Boris N. Oreshkin

Asynchronous distributed particle filter via decentralized evaluation of Gaussian products

We present a distributed particle filtering algorithm for target tracking in sensor networks. Several existing algorithms rely on the establishment and maintenance of a spanning path or tree. This is challenging in networks with dynamic topologies induced by mobile nodes and changing wireless conditions; the algorithms are vulnerable to link or node failure. More recent algorithms employ consensus algorithms to improve robustness but they adopt suboptimal fusion rules leading to a significant deterioration in performance. In our algorithm, nodes run local particle filters and then approximate their local posteriors using Gaussian approximations. A global posterior approximation is then computed using a novel gossiping approach that implements the optimal fusion rule. The resultant protocol is simple, robust and efficient. We present simulation results demonstrating a significant performance improvement over the best-performing existing algorithm with similar communication and computation requirements.

Efficient Delay-Tolerant Particle Filtering

This paper proposes a novel framework for delay-tolerant particle filtering that is computationally efficient and has limited memory requirements. Within this framework the informativeness of a delayed (out-of-sequence) measurement (OOSM) is estimated using a lightweight procedure and uninformative measurements are immediately discarded. The framework requires the identification of a threshold that separates informative from uninformative; this threshold selection task is formulated as a constrained optimization problem, where the goal is to minimize state estimation error whilst controlling the computational requirements. We develop an algorithm that provides an approximate solution for the optimization problem. Simulation experiments provide an example where the proposed framework processes less than 40% of all OOSMs with only a small reduction in state estimation accuracy.

Distributed average consensus with increased convergence rate

The average consensus problem in the distributed signal processing context is addressed by linear iterative algorithms, with asymptotic convergence to the consensus. The convergence of the average consensus for an arbitrary weight matrix satisfying the convergence conditions is unfortunately slow restricting the use of the developed algorithms in applications. In this paper, we propose the use of linear extrapolation methods in order to accelerate distributed linear iterations. We provide analytical and simulation results that demonstrate the validity and effectiveness of the proposed scheme. Finally, we report simulation results showing that the generalized version of our algorithm, when a grid search for the unknown optimum value of mixing parameter is used, significantly outperforms the optimum consensus algorithm based on weight matrix optimization.

Spatial and probabilistic codebook template based head pose estimation from unconstrained environments

In unconstrained environments, head pose detection can be very challenging due to the joint and arbitrary occurrence of facial expressions, background clutter, partial occlusions and illumination conditions. Despite the wide range of head pose literature, most current methods can address this problem only up to a certain degree, and mostly for restricted scenarios. In this paper, we address the problem of head pose classification from real world images with large appearance variation. We represent each pose with a probabilistic and spatial template learned from facial codewords. The inference of the best template representing a test image is achieved probabilistically and spatially at the codebook. The experimental results are obtained from 5500 video frames collected under different illumination and background conditions. Our probabilistic framework is shown to outperform the current state-of-the-art in head pose classification.

Efficient delay-tolerant particle filtering through selective processing of out-of-sequence measurements

This paper proposes a novel algorithm for delay-tolerant particle filtering that is computationally efficient and has limited memory requirements. The algorithm estimates the informativeness of delayed (out-of-sequence) measurements (OOSMs) and immediately discards uninformative measurements. More informative measurements are then processed using the storage efficient particle filter proposed by Orguner et al. If the measurement induces a dramatic change in the current filtering distribution, the particle filter is re-run to increase the accuracy. Simulation experiments provide an example tracking scenario where the proposed algorithm processes only 30-40% of all OOSMs using the storage efficient particle filter and 1-3% of OOSMs by re-running the particle filter. By doing so, it requires less computational resources but achieves greater accuracy than the storage efficient particle filter.

Bootstrapping Particle Filters using Kernel Recursive Least Squares

Although particle filters are extremely effective algorithms for object tracking, one of their limitations is a reliance on an accurate model for the object dynamics and observation mechanism. The limitation is circumvented to some extent by the incorporation of parameterized models in the filter, with simultaneous on-line learning of model parameters, but frequently, identification of an appropriate parametric model is extremely difficult. This paper addresses this problem, describing an algorithm that combines kernel recursive least squares and particle filtering to learn a functional approximation for the measurement mechanism whilst generating state estimates. The paper focuses on the specific scenario when a training period exists during which supplementary measurements are available from a source that can be accurately modelled. Simulation results indicate that the proposed algorithm, which requires very little information about the true measurement mechanism, can approach the performance of a particle filter equipped with the correct observation model.

Uncertainty Driven Probabilistic Voxel Selection for Image Registration

This paper presents a novel probabilistic voxel selection strategy for medical image registration in time-sensitive contexts, where the goal is aggressive voxel sampling (e.g., using less than 1% of the total number) while maintaining registration accuracy and low failure rate. We develop a Bayesian framework whereby, first, a voxel sampling probability field (VSPF) is built based on the uncertainty on the transformation parameters. We then describe a practical, multi-scale registration algorithm, where, at each optimization iteration, different voxel subsets are sampled based on the VSPF. The approach maximizes accuracy without committing to a particular fixed subset of voxels. The probabilistic sampling scheme developed is shown to manage the tradeoff between the robustness of traditional random voxel selection (by permitting more exploration) and the accuracy of fixed voxel selection (by permitting a greater proportion of informative voxels).

Analysis of error propagation in particle filters with approximation

This paper examines the impact of approximation steps that become necessary when particle filters are implemented on resource-constrained platforms. We consider particle filters that perform intermittent approximation, either by subsampling the particles or by generating a parametric approximation. For such algorithms, we derive time-uniform bounds on the weak-sense

Optimization and analysis of distributed averaging with memory

L_p

Optimization over random and gradient probabilistic pixel sampling for fast, robust multi-resolution image registration

error and present associated exponential inequalities. We motivate the theoretical analysis by considering the leader node particle filter and present numerical experiments exploring its performance and the relationship to the error bounds.