Abstract:One of the crucial problems in visual tracking is how the object is represented. Conventional appearance-based trackers are using increasingly more complex features in order to be robust. However, complex representations typically not only require more computation for feature extraction, but also make the state inference complicated. We show that with a careful feature selection scheme, extremely simple yet discriminative features can be used for robust object tracking. The central component of the proposed method is a succinct and discriminative representation of the object using discriminative non-orthogonal binary subspace (DNBS) which is spanned by Haar-like features. The DNBS representation inherits the merits of the original NBS in that it efficiently describes the object. It also incorporates the discriminative information to distinguish foreground from background. However, the problem of finding the DNBS bases from an over-complete dictionary is NP-hard. We propose a greedy algorithm called discriminative optimized orthogonal matching pursuit (D-OOMP) to solve this problem. An iterative formulation named iterative D-OOMP is further developed to drastically reduce the redundant computation between iterations and a hierarchical selection strategy is integrated for reducing the search space of features. The proposed DNBS representation is applied to object tracking through SSD-based template matching. We validate the effectiveness of our method through extensive experiments on challenging videos with comparisons against several state-of-the-art trackers and demonstrate its capability to track objects in clutter and moving background.

Moving object tracking based on bidirectional two-dimensional principle component analysis

A New Detection and Tracking Algorithm for Moving Objects

Moving Object Tracking Research Based on Active Vision

Object Tracking Method Based on Incrementally Updating Linear Discriminant Subspace

Moving Object Detection Based on Two-dimensional PCA

Weighted Incremental Subspace Learning Algorithm Suitable for Object Tracking

Robust visual tracking via weighted incremental subspace learning

Research on 3D Object Recognition Algorithm Based on PCA and Moment Invariant

Object Tracking Using Incremental 2D-LDA Learning and Bayes Inference.

Tracking Algorithm Based on Principal Component Variation Detection of Histogram

Two-Dimension Principal Component Analysis-Based Motion Detection Framework with Subspace Update of Background

Detecting Moving Objects from Dynamic Background Combining Subspace Learning with Mixed Norm Approach

Object Tracking Algorithm based on Kalman Filter and Two Level Feature Matching

Application of Improved Bhattacharyya Coefficient Based Multi-Object Detection and Tracking Integrated Strategy

Tracking Randomly Moving Objects on Edge Box Proposals

Object Tracking Algorithm Based on Adaptive Background Information

Incremental Pairwise Discriminant Analysis Based Visual Tracking

Discriminative object tracking with subspace representation

Heavy-Tailed model for visual tracking via robust subspace learning

Efficient Discriminative Nonorthogonal Binary Subspace with its Application to Visual Tracking

Discriminative Nonorthogonal Binary Subspace Tracking