Skip to content

Latest commit

 

History

History
41 lines (30 loc) · 1.29 KB

README.md

File metadata and controls

41 lines (30 loc) · 1.29 KB

SIFT

Paper & associated research

Read the associated paper on Methods to improve SIFT

Goal

Scale invariant feature transform for image processing. This is an investigation of methods to improve SIFT algorithms.

Execution strategy

  1. Images are first transformed into a collection of vectors.
  2. Keypoints (of high contrast) for distinguishing features in the image are identified and stored in a DB
  3. Some unstable points can falsely record as keypoints. We avoid this using keypoint localization
  4. The resulting feature vector is denoised. It's much smaller and more accurate.

Results

Feature detection

feature detection

Keypoint localization

keypoint localization

Learnings

Keypoint localization eliminated a significant amount of noise. Varying the image's contrast produced more accurate results, even with keypoint localization.

Contributors

Shruti Appiah, Mira Sleiman

License

License: MIT

Copyright © 2017 Shruti Appiah