at least one fax characteristic selected from the group consisting of spam characteristics and non-spam characteristics; and

,

identifying said candidate fax image as spam fax where said candidate fax image tests favorably against at least one of said fax characteristics.

8. The method of claim 7, wherein said transforming step comprises the step of processing said candidate fax image in an optical character recognition processor to convert said candidate fax image into a text.

9. The method of claim 7, wherein said transforming step comprises the steps of:

accentuating edges in said candidate fax image; and

,

computing a set of vectors for said accentuated edges.

10. The method of claim 7, wherein said transforming step comprises the steps of:

blurring said candidate fax image; and

,

producing a robust hash value for said blurred candidate fax image.

11. The method of claim 7, wherein said transforming step comprises the steps of:

detecting a color value for a selection of pixels in a region of said candidate fax image; and

,

numerically encoding said candidate fax image with at least one numerical value accounting for a number of black pixels and a number of white pixels in said candidate fax image.

12. The method of claim 7, wherein said testing step comprises the steps of:

applying selected ones of said characteristics to said non-rasterized form of said candidate fax image to produce test results;

weighting selected ones of said test results; and

,

combining said weighted test results to produce a result for said testing.

13. The method of claim 7, further comprising the steps of:

establishing threshold values for each of said characteristics across which said candidate image is scored as having demonstrated each said characteristic;

weighting each score which exceeds a corresponding one of said threshold values; and

,

summing said weighted scores to produce a result for said testing step.

14. A machine readable storage having stored thereon a computer program for filtering spam fax, the computer program comprising a routine set of instructions for causing the machine to perform the steps of:

transforming a rasterized form of a candidate fax image into a non-rasterized form;

testing said non-rasterized form of said candidate fax image against at least one fax characteristic selected from the group consisting of spam characteristics and non-spam characteristics; and

,

identifying said candidate fax image as spam fax where said candidate fax image tests favorably against at least one of said fax characteristics.

15. The machine readable storage of claim 14, wherein said transforming step comprises the step of processing said candidate fax image in an optical character recognition processor to convert said candidate fax image into a text.

16. The machine readable storage of claim 14, wherein said transforming step comprises the steps of:

accentuating edges in said candidate fax image; and

,

computing a set of vectors for said accentuated edges.

17. The machine readable storage of claim 14, wherein said transforming step comprises the steps of:

blurring said candidate fax image; and

,

producing a robust hash value for said blurred candidate fax image.

18. The machine readable storage of claim 14, wherein said transforming step comprises the steps of:

detecting a color value for a selection of pixels in a region of said candidate fax image; and

,

numerically encoding said candidate fax image with at least one numerical value accounting for a number of black pixels and a number of white pixels in said candidate fax image.

19. The machine readable storage of claim 14, wherein said testing step comprises the steps of:

applying selected ones of said characteristics to said non-rasterized form of said candidate fax image to produce test results;

weighting selected ones of said test results; and

,

combining said weighted test results to produce a result for said testing.

20. The machine readable storage of claim 14, further comprising the steps of:

establishing threshold values for each of said characteristics across which said candidate image is scored as having demonstrated each said characteristic;

weighting each score which exceeds a corresponding one of said threshold values; and

,

summing said weighted scores to produce a result for said testing step.

21. A method for filtering spam fax, the method comprising the steps of:

testing a non-rasterized form of a candidate fax image against characteristics of known spam fax; and

,

identifying said candidate fax image as spam fax where said candidate fax image tests favorably to at least one of said characteristics of said known spam fax.

22. A method for filtering spam fax, the method comprising the steps of:

testing a non-rasterized form of a candidate fax image against characteristics of known spam fax; and

,

identifying said candidate fax image as a legitimate fax where said candidate fax image does not test favorably to at least one of said characteristics of said known spam fax.

23. A method for filtering spam fax, the method comprising the steps of:

testing a non-rasterized form of a candidate fax image against characteristics of known spam fax; and

,

assigning a probability to said candidate fax image that said candidate fax image is spam fax where said candidate fax image tests favorably to at least one of said characteristics of said known spam fax.