Download the this issue (complete journal, excluding cover) (~16 MB).
Download the
cover (Editorial Board Information) (~300 kB).
Table of Contents (~180 kB)
Jump to:
-
Subdivision of Curves and Surfaces: An Overview
-
Unsupervised Classification of Dynamic Froths
-
Fuzzy Artmap and Neural Network Approach to Online Processing of Inputs with Missing Values
-
On Visual Object Tracking Using Active Appearance Models
-
Identify Confidence Estimation of Manoeuvring Aircraft
-
A Note on Difference Spectra for Fast Extraction of Global Image Information
Subdivision of Curves and Surfaces: An Overview by B. Herbst, K.M. Hunter and E. Rossouw
Abstract: Subdivision schemes are widely used in various applications such as data-fitting, computer graphics and solid modelling. In this paper we present the basic ideas of subdivision schemes for curves; both interpolatory and corner-cutting schemes, as well as their adaptation to finite sequences. We conclude with examples of specific applications for these subdivision schemes and provide an example of surface subdivision.
Download Paper (~3 Mb)
Unsupervised Classification of Dynamic Froths by C. Forbes and C. de Jager
Abstract: Machine vision systems typically classify images of a floating froth surface into one of a distinct set of classes. This process typically involves having an experienced operator identify a set of froth classes. After this, a machine vision system is trained to identify these froth classes. Identifying these froth classes is particularly challenging for froths which have “dynamic” bubble size distributions. Using unsupervised clustering algorithms, it is possible to automatically learn these froth classes without user input. Validation of this technique is done by showing that the identified froth classes have statistically different relationships between the froth velocity and concrete grade.
Download Paper (~3 Mb)
Fuzzy Artmap and Neural Network Approach to Online Processing of Inputs with Missing Values by F.V. Nelwamondo and T. Marwala
Abstract: An ensemble based approach for dealing with missing data, without predicting or inputing the missing values is proposed. This technique is suitable for online operations of neural networks and as a result, is used for online condition monitoring. The proposed technique is tested in both classification and regression problems. An ensemble of Fuzzy-ARTMAPs is used for classification whereas an ensemble of multi-layer perceptrons is used for the regression problem. Results obtained using this ensemble-based technique are compared to those obtained using a combination of auto-associative neural networks and genetic algorithms and findings show that this method can perform up to9 % better in regression problems. Another advantage of the proposed technique is that it eliminates the need for finding the best estimate of the data, and hence, saves time.
Download Paper (~3.8 Mb)
On Visual Object Tracking Using Active Appearance Models by M.R. Hoffmann, B.M. Herbst and K.M. Hunter
Abstract: Active appearance models provide an elegant framework for tracking objects. Using them in a deterministic algorithm to perform tracking is not robust enough since no history is used of the object’s movement and position. We discuss two approaches to rectify this situation. Both techniques are based on the particular filter. The first technique initialises the active appearance model search algorithm with a shape estimate obtained from an active contour tracker. A combination of a particle filter and an active appearance model forms the foundation for the second technique. Experimental results indicate the effectiveness of these techniques.
Download Paper (~3.8 Mb)
Identify Confidence Estimation of Manoeuvring Aircraft by P.J. Hotzhausen and B.M. Herbst
Abstract: A radar system observes an aircraft once during each scan of the airspace, and uses these observations to construct a track representing a possible route of the aircraft. However when aircraft interact closely there is the possibility of confusing the identities of the tracks. In this study multiple hypothesis techniques are applied to extract an identify confidence from a track, given a set of possible tracks and observations. The system utilises numerous estimation filters internally and these are investigated and compared in detail. The Identity Confidence algorithm is tested using a developed radar simulation system, and evaluated successfully against a series of benchmark tests.
Download Paper (~2.3 Mb)
A Note on Difference Spectra for Fast Extraction of Global Image Information by B.J. van Wyk, M.A. van Wyk and F. van den Bergh
Abstract: The concept of an Image difference spectrum, a novel tool for the extraction of global image information, is introduced. It is shown that Image Difference Spectra are fast alternatives to granulometric curves, also referred to as pattern spectra. Image Difference Spectra are computationally easy to implement and are suitable for real-time applications.
Download Paper (~1 Mb)