An improved similarity measure in particle filters for robust object tracking

Abstract

Infrared object tracking plays a key role in many research fields, and there is a series of work on applying particle filter to this tracking problem. Most of the PF-based tracking algorithms utilize the Bhattacharyya coefficient as a similarity measure, however, its performance in infrared object tracking is limited due to insufficient discriminative power. In this paper, we present a combined similarity measure under the particle filter framework, which integrates the advantages of the Bhattacharyya coefficient, histogram intersection, and structural similarity. The experimental results are gained by using different infrared image sequences, which show that the proposed measure gives superior discriminative power and achieves more robust and stable tracking performance than the traditional approach.

Unifying Registration Based Tracking: A Case Study with Structural Similarity

“…The positive constants α, β, γ are used to assign relative weights to these components while C 1 , C 2 , C 3 are added to ensure their numerical stability with small denominators. Here, as in most practical implementations [69,41,42,43,68], it is assumed that α = β = γ = 1 and C 3 = C 2 2 so that Eq. 2 simplifies to:…”

“…It is reasonable, therefore, to expect it to perform well for tracking under challenging conditions too. As such, it has indeed been used for tracking before with particle filters [41,68], gradient ascent [43] and hybrid [42] SMs. All of these trackers, however, used imprecise SSMs with low degrees of freedom (DOF) -estimating only translation and scaling of the target patch.…”

Robust particle filter-anti-zero bias modification

“…during shopping). Article is also focused on the human tracking problem; however, in this approach the measurement model has been changed. As authors wrote, in vision tracking the measurement model is usually based on the Bhattacharyya coefficient (probabilistic distance between two histograms).…”