Geometric transformations

Affine transformations
Forward mapping
Interpolations schemes
A) Geometric transformations permit elimination of the
geometric distortion that occurs when an image is captured.
Geometric distortion may arise because of the lens or because of
the irregular movement of the sensor during image capture.
B) Geometric transformation processing is also essential in
situations where there are distortions inherent in the imaging
process such as remote sensing from aircraft or spacecraft. On
example is an attempt to match remotely sensed images of the
same area taken after one year, when the more recent image was
probably not taken from precisely the same position. To inspect
changes over the year, it is necessary first to execute a geometric
transformation, and then subtract one image to other. We might
also need to register two or more images of the same scene,
obtained from different viewpoints or acquired with different
instruments. Image registration matches up the features that are
common to two or more images. Registration also finds
applications in medical imaging.
 x x’
Suppose that the true
position of a pixel is
(x,y) and the distorted
position is (x’,y’)

A geometric transformation is a vector T that maps the pixel ( x,y)

to a new position (x’, y’)
x’ = Tx(x,y) : y’=Ty(x,y) (1)

The transformation equations Tx and Ty are either know in

advance - in the case of rotation, translation, scaling or can be
determined from known original and transformed images.
How can a geometrically distorted image be restored?

A geometric transform consists of two basis steps:

1. Pixel co-ordinate transformation, which maps the co-
ordinates of the input image pixel to the point in the output image.
The output point co-ordinates should be computed as continuous
values ( real numbers) as the position does not necessarily match
the digital grid after the transform.
2. The second step is to find the point in the digital raster which
matches the transformed point and determine its brightness value .
It brightness is usually computed as an interpolation of the
brightness of several points in the neighborhood .
 Affine transformations
An affine transformation is an important class of linear 2-D geometric
transformations which maps variables (e.g. pixel intensity values
located at position (x,y) in an input image) into new variables (e.g.
in an output image (x’,y’) by applying a linear combination of
translation, rotation, scaling and/or shearing (i.e. non-uniform
scaling in some directions) operations.

From equation (1) Tx and Ty are expressed as polynomials in x

and y.
If they are linear mapping function in x and y we apply an affine
transformation
 0
a b 0 0 
x '  a 0 x  a1 y  a 2 x' y' 1   a1 b1 0x y 1
a 2 b2 1
y'  b0 x  b1 y  b2
 Geometric Processes

• Transformation applied on the coordinates of the pixels

(i.e., relocate pixels).

• A geometric transformation has the general form

(x,y) = T{(v,w)}

where (v,w) are the original pixel coordinates and

(x,y) are the transformed pixel coordinates.
Geometric Transformations

affine transformation
t11 t12 0
[ x y 1]  [v w1]t 21 t 22 0
t 31 t 32 1
 Figure shows a hierarchy of planar
transformations which are important to
computer vision
 Translation

x’=x+5 ;  1 0 0
 
x' y' 1   0 1 0x y 1
y’=y+3 Tx  5 Ty  3 1
 
The treatment of elements near image edges varies with implementation.
Translation is used to improve visualization of an image, but also has a
role as a preprocessor in applications where registration of two or more
images is required.
Since the dimensions of the input image are well defined, the output
image is also a discrete space of finite dimension. If the new coordinates
x’,y’ are outside the image, the translate operator will normally ignore
them, although, in some implementations, it may link the higher
coordinate points with the lower ones so as to wrap the result around
back onto the visible space of the image. Most implementations fill the
 Translation Guidelines for Use
The translate operator takes two arguments,
Bx,By, which specify the desired horizontal
and vertical pixel displacements, respectively.
For example, consider the artificial image in
which the subject's center lies in the center of
the 300×300 pixel image. We can naively
translate the subject into the lower, right
corner of the image by defining a mapping
(i.e. a set of values) for (Bx,By) which will
take the subject's center from its present
position at x=150,y=150 to an output position
of x=300,y=300 , as shown in the second
image
 Translation Guidelines for Use

Translation has many applications of the cosmetic sort illustrated

above. However, it is also very commonly used as a preprocessor
in application domains where registration of two or more images
is required. For example, feature detection and spatial filtering
algorithms may calculate gradients in such a way as to introduce
an offset in the positions of the pixels in the output image with
respect to the corresponding pixels from the input image. In the
case of the Laplacian of Gaussian spatial sharpening filter, some
implementations require that the filtered image be translated by
half the width of the Gaussian kernel with which it was
convolved in order to bring it into alignment with the original.
 Translation Guidelines for Use
We again view the effects of mis-
alignment if we consider translating by
one pixel in the x and y directions and
then subtracting this result from the
original.

The resulting image, shown in contains a

description of all the places (along the direction
of translation) where the intensity gradients are
different; i.e. it highlights edges (and noise).
 Translation - Applications

Note that if we increase the translation

parameter too much, e.g., by 6 pixels in
each direction, as in the next image edges
become severely mis-aligned and blurred.
 Rotation
 cos  sin  0
x' y ' 1   sin  cos  0x y 1
 0 0 1

The rotation operator performs a geometric transform which maps

the position (x,y) of a picture element in an input image onto a
position (x’y’) in an output image by rotating it through a user-
specified angle about an origin . In most implementations,
output locations which are outside the boundary of the image are
ignored. Rotation is most commonly used to improve the visual
appearance of an image, although it can be useful as a
preprocessor in applications where directional operators are
involved.
Point Transformation
 Example. Consider rotating an image by 30 degrees
clockwise. Note that cos(30) is .866 and sin(30) is -.5.
 The transformation is given by

 Consider relocating the sample at (10, 20)

10
20
1

16
 Forward mapping
• Transformed pixel coordinates might not lie within the
bounds of the image.
• Transformed pixel coordinates can be non-integer.
• There might be no pixels in the input image that map to
certain pixel locations in the transformed image

No one-to-one
correspondence!
 Forward mapping (cont’)

• An example of holes due to image rotation, implemented

using the forward transformation.
 Inverse Mapping
• To guarantee that a value is generated for every pixel in
the output image, we must consider each output pixel in
turn and use the inverse mapping to determine the position
in the input image.
• To assign intensity values to these locations, we need to
used some form of intensity interpolation.
 Interpolation
• Interpolation — process of using known data to estimate unknown
values.

• Interpolation can be used to increase (or decrease) the number of

pixels in a digital image.
e.g., some digital cameras use interpolation to produce a larger image than
the sensor captured or to create digital zoom
http://www.dpreview.com/learn/?/key=interpolation
 Interpolation (cont’d)

Zero-order interpolation: nearest-neighbor

Interpolation (cont’d)

First-order interpolation: average

 Interpolation (cont’d)
Bilinear interpolation
I(x,y) = ax + by + cxy + d

he 4 unknowns (a,b,c,d) can be determined from 4 equ

ormed by the 4 nearest neighbors.
Rotation- Applications
Rotating this image
through a value which is
not an integer multiple of
90 degrees (e.g. in this
case equals 45 degrees)
rotates part of the image
off the visible output and
leaves many empty pixel
values, as seen in the next
image
 Rotation- Applications
Like translation, rotation may be employed in the early stages of more
sophisticated image processing operations. For example, there are
numerous directional operators in image processing (e.g. many edge
detection and morphological operators) and, in many
implementations, these operations are only defined along a limited set
of directions: 0, 45, 90, etc. A user may construct a hybrid operator
which operates along any desired image orientation direction by first
rotating an image through the desired direction, performing the edge
detection , and then rotating the image back to the original orientation.
(See Figure .)
Rotation- Applications
We can perform edge
detection along the
opposite direction to that
shown in the image by
employing a 180 degree
rotation in the edge
detection algorithm. The
result is shown in
 Geometric Scaling
The scale operator performs a geometric transformation which can be
used to shrink or zoom the size of an image (or part of an image). Image
reduction, commonly known as subsampling, is performed by
replacement (of a group of pixel values by one arbitrarily chosen pixel
value from within this group) or by interpolating between pixel values in
a local neighborhoods. Image zooming is achieved by pixel replication
or by interpolation. Scaling is used to change the visual appearance of an
image, to alter the quantity of information stored in a scene
representation, or as a low-level preprocessor in multi-stage image
processing chain which operates on features of a particular scale.

Scaling compresses or expands an image along the

coordinate directions. As different techniques can be used
to subsample and zoom
 Geometric Scaling
Figure : illustrates the two methods of sub-sampling. In the first, one
pixel value within a local neighborhood is chosen (perhaps randomly) to
be representative of its surroundings. (This method is computationally
simple, but can lead to poor results if the sampling neighborhoods are
too large.) The second method interpolates between pixel values within a
neighborhood by taking a statistical sample (such as the mean) of the
local intensity values.
 Geometric Scaling
An image (or regions of an image) can be zoomed either through pixel
replication or interpolation. Figure 2 shows how pixel replication simply
replaces each original image pixel by a group of pixels with the same
value (where the group size is determined by the scaling factor).
Alternatively, interpolation of the values of neighboring pixels in the
original image can be performed in order to replace each pixel with an
expanded group of pixels.