search

sumeyye-agac / homography-and-image-stitching-from-scratch

Implemention of a python program which can stitch input images together to create a single panoramic image, called blended image
7 stars 0 forks source link

How to produce the points #1

Open MahejabeenNidhi opened 10 months ago

MahejabeenNidhi commented 10 months ago

Hi, thank you for your work. I am still learning homography and image stitching.

Could you please let me know how you were able to generate the points in the .npy files?

Best.

sumeyye-agac commented 9 months ago

Hi, thank you for your work. I am still learning homography and image stitching.

Could you please let me know how you were able to generate the points in the .npy files?

Best.

Hi, thank you for your interest. The points in the .npy files are generated through user interaction using the ginput function. Users click on distinctive features in the images, such as corners or key landmarks. Then, these points are stored in the .npy files for subsequent use in homography calculations.

For instance, when creating "points_parisa-paris_ab.npy", we aim to select coordinates of key points in image paris_a, considering common image regions with image paris_b as much as possible. Similarly, in "points_parisb-paris_ab.npy", we select coordinates of key points in image paris_b, considering common image regions with image paris_a as much as possible. Therefore, there is not just one set of points .npy for image paris_b. The points selected for stitching with image paris_a ("points_parisb-paris_ab.npy") differ from those selected for stitching with image paris_c ("points_parisb-paris_bc.npy").

For additional insights on how to select points, you can also take a look at the visuals on this link