I used same exposure for both parts. I just changed perspective around this banner in my room.
I used the following structure to create my matrix for homographies. I used least squares in order to get the actual 3x3 matrix with 8 entries.
The above example shows the 3x3 matrix for my above 2 pictures, warping image 2 to image 1.
I am warping images similar to how we did in project 3. I get a list of coordinates in the particular form of
x1, x2, x3 .... y1, y2, y3 .... 1 , 1 , 1 ....
and then multiply this by the homography matrix.
The warp I created looks like this:
In this part, we took slanted images and then applied our warp such that we can "straighten" out the object we have a picture of
In this part, I took pictures of parts of my apartment and created panoramas based on the images. The first set is a TV and dresser, the second set is a view from our balcony, and the last set is from a beer bottle and 2 post it notes. I used averaging to merge the two images instead of alpha feathering