Alignment using brute force and pyramid algorithms

Jordan Kaye - 10/04/10

Visual results with times

Pyramid time: 0.1353 seconds

Brute force time: 0.2328 seconds

Pyramid time: 0.1407 seconds

Brute force time: 0.1527 seconds

Pyramid time: 0.1434 seconds*

Brute force time: 0.1538 seconds

Pyramid time: 0.1435 seconds

Brute force time: 0.1497 seconds

Pyramid time: 0.1439 seconds

Brute force time: 0.1467 seconds

Pyramid time: 0.1453 seconds

Brute force time: 0.1496 seconds

Pyramid time: 0.1490 seconds

Brute force time: 0.1554 seconds

Pyramid time: 0.1514 seconds

Brute force time: 0.1474

Pyramid time: 0.1515 seconds*

Brute force time: 0.1931 seconds

Pyramid time: 0.1522 seconds

Brute force time: 0.1590 seconds

Pyramid time: 0.1532 seconds

Brute force time: 0.1509

Pyramid time: 7.6407 seconds

Brute force time: N/A

Pyramid time: 7.6497 seconds

Brute force time: N/A

Pyramid time: 7.7304 seconds

Brute force time: N/A

Pyramid time: 7.8435 seconds

Brute force time: N/A

Pyramid time: 7.9289 seconds

Brute force time: N/A

Pyramid time: 8.0622 seconds

Brute force time: N/A

 

Images that could not be aligned

For the brute force algorithm, any of the images ending in "u" could not be aligned due to their size. After a long period of time attempting the computation, Matlab would overload and shut down.

For the pyramid based algorithm, the memory problem with the "u" images was eliminated (although the average time for these images was much longer than the "v" images); however, some images were not correctly aligned. This was most likely due to anomalies in the images that could not be recognized by the algorithm. The two images that could not be aligned (denoted with a * in the above table) both contain many areas of similarity that likely made it difficult for the algorithm to find an ideal displacement for the red and blue layers over the green anchor layer.