All Packages  Class Hierarchy  This Package  Previous  Next  Index

Class feature.AffineModel 

java.lang.Object
   |
   +----feature.AffineModel
public class AffineModel
extends Object
Realization of Affine motion estimation. All methods are static public functions.

Author:
Di Zhong (Columbia University)

Constructor Index

o AffineModel()

Method Index

o affineEstimate(float[][][], Point, boolean[][], Point)
Estimate 6-parameter affine model by linear regression.
o compensate(boolean[][], Point, float[], Dimension)
Compensate region using affine model.
o compensate(Mask, float[], Dimension)
Compensate region using affine model.
o translate(boolean[][], Point, int, int, Dimension)
Compensate region using average motion Return compensated region mask and new upper left corner

Constructors

o AffineModel 
 public AffineModel()

Methods

o affineEstimate 
 public static final float[] affineEstimate(float flow[][][],
                                            Point ul,
                                            boolean mask[][],
                                            Point maskul)
Estimate 6-parameter affine model by linear regression. In addtion, return average motion vector. Return values 0-5 are affine parameters; 6-7 are average x&y displacements.

Parameters:
flow - optical flow
ul - upper-left corner of optical flow
mask - region mask (only masked pixels will be used for estimation)
maskul - upper-left corner of mask
o translate 
 public static final boolean[][] translate(boolean m0[][],
                                           Point ul,
                                           int dx,
                                           int dy,
                                           Dimension frame)
Compensate region using average motion Return compensated region mask and new upper left corner

Parameters:
m0 - original mask of region
ul - up-left corner of m0, will be update to up-left coner of compensated region
dx - x-displacement
dy - y-displacement
frame - size of whole image
See Also:
Mask
o compensate 
 public static final boolean[][] compensate(boolean m0[][],
                                            Point ul,
                                            float a[],
                                            Dimension frame)
Compensate region using affine model. Return compensated region mask and new upper left corner

Parameters:
m0 - original mask of region
ul - up-left corner of m0, will be update to up-left coner of compensated region
a - 6-parameter affine model
frame - size of whole image
See Also:
Mask
o compensate 
 public static final Mask compensate(Mask m0,
                                     float a[],
                                     Dimension frame)
Compensate region using affine model. Return compensated region mask

Parameters:
m0 - original mask of region
a - 6-parameter affine model
frame - size of whole image
See Also:
Mask 
All Packages  Class Hierarchy  This Package  Previous  Next  Index