Panorama - Linear stitching
- Identify the key features and feature descriptions using SIFT
- Calculate Similarity (distance) between these key features of two images
- Apply Lowe's test to identify best matches
- Calculate homograpy
- Warp Perpective and align images
Line-Circle Detection - Hough Transformation: Hough Transformation is a voting based algorithm in which we create parameter space with p and theta and if the pixel point falls in that space then its incremented by 1 the one with more votes are the outstanding lines. So if the length of lines are small then it will have lesser votes as it contains less pixel points
- Convert image to binary image
- Create hough transform matrix with parameter range
- Accumulate pixel points in the range
- Search for lines with the preper threshold - number of votes
- Apply angles to filter out wrong lines and circles
- Remove overlaps with the higher voted lines
- plot on the image the identified lines and circles.
Note: Code here assumes the x axis is vertical and y axis - horizontal But cv2 hough transforms considers x-axis as horizontal and y axis as vertical.