Maneesha Singh

Explosives detection systems (EDS) for aviation security

The detection of explosives and illicit material for the purposes of aviation security is an important area for preventing terrorism and smuggling. A number of different methods of explosive detection have been developed in the past that can detect such material from a very small up to a very large quantity. For the purposes of aviation security, the checks are performed on passengers, their carry on luggage, checked baggage, and cargo containers. Similar technology is used in post-offices for detecting dangerous substances in mail. In this paper we review some of these technologies and in particular discuss the application of computers for the analysis of data and images generated from security equipment.

Spatial texture analysis: a comparative study

In this paper we compare some of the traditional, and some fairly new, techniques of texture analysis on the MeasTex and VisTex benchmarks to illustrate their relative abilities. The methods considered include autocorrelation (ACF), co-occurrence matrices (CM), edge frequency (EF), Laws masks (LM), run length (RL), binary stack method (BSM), texture operators (TO), and texture spectrum (TS). In addition, we illustrate the advantage of using feature selection on a combined set that improves the overall recognition performance.

A dynamic classifier selection and combination approach to image region labelling

Abstract In this paper we propose a ‘bank of classifiers’ approach to image region labelling and evaluate dynamic classifier selection and classifier combination approaches against a baseline approach that works with a single best classifier chosen using a validation set. In this analysis, image segmentation, feature extraction, and classification are treated as three separate steps of analysis. The classifiers used are each trained with a different texture feature representation of training images. The paper proposes a new knowledge-based predictive approach based on estimating the Mahalanobis distance between test sample feature values and the corresponding probability distribution function from training data that selectively triggers classifiers. This approach is shown to perform better than probability-based classifier combination (all classifiers are triggered but their decisions are fused with combination rules), and single classifier, respectively, based on classification rates and confusion matrices. The experiments are performed on the natural scene analysis application.

A knowledge-based framework for image enhancement in aviation security

The main aim of this paper is to present a knowledge-based framework for automatically selecting the best image enhancement algorithm from several available on a per image basis in the context of X-ray images of airport luggage. The approach detailed involves a system that learns to map image features that represent its viewability to one or more chosen enhancement algorithms. Viewability measures have been developed to provide an automatic check on the quality of the enhanced image, i.e., is it really enhanced? The choice is based on ground-truth information generated by human X-ray screening experts. Such a system, for a new image, predicts the best-suited enhancement algorithm. Our research details the various characteristics of the knowledge-based system and shows extensive results on real images.

Colour image texture analysis: dependence on colour spaces

In this paper we investigate the role of colour spaces on texture analysis. We extract a range of correlogram and colour moment features for the VisTex colour texture benchmark in different colour spaces and find the average probabilistic distance of separation across different objects for different features and suggest the colour spaces that are best suited for the classification process. We also show the results of k-nearest neighbour classification for different features and their combined set.

Feature selection for face recognition based on data partitioning

Feature selection is an important consideration in several applications where one needs to choose a smaller subset of features from a complete set of raw measurements such that the improved subset generates as good or better classification performance compared to original data. In this paper, we describe a novel feature selection approach that is based on the estimation of classification complexity through data partitioning. This approach allows us to select the N best features from a given set in an order of their ability to separate data from different classes. In this paper, we perform our experiments on the ORL face database that consists of 400 images. The results show that the proposed approach outperforms the probability distance approach and is a viable method for implementing more advanced search methods of feature selection.

Image segmentation optimisation for X-ray images of airline luggage

Airline luggage contains a wide variety of objects and their automated image analysis require good quality image segmentation. Given the fact that such images are highly cluttered, it is non trivial task to optimise image segmentation algorithms. In this paper we present a methodology for optimising image segmentation algorithms based on image properties without manual intervention. The methodology computes image properties such as average edge gradient strength, inter- vs. intra-cluster distances using image colour features, and colour purity of resultant regions, to train a neural network that maps these to ground-truth labelling on the acceptability (good or bad) of the solution (resultant segmentation). We show that on unseen test data, this methodology performs extremely well by correctly predicting the optimal parameters of image segmentation algorithms used.

IMAGE RETRIEVAL USING SPATIAL CONTEXT

Abstract This paper presents an image retrieval system based on modelling the spatial relationship between image contents. We model the relationship between image objects to help image retrieval. This approach first detects objects and determines their label in an image and then codes their Spatial relationship using binary patterns. Image retrieval is based on a matching score that computes similarity based on these binary patterns. The paper shows good results on a publicly available scene analysis benchmark.

A Comparison of Texture Teatures for the Classification of Rock Images

Texture analysis plays a vital role in the area of image understanding research. One of the key areas of research is to compare how well these algorithms rate in differentiating between different textures. Traditionally, texture algorithms have been applied mostly on benchmark data and some studies have found certain algorithms are better suited for differentiating between certain types of textures. In this paper we compare 7 well-established image texture analysis algorithms on the task of classifying rocks.

Progress in Pattern Recognition

The field of pattern recognition has emerged as one of the most challenging and important endeavours in the area of information technology research. Research in the area of pattern recognition has benefits for improving many areas of human endeavour, including medicine, the economy, the environment, and security. This book presents some of the latest advances in the area of pattern recognition theory and applications. The first half of the book discusses novel pattern classification and matching schemes, and the second half describes the application of novel tools in biometrics and digital multimedia. The applications included, such as face/iris recognition, handwriting recognition, surveillance, human dynamics, sensor fusion, etc., provide a detailed insight into how to build real pattern recognition systems and how to evaluate them. Given the dynamic nature of technology evolution in this area, this book provides the latest algorithms and concepts that can be used to build real systems. Features and topics: Provides state-of-the art algorithms, as well as presents cutting-edge applications within the field Introduces achievements in theoretical pattern recognition, including statistical and Bayesian pattern recognition, structural pattern recognition, neural networks, classification and data mining, evolutionary approaches to optimisation, and knowledge based systems Offers insights and support to practitioners concerned with the state-of-the art technology in the area Progress in Pattern Recognition addresses the needs of postgraduate students, researchers, and practitioners in the areas of computer science, engineering and mathematics where pattern recognition techniques are widely used. Professor Sameer Singh is Director of the Research School of Informatics, Loughborough University, UK, and serves as Editor-in-Chief of the Springer journal, Pattern Analysis and Applications