• View
• Edit
• History
• Print

SIFT

This page documents the Scale Invariant Feature Transform (SIFT) pipeline. The goal is to use SIFT as a target application to be accelerated by the Dreamer architecture. Ideally implementation in Chisel will allow eventual deployment on a quadrotor, and along the way interesting design space exploration.

At a high level, SIFT turns an image stream into a collection of keypoints, or features, that are easily identifiable from different perspectives. The main paper for the SIFT algorithm is (link):

• Lowe, David G. (2004). "Distinctive image features from scale-invariant key points." International Journal of Computer Vision 60(2): 91-110.

A lot of this work was started in making the project for CS150 FA13. The SIFT lecture slides (1,2) provide a lot of detail on our constraints for the streaming hardware interface. You can find tools for testing SIFT feature detection in the Project section of the class website.

The CS150 project is based on a reference hardware implementation (link):

• Bonato, V.; Marques, E.; Constantinides, G.A.,"A Parallel Hardware Architecture for Scale and Rotation Invariant Feature Detection," Circuits and Systems for Video Technology, IEEE Transactions on , vol.18, no.12, pp.1703,1712, Dec. 2008 doi: 10.1109/TCSVT.2008.2004936