Computer Vision - an assignment based course

---

Lecturer: Simon Korman

Assignees:

Daniel Poroshan - https://github.com/MacDucky

Nir Segal - https://github.com/NirSeg

---

• Computer Vision - an assignment based course
  • Assignment 1
    • Puzzle solving (via image stitching)
    • Objective:
      • SIFT
      • RANSAC
  • Assignment 2

---

Assignment 1

Puzzle solving (via image stitching)

Objective:

---

Given:

• N square puzzle pieces
• An affine or homography transformation for the first puzzle piece
• Final image dimensions of the solved puzzle

Figure out a way to transform and stitch together all the puzzle pieces seamlessly into one single piece.

Input example:

      

      Affine Transformation for the first piece:

0.072873879065059	-1.33125760466063	558.492032374654
1.46728322166917	0.138337505192396	5.04480717894527
-3.13063377305189e-18	0	1

      note that the last row is approximately [0,0,1] therefore this is trully an affine transformation

      Final image size: 521x760

---

To achieve this goal, we've utilized 2 main concepts:

• Keypoint matching in the image with SIFT algorithm
• RANSAC image fitting and refitting

SIFT

RANSAC

---

Assignment 2