IMAGE PROCESSING METHOD AND SYSTEM FOR EXTRACTING DISTORTED CIRCULAR IMAGE ELEMENTS

• US 20160283821A1
• Filed: 03/25/2016
• Published: 09/29/2016
• Est. Priority Date: 03/26/2015
• Status: Active Grant

• Current Assignee: HL Mando Corporation (Halla Holdings Corporation)
• Original Assignee: Mando Corporation
• Inventors: ABHAU, Jochen
• Granted Patent: US 9,953,238 B2
• Time in Patent Office: Days
• US Class Current: 1/1
• CPC Class Codes: G06T 2207/10024 Color image
• CPC Class Codes: G06T 2207/20061 Hough transform
• CPC Class Codes: G06T 2207/20076 Probabilistic image processing
• CPC Class Codes: G06T 2207/30252 Vehicle exterior; Vicinity of vehicle
• CPC Class Codes: G06T 7/11 Region-based segmentation
• CPC Class Codes: G06V 10/42 Global feature extraction by analysis of the whole pattern, e.g. using frequency domain transformations or autocorrelation
• CPC Class Codes: G06V 10/56 relating to colour
• CPC Class Codes: G06V 20/582 of traffic signs

Abstract

The invention relates to an image processing method for automatically extracting distorted circular image elements (109; 601) from an image (101), comprising the steps of determining, by means of a circle detection (102), at least one circular image section (103; 301, 301′) as a first approximation of at least one distorted circular image element (109; 601), then determining, via a statistical colour analysis of the at least one determined circular image section (301, 301′), whether the circular image section (301, 301′) belongs to one of one or more predefined colour classes and, in case thereof, assigning the circular image section to the respective colour class, and, finally, determining an exact shape (109; 601) of the at least one distorted circular element by starting with at least one of the one or more determined circular image sections assigned to one of the one or more colour classes and varying the image section to maximize a value of a colour-class specific function of the image section. The invention furthermore relates to an image processing system which is configured for performing the steps according to the image processing method described above, and to a use of the image processing method or of the image processing system for determining ground truth values of a given input data set of images.

Claims

1. An image processing method for automatically extracting distorted circular image elements from an image, comprising the steps of

(a) determining, by means of a circle detection, at least one circular image section as a first approximation of at least one distorted circular image element,

(b) determining, via a statistical colour analysis of the at least one determined circular image section, whether the circular image section belongs to one of one or more predefined colour classes and, in case thereof, assigning the circular image section to the respective colour class, and

(c) determining an exact shape of the at least one distorted circular element by starting with at least one of the one or more determined circular image sections assigned to one of the one or more colour classes and varying the image section to maximize a value of a colour-class specific function of the image section.

2. The image processing method of claim 1, wherein the circle detection is a Hough circle detection.

3. The image processing method of claim 1, wherein the colour-class specific function assigns a colour average dependent value to each image section.

4. The image processing method according to claim 2, wherein the colour average dependent value is a weighted sum of brightness gradients of an outline of the respective image section and/or of pixel values of at least one colour channel over at least a part of the respective image section.

5. The image processing method according to claim 1, wherein the distorted circular image elements are distorted round traffic signs.

6. The image processing method according to claim 1, wherein, prior to step (c), the circular image sections determined by the circle detection are grouped according to their radiuses and centre points, only one representative or a proper subset of representatives per group is selected and the determining of the exact shape is restricted to the selected representatives.

7. The image processing method according to claim 1, wherein, in step (c), the at least one image section is varied iteratively.

8. The image processing method of claim 1, wherein the varying of the image section is restricted to elliptical shapes.

11. A use of the image processing method of claim 1 for determining ground truth values of a given input data set of images, the input data set and determined values being used or suitable for training, testing, calibrating, analysing and/or improving another image processing method or system.

9. An image processing system for automatically extracting distorted circular image elements, including an evaluation unit, the evaluation unit being configured for analysing an image, by performing the following steps:

(a) determining, by means of a circle detection, at least one circular image section of the image, as a first approximation of at least one distorted circular image element,

(b) determining, via a statistical colour analysis of the at least one determined circular image section whether the circular image section belongs to one of one or more predefined colour classes and, in case thereof, assigning the circular image section to the respective colour class, and

(c) determining an exact shape of the at least one distorted circular element by starting with at least one of the one or more determined circular image sections assigned to one of the one or more colour classes and varying the image section to maximize a value of a colour-class specific function of the image section.

10. The image processing system according to claim 9, wherein the circle detection is a Hough circle detection.

12. The image processing system of claim 9, wherein the colour-class specific function assigns a colour average dependent value to each image section.

13. The image processing system of claim 10, wherein the colour average dependent value is a weighted sum of brightness gradients of an outline of the respective image section and/or of pixel values of at least one colour channel over at least a part of the respective image section.

14. The image processing system of claim 9, wherein the distorted circular image elements are distorted round traffic signs.

15. The image processing system of claim 9, wherein the circular image sections determined by the circle detection are grouped according to their radiuses and centre points, only one representative or a proper subset of representatives per group is selected and the determining of the exact shape is restricted to the selected representatives.

16. The image processing system of claim 9, wherein the at least one image section is varied iteratively.

17. The image processing system of claim 9, wherein the varying of the image section is restricted to elliptical shapes.