Lab 1 – INT 302: Image Processing

Start Date: 2025-03-11      Deadline: 2025-03-25

15% of the final marks

Late Submission Policy: 5% of the total marks available for the assessment shall be deducted from the assessment mark for each working day after the submission date, up to a maximum of five working days.

Objectives:

1-    Introducing the image processing capabilities of Matlab and its Image Processing Toolbox.

2-    Learn to read and display different images.

3-    Learn basic image processing steps.

4-    Master different image enhancement techniques

Download:

Download the files of Lab1_material.zip from the Learning Mall, unzip the file into a folder Lab1_material, which contains “lenna512color.bmp”and “lenna512.bmp”.

Tasks:

1. Task1 (15 marks)

Download image “lenna512color.bmp” . Use the functions imread to load the image into  Matlab, and conduct the following questions, please specify the intermediate process of how you conduct the questions:

(1)  Display  the  image  and  decouple  the  image  into  RGB  components.  Write  the transformation code by yourself. Please plot both original image and the decoupled components. Please describe your observation about the plotted images. (5’)

(2) Change the color space into CMY and show the images of CMY components. Write the  transformation  code  by  yourself.  Please  plot  both  original  image  and  the decoupled components. Please describe your observation about the plotted images. (5’)

(3) Change the original image into gray level and show the gray image. Write the transformation code by yourself. Please describe your observation about the plotted images. (5’)

2. Task 2 (35 marks)

In this task, we use the monochrome image Lenna (i.e., “lenna512.bmp”) to do the following sub tasks, and let’s call the original image Lenna as I0.

(1) Write a function to measure the Peak Signal to Noise Ratio (PSNR) between two gray images in dB. For the peak value use 255. (5’)

where mse is the mean square error, and it is evaluated as:

(2) I0 -> down-sampling to I1  with 1/2 size of I0  (both horizontally and vertically) using nearest neighbor interpolation (implement it by yourself). Display it and compare to the original image. Explain your findings in the report; (10’)

(3)  I1->  up-sampling  to   I1’  with  the   same  size  of  I0    using  (a)nearest   neighbor interpolation;  (b)  bilinear  interpolation;  (c)  bicubic  interpolation.    Display  it  and compare to the original image. Explain your finding in the report.  (10’)

(4) Calculate the PSNR between the original image I0  and the up-sampled images, i.e., nearest,   bilinear,   and   bicubic,   respectively.   Compare   the   results   of   different interpolation methods. Explain your finding in the report. (10’)

(Note: for the bilinear and bicubic interpolation, you use the matlab function directly).

3. Task 3 (12 marks)

In this task, we use the monochrome image Lenna (i.e., lenna512.bmp) to do the  following sub tasks. You need to conduct power-law transform on the input image,  where power-law transform. is  s  = cr-, please plot the transformed images with c=1, Y=0.04, 0.4, 5, 25. Please write one function to generate this image instead of calling  matlab function directly and explain your finding in the report. (12’)

4. Task 4 (38 marks)

1) Add AWGN noise to the image Lenna (i.e., lenna512.bmp) and display the noisy image. Name it as im_SP. Please write one function to generate this image instead of calling matlab function directly (you can use Matlab function to generate uniform random numbers, e.g., rand()). (the  μ   and  σ  of the gaussian,  please set as the following: treat your ID number as a number set, compute the  μ   and  σ  for the gaussian. e.g., ID=1234567,  μ   =(1+2+3+4+5+6+7)/7=4, and the corresponding  σ .     please write down the computing process in your report, without this process, 5-mark deduction will be conducted.) and set k in the ppt slides as 0.1, 0.5, 0.7, 1. Plot the images with different k in your report and write your observation and analysis from the plotted images.

(10’)

2) Apply the median filter with a 3X3 window and a 5X5 window on the image im_SP (you  can  use  Matlab  function  medfilt2).  Display  and  evaluate  the  PSNR  of  the obtained images.  For each window size, comment on how effectively the noise is reduced while sharp edges and features in the image are preserved. (8’)

3) Implement the averaging filter 3X3 to filter the image im_SP by yourself (you are not allowed to use Imfilter and fspecial directly). Compute the PSNR and display the filtered image. You can use zero padding for the boundary pixels. (8’)

4) As you experimented with the averaging and median algorithms, what different “performance” did you notice? Which one is better for removing salt & pepper noise? and why? (12’)

Lab Report

Write a short report which should contain a concise description of your results and observations. Include listings of the Matlab scripts that you have written. Describe each of the images that you were asked to display.

Answer each question completely:

-   Do not attach the code at the end of the report, just put the useful code under each question.

-   The results maybe contain some figures, please add some index and caption of each figure.

Report format: Single column; Font size: #12; Page number: no more than 15;

Submission before 2025-03-25.

-   Electrical version to LM with a rar (ZIP) of all files

• Rar file name: INT302-Lab1-Name-studentID.rar/zip

•    One file with same file name of Rar/zip File: Report ( with studentID, name, Lab title on the homepage)

•    One folder: codes and other materials. (I can run it directly)

Marking scheme

80%-100%  Essentially complete and correct work.

60%-79%   Shows understanding, but contains a small number of errors or gaps.

40%-59%  Clear evidence of a serious attempt at the work, showing some understanding, but with important gaps.

20%-39%   Scrappy work, bare evidence of understanding or significant work omitted.

<20% No understanding or little real attempt made.