I took these pictures of Seattle from the top of the Space Needle this past summer. There is significant overlap between the pictures, and there are a lot of details that should be easy to align!
I took these mountain pictures in Utah. The features of the mountain should provide good details to base correspondences off of.
This, as you can tell, is the Golden Gate Bridge in San Francisco.
For this part of the project, we had to compute the homography between the two images. To do this, I chose corresponding points on each image, and then used Least-Squares to find the matrix that would most closely transform the points from image A into the corresponding points from image B.
The next step for this project was using the homography to warp one image so that the corresponding
points aligned. This was done by multiplying the pixels in one image by the homography matrix to
get the resulting warped image.
To check that this process was working, I used the following photos (with clear square tiles)
and warped the corners of the tiles to the corner of a square.
The next step for this project was blending the warped images into a panoramic. To accomplish
this, I used the already calculated correspondences to determine how much to shift the warped image.
Then, after the warped image was shifted, I used an alpha channel with linear feathering to blend
the aligned images together, setting
alpha to be one at the center of the image and decreasing until it reaches 0 at the edges of the images.