Face-space has become established as an effective model for representing the dimensions of variation that occur in collections of human faces. For example, a change of expression from neutral to smiling can be represented by one axis in a face space. Principal components can be used to determine the axes of a face-space, however, standard principal components are based entirely on the data set from which they are computed, and do not express any domain specific information about the application of interest. In this paper, we propose a face-space analysis that combines the variance criterion used in principal components with some prior knowledge about the task-driven experiment. The priors are based on measuring eye movements of participants to frontal 2D faces during separate gender and facial expression categorization tasks. Our findings show that saccades to faces are task-driven, especially from 500 to 1000 milliseconds, and automatic recognition performance does not improve with additional exposure time.

Modelling based on probabilistic inference can be used to estimate the quality of information delivered by a military sensor network. Different modelling tools have complementary characteristics that can be leveraged to create an accurate model open to intuitive and efficient querying. In particular, stochastic process models can be used to abstract away from the physical reality by describing it as components that exist in discrete states with probabilistically invoked actions that change the state. The quality of information may be assessed by using the model to compute the probability that reports made by the network to its users are correct. In contrast, dynamic Bayesian network models, which have been used in a variety of military applications, are a more suitable vehicle for understanding the overall network performance and making inferences about the quality of information. Queries can be made simply by instantiating some variables and computing the probability distributions over others. We show that it is possible to combine both modelling tools by constructing a Bayesian network over the state variables of the process algebra model. The sparsity of the resulting Bayesian network allows fast propagation of probabilities, and hence interactive querying for the quality of information.

In many biometric pattern-recognition problems, the number of training examples per class is limited, and consequently the sample group covariance matrices often used in parametric and nonparametric Bayesian classifiers are poorly estimated or singular. Thus, a considerable amount of effort has been devoted to the design of other covariance estimators, for use in limited-sample and high-dimensional classification problems. In this paper, a new covariance estimate, called the maximum entropy covariance selection (MECS) method, is proposed. It is based on combining covariance matrices under the principle of maximum uncertainty. In order to evaluate the MECS effectiveness in biometric problems, experiments on face, facial expression, and fingerprint classification were carried out and compared with popular covariance estimates, including the regularized discriminant analysis and leave-one-out covariance for the parametric classifier, and the Van Ness and Toeplitz covariance estimates for the nonparametric classifier. The results show that, in image recognition applications whenever the sample group covariance matrices are poorly estimated or ill posed, the MECS method is faster and usually more accurate than the aforementioned approaches in both parametric and nonparametric Bayesian classifiers.

We propose a visual clue for navigation, in which the local 3D information or (p, q) is summarised in a single structure, called the pq-histogram. The pq-histogram is obtained by discretising in logarithmic scale the local depth (p, q). Since this histogram gives a global view of the scene, it could be used as a fast clue for navigation. An arrangement of four pq-histograms, one for each image quadrant, can advise many orientations. The potential application of the pq-histogram is illustrated in two domains. Firstly, semi-automatic navigation of an endoscope inside the human colon. Secondly, mobile robot navigating in certain environments, such as corridors and mines. In both cases, the method was tested with real images with encouraging results.

In the construction of a Bayesian network from observed data, the fundamental assumption that the variables starting from the same parent are conditionally independent can be met by introduction of hidden node (C.K. Kwoh and D.F. Gillies, 1994). We show that the conditional probability matrices for the hidden node for a triplet, linking three observed nodes, can be determined by the gradient descent method. As in all operational research problems, the quality of the result depends on the ability to locate a feasible solution for the conditional probabilities. C.K. Kwoh and D.F. Gillies (1995) presented a paper in which they detailed the methodologies for estimating the initial values of unobservable variables in Bayesian networks. We present the concept of determining the best conditional matrices as an estimation problem. The discrepancies between the observed and predicted values are mapped into a monotonic function where its gradients are used for adjusting the parameters to be estimated. We present our investigation of choosing among various popular error cost functions for training the networks with hidden nodes and determined that both cross entropy and sum of squared error cost functions work equally well for our implementation.

Abstract A simulator has been built to teach doctors the skill of handling a flexible endoscope for gastrointestinal investigations. Trainees use a dummy endoscope in which the control actions are transduced into voltages and sensed by the computer. The simulator computes the position and viewing direction of the endoscope within an internal model of the human colon. Then a renderer draws the view, reproducing as far as possible what would be seen during a real colonoscopy. Since the system must generate at least ten frames per second for realistic animation, standard rendering techniques, such as ray tracing, could not be used. Consequently a new method was devised, based on identifying coherent regions along each scan line which can be rendered by table lookup. The method allows shaded Lambertian surfaces to be drawn at a frame rate of 15 per second, using modest computing resources. Although several approximations were required in the analysis, the computer images of the internal surfaces of the human colon present a high degree of visual realism.

A simulator has been designed and implemented for teaching the skills required for fibrendoscopic investigation of the colon. The student using the simulator manipulates a dummy endoscope control head in which the rotary and linear movements are converted to voltages and interfaced to the computer via analogue to digital converters. The simulator is based on a three dimensional computer model of the colon and endoscope. The program reads the endoscope controls, determines the position of the tip of the endoscope, and using this as viewpoint draws the appropriate view. In the worst case, the whole cycle takes 0.185 seconds on an ordinary IBM PC, giving a frame rate of just under 6/sec.

A theory of abstract programs is introduced which permits the synthesis of all control structures of a given size and facilitates the analysis and restructuring of programs. The paper illustrates the use of the proposed methods by restructuring a wellknown algorithm (Quicksort).