How can machines perceive the dynamic world around us? In this video, we discuss an influential Lucas-Kanade tracking method. The core algorithm and its variants are used in a wide variety of computer vision applications. Stay until the end to learn about the inspiring story behind this seminal paper!

Reference: 
- Bruce D Lucas and Takeo Kanade, An iterative image registration technique with an application to stereo vision, IJCAI 1981.

Learn more about detecting good features to track: https://www.youtube.com/watch?v=vSyBDHFIXKg

Credits:
- Teaser image: https://www.digitalcameraworld.com/tutorials/how-to-create-multi-frame-action-sequences
- Full talk by Prof. Takeo Kanade: https://www.youtube.com/watch?v=0B9hhOTYa_I

Jia-Bin Huang

An explanation of the Lucas-Kanade optical flow algorithm for feature tracking in computer vision. Covers the core brightness constancy assumption, Taylor expansion approximation, the aperture problem, and the least-squares solution using overdetermined systems. Also explains two extensions for handling large motions: iterative refinement and coarse-to-fine Gaussian pyramid search. Ends with an anecdote from Takeo Kanade about how Bruce Lucas convinced him to publish what became one of the most cited papers in computer vision history.

How Do Computers See Motion? Lucas-Kanade Method Explained