Abstract
Many studies argue that integrating multiple cues in an adaptive way increases tracking performance. However, what is the definition of adaptiveness and how to realize it remains an open issue. On the premise that the model with optimal discriminative ability is also optimal for tracking the target, this work realizes adaptiveness and robustness through the optimization of multi-cue integration models. Specifically, based on prior knowledge and current observation, a set of discrete samples are generated to approximate the foreground and background distribution. With the goal of optimizing the classification margin, an objective function is defined, and the appearance model is optimized by introducing optimization algorithms. The proposed optimized appearance model framework is embedded into a particle filter for a field test, and it is demonstrated to be robust against various kinds of complex tracking conditions. This model is general and can be easily extended to other parameterized multi-cue models.
Highlights
The goal of visual tracking is to obtain the state of interest target including the location and motion data
Tracking performance depends primarily on how discriminative the appearance model is in distinguishing an object from its surroundings
The scene is very complex and contains illumination changes, similar objects, partial occlusion, abrupt scene changes, etc.; these factors make it difficult to find a good margin that allows for a clear classification between the two classes. (ii) The complexity and variety of the target’s appearance: Targets, especially non-rigid targets, always change their shape and show complex inner structural deformation, which challenges appearance modeling methods
Summary
The goal of visual tracking is to obtain the state of interest target including the location and motion data. Given a candidate feature set and integration model, we combine prior knowledge and current observation, and define an evaluation function of a visual model, realizing optimization by optimizing the model parameters. We cannot obtain the target’s real state and cannot observe the target and its background directly in real problems; a Monte Carlo simulation method is employed to generate samples and approximate the distribution of foreground and background pixels. These samples are not generated randomly but are associated with the prior knowledge.
Sign-in/Register to view highlights & summary Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
