Abstract:Multiple feature object representation has been proved as a robust approach for visual tracking. Different types of situations such as occlusion, rotation and illumination may occur during tracking, especially long sequences. Robust tracking could be obtained as multiple features could complement each other. In this paper, we cast visual tracking as a novel multi-task sparse learning problem and exploit various types of visual features, such as intensity, color, texture and edge, where each feature can be sparsely represented by a linear combination of atoms from an adaptive feature template. We use an on-line feature selection mechanism based on the two-class variance ratio measure, applied to log likelihood distributions computed with respect to a given feature from samples of object and background pixels. The proposed method is integrated in a particle filtering framework. We jointly consider the underlying relationship across different particles, and tackle it in a unified robust multi-task formulation. In addition, to capture the frequently emerging outlier tasks, we make fully use of a decomposition model which enables a more robust and accurate approximation. We show that the proposed model can be efficiently solved using the Alternating direction method of multipliers (ADMM) with a small number of closed-form updates. Four types of features are implemented and tested on numerous benchmark video sequences. Both the qualitative and quantitative results demonstrate the superior performance of the proposed approach compared to 9 state of-the-art trackers.

Measuring Particles In Joint Feature-Spatial Space

The improved particle filter for object tracking

Contour tracking using Gaussian particle filter

Error Sensitive Gaussian Particle Filter and Its Evaluation for Contour Tracking

Visual tracking using the kernel based particle filter and color distribution

Visual Object Tracking Based on Motion-Adaptive Particle Filter under Complex Dynamics

An Improved Gaussian Mixture Particle Filter Based Targets Tracking Algorithm for Bearing-Only Tracking System

Joint Feature-Spatial-Measure Space: A New Approach to Highly Efficient Probabilistic Object Tracking

Gaussian Particle Filter Based Pose and Motion Estimation

Visual Tracking Via Subspace Motion Model

Particle Filtering Algorithm Based on Dynamic Multi-Feature Fusion

Adaptive Object Tracking Based on an Effective Appearance Filter

Improved Object Tracking Algorithm Based Onparticle Filter and Galerkin's Method

Visual tracking in continuous appearance space via sparse coding

Embedding Ensemble Tracking in a Stochastic Framework for Robust Object Tracking

Adaptive Tracking Algorithm Based on Particle Grouped Optimization

Particle Filters for Tracking Target with a Camera

Particle filter based on Particle Swarm Optimization resampling for vision tracking

Multisensor Multitarget Tracking Methods Based on Particle Filter.

Visual tracking via robust multi-task multi-feature joint sparse representation

Interacting multiple model particle filter to adaptive visual tracking.