Image Stitching

1
Image Stitching

• By Chris Dapkus Zach Goff

2
Ideal Images

• An ideal image is one taken where the focal point
  is at infinity
• With this situation, stitching images together
  only involves x and y translation and rotation

3
X Translation Results

• Given 2 images, the first column of the second
  image is compared with every column in the second
  image finding the error between them
• The match with the smallest error becomes the
  stitching point
• The second image is then cut and attached to the
  first image

4
Resulting Image
5
XY Translation Results

• The same error process can be applied in x and y
  directions, giving a 3D error plot

6
Resulting Image
7
Image Warping

• Camera lenses and perspective distorts 2d
  representation
• To match features exactly, images must be warped.
• Corresponding control points between 2 images are
  selected to control warping
• Image is warped using homography

8
Homography Matrix
9
Warped Image
10
Warped Image
11
Comparing Methods
12
Potential Problem