Project  5.1: Image Warping and Mosaicing

Shoot the Pictures

Here are the digitized input images I used. The top row is the an outdoor view. It had a stright wire pole and building edges that can  be used to make sure the alignment is correct. The bottom row is an indoor view. It has a straight line at the top as well other sharp corners.

Outside Image 1	Outside Image 2	Outside Image 3
Inside Image 1	Inside Image 2	Inside Image 3

Recover Homographies

I used 10 corresponding points for each image when calculating the transofrmation between two images.  Because I used 10 points, in order to avoid an overdetermined system, I solved the homography using least-squares.

Warp The Image / Image Rectification

Here are the results of my warping. The first image is warping from a trapezoid, or a square in different perspective, to a sqaure. The second is a slanted building into somewhat of a rectangular building.

Blending Into Mosaic

Below is the result of blending mulitple of my warped images together into one image. I used a linear blending technique and whited out the edges.

What I Learned

The most interesting thing that I learned was about creating and applying homography transformations uniformly to an entire image. It's very interesting how you can apply one transformation for an entire image, just to switch perspective.