We present a dynamic causal model that can explain context-dependent changes in neural responses, in the rat barrel cortex, to an electrical whisker stimulation at different frequencies. Neural responses were measured in terms of local field potentials. These were converted into current source density (CSD) data, and the time series of the CSD sink was extracted to provide a time series response train. The model structure consists of three layers (approximating the responses from the brain stem to the thalamus and then the barrel cortex), and the latter two layers contain nonlinearly coupled modules of linear second-order dynamic systems. The interaction of these modules forms a nonlinear regulatory system that determines the temporal structure of the neural response amplitude for the thalamic and cortical layers. The model is based on the measured population dynamics of neurons rather than the dynamics of a single neuron and was evaluated against CSD data from experiments with varying stimulation frequency (1-40 Hz), random pulse trains, and awake and anesthetized animals. The model parameters obtained by optimization for different physiological conditions (anesthetized or awake) were significantly different. Following Friston, Mechelli, Turner, and Price (2000), this work is part of a formal mathematical system currently being developed (Zheng et al., 2005) that links stimulation to the blood oxygen level dependent (BOLD) functional magnetic resonance imaging (fMRI) signal through neural activity and hemodynamic variables. The importance of the model described here is that it can be used to invert the hemodynamic measurements of changes in blood flow to estimate the underlying neural activity.

We describe a Bayesian architecture to estimate the position and pose of a 3D object. The system starts with knowledge of the 3D structure of the object and the prior probability distribution of its position and orientation in the workspace. This information is used to guide the search for focus features in the image, and the information recovered from the image processing is used to refine the estimates of the x-y position and pose of the object. The results of intermediate stages of processing is propagated using a Bayesian methodology. After iteration around the network, the peaks of the final probability distributions are used to estimate the position and pose, and the widths of the distributions provide a measure of confidence. The results of the study suggest that grey-level image processing algorithms and a simple 3D model, embedded in a Bayesian statistical reasoning architecture can provide a highly effective, albeit specialised object localisation system. keywo...

We describe an edge-based stereo correspondence algorithm in the model-driven vision system. A constraint derived from the 3D model is used to prune false alarms and speed up the matching process. This constraint is based on computional considerations and experimental and psychological observations concerning vertical disparities. An edge constraint is also presented. A numbering scheme is used to facilitate the implementation of the correspondence algorithm using dynamic programming. Experiments on natural images show that the correspondence of an edge can usually be achieved in a few seconds. The computed disparities are further incorporated into object model to refine the estimates of object pose. Keywords: Correspondence, Constraints, Cost function, Model-driven vision, Dynamic programming 1 Introduction The work reported here is a part of an ongoing research project "model-driven stereo vision under variable camera geometry", as shown in Figure 1. Three major stages c...

yuanOaivru.Sheffield.ac.uk We use a stereo disparity predictor, implemented as layered neural nets in the PILUT architecture, to encode the disparity flow field for the ground plane at various viewing positions over the work space. A deviation of disparity, computed using a correspondence algorithm, from its prediction may then indicate a potential obstacle. A casual bayes net model is used to estimate the probability that a point of interest lies on the ground plane. 1

This paper documents the details of a development environment called ANIT which provides tools to enable researchers with no knowledge of parallel systems to contribute to and operate whole systems. These tools extend from a simple compilation environment to X based control and display tools for all the modules of the systems. ANIT meets the need to bind Tma[10] (AIVRU's software vision system) and future projects together in a way which facilitates the inheritance of existing working code in an evolving system. The target architecture for ANIT is a mixed transputer Unix Host network incorporating specialised frame-rate vision hardware. To achieve this requirement ANIT has been given several goals, including: (i) Definition and support for a structure for each processing module so as to simplify system modeling; (ii) Definition and provision of a simple but powerful method by which modules communicate, thereby minimizing the knowledge and code needed to produce a communicating module and hence protecting the user from the typical time-consuming overhead in development time associated with solving inter-module communication problems; and (iii) Isolation of the user/programmer from the hardware to prevent dependencies on ever changing architectures, and specifically removing from the C programmer any need to know how to program the transputers even though they comprise the hardware on which his system is executed, the latter facility is supported by IIPC (AIVRU's in-house run-time environment for Transputer-Sun networks) [3] which is wholly integrated with ANIT. We report demonstrations of the value of ANIT/IIPC in several projects involving complex control systems for autonomous vehicles with visual guidance from a 4 degrees of freedom stereo 'eye/head' rig. The value of ANIT for other applications, such as industrial inspection and assembly and intelligent surveilance are discussed.

A method of vehicle control is described. Visual feedback from a micro-saccadic tracker is used to provide a docking competence. The vehicle is thereby capable of arriving at goal position with a specified head-ing angle. The vehicle uses a beacon to estimate it's trajectory and position in the ground plane. This information is used to control the vehicle to move to a "via-point". Complex trajectories can be achieved composed of multiple via-points. This is achieved without the need to plan the path between the via-points. The algorithm is therefore com-putationally cheap. 1

We introduce a novel form of shape representation based on recording the distribution of pairwise geometric relationships between local shape features. It is shown that the geometric histograms used to record these distributions can be easily and robustly acquired from image data and can support recognition even when the shape extracted from the image is badly degraded by fragmentation noise and occlusion. Moreover, the processing involved in establishing correspondences between model and image features is both simple and parallel and has many advantages over previous search based methods. 1

Representation of arbitrary shape for the purposes of visual object recognition is an unsolved problem. This paper outlines some theoretical properties of shape representations based on stored geometric constraints that make them suitable for input to a 3D object recognition system. The construction of such representations is detailed together with an evaluation of their efficacy based on experimental findings.>

In this paper a planar curve segmentation method based on the analysis of curvature and mean polygonal area is proposed. A rough segmentation of the curve is first produced by a high/low curvature classifier and then refined by a high/low/zero classifier and a knot/straight/convex/concave shape identifier so as to obtain a set of critical knots and shape descriptors for all segments. A straight line or conic approximation is then applied to each segment to generate a symbolic representation of the curve.

Abstract Many stereo correspondence algorithms require relative camera geometry, as the epipolar constraint is fundamental to their matching processes. We intend to build a eye/head camera rig to mount on the mobile platform COMODE to enhance the abilities of the TINA system to recover 3D geometry from its environment. Thus we need to be able to associate camera geometry with particular head configurations. Generic calibration of such a system would require the ability to compute camera geometry from arbitrary stereo images. A system which solves this problem using an established corner detector combined with a robust stereo matching algorithm and a variational solution for the camera geometry is described.