KNN Video Matting

The Hong Kong University of Science and Technology
Department of Computer Science and Engineering

Title: "KNN Video Matting"

by

LI Dingzeyu

Abstract

We demonstrate how the nonlocal principle benefits video matting via the 
KNN Laplacian, which comes with a straightforward implementation using 
motion-aware K nearest neighbors. We utilize motion information by turning 
the aperture problem to our advantage. In hindsight, the fundamental 
problem to solve in video matting is to produce spatio-temporally coherent 
clusters of moving foreground pixels. When used as described, the 
motion-aware KNN Laplacian is effective in addressing this fundamental 
problem, as demonstrated by sparse user markups typically on only one 
frame in a variety of challenging examples featuring ambiguous foreground 
and background colors, changing topologies with disocclusion, significant 
illumination changes, fast motion, and motion blur. When working with 
existing Laplacian-based systems, we expect our Laplacian can benefit them 
immediately with an improved clustering of moving foreground pixels.

Date            : 11 May 2013 (Sat)

Time            : 13:10 to 13:50

Venue           : Room 5506

Advisor         : Prof. C.K. TANG

2nd Reader      : Dr. Pedro SANDER