Lab 4: Transformation Functions and Geometric Transforms.: Objective

Lab 4: Transformation functions and geometric transforms.
Objective:
The objective of this lab is:
 To apply Log transformation on images and visualize results.
 To apply power-law transform to correct gamma in images.
 To apply various geometric transformations (like translation, rotation, scaling, sheering and
affine) on images and see their effects.
Theory:
Log Transform:
Applying a simple Logarithmic transformation on image can be used to process darker
images to a relatively natural images (see Figure 1) while inverse logarithmic transform can also
be used to transform saturated images to a natural ones.
In principal, a simple log transformation can be written as:
Where c is a scaling constant which can be calculated based on the maximum allowed output
value (255 for an 9-bit image) and the maximum value max(Iinput(i,j)) present in the image:
The effect of the logarithmic transform is to increase the dynamic range of the dark regions in an
image and decrease the dynamic range in the light regions.
Figure 1: Application of Log transform.

Similarly, an inverse logarithmic transformation function will increase dynamic range in the light
regions while decreasing the dynamic range in dark regions thus can be used to process
saturated and washed out images.
Fig 2: Some basic intensity transformation functions.
Power-Law transformations:
The nth power and nth root curves shown in fig. 2 can be given by the expression:
s = crγ
This transformation function is also called as gamma correction. For various values of γ different
levels of enhancements can be obtained. This technique is quite commonly called as Gamma
Correction. If you notice, different display monitors display images at different intensities and
clarity. That means, every monitor has built-in gamma correction in it with certain gamma ranges
and so a good monitor automatically corrects all the images displayed on it for the best contrast
to give user the best experience.
The difference between the log-transformation function and the power-law

functions is that using the power-law function a family of possible transformation
curves can be obtained just by varying the λ.
These are the three basic image enhancement functions for grey scale images that can be
applied easily for any type of image for better contrast and highlighting. Using the image
negation formula given above, it is not necessary for the results to be mapped into the grey scale
range [0, L-1]. Output of L-1-r automatically falls in the range of [0, L-1]. But for the Log and
Power-Law transformations resulting values are often quite distinctive, depending upon control
parameters like λ and logarithmic scales. So the results of these values should be mapped back to
the grey scale range to get a meaningful output image. For example, Log function s = c log(1 + r)
results in 0 and 2.41 for r varying between 0 and 255, keeping c=1. So, the range [0, 2.41] should
be mapped to [0, L-1] for getting a meaningful image.
Geometric Transformation:
Unlike intensity transformations where gray-levels of image are transformed, a geometric
transformation such as translation can be applied on image facilitate both synthetic and realistic
image morphing. In a geometric transform, image pixel location x and y are transformed to x’
and y’.
Translation of an image can be expressed as adding an arbitrary value tx in x and ty in y,
which can be conveniently written as:
x’ = x + tx
y’ = y + ty
Applying the translation operation can result in moving an image from one location to
another. For further simplicity we can write aforementioned equations in a matrix form as:
In the case of translation, the values of the affine matrix can be found in Figure 3. Similarly other
operations such as scaling, sheering and rotation can be performed on image for various digital
image processing tasks.
NOTE: Opencv provide a convenient function (cv2.warpAffine()) which takes input image I,
affine_matrix A and size of the output image (width, height) as arguments and after applying the
transformation the output image is returned.
Fig 3: Geometric transformations in the form of affine matrix
Lab Tasks:
Task 1:
1. Use python notebook and opencv’s imread function and load an image (e.g. dark.tif)
2. Visualize histogram by plotting the histogram.
3. Apply log transform on image and visualize the output histogram.
4. Try to experiment by changing the scaling constant c
5. Summarize your findings by providing a figure containing 4 plots (input image, histogram of
input image, output image and histogram of output image)
Task 2:
 Read “aerial.tif” image in python notebook and store that in “img” variable.
 Read the documentation of Interact from ipywidgets and create a slider named gamma.
 By using the value of gamma, apply a power-law transform on image and figure out which value
of gamma can result in a contrast enhanced output.
 HINT: since you have to apply this transformation to each pixel, nested for loops including the
power-law operation should do the trick.
 Summarize your findings by providing a figure containing 4 plots (input image, histogram of
input image, output image and histogram of output image)
Task 3:
 Read “messi5.jpg” image and apply following transformations using cv2.warpaffine() function:
 Translation
 Rotation
 Sheering in x
 Sheering in y
 Random affine transform
 Summarize your findings by providing a figure which contains images before and after geometric
transformation.
Task 4:
 You may have noticed that in task 2 when using interact with some real-time processing, the
system takes some time to process.
 This is usually caused by applying exponential mathematical operation withing nested for loops.
 You can use the concept of mapping to apply the transformation function in an efficient manner!
 Try out both techniques (i.e. power-law in nested for loops and mapping) by increasing the size
of input image and calculate the time take by each.
 Conclude your findings in the form of graph in which x-axis should indicate the increasing size of
image and y-axis should indicate time taken for processing.

Lab 4: Transformation Functions and Geometric Transforms.: Objective

Uploaded by

Copyright:

Available Formats

Lab 4: Transformation Functions and Geometric Transforms.: Objective

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Lab 4: Transformation Functions and Geometric Transforms.: Objective

Uploaded by

Copyright:

Available Formats

Lab 4: Transformation functions and geometric transforms.

Figure 1: Application of Log transform.

Fig 2: Some basic intensity transformation functions.

The difference between the log-transformation function and the power-law

You might also like