Method of detecting feature points of an object in a system for motion detection
A method of detecting feature points of an object in a system for motion detection includes obtaining a first image of the object from a first camera and a second image of the object from a second camera, extracting a foreground image from each of the first image and the second image, based on an assumption that the foreground image is a T-pose image, segmenting the foreground image into a first set of sections, identifying a first set of feature points associated with the first set of sections, obtaining a T-pose image with a set of predetermined feature points, and determining whether the foreground image is a T-pose image by comparing the first set of feature points with the set of predetermined feature points.
1. A method of detecting feature points of an object in a system for motion detection, the method comprising:
obtaining a first image of the object from a first camera and a second image of the object from a second camera;
extracting a foreground image from each of the first image and the second image;
based on an assumption that the foreground image from each of the first image and the second image is a T-pose image, segmenting the foreground image from each of the first image and the second image into a first set of sections;
identifying a first set of feature points associated with the first set of sections;
obtaining a T-pose image with a set of predetermined feature points; and
determining whether the assumed T-pose foreground image is a T-pose image by comparing the relative position of feature points associated with the assumed T-pose foreground image with the relative position of the set of predetermined feature points from the obtained T-pose image based on the spatial relationship of the feature points.
2. The method of claim 1 , when it is determined that the assumed T-pose foreground image from each of the first image and the second image is not a T-pose image, further comprising:
identifying a highest portion of the foreground image from each of the first image and the second image;
providing a check box of a predetermined size; and
determining whether the highest portion of the foreground image from each of the first image and the second image in the check box reaches a predetermined ratio.
3. The method of claim 2 , when it is determined that the highest portion of the foreground image from each of the first image and the second image in the check box does not reach a predetermined ratio, further comprising:
obtaining a head template.
4. The method of claim 3 further comprising:
adjusting the size of the obtained T-pose image from each of the first image and the second image.
5. The method of claim 3 further comprising:
identifying a trunk portion of the foreground image based on the obtained T-pose image from each of the first image and the second image; and
identifying a shoulder portion of the foreground image from each of the first image and the second image based on the trunk portion.
6. The method of claim 5 further comprising:
identifying feature points associated with the trunk portion and the shoulder portion; and
forming vectors based on the feature points associated with the trunk portion and the shoulder portion.
7. The method of claim 6 further comprising:
segmenting the foreground image from each of the first image and the second image into a second set of sections based on the vectors.
8. The method of claim 1 further comprising:
determining whether a feature point is identified in the first and second images; and
when it is determined that a feature point is not identified, conducting a search for regions of interest in disparity images associated with the feature point.
9. The method of claim 8 further comprising:
determining whether a feature point is close to another feature point; and
when it is determined that a feature point is close to another feature point, conducting a search for regions of interest in disparity images associated with the feature point.
10. The method of claim 9 further comprising:
calculating the depth of each pixels in the regions of interest in the disparity images associated with the feature point.
11. The method of claim 10 , wherein calculating the depth includes calculation in an equation below:
Z
=
L
×
f
dx
,
where Z is the depth of the feature point, f is a focal length of each of the first and second cameras, L is the center-to-center distance between the first and second cameras, and dx is the distance of the object in disparity images.
12. The method of claim 10 further comprising:
identifying an image of motion pixels by comparing the foreground image from each of the first image and the second image with a previous foreground image from each of the first image and the second image; and
conducting a search in an area of a predetermined size in the image of motion pixels, wherein the area corresponds to an area in the previous foreground image from each of the first image and the second image where a feature point similar to the feature point at issue has been identified.
13. The method of claim 10 further comprising:
conducting a search in an area of the foreground image from each of the first image and the second image that corresponds to an area in a previous image where a similar feature point is located.
14. A method of detecting feature points of an object in a system for motion detection, the method comprising:
obtaining an image of the object from a camera;
extracting a foreground image from the image;
based on an assumption that the foreground image is a T-pose image, segmenting the foreground image into a first set of sections;
identifying a first set of feature points associated with the first set of sections;
comparing the relative position of the first set of feature points associated with the assumed T-pose foreground image with the relative position of the set of predetermined feature points from an obtained T-pose image;
identifying a trunk portion and a shoulder portion in the foreground image;
segmenting the foreground image into a second set of sections based on the trunk portion and the shoulder portion; and
identifying a second set of feature points associated with the second set of sections.
15. The method of claim 14 , before identifying the trunk portion and the shoulder portion, further comprising:
obtaining a head template.
16. The method of claim 14 further comprising:
identifying feature points associated with the trunk portion and the shoulder portion; and
forming vectors based on the feature points associated with the trunk portion and the shoulder portion.
17. The method of claim 16 further comprising:
segmenting the foreground image into a second set of sections based on the vectors.
18. The method of claim 14 further comprising:
determining whether a feature point is identified in the first and second images; and
when it is determined that a feature point is not identified, conducting a search for regions of interest in disparity images associated with the feature point.
19. The method of claim 14 further comprising:
determining whether a feature point is close to another feature point; and
when it is determined that a feature point is close to another feature point, conducting a search for regions of interest in disparity images associated with the feature point.
20. The method of claim 14 further comprising:
identifying an image of motion pixels by comparing the foreground image with a previous foreground image; and
conducting a search in an area of a predetermined size in the image of motion pixels, wherein the area corresponds to an area in the previous foreground image where a feature point similar to the feature point at issue has been identified.