Exploiting a geometrically sampled grid in the steered response power algorithm for localization improvement.

The steered response power phase transform (SRP-PHAT) is a beamformer method very attractive in acoustic localization applications due to its robustness in reverberant environments. This paper presents a spatial grid design procedure, called the geometrically sampled grid (GSG), which aims at computing the spatial grid by taking into account the discrete sampling of time difference of arrival (TDOA) functions and the desired spatial resolution. A SRP-PHAT localization algorithm based on the GSG method is also introduced. The proposed method exploits the intersections of the discrete hyperboloids representing the TDOA information domain of the sensor array, and projects the whole TDOA information on the space search grid. The GSG method thus allows one to design the sampled spatial grid which represents the best search grid for a given sensor array, it allows one to perform a sensitivity analysis of the array and to characterize its spatial localization accuracy, and it may assist the system designer in the reconfiguration of the array. Experimental results using both simulated data and real recordings show that the localization accuracy is substantially improved both for high and for low spatial resolution, and that it is closely related to the proposed power response sensitivity measure.

How a visual surveillance system hypothesizes how you behave.

In the last few years, the installation of a large number of cameras has led to a need for increased capabilities in video surveillance systems. It has, indeed, been more and more necessary for human operators to be helped in the understanding of ongoing activities in real environments. Nowadays, the technology and the research in the machine vision and artificial intelligence fields allow one to expect a new generation of completely autonomous systems able to reckon the behaviors of entities such as pedestrians, vehicles, and so forth. Hence, whereas the sensing aspect of these systems has been the issue considered the most so far, research is now focused mainly on more newsworthy problems concerning understanding. In this article, we present a novel method for hypothesizing the evolution of behavior. For such purposes, the system is required to extract useful information by means of low-level techniques for detecting and maintaining track of moving objects. The further estimation of performed trajectories, together with objects classification, enables one to compute the probability distribution of the normal activities (e.g., trajectories). Such a distribution is defined by means of a novel clustering technique. The resulting clusters are used to estimate the evolution of objects' behaviors and to speculate about any intention to act dangerously. The provided solution for hypothesizing behaviors occurring in real environments was tested in the context of an outdoor parking lot

An adaptive high-order neural tree for pattern recognition.

A new neural tree model, called adaptive high-order neural tree (AHNT), is proposed for classifying large sets of multidimensional patterns. The AHNT is built by recursively dividing the training set into subsets and by assigning each subset to a different child node. Each node is composed of a high-order perceptron (HOP) whose order is automatically tuned taking into account the complexity of the pattern set reaching that node. First-order nodes divide the input space with hyperplanes, while HOPs divide the input space arbitrarily, but at the expense of increased complexity. Experimental results demonstrate that the AHNT generalizes better than trees with homogeneous nodes, produces small trees and avoids the use of complex comparative statistical tests and/or a priori selection of large parameter sets.

Invariant feature extraction and neural trees for range surface classification.

In this paper, a neural tree-based approach for classifying range images into a set of nonoverlapping regions is presented. An innovative procedure is applied to extract invariant surface features from each pixel of the range image. These features are: 1) robust to noise, and 2) invariant to scale, shift, rotations, curvature variations, and direction of the normal. Then, a generalized neural tree is used to classify each image point as belonging to one of the six surface models of differential geometry, i.e., peak, ridge, valley, saddle, pit, and flat. Comparisons with other methods and experiments on both synthetic and real three-dimensional range images are proposed.

Generalized neural trees for pattern classification.

In this paper, a new neural tree (NT) model, the generalized NT (GNT), is presented. The main novelty of the GNT consists in the definition of a new training rule that performs an overall optimization of the tree. Each time the tree is increased by a new level, the whole tree is reevaluated. The training rule uses a weight correction strategy that takes into account the entire tree structure, and it applies a normalization procedure to the activation values of each node such that these values can be interpreted as a probability. The weight connection updating is calculated by minimizing a cost function, which represents a measure of the overall probability of correct classification. Significant results on both synthetic and real data have been obtained by comparing the classification performances among multilayer perceptrons (MLPs), NTs, and GNTs. In particular, the GNT model displays good classification performances for training sets having complex distributions. Moreover, its particular structure provides an easily probabilistic interpretation of the pattern classification task and allows growing small neural trees with good generalization properties.

Visual inspection of sea bottom structures by an autonomous underwater vehicle.

This paper describes a vision-based system for inspections of underwater structures, e.g., pipelines, cables, etc., by an autonomous underwater vehicle (AUV). Usually underwater inspections are performed by remote operated vehicles (ROVs) driven by human operators placed in a support vessel. However, this task is often challenging, especially in conditions of poor visibility or in presence of strong currents. The system proposed allows the AUV to accomplish the task in autonomy. Moreover, the use of a three-dimensional (3-D) model of the environment and of an extended Kalman filter (EKF) allows the guidance and the control of the vehicle in real time. Experiments done on real underwater images have demonstrated the validity of the proposed method and its efficiency in the case of critical and complex situations.

A hierarchical approach to feature extraction and grouping.

In this paper, the problem of extracting and grouping image features from complex scenes is solved by a hierarchical approach based on two main processes: voting and clustering. Voting is performed for assigning a score to both global and local features. The score represents the evidential support provided by input data for the presence of a feature. Clustering aims at individuating a minimal set of significant local features by grouping together simpler correlated observations. It is based on a spatial relation between simple observations on a fixed level, i.e., the definition of a distance in an appropriate space. As the multilevel structure of the system implies that input data for an intermediate level are outputs of the lower level, voting can be seen as a functional representation of the "part-of" relation between features at different abstraction levels. The proposed approach has been tested on both synthetic and real images and compared with other existing feature grouping methods.

A distributed probabilistic system for adaptive regulation of image processing parameters.

A distributed optimization framework and its application to the regulation of the behavior of a network of interacting image processing algorithms are presented. The algorithm parameters used to regulate information extraction are explicitly represented as state variables associated with all network nodes. Nodes are also provided with message-passing procedures to represent dependences between parameter settings at adjacent levels. The regulation problem is defined as a joint-probability maximization of a conditional probabilistic measure evaluated over the space of possible configurations of the whole set of state variables (i.e., parameters). The global optimization problem is partitioned and solved in a distributed way, by considering local probabilistic measures for selecting and estimating the parameters related to specific algorithms used within the network. The problem representation allows a spatially varying tuning of parameters, depending on the different informative contents of the subareas of an image. An application of the proposed approach to an image processing problem is described. The processing chain chosen as an example consists of four modules. The first three algorithms correspond to network nodes. The topmost node is devoted to integrating information derived from applying different parameter settings to the algorithms of the chain. The nodes associated with data-transformation processes to be regulated are represented by an optical sensor and two filtering units (for edge-preserving and edge-extracting filterings), and a straight-segment detection module is used as an integration site.