Apply inverse spatial transformation

```
[U,V] =
tforminv(T,X,Y)
```

`[U1,U2,...,U_ndims_in] = tforminv(T,X1,X2,...,X_ndims_out)`

`U = tforminv(T,X)`

`[U1,U2,...,U_ndims_in] = tforminv(T,X)`

`U = tforminv(T,X1,X2,...,X_ndims_out)`

`[`

applies the 2D-to-2D inverse spatial transformation defined in `U`

,`V`

] =
tforminv(`T`

,`X`

,`Y`

)`T`

to
coordinate arrays `X`

and `Y`

, mapping the point
`[X(k) Y(k)]`

to the point `[U(k) V(k)]`

.

Both `T.ndims_in`

and `T.ndims_out`

must equal 2.
`X`

and `Y`

are typically column vectors, but they can
have any dimensionality. `U`

and `V`

are the same size as
`X`

and `Y`

`[`

applies the `U1,U2,...,U_ndims_in`

] = tforminv(`T`

,`X1,X2,...,X_ndims_out`

)`ndims_out`

-to-`ndims_in`

inverse transformation
defined in `T`

to the coordinate arrays
`X1,X2,...,X_ndims_out`

. The transformation maps the point ```
[X1(k)
X2(k) ... X_ndims_out(k)]
```

to the point ```
[U1(k) U2(k) ...
U_ndims_in(k)]
```

.

The number of input coordinate arrays, `ndims_out`

, must equal
`T.ndims_out`

. The number of output coordinate arrays,
`ndims_in`

, must equal `T.ndims_in`

. The arrays
`X1,X2,...,X_ndims_out`

can have any dimensionality, but must be the same
size. The output arrays `U1,U2,...,U_ndims_in`

must be this size also.

applies the `U`

= tforminv(`T`

,`X`

)`ndims_out`

-to-`ndims_in`

inverse transformation
defined in `T`

to array `X`

.

When

`X`

is a 2-D matrix with dimensions*m*-by-`ndims_out`

matrix,`U`

is a 2-D matrix with dimensions*m*-by-`ndims_in`

.`tforminv`

applies the transformation to each row of`X`

.`tforminv`

maps the point`X`

(*k*, : ) to the point`U`

(*k*, : ).When

`X`

is an (*N*+1)-dimensional array,`tforminv`

maps the point`X`

(*k*_{1},*k*_{2}, … ,*k*_{N}, : ) to the point`U`

(*k*_{1},*k*_{2}, … ,*k*_{N}, : ).`size(X,N+1)`

must equal`ndims_out`

.`U`

is an (*N*+1)-dimensional array, with`size(U,I)`

equal to`size(X,I)`

for`I`

= 1, … ,*N*, and`size(U,N+1)`

equal to`ndims_in`

.

The syntax `U = tforminv(X,T)`

is an older form of this syntax that remains
supported for backward compatibility.

`[`

maps one (`U1,U2,...,U_ndims_in`

] = tforminv(`T`

,`X`

)*N*+1)-dimensional array to `ndims_in`

equally sized
*N*-dimensional arrays.

maps `U`

= tforminv(`T`

,`X1,X2,...,X_ndims_out`

)`ndims_out`

*N*-dimensional arrays to one (*N*+1)-dimensional
array.