On this project my goal is to build a Computer Vision System that stitch two camera views to generate a panorama.
The algorithm can be described as follows:
• Step 0: Read all the images from the images folder sorting by name
• For every two images:
o Step 1: Resize images for faster processing
o Step 2: Detect keypoints and extract local invariant descriptors from the two images
o Step 3: Match the descriptors between the two images
o Step 4: Use RANSAC algorithm to estimate a Homography Matrix using our matched feature vectors
o Step 5: Apply warping transformation using the Homography Matrix obtained from Step 4
o Step 6: Visualize keypoint correspondences between two images
o Step 7: Plot the images in only one window
o Step 8: Remove/Reduce black edges from the Panorama if exists
Sample 1: the panorama is not perfect so we need to apply step 8 to crop the image
Sample 2: the panorama is perfect so we dont need to apply step 8
RUN: type python stitch.py in the command line