Learning Objectives

  • Learn how to use Affine transforms

  • Differentiate translate, rotate, skew, shear


Affine Transforms#

Affine transformations allows us to use simple systems of linear equations to manipulate any point or set of points. It allows us to move, stretch, or even rotate a point or set of points. In the case of GIS, it is used to distort raster data, for instance satellite imagery, to fit a new projection or CRS.

../_images/warp.png

First some general properties of affine transforms:

  • Preserves

    • Points, straight linear & planes

    • Sets of parallel lines

    • Ratio of distances between points on same straight line

  • Distorts

    • Angle between lines

    • Distance between points

Types of Transformations#

There are four core ways that you can manipulate an image. These are called “Transforms”:

Transform

Description

Example

Translation

Moves a set of points some fixed distance in the x and y plane

translate image

Scale

Increases or decreases the scale, or distance between points in the x and y plane

scale image

Rotate

Rotates points around the origin, or some defined axis

rotate image

Shear

Shifts points in proportion to any given points x and y coordinate

shear image

Image Credit: Wikipedia

In combination we can warp any point, or set of points (e.g. raster image) into a new projection. This is the equivalent of reprojection for vector data. In order to implement these transforms we will need to learn about the math behind Affine Transforms! Yeah!

Simple Transform Examples#

To simplify things first lets think about how to do transformations of a single point in a 2d space.

For any location \( \mathbf{x} = (x,y)\) we can transform \( \mathbf{x} \) to \(\mathbf{\acute{x}}\) using simple linear adjustments. Here we can think of point \(\mathbf{x}\) as being stored as an array:

\[\begin{split} \begin{eqnarray} \mathbf{x} = \left[ \begin{array}{cc|r}x \\ y \end{array} \right] \mbox{can be transformed to } \mathbf{\acute{x}} = \left[ \begin{array}{cc|r}\acute{x} \\ \acute{y} \end{array} \right] \end{eqnarray} \end{split}\]

From here we can transform the values in the \(x\) and \(y\) position using scaler values \(a\) through \(f\):

\[\begin{split} \begin{eqnarray} \mathbf{\acute{x}} = \left[ \begin{array}{cc|r}ax+by+c \\ dx+ey+f \end{array} \right] \end{eqnarray} \end{split}\]

Note

Looking at the formula above, we are adjusting the value of \(x\) with \(ax+by+c\), so \(x\) can be multiplied (scaled) by some value \(a\), it can also be scaled based on the \(y\) value with \(+ by\), or simply adjusted up or down by the value \(+ c\)

To understand how this works, let’s walk through the basic transformations of \(\mathbf{x}\):

If we multiply \(x\) and \(y\) by one, by setting \(a,e = 1\), and zero out the effect of the other axis, by setting \(b,d = 0\), we have a simple case of translation, where \(x\) moves right by \(c\) and \(y\) up by the value of \(f\):

\[\begin{split} \begin{eqnarray} \mathbf{\acute{x}} = \left[ \begin{array}{cc|r}1x+0y+c \\ 0x+1y+f \end{array} \right] \end{eqnarray} \end{split}\]
../_images/translate_coord.png

Fig. 33 Translate a coordinate#

We can scale \(x\) and \(y\) by setting \(a\) and \(e\) to 0.5 (or some fraction), and setting all other values to zero (\(b,d,c,f = 0\)):

\[\begin{split} \begin{eqnarray} \mathbf{\acute{x}} = \left[ \begin{array}{cc|r}ax \\ ey \end{array} \right] \end{eqnarray} \end{split}\]
../_images/scale_coord.png

Fig. 34 Scale a coordinate#

We can rotate a point around the origin by setting \(a,e = \cos\theta\), \(b = -\sin\theta\) and \(c,f = 0\):

\[\begin{split} \begin{eqnarray} \mathbf{\acute{x}} = \left[ \begin{array}{cc|r} x\cos\theta - y\sin\theta \\ x\sin\theta+y\cos\theta \end{array} \right] \end{eqnarray} \end{split}\]

where \(\theta\) is the angle of rotation (counterclockwise) around the origin.

../_images/rotate_coord.png

Fig. 35 Rotate a coordinate#

Finally by adjusting \(x\) based on the value of \(y\) (and visa versa), we can achieve a shear transform:

\[\begin{split} \begin{eqnarray} \mathbf{\acute{x}} = \left[ \begin{array}{cc|r} x+by \\ y+dx \end{array} \right] \end{eqnarray} \end{split}\]
../_images/shear_coord.png

Fig. 36 Shear transform a coordinate#

Transforming Matrices#

It is often convenient to represent these equations as matrices. This allows us to easily chain together a series of operations. We can represent our transformed point \(\mathbf{\acute{x}}\) as follows:

\[\begin{split} \begin{eqnarray} \left[ \begin{array}{cc|r} \acute{x} \\ \acute{y} \end{array} \right] = \left[ \begin{array}{cc|r} ax+by \\ dx+ey \end{array} \right] = \left[ \begin{array}{cc|r} a \ \ b \\ d \ \ e \end{array} \right] \left[ \begin{array}{cc|r} x \\ y \end{array} \right] \end{eqnarray} \end{split}\]

Note

If you aren’t familiar with how matrix multiplication works please watch this.

In this new context we can easily do a scale, rotate or shear transform by replacing the matrix of \(a,b,d,e\) with:

\[\begin{split} \begin{eqnarray} \mbox{Rotate: } \left[ \begin{array}{cc|r} \cos\theta \ \ -\sin\theta \\ \sin\theta \ \ \cos\theta \end{array} \right] \end{eqnarray} \end{split}\]
\[\begin{split} \begin{eqnarray} \mbox{Scale: } \left[ \begin{array}{cc|r} S_{x} \ \ 0 \\ 0 \ \ S_{y} \end{array} \right] \end{eqnarray} \end{split}\]
\[\begin{split} \begin{eqnarray} \mbox{Shear: } \left[ \begin{array}{cc|r} 1 \ \ r_{x} \\ r_{y} \ \ 1 \end{array} \right] \end{eqnarray} \end{split}\]

But wait, what about the easiest transform, “translation”? Unfortunately that makes things a little more complicated! But not that complicated.

In order to be able to perform a translate in matrix form we need to extend our matrices, adding one row along the bottom. In the following form we can now perform all the basic transformations to calculate \( \mathbf{\acute{x}}\):

\[\begin{split} \begin{eqnarray} \left[ \begin{array}{cc|r} \acute{x} \\ \acute{y} \\ 1 \end{array} \right] = \left[ \begin{array}{cc|r} a \ \ b \ \ c \\ d \ \ e \ \ f \\ 0 \ \ 0 \ \ 1 \end{array} \right] \left[ \begin{array}{cc|r} x \\ y \\ 1 \end{array} \right] = \left[ \begin{array}{cc|r} ax+by+c \\ dx+ey+f \\ 1 \end{array} \right] \end{eqnarray} \end{split}\]

Now that we have scalers \(c\) and \(f\) all the transforms are possible. We do however, need to update our previous operations:

\[\begin{split} \begin{eqnarray} \mbox{Translate: } \begin{bmatrix} 1 & 0 & \Delta x \\ 0 & 1 & \Delta y \\ 0 & 0 & 1 \end{bmatrix} \end{eqnarray} \end{split}\]

Where \(\Delta x\) shifts in the \(x\) axis and \(\Delta y\) determines the shift in the \(y\) axis.

\[\begin{split} \begin{eqnarray} \mbox{Rotate: } \begin{bmatrix} \cos\theta & -\sin\theta & 0 \\ \sin\theta & \cos\theta & 0 \\ 0 & 0 & 1 \end{bmatrix} \end{eqnarray} \end{split}\]
\[\begin{split} \begin{eqnarray} \mbox{Scale: } \begin{bmatrix} S_{x} & 0 & 0 \\ 0 & S_{y} & 0 \\ 0 & 0 & 1 \end{bmatrix} \end{eqnarray} \end{split}\]
\[\begin{split} \begin{eqnarray} \mbox{Shear: } \begin{bmatrix} 1 & r_{x} & 0 \\ r_{y} & 1 & 0 \\ 0 & 0 & 1 \end{bmatrix} \end{eqnarray} \end{split}\]

Numeric Examples#

Translate#

Now let’s assume we have a point a at (-2,-2) for (x,y). For simplicity sake lets assume we want to move it up to the origin by adding 2 to both x and y.

Let’s start by defining our point in our matrix form:

\[\begin{split} \begin{eqnarray} \begin{bmatrix} x \\ y \\ 1 \end{bmatrix} = \begin{bmatrix} -2 \\ -2 \\ 1 \end{bmatrix} \end{eqnarray} \end{split}\]

Let’s get our transform matrix \(M\) to perform our translate, where \(\Delta x, \Delta y = 2\) because we want to move it up and to the right:

\[\begin{split} \begin{eqnarray} \begin{bmatrix} 1 & 0 & 2 \\ 0 & 1 & 2 \\ 0 & 0 & 1 \end{bmatrix} \end{eqnarray} \end{split}\]

We can then multiply the two:

\[\begin{split} \begin{eqnarray} \begin{bmatrix} 1 & 0 & 2 \\ 0 & 1 & 2 \\ 0 & 0 & 1 \end{bmatrix} \begin{bmatrix} -2 \\ -2 \\ 1 \end{bmatrix} = \begin{bmatrix} -2 \times 1 + -2 \times 0 + 1 \times 2 \\ -2 \times 0 + -2 \times 1 + 1 \times 2 \\ -2 \times 0 + -2 \times 0 + 1 \times 1 \end{bmatrix} = \begin{bmatrix} 0 \\ 0 \\ 1 \end{bmatrix} \end{eqnarray} \end{split}\]

Congrats you reached (0,0), just like you always dreamed!

../_images/translate_ex.png

Fig. 37 Moving a point#

Note

Remember the bottom row can be ignored because \( \begin{bmatrix} x \\ y \\ 1 \end{bmatrix}\)

Rotate#

All the transformations follow the same procedure, let’s try rotation just to make sure that we have it figured out. Let’s rotate our point at (-2,-2) by 180 degrees around the origin:

\[\begin{split} \begin{eqnarray} \begin{bmatrix} \cos{180} & -\sin{180} & 0 \\ \sin{180} & \cos{180} & 0 \\ 0 & 0 & 1 \end{bmatrix} \begin{bmatrix} -2 \\ -2 \\ 1 \end{bmatrix} = \end{eqnarray} \end{split}\]
\[\begin{split} \begin{eqnarray} \begin{bmatrix} -2 \times \cos{180} + -2 \times -\sin{180} + 1 \times 0 \\ -2 \times \sin{180} + -2 \times \cos{180} + 1 \times 0 \\ 1 \times 0 + 1 \times 0 + 1 \times 1 \end{bmatrix} = \begin{bmatrix} 2 \\ 2 \\ 1 \end{bmatrix} \end{eqnarray} \end{split}\]
../_images/rotate_ex.png

Fig. 38 Rotate a point#