Integrating the Projective Transform with Particle Filtering for Visual Tracking

Visual tracking of humans or objects in motion is a challenging problem when observed data undergo appearance changes (e.g., due to illumination variations, occlusion, cluttered background, etc.). Moreover, tracking systems are usually initialized with predefined target templates, or trained beforehand using known datasets. Hence, they are not always efficient to detect and track objects whose appearance changes over time. In this paper, we propose a multimodal framework based on particle filtering for visual tracking of objects under challenging conditions (e.g., tracking various human body parts from multiple views). Particularly, the authors integrate various cues such as color, motion and depth in a global formulation. The Earth Mover distance is used to compare color models in a global fashion, and constraints on motion flow features prevent common drifting effects due to error propagation. In addition, the model features an online mechanism that adaptively updates a subspace of multimodal templates to cope with appearance changes. Furthermore, the proposed model is integrated in a practical detection and tracking process, and multiple instances can run in real-time. Experimental results are obtained on challenging real-world videos with poorly textured models and arbitrary non-linear motions.

Download Full-text

Visual Tracking Using Multimodal Particle Filter

International Journal of Natural Computing Research ◽

10.4018/ijncr.2014070104 ◽

2014 ◽

Vol 4 (3) ◽

pp. 69-84 ◽

Cited By ~ 1

Author(s):

Tony Tung ◽

Takashi Matsuyama

Keyword(s):

Visual Tracking ◽

Particle Filtering ◽

Body Parts ◽

Tracking Process ◽

Global Formulation ◽

Flow Features ◽

Target Templates ◽

Textured Models ◽

Model Features ◽

Appearance Changes

Visual tracking of humans or objects in motion is a challenging problem when observed data undergo appearance changes (e.g., due to illumination variations, occlusion, cluttered background, etc.). Moreover, tracking systems are usually initialized with predefined target templates, or trained beforehand using known datasets. Hence, they are not always efficient to detect and track objects whose appearance changes over time. In this paper, we propose a multimodal framework based on particle filtering for visual tracking of objects under challenging conditions (e.g., tracking various human body parts from multiple views). Particularly, the authors integrate various cues such as color, motion and depth in a global formulation. The Earth Mover distance is used to compare color models in a global fashion, and constraints on motion flow features prevent common drifting effects due to error propagation. In addition, the model features an online mechanism that adaptively updates a subspace of multimodal templates to cope with appearance changes. Furthermore, the proposed model is integrated in a practical detection and tracking process, and multiple instances can run in real-time. Experimental results are obtained on challenging real-world videos with poorly textured models and arbitrary non-linear motions.

Download Full-text

Deep learning assisted robust visual tracking with adaptive particle filtering

Signal Processing Image Communication ◽

10.1016/j.image.2017.09.001 ◽

2018 ◽

Vol 60 ◽

pp. 183-192 ◽

Cited By ~ 12

Author(s):

Xiaoyan Qian ◽

Lei Han ◽

Yuedong Wang ◽

Meng Ding

Keyword(s):

Deep Learning ◽

Visual Tracking ◽

Particle Filtering

Download Full-text

Robust Visual Tracking Using Particle Filtering on SL(3) Group

Applied Mechanics and Materials ◽

10.4028/www.scientific.net/amm.457-458.1028 ◽

2013 ◽

Vol 457-458 ◽

pp. 1028-1031

Author(s):

Ying Hong Xie ◽

Cheng Dong Wu

Keyword(s):

Dynamic Model ◽

Visual Tracking ◽

Particle Filtering ◽

Transformation Process ◽

Covariance Matrices ◽

Target Area ◽

Observation Model ◽

Tracking Method ◽

Geometric Deformation ◽

Projection Transformation

Considering the process of objects imaging in the camera is essentially the projection transformation process. The paper proposes a novel visual tracking method using particle filtering on SL(3) group to predict the changes of the target area boundaries of next moment, which is used for dynamic model. Meanwhile, covariance matrices are applied for observation model. Extensive experiments prove that the proposed method can realize stable and accurate tracking for object with significant geometric deformation, even for nonrigid objects.

Download Full-text