|
Developer Documentation |
|
Centre for Vision, Speech & Signal Processing |
Comments:
This method computes the solution to the motion constraint equation over an
image patch (as described here).
The covariance matrix for the motion vector and the variance of the fit are
also available. The default noise parameter value = 1.0.
Variables:
- PairC> image;
-
- NumImageC grad_grad;
-
- NumImageC grad_t;
-
- NumImageC dt_sq;
-
- Matrix2d2C cov;
-
covariance of motion vector estimation
- RealT var;
-
variance of motion estimation errors
- Vector2dC lambda;
-
eigenvalues of gradient outer product
- IntT region;
-
size of region
- RealT noise;
-
to help bad matrices invert nicely
- BooleanT verbose;
-
Methods:
- LMSRegionMatchC(const PairC<NumImageC<RealT>> & im,BooleanT verbose = false)
-
Constructor from filtered image pair
- Vector2dC Estimate(const NumImageC<IntT> & mask,IntT label)
-
Estimates motion for region with mask value "label"
- void SetNoiseLevel(RealT Noise)
-
Sets noise level for matrix inversion
- Matrix2d2C MotionCov() const
-
Returns matrix of motion estimate covariance
- RealT ErrorVar() const
-
Returns motion constraint equation error variance
- Vector2dC Eigenvalues() const
-
Returns eigenvalues of spatial gradient outer product
|
Programmer:Bill Christmas, Documentation by CxxDoc: Tue Mar 20 10:49:27 2001
|