Method and apparatus for distortion correction of scanned images

• US 5,497,236 A
• Filed: 06/23/1993
• Issued: 03/05/1996
• Est. Priority Date: 06/23/1993
• Status: Expired due to Term

• Current Assignee: Ricoh Company Limited, Ricoh Corporation (Ricoh Company Limited)
• Original Assignee: Ricoh Company Limited, Ricoh Corporation (Ricoh Company Limited)
• Inventors: Wolff, Gregory J., Stork, David G.
• Primary Examiner(s): Wong, Peter S.
• Assistant Examiner(s): FRAHM, ERIC S
• Application Number:
  US08/082,118
• Time in Patent Office: 986 Days
• Field of Search: 358/296, 358/300, 358/406, 358/443, 358/447, 358/471, 358/474, 358/482, 358/486, 358/493, 358/497, 358/498, 358/504, 382/254, 382/255, 382/319, 355/208, 355/233
• US Class Current: 358/296
• CPC Class Codes: G06T 3/06 Topological mapping of higher dimensional structures onto lower dimensional surfaces
• CPC Class Codes: H04N 1/047 Detection, control or error compensation of scanning velocity or position H04N1/0402 and H04N1/17 take precedence
• CPC Class Codes: H04N 1/1013 with sub-scanning by translatory movement of at least a part of the main-scanning components H04N1/107 takes precedence
• CPC Class Codes: H04N 1/193 using electrically scanned linear arrays , e.g. linear CCD arrays
• CPC Class Codes: H04N 2201/02425 Self-adjusting arrangements, e.g. compensating for temperature fluctuations
• CPC Class Codes: H04N 2201/03162 Original guide plate
• CPC Class Codes: H04N 2201/0434 specially adapted for scanning pages of a book
• CPC Class Codes: H04N 2201/04703 using the scanning elements as detectors, e.g. by performing a prescan
• CPC Class Codes: H04N 2201/04729 in the main-scan direction
• CPC Class Codes: H04N 2201/04731 in the sub-scan direction
• CPC Class Codes: H04N 2201/04734 Detecting at frequent intervals, e.g. once per line for sub-scan control
• CPC Class Codes: H04N 2201/04743 by detecting the image directly
• CPC Class Codes: H04N 2201/04787 by changing or controlling the addresses or values of pixels, e.g. in an array, in a memory, by interpolation
• CPC Class Codes: H04N 2201/04789 in the main-scan direction
• CPC Class Codes: H04N 2201/04791 in the sub-scan direction

Abstract

An improved method and apparatus for correcting for splay is provided. A document distorted by the curvature of a page of text away from a platen is converted to a digital image. The digital image is the manipulated to remove the distortion by fitting the lines of text in an unsplayed portion to a skew line, which represents the deviation of lines of text in the digital image from horizontal. Then the splay is determined for each line of text. Once the skew and the splay are determined, an inverse transformation is done to straighten the lines of text. A horizontal stretching is also applied to the text to correct for the projection angle of the original document.

Claims

1. A distortion correction apparatus for correcting distortion in a distorted image of a page caused by the page not being fully aligned with an image capture surface at the time the distorted image is captured, comprising:

input means, for accepting the distorted image and formatting the distorted image into a form processable by the distortion correction apparatus;

a line locator coupled to said input means, for identifying paths of each of a plurality of lines within the distorted image;

a coefficient generator coupled to said line locator, for generating, from said paths, coefficient values representative of distortion of said plurality of lines;

a transform means coupled to said coefficient generator and said input means, for transforming the distorted image into a corrected image according to a transformation dictated by a predetermined relation among said coefficient values; and

output means coupled to said transform means, for outputting said corrected image, wherein said corrected image comprises an image of the page with less spatial distortion than the spatial distortion in the distorted image.

2. The apparatus of claim 1, wherein said line locator locates lines of text and further comprises:

a column identifier for identifying columnar regions in the distorted image, wherein said column identifier outputs a text image comprising a column of text lines;

a text smoother for generating a banded image from said text image, wherein each band in said banded image corresponds to one of said text lines in said text image; and

an edge detector for generating an edge line array from said banded image, wherein each edge line in said edge line array corresponds to one of said bands in said banded image, and wherein said edge line array represents said paths of each of said plurality of text lines.

3. The apparatus of claim 1, wherein said coefficient generator further comprises:

an apportioning means for apportioning each of said paths into a splayed portion and an unsplayed portion separated by a break point;

a first modelling means coupled at least to said apportioning means, for modelling said unsplayed portion as a straight line; and

a second modelling means coupled at least to said apportioning means, for modelling said splayed portion as an exponential curve.

4. The apparatus of claim 3, wherein said apportioning means comprises a break point locator which locates said break point by fitting a straight line to an image line, said break point being a point of intersection of said straight line and said image line which is closer to the more distorted end of said image line than any other point of intersection.

5. The apparatus of claim 1, wherein the distorted image is captured using an optical sensor in a photocopying machine.

6. The apparatus of claim 1, wherein said transform means comprises a vertical transform means and a horizontal transform means.

7. The apparatus of claim 1, wherein said input means comprises means for identifying columns of text and for outputting said columns in series to said line locator, whereby each column of text in the distorted image is separately corrected for distortion.

8. The apparatus of claim 1, wherein said input means formats the distorted image into a two-dimensional array of digital pixel values.

9. The apparatus of claim 1, wherein said predetermined relation is a modelling of an image line comprising a portion modelled by a straight line and another portion modelled by an exponential curve.

10. A distortion correction apparatus for correcting distortion in a distorted image of a page caused by the page not being fully aligned with an image capture surface at the time the distorted image is captured, comprising:

a column identifier for identifying columnar regions in the distorted image, wherein said column identifier outputs a text image comprising a column of lines of text;

a text smoother for generating a banded image from said text image, wherein each band in said banded image corresponds to one of said lines of text in said text image;

an edge detector for generating an edge line array from said banded image, wherein each edge line in said edge line array corresponds to one of said bands in said banded image;

a coefficient generator for generating from said edge line array coefficient values representative of distortion in said text image;

a transform means coupled to said coefficient generator and said column identifier, for transforming said text image into a corrected image according to a transformation dictated by a predetermined relation among said coefficient values; and

output means coupled to said transform means, for outputting said corrected image, wherein said corrected image comprises an image of the page with less spatial distortion than the spatial distortion in the distorted image.

11. A method of correcting for distortion in an image of a text page where said distortion is caused by the text page not being fully aligned with an image capture surface at the time the distorted image is captured, comprising the steps of:

capturing the distorted image in a form processable by a distortion correction apparatus;

locating a text line path for each of a plurality of text lines within the distorted image;

identifying a break point for each of said plurality of text lines, where a break point for a given text line separates an undistorted portion of said given text line from a distorted portion of said given text line;

fitting a straight line to said undistorted portion;

fitting an exponential curve to said distorted portion;

generating an inverse transform for said given text line, where said inverse transform derives from coefficients of said break point, said straight line, and said exponential curve for said given line, and said inverse transform of said given text line better approximates a straight line of text than said given text line before said inverse transform is applied;

transforming said given text line according to said inverse transform; and

outputting a corrected image comprising at least a transform of said given text line, wherein said corrected image comprises an image of the page with less spatial distortion than the spatial distortion in the distorted image.

12. The method of claim 11, wherein said step of locating a text line path further comprises the steps of:

identifying columnar regions in the distorted image;

separating said columnar regions into at least one text image;

smoothing said at least one text image into a banded image, wherein each band in said banded image corresponds to a text line in said at least one text image; and

locating an edge line for each of said bands, wherein said edge line represents said text line path.

13. The method of claim 11, wherein said step of generating an inverse transform further comprises:

apportioning each of said text line locations into a splayed portion and an unsplayed portion separated by a break point;

modelling said unsplayed portion as a straight line; and

modelling said splayed portion as an exponential curve.

14. The method of claim 13, wherein said step of apportioning further comprises the steps of locating said break point by fitting a straight line to said text line, said break point being a point of intersection of said straight line and said text line which is closer to the more distorted end of said text line than any other point of intersection.

15. The method of claim 11, further comprising the step of initially capturing the distorted image using an optical sensor in a photocopying machine.

16. The method of claim 11, wherein said step of transforming a text line further comprises the steps of:

vertically transforming said text line into a vertically transformed text line; and

horizontally transforming said vertically transformed text line.

17. A method of correcting for distortion in an image of a page photocopied from a bound book, the bound book preventing the page from being laid flat on a platen glass of a photocopier, comprising the steps of:

a) capturing an image of the page using conventional photocopying techniques, thereby producing a copied image;

b) locating at least one column of text in said copied image, said at least one column of text characterized by a region of relatively high variance in pixel values between regions of relatively low variance in pixel values;

c) transforming each line of text in said at least one column of text, said step of transforming comprising the steps of;

1) identifying an edge line corresponding to a line of text to be transformed;

2) identifying a skew line for said edge line, said skew line being a straight line of best fit to points comprising said edge line;

3) locating a break point on said edge line, said break point being a point which separates an unsplayed portion of said edge line from a splayed portion, said unsplayed portion being substantially modelled by said skew line, and said splayed portion being substantially nonlinear;

4) fitting an exponential curve to said splayed portion, wherein said exponential curve is represented by at least one coefficient;

5) fitting a horizontal linear correction to deviations of said at least one coefficient from a linear function of a vertical position of said splayed portion;

6) transforming said line of text according to a vertical inverse transform derived from said skew line, said exponential curve, and said horizontal linear correction, to form a vertically corrected line of text;

7) horizontally correcting said vertically corrected line of text according to a predetermined function of said at least one coefficient derived in said step of fitting an exponential curve, to form a transformed line of text; and

d) assembling each transformed line of text into a corrected image, whereby said corrected image comprises an image of the page with less spatial distortion than the copied image.