(ANSI standard) program that can extract object edges

Assignment #2

Marks: 20 marks

OBJECTIVE
Design a C or C++ (ANSI standard) program that can extract object edges and detect the
presence of straight lines.
BACKGROUND
Automated visual quality control and visual monitoring of industrial assembly lines are
the application areas where computer vision has proven its efficiency. Vision systems can
operate continuously and provide measurement of critical parameters such as position
and orientation of objects, their shape, colour, etc. An obvious advantage of such systems
is the possibility of controlling image acquisition process optimising illumination,
position of cameras and background colour. As a result, images used for automated visual
inspection have good quality in terms of contrast and SNR.
Your program shall include the following stages:
1. Rescale the image down by the factor of 2, to reduce computational complexity of the
subsequent stages
2. Extract boundaries of objects using 3x3 gradient operators
3. Save the generated b/w image gradient.bmp
4. Detect the presence of straight lines and measure their orientation with the precision 3o
5. Extract boundaries of objects using binary morphology
6. Save the generated b/w image morphology.bmp
7. Detect the presence of straight lines and measure their orientation with the precision 3o
8. Compare performance and efficiency of both methods and reflect your observations,
findings and conclusions in the report assignment2report.txt
DESIGN SPECIFICATION
You need to use OpenCV library to implement the solution.
It can be tested using asm2.bmp test image. Since information about colour is not
required, it contains only 8bit/pix luminance component. The image size is 640x480.
Your program shall produce:
1. Two 8bit/pix black/white (b/w) images gradient.bmp and morphology.bmp
( size 320x240), which represent object boundaries obtained by two boundary
extraction methods. The level of background should be set to 0 (black), while the pixels
corresponding to boundaries are set to 255 (white).
Comments: You may try to apply a LPF before the gradient-based method to improve
overall accuracy of edge detection. If needed, you can also apply binary morphological

methods ( erosion, dilation, etc) to smooth boundaries obtained as a thresholded output of
the gradient-based method.
Considering the second solution, before applying the morphological boundary detection
method, you need to binarize the input image first. The test image provided has a very
high contrast and therefore selection of a threshold is not very difficult. For asm2.bmp
the threshold can be set around 90-100.
2. For each boundary detection method you need to printout parameters of all detected
lines in the following format:
Line detection in a b/w image produced by gradient operators:
Line 1 is detected. Orientation = … degrees
Line 2 is detected. Orientation = … degrees
or
No straight lines detected
Having implemented and tested the program, you should investigate some practical
aspects of boundary extraction and straight-line detection. For example:
- You can compare accuracy and computational complexity of two boundary detection
methods
- You can evaluate efficiency of edge detection with LPF pre-processing and without it
- You can measure computational complexity and execution time of Hough Transform
- You can analyze sensitivity of your system to noise by running the program with the
second test image asm2n.bmp
Your observations, findings and conclusions must be summarized in the report
assignment2report.txt
SUBMISSION
In this assignment you have to submit:
- assignment2.c (or assignment2.cpp) source file with appropriate comments
- assignment2report.txt (doc, or pdf) report
The report must provided a 1-2 page description of the implemented solution, its analysis
regarding sensitivity to noise and computational complexity, your observations and
conclusions, which show your understanding of the theoretical aspects of edge detection,
line detection and their practical implementation.
All submitted files must include an assignment header with your name, your student
number and your email address. Anonymous submissions without these details will
not be assessed and will not be marked.
Do not add a separate header *.h file to your source code as it makes evaluation and
marking of the assignment more complex.

Add the two files to a “.zip” archive for submission. On the Moodle site for the subject
you will find a section with the title “Assignment Submission”. Under “Assignment 2”
upload your “.zip” file. You are allowed to submit up to 3 times before the deadline.