Statistical Edge Detection of Concealed Weapons Using Artificial Neural Networks. Ian Williams a , David Svoboda b , Nicholas Bowring a , and Elizabeth Guest c . a Department of Engineering and Technology, Manchester Metropolitan University, Manchester, UK; b Centre for Biomedical Image Analysis, Faculty of Informatics, Masaryk University, Brno, CZ; c School of Computing, Leeds Metropolitan University, Leeds, UK ABSTRACT A novel edge detector has been developed that utilises statistical masks and neural networks for the optimal detection of edges over a wide range of image types. The failure of many common edge detection techniques has been observed when analysing concealed weapons X-ray images, biomedical images or images with significant levels of noise, clutter or texture. This novel technique is based on a statistical edge detection filter that uses a range of two-sample statistical tests to evaluate any local image texture differences and by applying a pixel region mask (or kernel) to the image analyse the statistical properties of that region. The range and type of tests has been greatly expanded from the previous work of Bowring et al. 1 This process is further enhanced by applying combined multiple scale pixel masks and multiple statistical tests, to Artificial Neural Networks (ANN) trained to classify different edge types. Through the use of Artificial Neural Networks (ANN) we can combine the output results of several statistical mask scales into one detector. Furthermore we can allow the combination of several two sample statistical tests of varying properties (for example; mean based, variance based and distribution based). This combination of both scales and tests allows the optimal response from a variety of statistical masks. From this we can produce the optimum edge detection output for a wide variety of images, and the results of this are presented. Keywords: Edge Detection, Segmentation, Neural Networks, Statistics, Cluttered Images, X-ray Baggage screening. 1. INTRODUCTION Most traditional edge detectors work on the assumption that an edge in a digital image is a discrete change in the intensity profile of the neighbouring pixels. 2 Where this is assumed the derivative of the pixel grey levels can be computed to accurately determine the edge location, and many of the earliest derivative based edge filters were based in this principle. Like the Roberts, 3 and Sobel 4 edge detectors. Gradient based edge detectors perform effectively on many synthetic or real images where the pixel intensity change is clear, or on images with very little noise. If the edge intensity difference becomes corrupted with noise or is eliminated completely by image texture or excess clutter as in figure 1, these derivative based edge detectors perform more poorly as discussed by Lim. 5 In these situations an image pre-processing noise suppression stage such as Gaussian and median filtering as detailed in Gonzalez, 6 or nonlinear filtering techniques described by Perona, 7 can be applied to reduce the noise effects. Underpinned by a pre-processing smoothing stage, Canny 8 introduced an analytically optimal step edge detector based on the first derivative of a Gaussian filter. Since it’s development, Canny’s edge detector has been seen as the benchmark and consequently sets the standard for all newly developed detectors, as discussed by Lim. 9 However, using Gaussian smoothing prior to any edge detection is not always ideal and any threshold values used in post processing steps may not always be accurate. For further information please send any correspondence to Ian Williams or Nicholas Bowring. Ian Williams.: E-mail: i.williams@mmu.ac.uk, Telephone: +44 (0)161 244 1666 Nicholas Bowring.: E-mail: n.bowring@mmu.ac.uk, Telephone: +44 (0)161 244 2246 Image Processing: Algorithms and Systems VI, edited by Jaakko T. Astola, Karen O. Egiazarian, Edward R. Dougherty, Proc. of SPIE-IS&T Electronic Imaging, SPIE Vol. 6812, 68121J, © 2008 SPIE-IS&T · 0277-786X/08/$18 SPIE-IS&T Vol. 6812 68121J-1 2008 SPIE Digital Library -- Subscriber Archive Copy