Method and apparatus for tracking target

Patent metadata

Field	Value
Publication number	US-11915432-B2
Application number	US-202117146809-A
Country	US
Kind code	B2
Filing date	Jan 12, 2021
Priority date	Jan 16, 2020
Publication date	Feb 27, 2024
Grant date	Feb 27, 2024

Abstract

Disclosed is a target tracking method and apparatus. The target tracking apparatus includes a processor configured to obtain a first depth feature from a target region image and obtain a second depth feature from a search region image, obtain a global response diagram between the first depth feature and the second depth feature, acquire temporary bounding box information based on the global response diagram, updated the second depth feature based on the temporary bounding box information, obtain local feature blocks based on the first depth feature, obtain a local response diagram based on the local feature blocks and the updated second depth feature, and determine output bounding box information based on the local response diagram.

First claim

What is claimed is:

1. A method of tracking a target, the method comprising:
obtaining a first depth feature from a target region image and obtaining a second depth feature from a search region image;
obtaining a global response diagram between the first depth feature and the second depth feature;
acquiring temporary bounding box information based on the global response diagram;
updating the second depth feature based on the temporary bounding box information;
obtaining local feature blocks based on the first depth feature;
obtaining a local response diagram based on the local feature blocks and the updated second depth feature; and
determining output bounding box information based on the local response diagram.

2. The method of claim 1 , wherein the obtaining of the local feature blocks comprises obtaining the local feature blocks by dividing the first depth feature or a third depth feature extracted from the first depth feature, and
the obtaining of the local response diagram comprises obtaining the local response diagram based on correlations between the second depth feature or a fourth depth feature extracted from the updated second depth feature and the local feature blocks.

3. The method of claim 2 , wherein the obtaining of the local response diagram based on the correlations comprises:
obtaining local sub-response diagrams based on the correlations between the second depth feature or the fourth depth feature and the respective local feature blocks; and
obtaining the local response diagram by synthesizing the local sub-response diagrams.

4. The method of claim 3 , wherein the obtaining of the local response diagram by synthesizing the local sub-response diagrams comprises:
classifying the local feature blocks into target feature blocks and background feature blocks; and
obtaining the local response diagram by synthesizing the local sub-response diagrams based on a result of the classification.

5. The method of claim 4 , wherein the classifying comprises classifying the local feature blocks into the target feature blocks and the background feature blocks based on overlap ratios of the respective local feature blocks to a temporary bounding box.

6. The method of claim 3 , wherein the output bounding box information comprises coordinate offsets between coordinates of the center of a temporary bounding box included in the temporary bounding box information and coordinates of the center of an output bounding box and a size offset between the size of the output bounding box and a preset size, and
the determining of the output bounding box information comprises:
outputting the temporary bounding box information as the output bounding box information, in response to a sum of absolute values of the coordinate offsets being greater than a threshold; and
outputting a result of adding the coordinates of the center of the temporary bounding box and the coordinate offsets and a result of adding the size of the temporary bounding box and the size offset as the output bounding box information, in response to the sum of absolute values of the coordinate offsets being less than or equal to the threshold.

7. The method of claim 6 , wherein the acquiring of the temporary bounding box information comprises outputting coordinates with the highest correlation in the global response diagram of a current frame as coordinates of the center of the temporary bounding box of the current frame, and outputting the size of an output bounding box estimated in a previous frame as the size of the temporary bounding box of the current frame.

8. The method of claim 2 , wherein the obtaining of the local feature blocks by dividing the first depth feature or the third depth feature comprises dividing the first depth feature or the third depth feature based on any one of the local feature blocks not overlapping, the local feature blocks overlapping, and a preset block distribution.

9. A non-transitory computer-readable storage medium storing instructions that, when executed by a processor, cause the processor to perform the method of claim 1 .

10. An apparatus for tracking a target, the apparatus comprising:
a processor configured to:
obtain a first depth feature from a target region image and obtain a second depth feature from a search region image,
obtain a global response diagram between the first depth feature and the second depth feature,
acquire temporary bounding box information based on the global response diagram,
update the second depth feature based on the temporary bounding box information,
obtain local feature blocks based on the first depth feature,
obtain a local response diagram based on the local feature blocks and the updated second depth feature, and
determine output bounding box information based on the local response diagram.

11. The apparatus of claim 10 , wherein the processor is further configured to:
obtain the local feature blocks by dividing the first depth feature or a third depth feature extracted from the first depth feature; and
obtain the local response diagram based on correlations between the second depth feature or a fourth depth feature extracted from the updated second depth feature and the local feature blocks.

12. The apparatus of claim 11 , wherein the processor is further configured to divide the first depth feature or the third depth feature based on any one of the local feature blocks not overlapping, the local feature blocks overlapping, and a preset block distribution.

13. The apparatus of claim 11 , wherein the processor is further configured to:
obtain local sub-response diagrams based on the correlations between the second depth feature or the fourth depth feature and the respective local feature blocks; and
obtain the local response diagram by synthesizing the local sub-response diagrams.

14. The apparatus of claim 13 , wherein the processor is further configured to:
classify the local feature blocks into target feature blocks and background feature blocks; and
obtain the local response diagram by synthesizing the local sub-response diagrams based on a result of the classification.

15. The apparatus of claim 14 , wherein the processor is further configured to classify the local feature blocks into the target feature blocks and the background feature blocks based on overlap ratios of the respective local feature blocks to a temporary bounding box.

16. The method of claim 13 , wherein the output bounding box information comprises coordinate offsets between coordinates of the center of a temporary bounding box included in the temporary bounding box information and coordinates of the center of an output bounding box and a size offset between the size of the output bounding box and a preset size; and
the processor is further configured to:
output the temporary bounding box information as the output bounding box information, in response to a sum of absolute values of the coordinate offsets being greater than a threshold, and
output a result of adding the coordinates of the center of the temporary bounding box and the coordinate offsets and a result of adding the size of the temporary bounding box and the size offset as the output bounding box information, in response to the sum of absolute values of the coordinate offsets being less than or equal to the threshold.

17. The apparatus of claim 10 , wherein the obtaining of the local feature blocks is based on the first depth feature and the temporary bounding box.

18. An apparatus, the apparatus comprising:
a processor configured to:
extract a second d

Assignees

• Samsung Electronics Co Ltd

Inventors

• Xu Jingtao
• Yu Jiaqian
• Yoo Byung In
• Choi Chang Kyu
• Lee Hyunjeong
• Tan Hangkai
• Han Jaejoon
• Wang Qiang
• Chen Yiwei

Classifications

• G06T7/248 Primary

  involving reference images or patches · CPC title

• G06F18/241

  relating to the classification model, e.g. parametric or non-parametric approaches · CPC title

• G06T7/73

  using feature-based methods · CPC title

• G06V10/25

  Determination of region of interest [ROI] or a volume of interest [VOI] · CPC title

• G06V10/764

  using classification, e.g. of video objects · CPC title