AverageAffineTransform (1)

AverageAffineTransform ImageDimension output_affine_transform [-R reference_affine_transform] {[-i] affine_transform_txt [weight(=1)] ]}

• Usage: Compute weighted average of input affine transforms.

For 2D and 3D transform, the affine transform is first decomposed into scale x shearing x rotation. Then these parameters are averaged, using the weights if they provided. For 3D transform, the rotation component is the quaternion. After averaging, the quaternion will also be normalized to have unit norm. For 2D transform, the rotation component is the rotation angle. The weight for each transform is a non-negative number. The sum of all weights will be normalized to 1 before averaging. The default value for each weight is 1.0.

All affine transforms is a "centerd" transform, following ITK convention. A reference_affine_transform defines the center for the output transform. The first provided transform is the default reference transform Output affine transform is a MatrixOffsetBaseTransform. -i option takes the inverse of the affine mapping.

• For example: 2 output_affine.txt -R A.txt A1.txt 1.0 -i A2.txt 2.0 A3.txt A4.txt 6.0 A5.txt

This computes: (1*A1 + 2*(A2)^-1 + A3 + A4*6 + A5 ) / (1+2+1+6+5)