Methods for extraction and recognition of pattern in an image method for image abnormality judging, and memory medium with image processing programs

• US 6,272,244 B1
• Filed: 08/05/1998
• Issued: 08/07/2001
• Est. Priority Date: 08/06/1997
• Status: Expired due to Fees

• Current Assignee: Nippon Telegraph and Telephone Corporation
• Original Assignee: Nippon Telegraph and Telephone Corporation
• Inventors: Takahashi, Yasuko, Ohtsuka, Sakuichi, Shio, Akio
• Primary Examiner(s): Johns, Andrew W.
• Assistant Examiner(s): AZARIAN, SEYED H
• Application Number:
  US09/129,410
• Time in Patent Office: 1,098 Days
• Field of Search: 382/190, 382/232, 382/284, 382/156, 382/157, 382/199, 382/205, 382/206, 382/209, 382/217, 382/218, 382/275, 348/475, 235/494
• US Class Current: 382/190
• CPC Class Codes: G06T 7/0004 Industrial image inspection
• CPC Class Codes: G06T 7/74 involving reference images or patches
• CPC Class Codes: G06V 20/625 License plates
• CPC Class Codes: G06V 20/63 Scene text, e.g. street names

Abstract

A method for extracting a pre-selected pattern from an image by a programmed computer is able to identify and extract the pattern even under adverse conditions such as presence of noise or shadow in the image or the target pattern is partially shielded. The steps are: storing reference patterns for referencing; storing input images; applying filtering processes to reference patterns and input images; integrating filtered results of the reference pattern and filtered results of the input image to generate integrated results; and extracting a reference pattern according to the integrated results and specifying a location of the reference pattern in the input image. An image abnormality detection can also be performed using the basic method to identify a normal scene and detect a deviation from normalcy even under adverse conditions.

Claims

1. A method for extracting a pattern from an image recorded by a programmed computer comprising the steps of:

storing a reference for extraction;

storing input images;

converting pixel values of said input image for correcting any conflicts in converting analog camera data of an image to digital data, and gradient (∇

=(∂

/∂

x, ∂

/∂

y), directional differential) processing;

integrating the gradient of said reference pattern and the gradient of said input image to generate integrated results; and

extracting the reference pattern similar to said input image according to said integrated results.

2. A method according to claim 1, wherein said converting step is carried out to separate surface reflectivity effects from intensity effect.

3. A method according to claim 1, wherein said step of integrating includes a convolution of an inner product of a reference pattern filter output and an input image filter output.

4. A method according to claim 2, wherein said step of integrating includes a convolution of an inner product of a reference pattern filter output and an input image filter output.

5. A method according to claim 1, wherein said step of integrating includes a convolution of an outer product of a reference pattern filter output and an input image filter output.

6. A method according to claim 2, wherein said step of integrating includes a convolution of an outer product of a reference pattern filter output and an input image filter output.

7. A method according to claim 1, wherein said step of integrating includes a difference of a convolution of a reference pattern filter output and an input image filter output.

8. A method according to claim 2, wherein said step of integrating includes a difference of a convolution of a reference pattern filter output and an input image filter output.

9. A method according to claim 2, wherein said step of integrating includes a step which gives the contribution which shows the degree of the difference between a plurality of reference patterns.

10. A method according to claim 3, wherein said step of integrating includes a step which gives the contribution which shows the degree of the difference between a plurality of reference patterns.

11. A method according to claim 4, wherein said step of integrating includes a step which gives the contribution which shows the degree of the difference between a plurality of reference patterns.

12. A method according to claim 9, wherein said step of weighting relates to a numerical technique of expressing a pattern shape factor common to a plurality of reference patterns, and prompts an extraction of a category of similar reference patterns instead of extracting one reference pattern according to an integration result.

13. A method according to claim 10, wherein said step of weighting relates to a numerical technique of expressing a pattern shape factor common to a plurality of reference patterns, and prompts an extraction of a category of similar reference patterns instead of extracting one reference pattern according to an integration result.

14. A method according to claim 11, wherein said step of weighting relates to a numerical technique of expressing a pattern shape factor common to a plurality of reference patterns, and prompts an extraction of a category of similar reference patterns instead of extracting one reference pattern according to an integration result.

15. A recording medium for executing extraction programs, for extracting a pattern from an image recorded by a computer, comprised by the steps of:

storing reference patterns for extraction;

storing input images;

comprising a step of a logarithmic conversion with a table of pixel values of said input image for correcting any conflicts in converting analog camera data of an image to digital data, and a step of applying at least one processing step selected from the group consisting of a convolution and a gradient (∇

=(∂

/∂

x, ∂

/∂

y), directional differential) processing;

applying a filter means to reference patterns and input images;

integrating filtered results of said reference pattern and filtered results of said input image to generate integrated results; and

extracting a reference pattern according to said integrated results and specifying a location of said reference pattern in said input image.

17. A recording medium according to claim 15, wherein said step of integrating includes a convolution of an inner product of a reference pattern filter output and an input image filter output thereby integrating said reference pattern filter output and said image filter output thereby integrating said reference pattern filter output and said image filter output.

19. A recording medium according to claim 15, wherein said step of integrating includes a convolution of an outer product of a reference pattern filter output and an input image filter output thereby integrating said reference pattern filter output and said image filter output.

21. A recording medium according to claim 15, wherein said step of integrating includes a difference of a convolution of a reference pattern filter output and an input image filter output thereby integrating said reference pattern filter output and said image filter output.

23. A recording medium according to claim 15, wherein said step of integrating includes a step of weighting according to a degree of contribution of a coordinate point in a reference pattern.

25. A recording medium according to claim 17, wherein said step of integrating includes a step of weighting according to a degree of contribution of a coordinate point in a reference pattern.

28. A recording medium according to claim 23, wherein said step of weighting relates to a numerical technique of expressing a pattern shape factor common to a plurality of reference patterns, and prompts an extraction of a category of similar reference patterns instead of extracting one reference pattern according to an integration result.

29. A recording medium according to claim 25, wherein said step of weighting relates to a numerical technique of expressing a pattern shape factor common to a plurality of reference patterns, and prompts an extraction of a category of similar reference patterns instead of extracting one reference pattern according to an integration result.

16. A recording medium for executing extraction program, for extracting a pattern from an image recorded by a computer, comprised by the steps of:

storing reference patterns for extraction;

storing input images;

comprising a step of a logarithmic conversion of pixel values of said input image for correcting any conflicts in converting analog camera data of an image to digital data, and a step of applying at least one processing step selected from the group consisting of a convolution and a gradient (∇

=(∂

/∂

x, ∂

/∂

y), directional differential) processing;

applying filter means to reference patterns and input images;

integrating filtered results of said reference pattern and filtered results of said input image to generate integrated results; and

extracting a reference pattern according to said integrated results and specifying a location of said reference pattern in said input image.

18. A recording medium according to claim 16, wherein said step of integrating includes a convolution of an inner product of a reference pattern filter output and an input image filter output thereby integrating said reference pattern filter output and said image filter output thereby integrating said reference pattern filter output and said image filter output.

20. A recording medium according to claim 16, wherein said step of integrating includes a convolution of an outer product of a reference pattern filter output and an input image filter output thereby integrating said reference pattern filter output and said image filter output.

22. A recording medium according to claim 16, wherein said step of integrating includes a difference of a convolution of a reference pattern filter output and an input image filter output thereby integrating said reference pattern filter output and said image filter output.

24. A recording medium according to claim 16, wherein said step of integrating includes a step of weighting according to a degree of contribution of coordinate point in a reference pattern.

26. A recording medium according to claim 18, wherein said step of integrating includes a step of weighting according to a degree of contribution of a coordinate point in a reference pattern.

27. A recording medium according to claim 24, wherein said step of weighting relates to a numerical technique of expressing a pattern shape factor common to a plurality of reference patterns, and prompts a n extraction of a category of similar reference patterns instead of extracting one reference pattern according to an integration result.

30. A recording medium according to claim 26, wherein said step of weighting relates to a numerical technique of expressing a pattern shape factor common to a plurality of reference patterns, and prompts an extraction of a category of similar reference patterns instead of extracting one reference pattern according to an integration result.

31. A method for recognizing a target pattern, in stored input images, which is similar to a plurality of pre-stored reference patterns and associated pattern locations, comprising the steps of:

cropping of target regions for determining interim target regions;

computing local similarities between said target pattern and said reference patterns in obtaining intermediate data by applying a local computation to said input image;

logarithmic converting of pixel values of said input image for correcting any conflicts in converting analog camera data of an image to digital data;

computing a degree of resemblance by comparing the gradient of said target pattern and said reference patterns;

extracting local patterns according to results of said degree of resemblance to enable selecting similar reference patterns;

restoring extracted regions so as to match with an original input image by verifying extracted pattern locations with reference pattern locations;

deciding on a matching reference pattern for each extracted region; and

outputting recognition results and associated reference pattern locations according to the results of comparison.

32. A method according to claim 31, wherein said target pattern in stored input images include numbers on license plates;

and said reference patterns are characters including numerals, dots, kanji, hiragana and alphabet letters.

33. A computer-readable recording medium for executing a program to recognize a target pattern in stored input images, comprised by the steps of:

cropping of target regions for determining interim target regions;

computing local similarities between said target pattern and said reference patterns in obtaining intermediate data by applying a local computation to said input image;

logarithmic converting of pixel values of said input image for correcting any conflicts in converting analog camera data of an image to digital data;

computing a degree of resemblance by comparing the gradient of said target pattern and said reference patterns;

extracting local patterns according to results of said degree of resemblance to enable selecting similar reference patterns;

restoring extracted regions so as to match with an original input image by verifying extracted pattern locations with reference pattern locations;

deciding on a matching reference pattern for each extracted region; and

outputting recognition results and associated reference pattern locations according to results of comparison.

34. A method for detecting an image abnormality, by comparing a comparison image with a reference image so as to detect a change, comprising the steps of:

storing a plurality of reference images to be referenced against said comparison image;

storing a plurality of comparison images for comparing with said plurality of reference images;

specifying a scanning area in said comparison image and in said reference image for detection of an abnormality;

applying filter means, to said reference image and said comparison image so as to obtain intermediate data, including a step of applying a logarithmic conversion with a table of pixel values of said input image for correcting any conflicts in converting analog camera data of an image to digital data and a step of applying at least one processing step selected from the group consisting of a convolution and a gradient (∇

=(∂

/∂

x, ∂

/∂

v), directional differential) processing;

computing a degree of resemblance of comparison image filtered results and reference image filtered results using similarity identification means selected from the group consisting of an inner product, an outer product, and correlation; and

identifying an abnormality when said degree of resemblance is not more than a specific value, and issuing an abnormality warning as a result of a recognition process.

35. A recording medium for executing an image abnormality detection programs comprised by the steps of:

storing a plurality of reference images to be referenced against said comparison image;

storing a plurality of comparison images for comparing with said plurality of reference images;

specifying a scanning area in said comparison image and in said reference image for detection of an abnormality;

applying filter means, to said reference image and said comparison image so as to obtain intermediate data, including a step of applying a logarithmic conversion with a table of pixel values of said input image for correcting any conflicts in converting analog camera data of an image to digital data and a step of applying at least one processing step selected from the group consisting of a convolution and a gradient (∇

=(∂

/∂

x, ∂

/∂

y), directional differential) processing;

computing a degree of resemblance of comparison image filtered results and reference image filtered results using similarity identification means selected from the group consisting of an inner product, an outer product, and correlation; and

identifying an abnormality when said degree of resemblance is not more than a specific value, and issuing an abnormality warning as a result of a recognition process.