Partial Occlusion Handling for Visual Tracking via Robust Part Matching

Abstract

Part-based visual tracking is advantageous due to its robustness against partial occlusion. However, how to effectively exploit the confidence scores of individual parts to construct a robust tracker is still a challenging problem. In this paper, we address this problem by simultaneously matching parts in each of multiple frames, which is realized by a locality-constrained low-rank sparse learning method that establishes multi-frame part correspondences through optimization of partial permutation matrices. The proposed part matching tracker (PMT) has a number of attractive properties. (1) It exploits the spatial-temporal localityconstrained property for robust part matching. (2) It matches local parts from multiple frames jointly by considering their low-rank and sparse structure information, which can effectively handle part appearance variations due to occlusion or noise. (3) The proposed PMT model has the inbuilt mechanism of leveraging multi-mode target templates, so that the dilemma of template updating when encountering occlusion in tracking can be better handled. This contrasts with existing methods that only do part matching between a pair of frames. We evaluate PMT and compare with 10 popular state-of-the-art methods on challenging benchmarks. Experimental results show that PMT consistently outperform these existing trackers.

Robust Visual Tracking via Exclusive Context Modeling

“…By forming the Lagrange function of (9) and taking its minimum with respect to the primal variables r, z, c g ∀g, we obtain its dual problem in (10), where g(u 0 , u g ) is the dual function expressed in (11) and (u 0 , u g ∀g ∈ G) are the dual variables corresponding to the primal equality and inequality constraints, respectively min u 0 ,u g ≥0∀g g(u 0 , u g ).…”

“…Recently, sparse representation has been successfully applied to visual tracking [1], [5]- [10]. In this case, the tracker represents each target candidate as a sparse linear combination of dictionary templates that can be dynamically updated to maintain an up-to-date target appearance model.…”

Consistent multi-layer subtask tracker via hyper-graph regularization

“…In [37] , low-rank sparse learning is adopted to consider the correlations among particles for robust tracking. Part Matching Tracker (PMT) is developed for robust visual tracking in [31] , which realizes part matching among multiple frames by optimizing a partial permutation matrix for each frame, using locality-constrained low-rank and sparsity of matched parts as criteria. But, this method ignores the discriminative ability from its surrounding context information.…”

“…is the most representative work, and some extensions [5,[29][30][31][32] are developed to improve the l 1 tracker in terms of both speed and accuracy. In [29] , APG based solution is used to improve the l 1 tracker.…”

Inter-Person Occlusion Handling with Social Interaction for Online Multi-Pedestrian Tracking

“…One of the most challenging issues is the handling of inter-person occlusion, and it is one of the key aspects considered for tracking. Many reported methods [1]- [3] consider the changes in appearance during occlusion. Others [4] focus on solving the issue of missed targets by utilizing scene knowledge caused by occlusion.…”