Robust Visual Tracking via Incremental Lorentzian Discriminant Projection

摘要

This paper develops a novel visual tracking method using Incremental Lorentzian Discriminant Projection (ILDP). A subspace based tracking method is promising, especially in handling appearance, illumination and expression variation. Inspired by the promising discriminant characteristic of Lorentzian Discriminant Projection (LDP), we apply a variant of it to visual tracking problem and incrementally update the projected space. Our method is able to track the target with large variation, such as large appearance, illumination, expression, pose and scale variation indoor and outdoor. Various experiments are conducted to demonstrate the high performance of our method.