Abstract:The work proposed in this paper, addresses the issue of robust tracking scheme by further studying the problem occlusion that causes tracker to drift. The proposed work addresses the problem within context of real-time tracking for surveillance applications. Firstly, we studied the occlusion and drift problems and how it is linked to the visual object tracking framework. Secondly we proposed a robust tracking scheme that can handle occlusion and drift problems as well as other visual object tracking challenges to predict the target object position when occlusion is occurred. The proposed scheme adopts an efficient integration of motion modeling via particle-kalman-filter (PKF) into the kernelized correlation filter (KCF) tracking framework to achieve an efficient and robust tracking scheme that mitigate the problem of tracker drift. In the proposed tracking scheme KCF acts as our basic tracker due to its better performance and high efficiency but the tracker lags behind other state-of-the-art when there are problems like occlusion and illumination variation causing it to drift. When the occlusion occurs, the PKF will be used to predict the target object location and will use the available position and state of the target object before occlusion is occurred. An experimental result on publicly available dataset demonstrates that the proposed scheme achieves a competitive performance as compared with other state-of-the-art trackers.

Robust Occlusion Handling in Object Tracking

Robust Human Tracking Via Key Face Information.

Robust and Accurate Object Tracking under Various Types of Occlusions

Robust Object Tracking with Scene-Adaptive Scheme in Occlusion.

Robust object tracking with occlusion handle

Robust Visual Tracking With Context-Based Active Occlusion Recognition

Real-Time Patch-Based Tracking with Occlusion Handling.

A Robust Object Tracking Method for Surveillance Applications to Handle Occlusion

Robust Object Tracking Based on Motion Consistency.

Towards Occlusion Handling in Visual Tracking-by-Detection Systems

A Robust Tracking Algorithm Based on Feature Fusion and Occlusion Judgment

Occlusion Detection Via Correlation Filters for Robust Object Tracking

Partial Occlusion Robust Object Tracking Using an Effective Appearance Model

Robust Visual Tracking with Contiguous Occlusion Constraint

Robust Compressive Tracking under Occlusion

Efficient Block-Division Model for Robust Multiple Object Tracking.

Object Tracking with Serious Occlusion Based on Occluder Modeling

A novel occlusion-adaptive object tracking method

An improved object tracking algorithm based on adaptive weighted strategy and occlusion detection mechanism

An Anti-Occlusion Algorithm for Object Tracking

Robust Object Tracking with Bidirectional Corner Matching and Trajectory Smoothness Algorithm.