BEng (Hons) Electronic Engineering. Examinations for / Semester 1

Transcription

1 BEng (Hons) Electronic Engineering Cohort: BEE/13B/FT Examinations for / Semester 1 Resit Examinations for BEE/10B/FT & BEE/12/FT MODULE: DIGITAL IMAGE PROCESSING MODULE CODE: SCG4101C Duration: 2 Hours 30 Minutes Instructions to Candidates: 1. Answer all FOUR questions. 2. Questions may be answered in any order but your answers must show the question number clearly. 3. Always start a new question on a fresh page. 4. All questions carry equal marks. 5. Total marks 100. This Exam paper contains 4 questions and 8 pages. Page 1 of 8

2 ANSWER ALL QUESTIONS QUESTION 1: (25 MARKS) (a) The resolution of an image source (e.g. a camera) can be specified in terms of three quantities; Spatial resolution, Temporal resolution and Bit resolution. Explain each of them. (3 marks) (b) Visual significance of individual pixel bits in an image can be assessed in a subjective but useful manner by the technique of bit-plane splicing. i. Explain the concept of bit-plane slicing. ii. iii. Using MATLAB codes, explain how bit-plane slicing is achieved. Using MATLAB codes, explain how image reconstruction using n bit planes is carried out. (2+3+3 marks) (c) The main barrier to effective image processing and signal processing in general is noise. i. What is noise in digital images? ii. Explain with explanation four sources from which digital noise may originate. (2+4 marks) (d) MATLAB is widely used in digital image processing. The following is an extract of MATLAB code as applied to image D. Explain each line of code. D=imread( onion.png ); Dred1=D(:,:,1); Dgreen1=D(:,:,2); Dblue1=4D(:,:,3); subplot(2,2,1); imshow(d); axis image; subplot(2,2,2); imshow(dred); title( red ); subplot(2,2,3); imshow(dgreen); title( green ); subplot(2,2,4); imshow(dblue); title( blue ); (3 marks) Page 2 of 8

3 (e) Image B in the figure below is the result of a basic point transform. Image A Image B Name the operation performed on Image A to obtain Image B. (1 mark) (f) Digitising an analog signal into a digital signal requires two basic steps; Sampling and Quantisation. Differentiate between Sampling and Quantisation. (4 marks) QUESTION 2: (25 MARKS) (a) There are two broad categories of image enhancement techniques; Spatial domain techniques and Frequency domain techniques. Distinguish between these two techniques. (3 marks) (b) The transformation map plot below depicts various curves that fall under the Power Law Transformation. Page 3 of 8

4 i) The process to correct the power law phenomena is referred to as Gamma Correction. Explain the result of modifying gamma as follows: a. Set Gamma =1 b. Set Gamma > 1 c. Set Gamma < 1 (6 marks) ii) Give an example with explanation where each of the following changes to gamma would be desirable. a. Set Gamma > 1 b. Set Gamma < 1 (2 marks) (c) Basic arithmetic operations can be performed quickly and easily on image pixels for a variety of effects and applications. With the help of an example explain one such application. (2 marks) Page 4 of 8

5 (d) i) Explain what you understand by Histogram Equalisation. ii) Explain why images cannot be reconstructed from histograms. iii) Match the Images (1-4) below to their corresponding pixel histogram (A- D). iv) The table below shows the intensity distribution of a 3-bit image (L=8) of size 64 x 64 (MN =4096). Perform histogram equalization to transform it into a histogram equalised image. where r k is the k th intensity value and n k is the number of pixels in the image with intensity r k ( marks) Page 5 of 8

6 QUESTION 3: (25 MARKS) (a) i) Dilation and Erosion are two primitive operators, which may be used to define other morphological operations. (a) Explain how erosion is performed. (b) Explain how dilation is performed. (2+2 marks) ii) Binary image, X and structuring element, B, are given as follows Calculate Y1= XΘB, where Θ denotes the morphological erosion operator and Y2= X B where denotes the morphological dilation operator; (5 marks) iii) (a) Explain the concept of Morphological Opening. (b) Explain the concept of Morphological Closing. (c) Use the rolling ball analogy to explain the result of Opening. (d) Use the rolling ball analogy to explain the result of Closing. ( marks) Page 6 of 8

7 (b) i. Median Filtering is a non-linear method. Give one use of median filtering. ii. A 4 x 4 gray-scale image is given below: Filter the image with a 3x3 median filter, after zero padding. (3+5 marks) QUESTION 4: (25 MARKS) (a) In 1952, D. A. Huffman developed a code construction method that can be used to perform lossless compression. i) Explain the concept underlying the Huffman codes. ii) The table below shows different symbols with their corresponding occurring probabilities. Create a Huffman tree and Huffman table for table above. The table should show the code word for each symbol and the corresponding code-word length. Page 7 of 8

8 iii) Given the following Huffman codes, encode the string goat. Huffman Code Character 00 a 01 o 10 t 110 d 1110 g 1111 c (4+4+2 marks) (b) i. What is Edge Detection? ii. Explain how Edge Detection is achieved? iii. Distinguish between the Roberts Cross Edge and Sobel Edge Detector. (2+5+8 marks) ***END OF QUESTION PAPER*** Page 8 of 8