CV Workshop 4

MISCADA – computer vision

Workshop 4

Part 1: Smoothing filters (15 min)

• Revise the lecture where I introduced the Gaussian and other smoothing filters
• Check the filters implemented in the Pillow library https://pillow.readthedocs.io/en/stable/reference/ImageFilter.html
• Implement code that uses BLUR, SMOOTH and SMOOTH_MORE
• Using Matplotlib, display all the results, including the original image you have used

Part 2: moving shape on an image (15 min)

• Please study the code provided in the next slide
• It makes use of the ipywidgets library, which provides interactivity
• Do implement code that moves a transparent rectangle on the image
• This will have to be controlled by two sliders, for horizontal and vertical movement

Part 3: sliding filter (30 min)

• Use the code I provided in the previous slide
• This time, instead of having a transparent rectangle you will implement a sliding filter
• The two sliders will control the horizontal and vertical position of the sliding window
• The sliding window will be filtered, and the original image in the rectangular area will be replaced with the filtered pixels
• You can see how an image can be cropped using https://learnopencv.com/cropping-an-image-using-opencv/#:~:text=There%20is%20no%20specific%20function,And%20it’s%20done!

Part 4: edge filters (10 min)

• Study the edge filters implemented in the Pillow library https://pillow.readthedocs.io/en/stable/reference/ImageFilter.html
• Using the code developed for the previous problem, use the sliding window to extract edge information in the moving window

Part 5: Interest point detection (30 min)

• Go back to the lecture that covered the corner detectors
• Using the OpenCV library, study how corners can be extracted using the Harris function https://docs.opencv.org/3.4/dc/d0d/tutorial_py_features_harris.html
• Run the Harris corner detector on the shape image Implement Harris with the sliding window as described in the previous exercises
• Use the circle drawing function provided in OpenCV https://www.geeksforgeeks.org/python-opencv-cv2-circle-method/