Method and apparatus for recognizing edges in a camera image, and vehicle

Patent metadata

Field	Value
Publication number	US-10719938-B2
Application number	US-201816000060-A
Country	US
Kind code	B2
Filing date	Jun 5, 2018
Priority date	Jun 8, 2017
Publication date	Jul 21, 2020
Grant date	Jul 21, 2020

Abstract

A method of recognizing edges in a camera image (A) produced by a camera, includes steps of: assigning (S 1 ) a respective scaling size to each pixel (aij) of the camera image (A) as a function of an object distance, relative to the camera, of a spatial region imaged in the respective pixel (aij); producing (S 2 ) an integral image (B) from the camera image (A); applying (S 3 ) Haar wavelet transforms to the pixels (bij) of the integral image (B) to generate a gradient image (C), wherein a scaling of the Haar wavelets is determined for each pixel (bij) of the integral image (B) taking account of the scaling size assigned to the corresponding pixel (aij) of the camera image (A); and recognizing (S 4 ) edges in the camera image (A) using the generated gradient image (C).

First claim

The invention claimed is:

1. A method of recognizing edges in a camera image of an environment scene, wherein the camera image has plural pixels and has been produced by a camera, the method comprising the steps:
assigning a respective scaling size to each respective pixel of the camera image as a function of an object distance, relative to the camera, of a spatial region of the environment scene imaged in the respective pixel of the camera image;
producing an integral image from the pixels of the camera image;
applying Haar wavelet transforms to pixels of the integral image to generate a gradient image, wherein a respective scaling of Haar wavelets applied in the Haar wavelet transforms is determined for each respective pixel of the integral image dependent on the respective scaling size assigned to the respective pixel of the camera image that corresponds to the respective pixel of the integral image; and
recognizing edges in the camera image using the gradient image.

2. The method according to claim 1 , wherein a plurality of the gradient images is generated, wherein the respective scaling of the Haar wavelets applied in the Haar wavelet transforms to generate the plurality of the gradient images additionally depends on a respective global scaling factor which is determined differently for each of the plurality of gradient images.

3. The method according to claim 2 , wherein the global scaling factors differ from one another by predefined factors.

4. The method according to claim 2 , wherein a homogeneity of regions adjacent to the edges is determined based on the plurality of gradient images.

5. The method according to claim 4 , wherein a respective corresponding gradient strength around the edges is determined respectively for each of the plurality of gradient images, and the homogeneity is determined taking account of the gradient strengths.

6. The method according to claim 1 , wherein the respective scaling of the Haar wavelets is selected so that a dimension of the respective spatial region imaged in the respective pixel of the camera image that corresponds to the respective pixel of the integral image for which the respective scaling is determined during the Haar wavelet transforms is smaller than a predefined value.

7. The method according to claim 1 , wherein the object distances of the spatial regions imaged in the pixels of the camera image are estimated based on an alignment of the camera relative to a ground.

8. An apparatus, including a microprocessor, for recognizing edges in a camera image of an environment scene, wherein the camera image has plural pixels and has been produced by a camera, wherein the apparatus is configured:
to assign a respective scaling size to each respective pixel of the camera image as a function of an object distance, relative to the camera, of a spatial region of the environment scene imaged in the respective pixel of the camera image;
to produce an integral image from the pixels of the camera image;
to apply Haar wavelet transforms to pixels of the integral image to generate a gradient image, wherein a respective scaling of Haar wavelets applied in the Haar wavelet transforms is determined for each respective pixel of the integral image dependent on the respective scaling size assigned to the respective pixel of the camera image that corresponds to the respective pixel of the integral image; and
to recognize edges in the camera image using the generated gradient image.

9. The apparatus according to claim 8 , which is further configured to generate a plurality of the gradient images, wherein the respective scaling of the Haar wavelets applied in the Haar wavelet transforms to generate the plurality of the gradient images additionally depends on a respective global scaling factor which is determined differently for each of the plurality of gradient images.

10. The apparatus according to claim 9 , wherein the global scaling factors differ from one another by predefined factors.

11. The apparatus according to claim 9 , which is further configured to determine a homogeneity of regions adjacent to the edges based on the plurality of gradient images.

12. The apparatus according to claim 11 , which is further configured to determine a respective corresponding gradient strength around the edges respectively for each of the plurality of gradient images, and to determine the homogeneity taking account of the gradient strengths.

13. A vehicle comprising the apparatus according to claim 8 in combination with a vehicle body, and a vehicle camera device including the camera configured to produce the camera image of the environment scene of surroundings of the vehicle, wherein the apparatus is configured to recognize the edges in the camera image.

14. The method according to claim 1 , wherein the step of producing the integral image comprises determining each respective one of the pixels of the integral image as a respective sum of all of the pixels of the camera image between a position of an origin point of the camera image and a position of a respective one of the pixels of the camera image corresponding to the respective pixel of the integral image.

15. The method according to claim 1 , wherein the gradient image comprises gradient image pixels, and the step of recognizing edges in the camera image using the gradient image comprises identifying that the edges exist in the camera image at locations corresponding to the gradient image pixels of which respective values are greater than a predefined threshold value.

16. The method according to claim 4 , wherein the step of recognizing the edges in the camera image using the gradient image comprises identifying the edges only in those of the pixels of the camera image for which the respective homogeneity determined for the regions adjacent to those pixels is higher than a threshold homogeneity.

17. The method according to claim 6 , wherein the dimension of the respective spatial region is a surface area, a width, a diameter or a circumference of the respective spatial region.

18. The method according to claim 6 , wherein the predefined value is less than a dimension of a pertinent physical feature to be recognized in the camera image based on the recognizing of the edges.

19. The method according to claim 1 , further comprising an initial step of producing the camera image of the environment scene with the camera.

20. The method according to claim 19 , wherein the camera is a vehicle camera, and wherein the recognizing of the edges in the camera image involves recognizing lane markings on a road in the environment scene.

Assignees

• Conti Temic Microelectronic Gmbh

Inventors

• Roth Axel
• Hamidi Dimitri
• Cheng Bingxiang
• Bopp Thomas

Classifications

• G06V10/44

  Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components · CPC title

• G06V20/588

  Recognition of the road, e.g. of lane markings; Recognition of the vehicle driving pattern in relation to the road · CPC title

• G06T7/168

  involving transform domain methods · CPC title

• G06T2207/10024

  Color image · CPC title

• G06T3/4084

  in the transform domain, e.g. fast Fourier transform [FFT] domain scaling · CPC title