Object or shape information representation method

• US 9,256,802 B2
• Filed: 11/11/2011
• Issued: 02/09/2016
• Est. Priority Date: 11/26/2010
• Status: Active Grant

• Current Assignee: NEC Corporation, Tohoku University National Universities Corporation
• Original Assignee: NEC Corporation, Tohoku University National Universities Corporation
• Inventors: Matsuda, Yuma, Ogawa, Masatsugu, Yano, Masafumi, Kawakami, Susumu
• Primary Examiner(s): Desire, Gregory M
• Application Number:
  US13/988,655
• Publication Number:
  US 20130236108A1
• Time in Patent Office: 1,551 Days
• US Class Current: 1/1
• CPC Class Codes: G06V 10/46 Descriptors for shape, contour or point-related descriptors, e.g. scale invariant feature transform [SIFT] or bags of words [BoW]; Salient regional features colour feature extraction G06V10/56
• CPC Class Codes: G06V 10/471 using approximation functions

Abstract

An information representation method for representing an object or a shape includes: dividing a contour shape of an entirety or a part of the object or the shape into one or a plurality of curves; and representing the contour shape of the object or the shape by parameters including a degree of curvature and a positional relationship of each curve obtained by the dividing. Therefore, there is provided an information representation method for an object or a shape, which is capable of robust object recognition against a change in image by geometric transformations and occlusions.

Claims

1. An image data processing method for processing, by a processing unit, image data of an object or a shape by comparing the image data with comparison image data stored in an image information memory to determine a degree of similarity of the object or the shape and the comparison image data, comprising:

extracting contour information of an entirety or a part of the object or the shape from the image data;

stepwise smoothing the contour information of the entirety or the part of the object or the shape over a plurality of stages to obtain smoothed contour lines on each stage;

processing the contour information represented by each smoothed contour line on each stage to divide a contour shape represented by the contour information into one or a plurality of divided curves at each inflection point of each smoothed contour line by calculating a degree of curvature of each curve to store the degree of the curvature as information of each divided curve in the information memory; and

obtaining a feature amount of each divided curve on each of the stages to store the feature amount of the smoothed contour lines in the information memory to compare the feature amount with the corresponding feature amount of the comparison image data and to determine the degree of the similarity of the image data with the comparison image data.

2. An image data processing method according to claim 1, wherein the degree of curvature of the curve is represented by an area calculated by using a plurality of points on the curve or a plurality of points in positions which are determinable by a shape of the curve.

3. An image data processing method according to claim 1, wherein the degree of curvature of the curve is represented by any two or more of an area calculated by using a plurality of points on the curve or a plurality of points in positions which are determinable by a shape of the curve, an arc length of the curve, and a curvature of the curve.

4. An image data processing method according to claim 1, wherein the degree of curvature of the curve is represented by the following two:

an arc length of the curve and a curvature of the curve.

5. An image data processing method according to claim 1, wherein an arc length of the curve and a curvature of the curve are respectively an affine arc length and an affine curvature.

6. An image data processing method according to claim 1, wherein positions of a plurality of points on the curve or a plurality of points in positions which are determinable by the shape of the curve are invariant positions with respect to an affine geometric transformation.

7. An image data processing method according to claim 1, wherein a division point at which the division is carried out includes an inflection point at which a sign of a curvature on a contour changes.

8. An image data processing method according to claim 1, wherein a method used for the dividing uses a division method on two stages involving:

carrying out global division based on inflection points at which a sign of a curvature on a contour changes; and

further locally dividing a region between the inflection points.

9. An image data processing method according to claim 1, wherein an area defined by the curve is derived by using at least three points out of division points for dividing the curve into a plurality of parts based on an affine arc length of the curve as an index, and end points of the curve.

10. An image data processing method according to claim 1, wherein a positional relationship between the one or the plurality of curves is represented by an invariable with respect to an affine geometric transformation.

11. An image data processing method according to claim 1, wherein a positional relationship of the curve is a positional relationship among the one or the plurality of curves obtained by the dividing or a positional relationship among the one or the plurality of curves obtained by the dividing in a plurality of the contour shapes.

12. An image data processing method according to claim 1, wherein the positional relationship of the curve is represented by one or more indices selected from objective indices including a sequence or coordinates of the each curve.

13. An image data processing method according to claim 1, wherein the each curve obtained by the dividing is represented by parameters including a direction in addition to the degree of curvature and the positional relationship of the curve.

14. An image data processing method according to claim 1, wherein the each curve obtained by the dividing is represented by a plurality of pieces of stratified information.

15. An image data processing method according to claim 1, wherein, when two of the objects or the shapes are matched to each other, the image data processing method accompanies a geometric transformation.

16. A non-transitory computer-readable program recording medium which stores a program for causing a computer to execute the method of claim 1.

17. An image data processing system for processing image data of an object or a shape by comparing the image data with comparison image data stored in an image information memory to determine a degree of similarity of the object or the shape and the comparison image data, comprising:

a contour extractor configured to extract a contour information of an entirety or a part of the object or the shape from the image data;

a smoothing unit configured to stepwise smooth the contour information over a plurality of stages to obtain smoothed contour lines on each stage;

a processing unit configured to process the contour information represented by each smoothed contour line on each stage to divide a contour shape of the smoothed contour lines into one or a plurality of curves at each inflection point of each smoothed contour line by calculating a degree of curvature of each curve and to store the degree of the curvature as information of each divided curve in the information memory; and

a calculation unit configured to obtain a feature amount of each divided curve on each of the stages to store the feature amount of the smoothed contour lines in the information memory to compare the feature amount with the corresponding feature amount of the comparison image data and to determine the degree of the similarity of the image data with the comparison image data.