Task 1: Edge Analysis

• The notebook containing my solution to this task is Edge_Analysis_Notebook.ipynb
• The photo I took for this experiment: Original_Images/Edge_Analysis/photo_elshan.jpg
• Pip dependencies: pip install numpy matplotlib opencv-python
• Canny for Edge Detection, Shi-Tomasi (goodFeaturesToTrack) for Corner Detection, Hough for Line Detecton, Manual Contour Fitting for Ellipse/Circle Detection.
• The parameters for each detection algorithm were manually tuned to yield better results.
• The results are stored at: Results/Edge_Analysis_Results

Task 2: Active Contour

• The scipt name: Contour_Interactive_Elshan.py
• The sample images to test the script are at: Original_Images/Active_Contour
• The results are stored at: Results/Active_Contour
• Pip dependencies: pip install numpy opencv-python scikit-image

How to use it:

1. Run the script from the command line with the image file path as an argument.

(For example, python Contour_Interactive_Elshan.py Original_Images/Active_Contour/Car_Contour_Sample.jpg)

3. Once the image window opens, use the mouse to draw an initial contour:
4. Left mouse button click and drag to draw.
5. Release the mouse button to finish drawing.
6. If you need to redraw the contour, press R to reset the image to its original state.
7. Press the Space Bar when you are done drawing the initial contour. The script will then process the image using the active contour algorithm, showing the contour adjustments gradually.
8. The final result will be displayed in a window, and the image with the contour drawn on it will be saved at the following path: Results/Active_Contour within the script's running directory. The file itself will be named according to the input file but with _contour_result.png appended to the original file name.

Task 3: Interest points

• The notebook containing my solution to this task is: Interest_Points_Notebook.ipynb
• I photographed the entrance of the ADA campus from 2 angles ('First_Frame.png', 'Second_Frame.png'); they are stored at Original_Images/ORB_Images
• I have employed the ORB algorithm setting its number of features to 1000 and used the Flann interest point matcher.
• Pip dependencies: pip install numpy matplotlib opencv-python