No description available.

No description available.

Random-access memory (RAM) testing to detect unrestricted pattern-sensitive faults (PSFs) is impractical due to the size of the memory checking sequence required. A formal model for restricted PSFs in RAMs called adjacent-pattern interference faults (APIFs) is presented. A test algorithm capable of detecting APIFs in RAMs requiring a minimum number of memory operations is then developed.

Feature selection is a critical part of any pattern classification problem. There are many methods for selecting a good set of features. However, for problems where features must be selected from a massive set, most of these methods have accuracy rates that are very low, or computational complexities that are very high. While for some pattern classification problems it might be reasonable to reduce a massive set of features by using application specific information, in problems such as dynamic signature verification this is not possible. Several existing feature selectors are evaluated including the Karhunen-Loeve, SELECT, exhaustive, accelerated, "n best features", sequential forward search, sequential backward search, and the "plus q - take away r" feature selection methods. Each of these methods has particular problems, making them poor candidates for selection of features from a massive set. A General-Purpose Reduction-Intensive (GPRI) feature selector is proposed in this thesis. The GPRI feature selector reduces a large set of features to a small final feature set. The time complexity of the GPRI method is close to the "n best features" method; however, the accuracy rates (obtained with the features selected) far exceeds the "n best features" feature selector. Thus, the GPRI feature selector is a viable candidate for selecting features in general environments where little application specific information is available. / M.S.

No description available.

The problem of interest as indicated is to develop a general purpose technique that is a combination of the structural approach, and an extension of the Finite Inductive Sequence (FI) technique. FI technology is pre-algebra, and deals with patterns for which an alphabet can be formulated.

Event understanding systems, responsible for automatically generating human relatable event descriptions from video sequences, is an open problem in computer vision research that has many applications in the sports domain, such as indexing and retrieval systems for sports video. Background modelling and shot classification of broadcast video are important steps in event understanding in video sequences. Shot classification seeks to identify shots, i.e. the labelling of continuous frame sequences captured by a single camera action such as long shot, close-up and audience shot, while background modelling seeks to classify pixels in an image as foreground/background. Many features used for shot classification are built upon the background model therefore background modelling is an essential part of shot classification. This dissertation reports on an investigation into techniques and procedures for background modelling and classification of shots in broadcast soccer videos. Broadcast video refers to video which would typically be viewed by a person at home on their television set and imposes constraints that are often not considered in many approaches to event detection. In this work we analyse the performances of two background modelling techniques appropriate for broadcast video, the colour distance model and Gaussian mixture model. The performance of the background models depends on correctly set parameters. Some techniques offer better updating schemes and thus adapt better to the changing conditions of a game, some are shown to be more robust to changes in broadcast technique and are therefore of greater value in shot classification. Our results show the colour distance model slightly outperformed the Gaussian mixture model with both techniques performing similar to those found in literature. Many features useful for shot classification are proposed in the literature. This dissertation identifies these features and presents a detailed analysis and comparison of various features appropriate for shot classification in broadcast soccer video. Once a feature set is established, a classifier is required to determine a shot class based on the extracted features. We establish the best use of the feature set and decision tree parameters that result in the best performance and then use a combined feature set to train a neural network to classify shots. The combined feature set in conjunction with the neural network classifier proved effective in classifying shots and in some situations outperformed those techniques found in literature. / Thesis (M.Sc.)-University of KwaZulu-Natal, Durban, 2012.

The first area of work is to assess image quality by measuring the similarity between edge map of a distorted image and that of its original version using classical edge quality evaluation metrics.  Experiments show that comparing edge maps of original and distorted images gives a better result than comparing the images themselves.  Based on redefined source and distortion models, a novel FR image quality assessment metric DQM is proposed, which is proved by subsequent experiments to be competitive with state-of-the-art metrics (SSIM, IFC, VIF, etc.).  The thesis also proposes several image quality metrics based on a framework for developing image quality assessment algorithms with the help of data-driven models (multiple linear regression, artificial neural network and support vector machine). Among them, CAM_BPNN and CAM_SVM perform better than SSIM and can even compete with its improved multi-scale version MSSIM.  Apart from the research about FR image quality assessment, a novel RR image quality assessment system is proposed, based on low-level features (corner, edge and symmetry).

Object localization is the task of locating objects in an image, typically by finding bounding boxes that isolate those objects. Identifying objects in images that have not had regions of interest labeled by humans often requires object localization to be performed first. The sliding window method is a common naïve approach, wherein the image is covered with bounding boxes of different sizes that form windows in the image. An object classifier is then run on each of these windows to determine if each given window contains a given object. However, because object classification algorithms tend to be computationally expensive, it is helpful to have an effective filter to reduce the number of times those classifiers have to be run. In this thesis I evaluate one promising approach to object localization: the objectness algorithm proposed by Alexe et al. Specifically, I verify the results given by Alexe et al., and further explore the weaknesses and strengths of their "objectness"

No description available.