Model-based dewarping method and apparatus

Abstract

An apparatus and method for processing a captured image and, more particularly, for processing a captured image comprising a document. In one embodiment, an apparatus comprising a camera to capture documents is described. In another embodiment, a method for processing a captured image that includes a document comprises the steps of distinguishing an imaged document from its background, adjusting the captured image to reduce distortions created from use of a camera and properly orienting the document is described.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

FIG. 1A depicts a prior art document scanner.

FIG. 1B depicts a prior art digital camera.

FIG. 2 depicts a general flowchart of a preferred method for processing a captured image.

FIG. 3 depicts a flowchart of another embodiment of a method for processing a captured image.

FIG. 4 depicts a flowchart of a method of performing segmentation in accordance with one of the implementations of the method of imaging a document disclosed herein.

FIG. 5 depicts a flowchart of one method of performing the random sample consensus step illustrated in FIG. 4.

FIG. 6 depicts a flowchart of one method of performing the outlier removal step illustrated in FIG. 4.

FIG. 7 depicts a flowchart for another method of performing segmentation in accordance with the method of imaging a document disclosed herein.

FIG. 8 depicts a flowchart of one method of performing the distortion removal steps illustrated in FIG. 2 and FIG. 3.

FIG. 9 depicts a flowchart of one method of performing the lines of text step illustrated in FIG. 3.

FIG. 10 depicts a flowchart of one method of determining whether a document is properly oriented in an upright manner in accordance with one implementation of the method of imaging a document disclosed herein.

FIG. 11 depicts one embodiment of an apparatus for capturing and processing an image including an imaged document.

FIG. 12 depicts a flowchart of one method of determining whether a document is oriented in an upright manner in accordance with one implementation of the method of imaging a document disclosed herein.

FIG. 13 depicts one embodiment of a system for processing a captured image.

FIG. 14 depicts a flowchart of a method of performing image dewarping according to the present invention.

FIG. 15 depicts a flowchart of a method for generating a curved coordinate net to an image of a warped document.

FIG. 16 depicts a flowchart of one method of performing the local direction enhancement step illustrated in FIG. 15.

FIG. 17 depicts a segmented original image to be corrected by the image dewarping technique of FIG. 14 and illustrates an alternative method of applying a curved coordinate net to an image of a warped document.

FIG. 18 depicts an example of an image of a warped document page with a curved coordinate net.

FIG. 19 depicts an image of a document with a stretched, non-proportional rectangular net produced after the stretching step of FIG. 14.

FIG. 20 depicts a projection diagram in 3D for illustrating the model used in a preferred embodiment of the present invention for performing the stretching step of FIG. 14.

FIG. 21 depicts a rectangular net following the adjustment step of FIG. 14.

FIG. 22 depicts a projection diagram for further illustrating a preferred model for performing the stretching step of FIG. 14.

FIG. 23 depicts a side view of an opened book for illustrating a preferred model for performing the adjustment step of FIG. 14.

FIG. 24 depicts a projection diagram for further illustrating a preferred model for performing the adjustment step of FIG. 14.

Claims

1. A method for processing a captured image that comprises an imaged document; said method comprising: detecting graphical information in the captured image relating to the transition between said imaged document and the remainder of said captured image;selecting one or more lines from said graphical information corresponding to edges of said imaged document;calculating corners of said imaged document based on intersection of one or more lines corresponding to edges of said imaged document;isolating said imaged document from background of said captured image based on one or more lines corresponding to edges of said imaged document;setting a curved coordinate net on said imaged document;stretching said curved coordinate net to a rectangle coordinate net;adjusting said rectangle coordinate net to a well-proportional rectangle coordinate net.

2. The method recited in claim 1, further comprising the steps of: mapping coordinates of pixels of said imaged document to coordinates corresponding to a non-distorted perspective of said imaged document based on said computed deviation.

3. The method recited in claim 2, further comprising the steps of: rotating said non-distorted imaged document according to said determination of format of said non-distorted imaged document.

4. A system for processing a captured image, said captured image comprising an imaged document, said system comprising: means for selecting one or more lines from said graphical information corresponding to edges of said imaged document;means for calculating corners of said imaged document based on intersection of one or more lines corresponding to edges of said imaged document;means for isolating said imaged document from background of said captured image based on one or more lines corresponding to edges of said imaged document;means for setting a curved coordinate net on said imaged document;means for stretching said curved coordinate net to a rectangle coordinate net; andmeans for adjusting said rectangle coordinate net to a well-proportional rectangle coordinate net.

5. A computer readable medium for processing captured image, the computer readable medium carrying one or more sequences of one or more instructions which, when executed by one or more processors, cause the one or more processors to perform the computer-implemented steps of: detecting graphical information in the captured image relating to the transition between said imaged document and the remainder of said captured image;selecting one or more lines from said graphical information corresponding to edges of said imaged document;calculating corners of said imaged document based on intersection of one or more lines corresponding to edges of said imaged document;isolating said imaged document from background of said captured image based on one or more lines corresponding to edges of said imaged document;setting a curved coordinate net on said imaged document;stretching said curved coordinate net to a rectangle coordinate net;adjusting said rectangle coordinate net to a well-proportional rectangle coordinate net.resampling pixels of said imaged document based on said computed deviations;

6. The method recited in claim 5, further comprising the steps of: mapping coordinates of pixels of said imaged document to coordinates corresponding to a non-distorted perspective of said imaged document based on said computed deviation.

7. The method recited in claim 6, further comprising the steps of: rotating said non-distorted imaged document according to said determination of format of said non-distorted imaged document.

8-9. (canceled)