IMAGE RECTIFICATION USING SPARSELY-DISTRIBUTED LOCAL FEATURES

• US 20140140627A1
• Filed: 11/20/2012
• Published: 05/22/2014
• Est. Priority Date: 11/20/2012
• Status: Active Grant

• Current Assignee: Eastman Kodak Company
• Original Assignee: Eastman Kodak Company
• Inventors: Wu, Hao, Spaulding, Kevin Edward
• Granted Patent: US 9,008,444 B2
• Time in Patent Office: Days
• US Class Current: 382/203
• CPC Class Codes: G06V 10/247 by affine transforms, e.g. correction due to perspective effects; Quadrilaterals, e.g. trapezoids
• CPC Class Codes: G06V 30/1478 of characters or characters lines

Abstract

This invention is a method for rectifying an input digital image including warped textual information. The method includes analyzing the input digital image to locate a plurality of local features, at least some of the local features including textual features. A sparse set of local image regions are located corresponding to reliable combinations of spatially-consecutive local features, and corresponding local orientations are determined. A global deformation function is formed by interpolating between the determined local orientations and is used to form a rectified image.

Claims

1. A method for rectifying an input digital image including warped textual information, the method implemented at least in part by a data processing system and comprising:

receiving the input digital image;

analyzing the input digital image to locate a plurality of local features, at least some of the local features including textual features;

locating a sparse set of local image regions corresponding to reliable combinations of spatially-consecutive local features;

determining local orientations of the identified local image regions;

forming a global deformation function based on the determined local orientations;

forming a rectified image by rectifying the input digital image responsive to the global deformation function; and

storing the rectified image in a processor-accessible memory.

2. The method of claim 1 wherein the analysis of the input digital image includes an edge detection operation or a thresholding operation.

3. The method of claim 1 wherein the step of locating the reliable combinations of spatially consecutive textual features includes performing morphological operations to connect groups of nearby textual features.

4. The method of claim 3 wherein the step of locating the reliable combinations of spatially consecutive textual features includes analyzing shapes of the connected groups to identify and eliminate any connected groups that are unlikely to correspond to reliable combinations of spatially consecutive textual features.

5. The method of claim 4 wherein the step of analyzing shapes of the connected groups includes determining an aspect ratio associated with each connected group and eliminating any connected groups that do not satisfy an aspect ratio criterion.

6. The method of claim 4 wherein the step of analyzing shapes of the connected groups includes determining an orientation associated with each connected group and eliminating any connected groups that do not satisfy an orientation criterion.

7. The method of claim 4 wherein the step of analyzing shapes of the connected groups includes determining a size of each connected group and eliminating any connected groups that do not satisfy a size criterion.

8. The method of claim 1 wherein the step of determining local orientations includes:

fitting an ellipse having a major axis to each identified local image region; and

determining an orientation corresponding to an angle of the major axis of the ellipse relative to a horizontal direction.

9. The method of claim 1 wherein the step of determining the local orientation for a particular local image region includes:

defining a set of candidate orientations;

determining projection profiles by integrating pixels in the local image region along paths in the direction of the candidate orientations;

determining associated energy values from the projection profiles; and

designating the candidate orientation having the maximum energy value to be the local orientation.

10. The method of claim 1 wherein the step of forming the global deformation function includes:

determining an orientation vector field by interpolating between the determined local orientations for a lattice of positions;

determining a set of streamlines responsive to the orientation vector field; and

forming the global deformation function by interpolating between the streamlines.

11. The method of claim 1 wherein the step of forming the global deformation function includes:

fitting a parametric orientation function to the determined local orientations;

determining a set of streamlines responsive to the parametric orientation function; and

forming the global deformation function by interpolating between the streamlines.

12. The method of claim 1 wherein the global deformation function is represented using a two-dimensional look-up table that stores displacement vectors for a lattice of image positions.

13. The method of claim 1 wherein the global deformation function is represented using a two-dimensional look-up table that stores warped image positions corresponding to a lattice of rectified image positions.

14. The method of claim 1 wherein the global deformation function is represented using a parametric function that computes displacement vectors as a function of position in the input digital image.

15. The method of claim 1 wherein the textual features correspond to characters, portions of characters or groups of characters.

16. The method of claim 1 wherein at least some of the local image regions correspond to words or groups of words.