Processing

Please wait...

Settings

Settings

1. WO2007001884 - DIRECT METHOD FOR MODELING NON-RIGID MOTION WITH THIN FILM PLATE SPLINE TRANSFORMATION

Note: Text based on automatic Optical Character Recognition processes. Please use the PDF version for legal matters

CLAIMS
1. A computer based method for estimating a thin plate spline (TPS) warp between two digital images of a sequence of digital images representing one or more physical objects, the digital images comprising pixels, the method comprising:
receiving first and second digital images of the sequence of digital images;
determining a fixed template region within said first image;
positioning a fixed set of reference points within said fixed template region;
determining a first set of pixels within said second image, said first set of pixels
corresponding to said fixed template region and including a set of projections of said fixed reference points;
determining a TPS parameter vector from said set of projections;
determining a back- warped image by warping said first set of pixels based on said
TPS parameter vector.
2. The method of claim 1, further comprising the steps of:
determining an error function based on a difference between said fixed template
region and said back-warped image; and
updating said TPS parameter vector based on said error function and using a gradient descent technique.
3. The method of claim 2, further comprising the step of:
re-determining said back-warped image based on said updated TPS parameter vector.

4. The method of claim 1, wherein said reference points are arranged on a rectangular grid.
5. The method of claim 1, wherein said fixed template region is determined
automatically.
6. The method of claim 1, further comprising the step of :
determining a first value of a regularization parameter that determines the degree to
which pixels in said template image contribute to an update of said set of
reference points;
determining a second value of said regularization parameter following said
determining a back-warped image; and
refining said back- warped image based on said updated value of said regularization
parameter.

7. The method of claim 1, wherein said fixed template region includes a non-rigid
object.
8. The method of claim 7, wherein a relative motion of said non-rigid object is
determined based on said TPS parameter vector.
9. The method of claim 1, wherein a set of reference points within said second digital image is determined based on said fixed set of reference points within said first digital image.
10. The method of claim 9, wherein a relative motion is determined based on said set of fixed reference points and said set of reference points within said second digital
image.
11. An apparatus for automatically estimating a thin plate spline (TPS) warp between two digital images of a sequence of digital images representing one or more physical
objects, the digital images comprising pixels, the apparatus comprising:
an input module configured to receive first and second digital images of the sequence of digital images; and
a processor module configured to:
determine a fixed template region within said first digital image,
position a fixed set of reference points within said fixed template region,
determine a first set of pixels within said second image, said first set of pixels
corresponding to said fixed template region and including a set of
projections of said fixed reference points,
determine a TPS parameter vector from said set of projections, and
determine a back-warped image by warping said first set of pixels based on said
TPS parameter vector.
12. An apparatus for automatically estimating a thin plate spline (TPS) warp between two digital images of a sequence of digital images representing one or more physical
objects, the digital images comprising pixels, the apparatus comprising:
means for receiving first and second digital images of the sequence of digital images; means for determining a fixed template region within said first image;
means for positioning a fixed set of reference points within said fixed template region; means for determining a first set of pixels within said second image, said first set of
pixels corresponding to said fixed template region and including a set of
projections of said fixed reference points;

means for determining a TPS parameter vector from said set of projections; and
means for determining a back-warped image by warping said first set of pixels based on said TPS parameter vector.
13. A computer program product, comprising a computer-readable medium having
computer program instructions embodied thereon to cause a computer processor to implement a method for estimating a thin plate spline (TPS) warp between two digital images of a sequence of digital images representing one or more physical objects, the digital images comprising pixels, the method comprising:
receiving first and second digital images of the sequence of digital images;
determining a fixed template region within said first image;
positioning a fixed set of reference points within said fixed template region;
determining a first set of pixels within said second image, said first set of pixels
corresponding to said fixed template region and including a set of projections of said fixed reference points;
determining a TPS parameter vector from said set of projections;
determining a back- warped image by warping said first set of pixels based on said
TPS parameter vector.