Method and apparatus for generating image information

Patent metadata

Field	Value
Publication number	US-10410315-B2
Application number	US-201715720966-A
Country	US
Kind code	B2
Filing date	Sep 29, 2017
Priority date	Mar 3, 2017
Publication date	Sep 10, 2019
Grant date	Sep 10, 2019

Abstract

A method and apparatus for generating image information. A specific embodiment of the method includes: acquiring a rotation angle and a scaling of a target image relative to a reference image, and acquiring an intrinsic parameter of a video camera collecting the reference image; generating a projective transformation matrix based on the rotation angle, the scaling and the intrinsic parameter; generating a coordinate offset of the each neighborhood pixel of the feature point in the reference image; generating, based on a coordinate of the feature point in the reference image, the coordinate offset of the each neighborhood pixel of the feature point in the reference image and the projective transformation matrix, coordinates of the each neighborhood pixel of the feature point in the reference image; and generating a feature point descriptor of the feature point.

First claim

What is claimed is:

1. A method for generating image information, comprising:
acquiring a rotation angle and a scaling of a target image relative to a reference image, and acquiring an intrinsic parameter of a video camera collecting the reference image;
generating a projective transformation matrix based on the rotation angle, the scaling and the intrinsic parameter;
generating, based on the projective transformation matrix and a coordinate offset of each neighborhood pixel of a preset feature point in the target image, a coordinate offset of each neighborhood pixel of the feature point in the reference image;
generating, based on a coordinate of the feature point in the reference image, the coordinate offset of the each neighborhood pixel of the feature point in the reference image and the projective transformation matrix, coordinates of the each neighborhood pixel of the feature point in the reference image; and
generating, based on the coordinates of the each neighborhood pixel of the feature point in the reference image, a feature point descriptor of the feature point.

2. The method according to claim 1 , wherein the generating a projective transformation matrix based on the rotation angle, the scaling and the intrinsic parameter comprises:
generating a rotation matrix, a translation matrix and an intrinsic parameter matrix based on the rotation angle, the scaling and the intrinsic parameter; and
generating the projective transformation matrix based on the rotation matrix, the translation matrix and the intrinsic parameter matrix.

3. The method according to claim 2 , wherein the generating, based on a coordinate of the feature point in the reference image, the coordinate offset of the each neighborhood pixel of the feature point in the reference image and the projective transformation matrix, coordinates of the each neighborhood pixel of the feature point in the reference image comprises:
generating, based on the coordinate of the feature point in the reference image, a third dimension of the projective transformation matrix and the coordinate offset of the each neighborhood pixel of the feature point in the reference image, non-normalized coordinates of the each neighborhood pixel of the feature point in the reference image; and
normalizing the non-normalized coordinates of the each neighborhood pixel of the feature point in the reference image, and generating normalized coordinates of the each neighborhood pixel of the feature point in the reference image.

4. The method according to claim 1 , wherein the generating, based on the coordinates of the each neighborhood pixel of the feature point in the reference image, a feature point descriptor of the feature point comprises:
generating the feature point descriptor of the feature point by using a random ferns algorithm to process the coordinates of the each neighborhood pixel of the feature point in the reference image.

5. The method according to claim 1 , the method further comprising:
reconstructing the reference image three-dimensionally based on the feature point descriptor of the feature point.

6. The method according to claim 1 , the method further comprising:
training, by using a machine leaning method based on the feature point descriptor of the feature point and a category of the feature point, to obtain a feature point classification model, wherein the feature point classification model is used to characterize a corresponding relation between the feature point descriptor of the feature point and the category of the feature point.

7. An apparatus for generating image information, comprising:
at least one processor; and
a memory storing instructions, which when executed by the at least one processor, cause the at least one processor to perform operations, the operations comprising:
acquiring a rotation angle and a scaling of a target image relative to a reference image, and acquiring an intrinsic parameter of a video camera collecting the reference image;
generating a projective transformation matrix based on the rotation angle, the scaling and the intrinsic parameter;
generating, based on the projective transformation matrix and a coordinate offset of each neighborhood pixel of a preset feature point in the target image, a coordinate offset of each neighborhood pixel of the feature point in the reference image;
generating, based on a coordinate of the feature point in the reference image, the coordinate offset of the each neighborhood pixel of the feature point in the reference image and the projective transformation matrix, coordinates of the each neighborhood pixel of the feature point in the reference image; and
generating, based on the coordinates of the each neighborhood pixel of the feature point in the reference image, a feature point descriptor of the feature point.

8. The apparatus according to claim 7 , wherein the generating a projective transformation matrix based on the rotation angle, the scaling and the intrinsic parameter comprises:
generating a rotation matrix, a translation matrix and an intrinsic parameter matrix based on the rotation angle, the scaling and the intrinsic parameter; and
generating the projective transformation matrix based on the rotation matrix, the translation matrix and the intrinsic parameter matrix.

9. The apparatus according to claim 8 , wherein the generating, based on a coordinate of the feature point in the reference image, the coordinate offset of the each neighborhood pixel of the feature point in the reference image and the projective transformation matrix, coordinates of the each neighborhood pixel of the feature point in the reference image comprises:
generating, based on the coordinate of the feature point in the reference image, a third dimension of the projective transformation matrix and the coordinate offset of the each neighborhood pixel of the feature point in the reference image, non-normalized coordinates of the each neighborhood pixel of the feature point in the reference image; and
normalizing the non-normalized coordinates of the each neighborhood pixel of the feature point in the reference image, and generating normalized coordinates of the each neighborhood pixel of the feature point in the reference image.

10. The apparatus according to claim 7 , wherein the generating, based on the coordinates of the each neighborhood pixel of the feature point in the reference image, a feature point descriptor of the feature point comprises:
generating the feature point descriptor of the feature point by using a random ferns algorithm to process the coordinates of the each neighborhood pixel of the feature point in the reference image.

11. The apparatus according to claim 7 , the operations further comprising:
reconstructing three-dimensionally the reference image based on the feature point descriptor of the feature point.

12. The apparatus according to claim 7 , the operations further comprising:
training, by using a machine leaning method based on the feature point descriptor of the feature point and a category of the feature point, to obtain a feature point classification model, wherein the feature point classification model is used to characterize a corresponding relation between the feature point descriptor of the feature point and the category of the feature point.

13. A non-transitory computer readable storage medium, storing a computer program thereon, the computer program, when executed by a processor, cause the processor to perform operations, the operations comprising:
acquiring a rotation angle and a scaling of a target image relative to a reference image, and acquiring an intrinsic parameter of a video

Assignees

• Baidu online network technology beijing co ltd

Inventors

• Chen Sili
• Lin Yuan
• Dong Yuhua
• Zheng Binbin
• Yan Yijie
• Wu Zhongqin

Classifications

• H04N13/117

  the virtual viewpoint locations being selected by the viewers or determined by viewer tracking · CPC title

• G06T7/33

  using feature-based methods · CPC title

• G06T3/60

  Rotation of whole images or parts thereof · CPC title

• G06T3/40

  Scaling of whole images or parts thereof, e.g. expanding or contracting · CPC title

• G06T15/10

  Geometric effects · CPC title