Abstract:Visual tracking can be formulated as a state estimation problem of target representation based on observations in image sequences. To investigate the integration of rough models from multiple cue and to explore computationally efficient algorithms, this paper formulates the problem of multiple cue integration and tracking to combine Interactive Multiple Model (IMM) with particle filter (IMM_PF). Interactive Multiple Model can estimate the multiple cue state of a dynamic system with several behavioral models that switch from one to another using model likelihoods and model transition probabilities. For the problem of visual tracking, the model of IMM is adopted to three target observation models: Corrected Background Weighted Histogram (CBWH), Completed Local Ternary Patterns (CLTP) and Histogram of Oriented Gradients (HOG). The probabilities of these models are corresponding to the weights of multiple cues. IMM_PF then dynamically adjusts the weights of different features. Compared with those state-of-the-art methods in the tracking literature, this algorithm can track the object accurately in conditions of rotation, abrupt shifts, as well as clutter and partial occlusions occurring to the tracking object with good robustness, as demonstrated by experimental results.

Interacting multiple model algorithm based on convolution particle filter

Using Interacting Multiple Model Particle Filter to Track Airborne Targets Hidden in Blind Doppler

Novel multiple-model probability hypothesis density filter for multiple maneuvering targets tracking

Interacting Multiple Model Tracking Algorithm of Multiple Maneuvering Targets Based on Particle Filter

Interacting multiple model algorithm based on H∞ filter

Multiple Model Bernoulli Particle Filter for Maneuvering Target Tracking

Improved Multiple Model Target Tracking Algorithm Based on Particle Filter

Interacting Multiple Model Gaussian Particle Filter

Interacting multiple model algorithm based on α-β/α-β-γ filters

Research of Interacting Multiple Model Particle Filter Based on Passive Multi-Sonar

A Multiple Model Tracking Algorithm Based on an Adaptive Particle Filter

Multiple target tracking based on particle filtering method

Interacting multiple model particle filter to adaptive visual tracking.

Maneuvering Target Tracking Algorithm with Multiple Passive Sensors in Clutter Environment

Robust visual tracking based on interactive multiple model particle filter by integrating multiple cues

Interacting Multiple Model Algorithm Based on /spl Alpha/-/Spl Beta///spl Alpha/-/Spl Beta/-/Spl Gamma/ Filters

Interacting Multiple Model Particle-type Filtering Approaches to Ground Target Tracking.

Efficient Particle Swarm Optimized Particle Filter Based Improved Multiple Model Tracking Algorithm

An improved UPF-based interacting multiple model target tracking method for dynamic acoustic array

An Improved IMMPF Object Tracking Algorithm

Research of Multiple Model Particle Filter Tracking Algorithms