Stuart Perry has over 20 years of experience conducting research into image processing, psychophysics, signal processing, image quality, and models for the quantification of image preference and aesthetics for both government, industry and academia. Following receiving a PhD from the University of Sydney in 1999, he began his research career studying the application of image processing to the detection of underwater objects in sector-scan sonar imagery for the Maritime Operations Division, Defence Science and Technology Organisation (DSTO). During this time he represented Australia on Computer-Aided Detection and Classification Specialist Group, Technical Panel 13 (Mine Warfare and High Frequency Acoustics), Maritime Group, The Technical Cooperation Program (TTCP).
From 2003 to 2016, he worked for Canon Information Systems Research Australia (CiSRA), a Canon group company and one of the Canon Group's largest R&D facilities outside of Japan. During this time he worked on camera white balancing technologies, and led research teams working on print quality measurement, document security and perceptual quality measurement for various consumer devices.
In 2016 he joined the FEIT's Perceptual Imaging Laboratory (PILab) conducting research into colour and perceptual quality in 3D environments.
- Member of Institute of Electrical and Electronic Engineers (IEEE) from 1996
- Member of Society for Photo-Optical Instrumentation Engineers (SPIE) from 1998
- Member of Society for Imaging Science and Technology (IS&T)
Membership on Scholarly Committees:
- Australian Point of Contact for the Computer-Aided Detection and Classification Specialist Group, Technical Panel 13 (Mine Warfare and High Frequency Acoustics), Maritime Group, The Technical Cooperation Program (TTCP). (2001–2003)
- Local Arrangements and Registration Chair: 2000 IEEE Workshop on Neural Networks in Signal Processing, December 2000, University of Sydney, NSW, Australia.
- Session Chair and Local Arrangements and Registration Chair: First IEEE Pacific Rim Conference on Multimedia, December 2000, University of Sydney, NSW, Australia.
- Organising Committee Member and Session Chair: First Computer-Aided Detection and Computer-Aided Classification Conference (CADCAC 2001), 12-14th of November 2001, Halifax, Nova Scotia, Canada.
Technical Committee Member and Session Chair, 2004 IEEE International Conference on Image Processing (ICIP 2004), 24th-26th of September 2004, Singapore.
- Technical Commitee Member and Session Chair, 2013 IEEE International Conference on Image Processing (ICIP 2013), 15th–18th of September 2013, Melbourne, Australia.
- Program Committee Member, Image Quality and System Performance , IS&T/SPIE Electronic Imaging, 2013–current.
- Program Committee Member, Optics, Photonics and Digital Technologies for Imaging Applications, SPIE Optics + Photonics, 2014-current.
- Committee Member, Standards Australia Committee MS-65, Australian Mirror Committee for ISO/TC 42 Photography.
- Committee Member, Standards Australia Committee IT-029-01, Australian Mirror Committee for ISO/IED JTC 1/SC 29 - Coding of audio, picture, multimedia and hypermedia information.
- “Electrical Engineering Foundation Award for Excellence in Teaching (Tutoring) 1998”, awarded by the School of Electrical and Information Engineering, University of Sydney, Australia.
- “IS&T Service Award 2014”, awarded by the Society for Imaging Science and Technology (IS&T) for contributions to the Data Analytics Task Force in 2013.
Australian Patent Applications:
2015201623 Choosing optimal images with preference distributions, Perry, Stuart William
2014201797 Method, apparatus and system for determining chromatic difference, Perry, Stuart William; Bonnier, Nicolas Pierre Marie Frederic
2013276980 Chroma structure affecting chroma perception, Woolfe, Geoffrey John; Bonnier, Nicolas Pierre Marie Frederic; Pakulski, Peter Jan; Perry, Stuart William; Rich, Anina Nicole; Williams, Mark Alexander; Weldon, Kimberly
2013273630 Observer preference model. Perry, Stuart William
2009238260 Forgery detection using finger print, Perry, Stuart William; Gupta, Amit Kumar
2009203182 Document authentication using handheld device, Fields, Andrew James; DeQiang, Eugene Cai; Gibson, Ian Richard; Perry, Stuart William
2008264191 Detecting and marking incidental test patches in a printed document for the purpose of print quality analysis, Yenson, Brendon; Degros, Francois; Perry, Stuart William
2008260092 Document authentication and workflow, Drake, Barry James; Gibson, Ian Richard; Amielh, Myriam Elisa Lucie; Hardy, Stephen James; Perry, Stuart William
2008252022 Colour printing with achromatic substance, Perry, Stuart William
2007254658 Positional alignment accuracy of a printer, Larkin, Kieran Gerard; Duggan, Matthew Christian; Perry, Stuart William
2007254655 Authenticating partially transparent medium, Rudkin, Scott Alexander; Ecob, Stephen Edward; Perry, Stuart William
2007254624 A method for recommending quality analysis techniques for test targets in the process of designing test charts, Degros, Francois; Perry, Stuart William; Tot, Robert; Duggan, Matthew Christian
2006202198 Method of reviewing multiple images, Dorrell, Andrew James; Gibson, Richard Ian; Perry, Stuart William; Chan, Woei
2005242227 Camera System Implementing Flash-No-Flash Processing Mode, Dorrell, Andrew James; Gibson, Ian Richard; Perry, Stuart William; Chan, Woei
2005203381 White balance adjustment, Dorrell, Andrew James; Perry, Stuart William; Chan, Woei
2004906703 Selection of Images for White Balance Adjustment, Dorrell, Andrew James; Perry, Stuart William; Chan, Woei
2004906020 Post-capture fill flash, Dorrell, Andrew James; Perry, Stuart William; Chan, Woei
2004904409 White Balance Adjustment, Dorrell, Andrew James; Perry, Stuart William; Chan, Woei
US Granted Patents:
7,551,797, White Balance Adjustment, Andrew James Dorrell, Stuart William Perry, Woei Chan
US Patent Applications:
US2015/0169982, Observer Preference Model, Stuart William Perry
Can supervise: YES
Stuart is currently interested in many aspects of image processing and enabling technologies that allow machines to sense and understand their environment. This includes, adaptive image processing, adaptive image restoration, image restoration, noise removal, filtering, general image processing, as well as the detection and classification of objects using statistical and machine learning techniques.
Stuart is very interested in psychophysics and mathematical models that describe the human visual system and model perceptual image quality and other subjective human responses to imagery and objects such as material appearance, aesthetics and three dimensionality. Stuart believes that perceptual image quality remains an unsolved problem and effective models of perceptual image quality have the potential to improve a variety of image processing algorithms and open up new applications. In addition, new aspects of human perception have begun to be examined such as aesthetics and material appearance. These new aspects as well as the continuing need for effective image quality/preference measures represent exciting new directions in the image processing field. In the last few years, Stuart's attention has been shifting to the design and statistical analysis of psychophysical experiments to support research into the human perception of imagery. He has a keen interest in experimental design methodologies, machine learning, big data techniques and statistical analysis and regression and how these tools might be applied to problems in human perception, including the perception of aesthetics, three-dimensionality, material properties and quality.
31256 Image Processing and Pattern Recognition
31261 Internetworking Project
Yap, KH, Guan, L, Perry, SW & Wong, HS 2009, Adaptive image processing: A computational intelligence perspective, second edition.
© 2018 by Taylor & Francis Group, LLC. Illustrating essential aspects of adaptive image processing from a computational intelligence viewpoint, the second edition of Adaptive Image Processing: A Computational Intelligence Perspective provides an authoritative and detailed account of computational intelligence (CI) methods and algorithms for adaptive image processing in regularization, edge detection, and early vision. With three new chapters and updated information throughout, the new edition of this popular reference includes substantial new material that focuses on applications of advanced CI techniques in image processing applications. It introduces new concepts and frameworks that demonstrate how neural networks, support vector machines, fuzzy logic, and evolutionary algorithms can be used to address new challenges in image processing, including low-level image processing, visual content analysis, feature extraction, and pattern recognition. Emphasizing developments in state-of-the-art CI techniques, such as content-based image retrieval, this book continues to provide educators, students, researchers, engineers, and technical managers in visual information processing with the up-to-date understanding required to address contemporary challenges in image content processing and analysis.
Yap, K-H, Guan, L, Perry, SW & Wong, HS 2009, Adaptive Image Processing A Computational Intelligence Perspective, Second Edition, CRC Press.
Emphasizing developments in state-of-the-art CI techniques, such as content-based image retrieval, this book continues to provide educators, students, researchers, engineers, and technical managers in visual information processing with the ...
Feng, X, Wan, W, Xu, RYD, Perry, S, Li, P & Zhu, S 2018, 'A novel spatial pooling method for 3D mesh quality assessment based on percentile weighting strategy', Computers & Graphics, vol. 74, pp. 12-22.View/Download from: UTS OPUS or Publisher's site
Feng, X, Wan, W, Xu, RYD, Perry, S, Zhu, S & Liu, Z 2018, 'A new mesh visual quality metric using saliency weighting-based pooling strategy', Graphical Models, vol. 99, pp. 1-12.View/Download from: UTS OPUS or Publisher's site
© 2018 Elsevier Inc. Several metrics have been proposed to assess the visual quality of 3D triangular meshes during the last decade. In this paper, we propose a mesh visual quality metric by integrating mesh saliency into mesh visual quality assessment. We use the Tensor-based Perceptual Distance Measure metric to estimate the local distortions for the mesh, and pool local distortions into a quality score using a saliency weighting-based pooling strategy. Three well-known mesh saliency detection methods are used to demonstrate the superiority and effectiveness of our metric. Experimental results show that our metric with any of three saliency maps performs better than state-of-the-art metrics on the LIRIS/EPFL general-purpose database. We generate a synthetic saliency map by assembling salient regions from individual saliency maps. Experimental results reveal that the synthetic saliency map achieves better performance than individual saliency maps, and the performance gain is closely correlated with the similarity between the individual saliency maps.
Pham, TQ, Perry, SW, Fletcher, PA & Ashman, RA 2011, 'Paper fingerprinting using alpha-masked image matching', IET Computer Vision, vol. 5, no. 4, pp. 232-243.View/Download from: UTS OPUS or Publisher's site
In this study, the authors examine the process of authenticating paper media using the unique fibre structure of a piece of paper (the so-called 'paper fingerprint') In particular, the authors look at methods to authenticate a paper fingerprint when text has been printed over the authentication zone. The authors show how alpha-masked correlation can be applied to this problem and develop a modification to this technique that is more closely matched to the requirements of this problem and produces an improvement in performance. They also investigate two methods of pixel inpainting to remove printed text or marks from the authentication zone and allow ordinary correlation to be performed. The authors show that these methods can perform as well as alpha-masked correlation. They derive confidence estimates for one of the matching processes. Finally, the authors investigate some methods for improving the robustness to forgery and rotation of the matching process. © 2011 The Institution of Engineering and Technology.
Perry, SW, Guan, L & Varjavandi, P 2006, 'Incorporating local statistics in image error measurement for adaptive image restoration', Optical Engineering, vol. 45, no. 3.View/Download from: Publisher's site
This paper presents an image restoration technique incorporating local statistical knowledge in the cost function. Instead of using a conventional grayscale-based error measurement such as the mean squared error, we compare local statistical information about regions in two images using a new error measure. Transient features such as edges and textures are more strongly emphasized than relatively homogeneous regions. With the addition of this local information, we attempt to provide a measure closer to human visual appraisal. We then extend the popular constrained squared-error cost function by incorporating this image error measure. Due to its nonlinear nature, conventional restoration algorithms cannot optimize this cost function efficiently. Therefore we seek an iterative approach. In particular, an extended neural network algorithm is proposed to perform the restoration. It is shown that this technique is efficient, effective, and robust. It compares favorably with other techniques when applied to both grayscale and color images. The results of a subjective survey comparing the proposed algorithm with a more conventional neural network algorithm are presented. The subjects tested in the survey overwhelmingly favored the results provided by the proposed method. © 2006 Society of Photo-Optical Instrumentation Engineers.
Perry, SW & Guan, L 2004, 'A recurrent neural network for detecting objects in sequences of sector-scan sonar images', IEEE Journal of Oceanic Engineering, vol. 29, no. 3, pp. 857-871.View/Download from: Publisher's site
This paper presents a system for detecting small man-made objects in sequences of sector-scan images formed using a medium-range sector-scan sonar. The detection of such objects is considered out to ranges of 200 m from the vessel and while the vessel is in motion. This paper extends previous work by making use of temporal information present in the data to improve performance. The system begins by cleaning the imagery, which is done by tracking objects on the sea bed in the imagery and using this information to obtain an improved estimate of the motion of the vessel. Once the vessel's motion is accurately known, the imagery is cleaned by temporally averaging the images after motion compensation. The detector consists of two stages. After the first detection stage has identified possible objects of interest, a bank of Kalman filters is used to track objects in the imagery and to supply sequences of feature vectors to the final detection stage. A recurrent neural network is used for the final detection stage. The feedback loops within the recurrent network allow the incorporation of temporal information into the detection process. The performance of the proposed system is shown to exceed the performances of other models for the final detection stage, including nonrecurrent networks that make use of temporal information supplied in the form of temporal feature vectors. The proposed detection system attains a probability of detection of 77.0% at a mean false-alarm rate of 0.4 per image. © 2004 IEEE.
Perry, SW & Guan, L 2004, 'Pulse-Length-Tolerant Features and Detectors for Sector-Scan Sonar Imagery', IEEE Journal of Oceanic Engineering, vol. 29, no. 1, pp. 138-156.View/Download from: Publisher's site
This paper presents a neural-network-based system to detect small man-made objects in sequences of sector-scan sonar images created using signals of various pulse lengths. The detection of such objects is considered out to ranges of 150 m by using an experimental sector-scan sonar system mounted on a vessel. The sonar system considered in this investigation has three modes of operation to create images over ranges of 200, 400, and 800 m from the vessel using acoustic pulses of a different duration for each mode. After an initial cleaning operation performed by compensating for the motion of the vessel, the imagery is segmented to extract objects for analysis. A set of 31 features extracted from each object is examined. These features consist of basic object size and contrast features, shape moment-based features, moment invariants, and features extracted from the second-order histogram of each object. Optimal sets of 15 features are then selected for each mode and over all modes using sequential forward selection (SFS) and sequential backward selection (SBS). These features are then used to train neural networks to detect man-made objects in each sonar mode. By the addition of a feature describing the sonar's mode of operation, a neural network is trained to detect man-made objects in any of the three sonar modes. The multimode detector is shown to perform very well when compared with detectors trained specifically for each sonar mode setting. The proposed detector is also shown to perform well when compared to a number of statistical detectors based on the same set of features. The proposed detector achieves a 92.4% probability of detection at a mean false-alarm rate of 10 per image, averaged over all sonar mode settings.
Lo, KW, Perry, SW & Ferguson, BG 2002, 'Aircraft flight parameter estimation using acoustical Lloyd's mirror effect', IEEE TRANSACTIONS ON AEROSPACE AND ELECTRONIC SYSTEMS, vol. 38, no. 1, pp. 137-151.View/Download from: Publisher's site
Perry, SW & Guan, L 2002, 'A pulse length tolerant neural networkbased detector for sectorscan sonar', The Journal of the Acoustical Society of America, vol. 112, no. 5, pp. 2307-2307.View/Download from: Publisher's site
Perry, SW & Wyber, RJ 2000, 'Hopfield neural network approach for the reconstruction of wide-bandwidth sonar data', Neural Networks for Signal Processing - Proceedings of the IEEE Workshop, vol. 2, pp. 876-885.
Sonar systems with small physical apertures are easier to mount on small vessels and remotely operated vehicles (ROVs). Such systems however are limited in terms of angular resolution. Although wide-bandwidth signals may be used to increase the range resolution of a sonar system, angular resolution is unaffected. Such limitations can be overcome if the region of interest in the underwater environment is insonified from a number of different angles, and this low resolution information reconstructed into a high resolution image of the region. This paper proposes a reconstruction approach based on a Hopfield neural network. This approach is shown to perform better than the Inverse Radon Transform for image reconstruction under both noisy and noise-less conditions. To verify these claims, results are presented using both real and simulated sonar data.
Sutton, JP, Sha, DD, Perry, S & Guan, L 1999, 'Enhancing mine signatures in sonar images using nested neural networks', Proceedings of SPIE - The International Society for Optical Engineering, vol. 3710, no. I, pp. 570-577.
An adaptive image regularization algorithm, based on the Network of Networks (NON) neural computing theory, is applied to enhance mine signatures. The algorithm, developed by Guan and Sutton (GS), uses vector connections among model neurons (pixels) to delineate dynamic boundaries corresponding to critical features of images. The boundaries subdivide large networks into many smaller networks, where each smaller network has, in many instances, attractor properties. In this report, the GS algorithm is applied to deblur and segment three sets of underwater mine data. The results suggest that the GS algorithm requires minimal training, performs well under inhomogeneous conditions and generates contours, which can be fed into other NoN architectures for further processing, including classification.
Perry, SW & Guan, L 1996, 'A partitioned modified Hopfield neural network algorithm for real-time image restoration', Real-Time Imaging, vol. 2, no. 4, pp. 215-224.
In this paper, two image partitioning schemes are examined. The first scheme examined avoids boundary conflicts by the use of four restoration phases. The second scheme examined requires a degree of synchronization of the processors restoring adjacent regions. Both schemes avoid conflicting boundary conditions by taking into account the local image formation properties. Without any loss of processing speed, or increase in the number of processors required to restore the image, synchronizing conditions are not required in the four-phase scheme to restore the image accurately, however can be used to maximize restoration efficiency. An improved modified Hopfield neural network-based algorithm is developed to be especially applicable to the problems of real-time image processing based on the described partitioning schemes. The proposed algorithm extends the concepts involved with previous algorithms to enable faster image processing and a greater scope for using the inherent parallelism of the neural network approach to image processing. The simulation in this investigation shows that the new algorithm is able to maximize the efficiency of the described partitioning methods. This paper also presents an example of an application of the proposed algorithm to restore images degraded by motion blur. © 1996 Academic Press Limited.
An adaptive scaled mean square error (SMSE) filter using a Hopfield neural-network-based algorithm is presented. We show the development of the original SMSE filter from the minimum mean square error (MMSE) filter and the parametric mean square error (PMSE) filter, both of which suffer from the oversmooth phenomena. The SMSE filter is more efficient than the PMSE filter in terms of noise removal as it does not take into account all the correlation factors used for image enhancement. To further improve the performance of the SMSE filter, an adaptive approach is introduced. The adaptive SMSE filter uses a mask operation technique. A user-defined mask is moved across the image and the filtering parameters are computed based on the local image statistics of the region below the mask. The original and the adaptive SMSE filters are implemented using a Hopfield neural-network-based algorithm. A number of experiments were performed to test the filter characteristics. © 1996 SPIE and IS&T.
A neural network algorithm for the restoration of images suffering space-variant distortion is introduced. Using multiple weighting matrices to represent space-variance, the algorithm provides high quality restorations in a computationally inexpensive fashion. © 1995, IEE. All rights reserved.
Perry, SW 2018, 'Image and Video Noise: An Industry Perspective' in Bertalmio, M (ed), Denoising of Photographic Images and Video Fundamentals, Open Challenges and New Trends, Springer, United Kingdom, pp. 207-234.View/Download from: UTS OPUS or Publisher's site
This unique text/reference presents a detailed review of noise removal for photographs and video. An international selection of expert contributors provide their insights into the fundamental challenges that remain in the field of denoising, examining how to properly model noise in real scenarios, how to tailor denoising algorithms to these models, and how to evaluate the results in a way that is consistent with perceived image quality. The book offers comprehensive coverage from problem formulation to the evaluation of denoising methods, from historical perspectives to state-of-the-art algorithms, and from fast real-time techniques that can be implemented in-camera to powerful and computationally intensive methods for off-line processing.
Topics and features: describes the basic methods for the analysis of signal-dependent and correlated noise, and the key concepts underlying sparsity-based image denoising algorithms; reviews the most successful variational approaches for image reconstruction, and introduces convolutional neural network-based denoising methods; provides an overview of the use of Gaussian priors for patch-based image denoising, and examines the potential of internal denoising; discusses selection and estimation strategies for patch-based video denoising, and explores how noise enters the imaging pipeline; surveys the properties of real camera noise, and outlines a fast approximation of nonlocal means filtering; proposes routes to improving denoising results via indirectly denoising a transform of the image, considering the right noise model and taking into account the perceived quality of the outputs.
This concise and clearly written volume will be of great value to researchers and professionals working in image processing and computer vision. The book will also serve as an accessible reference for advanced undergraduate and graduate students in computer science, applied mathematics, and related fields.
Matthews, L, Culpepper, J, Perry, S, Perin, G & Bone, D 2018, 'International Conference on Science and Innovation for Land Power 2018', International Conference on Science and Innovation for Land Power 2018, Department of Defence, Australian Government, Adelaide, SA, Australia.View/Download from: UTS OPUS
Recent research reveals that signature disruption strategies of detection delay and disguise can provide effective counter-surveillance techniques for contemporary low-altitude Uninhabited Aerial Vehicle (UAV) or drone detection platforms. As the first in a series of tiered tests, a virtual 3D model of selected 'scaled-up' HSV-based (Human Visual System based) algorithmic patterns and 3D biological nanostructures were found to disrupt a camera sensor when
mirrored in a physical surface. Further prototype and field tests will be conducted to corroborate these findings, with the ultimate aim of proposing an effective, controllable and disruptive mechanism to overhead UAV surveillance technology.
Nguyen, N, Le, TH, Perry, S & Nguyen, TT 2018, 'Pavement crack detection using convolutional neural network', SoICT 2018: Ninth International Symposium on Information and Communication Technology, Association for Computing Machinery, Danang, Vietnam, pp. 251-256.View/Download from: UTS OPUS or Publisher's site
Pavement crack detection is an important problem in road maintenance. There are many processing methods, including traditional and modern methods, solving this issue. Traditional methods use edge detection or some other digital image processing for crack detection, but these approaches are sensitive to many types of noise and unwanted objects on the road. For the purpose of increasing accuracy, image pre-processing methods are required for many of these techniques. Recently, some techniques that utilize deep learning to detect cracks in images have achieved high accuracy, without pre-processing. However, some of them are very complicated, some make use of manually collected data and some methods still need some form of pre-processing. In this paper, we propose a method that applies a convolutional neural networks to detect cracks in pavement images. Our research uses two data sets, one public data set and the other collected by ourselves. We also experimentally compare our method with some exiting methods and the experiments show that the proposed approach achieves high accuracy and generates stable models.
Cong, HP, Perry, SW, Vu, TA & Hoang, XV 2017, 'Joint exploration model based light field image coding: A comparative study', 2017 4th NAFOSTED Conference on Information and Computer Science NICS 2017 Proceedings, 2017 4th NAFOSTED Conference on Information and Computer Science, IEEE, Hanoi, Vietnam, pp. 308-313.View/Download from: UTS OPUS or Publisher's site
The recent light field imaging technology has been attracting a lot of interests due to its potential applications in a large number of areas including Virtual Reality, Augmented Reality (VR/AR), Teleconferencing, and E-learning. Light Field (LF) data is able to provide rich visual information such as scene rendering with changes in depth of field, viewpoint, and focal length. However, Light Field data usually associates to a critical problem — the massive data. Therefore, compressing LF data is one of the main challenges in LF research. In this context, we present in this paper a comparative study for compressing LF data with not only the widely used image/video coding standards, such as JPEG-2000, H.264/AVC, HEVC and Google/VP9 but also with the most recent image/video coding solution, the Joint Exploration Model. In addition, this paper also proposes a LF image coding flow, which can be used as a benchmark for future LF compression evaluation. Finally, the compression efficiency of these coding solutions is thoroughly compared throughout a rich set of test conditions.
We present an adaptive weighted temporal averaging filter with implicit motion-compensation for effective object enhancement in sector scan sonar image sequences. Visual blurring artifacts introduced by the temporal filtering process due to motion of the sonar platform are minimized by accurate motion estimation and compensation. An algorithm is proposed to perform object boundary extraction for better motion estimation. Motion estimation is performed directly on polar image sequences using cross-correlation followed by a Minimum Mean Square Error (MMSE) method. Each pixel of the filtered image is computed as the weighted average of the image pixel values over successive frames after motion compensation. The performance of the proposed filter is tested using real sector scan sonar image sequences and the results are compared with those obtained using the temporal averaging and motion compensated temporal averaging filters. © 2010 IEEE.
Pham, TQ, Perry, SW & Fletcher, PA 2009, 'Paper fingerprinting using alpha-masked image matching', DICTA 2009 - Digital Image Computing: Techniques and Applications, pp. 439-446.View/Download from: UTS OPUS or Publisher's site
In this paper, we examine the problem of authenticating paper media using the unique fibre structure of each piece of paper (the so-called "paper fingerprint"). In particular, we look at methods to authenticate paper media when text has been printed over the authentication zone. We show how alpha-masked correlation  can be applied to this problem and develop a modification to alpha-masked correlation that is more closely matched to the requirements of this problem and produces an improvement in performance. We also investigate two methods of pixel inpainting to remove printed text or marks from the authentication zone and allow ordinary correlation to be performed. We show that these methods can perform as well as alpha-masked correlation. Finally two methods of improving the robustness to forgery are investigated. © 2009 IEEE.
Perry, SW, Varjavandi, P & Guan, L 2004, 'Adaptive image restoration using a perception based error meausrement', Canadian Conference on Electrical and Computer Engineering, pp. 1585-1588.
This paper deals with image restoration; we have developed a novel, perceptually inspired image restoration method which takes human perception knowledge into consideration to reverse the effects of blur. Instead of using a conventional greyscale based error measurement such as the MSE, we compare local statistical information about regions in two images using a new error measure. The new method provides a better appraisal of image quality in terms of human vision. We extended the popular constrained least square error cost function by incorporating this novel image error measure. Using the well known Karush-Kuhn-Tucker theorem, we have mathematically verified that there exists an optimal solution to this non-linear constrained optimization problem in terms of the Hopfield neural network . We will show that the new restoration algorithm visually restores images as well as the previously presented LVMSE-based algorithm .
This paper presents a neural network based system to detect small man-made objects in sequences of sector scan sonar images. The detection of such objects is considered out to ranges of 150 metres using a forward-looking sonar system mounted on a vessel. After an initial cleaning operation performed by compensating for the motion of the vessel, the imagery was segmented to extract objects for analysis. A set of 31 features extracted from each object was examined. These features consisted of basic object size and contrast features, shape moment-based features, moment invariants, and features extracted from the second-order histogram of each object. The best set of 15 features was then selected using Sequential Forward Selection and Sequential Backward Selection. These features were then used to train a neural network to detect man-made objects in the image sequences. The detector achieved 97% accuracy at a mean false positive rate of 9 per frame.
Lo, KW, Perry, SW & Ferguson, BG 1999, 'An image processing approach for aircraft flight parameter estimation using the acoustical Lloyd's mirror effect', ISSPA 1999 - Proceedings of the 5th International Symposium on Signal Processing and Its Applications, pp. 503-506.View/Download from: Publisher's site
A time-frequency analysis of the output of an acoustic sensor located above the ground during the transit of an aircraft shows an interference (or fringe) pattern on the time-frequency plane. This interference pattern, referred to as the Lloyd's mirror effect, is caused by the temporal variations of the constructive/destructive interference frequencies of the direct and ground-reflected aircraft sound fields at the sensor. A model has been developed to describe the temporal variations of the destructive-interference frequencies for an aircraft in level flight over a hard ground. This paper describes two methods to estimate the aircraft flight parameters based on this model. In both methods, the time-frequency distribution of the sensor output is treated as an image. This image is pre-processed to enhance the destructive-interference pattern and then the flight parameters are extracted from the resultant image by optimising a cost function. The effectiveness of the methods is verified using real acoustic data. © 1999 IEEE.
Guan, L, Perry, S, Romagnoli, R, Wong, H & Kong, H 1998, 'Neural vision system and applications in image processing and analysis', ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings, pp. 1245-1248.View/Download from: Publisher's site
We present a computer vision system based on an integrated neural network architecture. In the low level vision subsystem, a network of networks-a biologically inspired network is used to recursively perform filtering, segmentation and edge detection; in the intermediate level and the high level, hierarchically structured arrays of self-organizing tree maps-extension of the popular self-organizing map are utilized to carry out image/feature analysis. The system has been applied to solve a number of real world problems. Some interesting and encouraging results are reported. © 1998 IEEE.
Perry, SW & Guan, L 1998, 'Perception based adaptive image restoration', ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings, pp. 2893-2896.View/Download from: Publisher's site
This paper presents an image restoration technique which uses a cost function based on a novel image error measure. The cost function presented here takes into account local statistical information of the image when performing restoration. It is shown that this technique compares favourably with other techniques, especially when applied to colour images. © 1998 IEEE.
Perry, SW & Guan, L 1998, 'Statistics-based weight assignment in a Hopfield neural network for adaptive image restoration', IEEE International Conference on Neural Networks - Conference Proceedings, pp. 922-925.
This paper investigates the assignment of weights to a Hopfield-based neural network in adaptive image restoration. The network is given a range of possible weights which are functions of a constraint factor to suppress noise in the restored image. Two methods for choosing the optimal weights are investigated. The first method is the traditional gradient descent method which is based on choosing the constraint value which best minimizes the neural network energy function for each pixel during each iteration of the algorithm. It is shown in this paper that, contrary to our intuition, this method will not produce optimal results. We then propose a second method which is based on selecting each neurons constraint value by considering local image statistics before restoration is commenced. It is shown that this method compares favourably with other neural network image restoration techniques.
Perry, SW & Guan, L 1997, 'Adaptive constraint restoration and error analysis using a neural network', Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), pp. 87-95.
© Springer-Verlag Berlin Heidelberg 1997. In this paper we present a restoration technique aimed at correcting image degradations by consideration of human visual criteria. A neural network model with an adaptive constraint factor is used. By considering local statistical information about regions within an image, the value of constraint factor can be selected which produces an optimal trade-off between noise suppression and edge preservation in each statistically homogeneous region. In addition a novel image error measure is presented which takes into account the statistical matching of homogeneous regions and its effect on human visual appraisal of image quality.
Guan, L, Perry, S & Wong, H 1997, 'Recursive low level vision system', Proceedings of the IEEE International Conference on Systems, Man and Cybernetics, pp. 637-648.
The paper addresses a very important, yet one of the difficult issues in computer vision and visualization - low level vision modeling. It proposes a novel low level vision model which recursively integrates adaptive filtering, segmentation and edge detection. The model has strong biological merits: a) the model architecture is based on a biologically inspired neural network - network of networks which simulates human visual cortex; b) evolutionary computation is applied to identify the hierarchy and clusters in the network. But the model does not constrain itself by the biological facts. Instead, it proposes that by using clustering method, adaptive filtering, segmentation and edge detection are naturally linked to one another. This is a completely computationally oriented concept and is the center of the. The feasibility of the concept will be demonstrated via a visual example.
Wong, EP, Guan, L & Perry, SW 1996, 'Neural network implementation of the SMSE filter for imaging processing', Proceedings of SPIE - The International Society for Optical Engineering, pp. 77-85.
This paper presents an implementation and enhancement of the SMSE (scaled mean square error) filter, using a Hopfield neural network based algorithm. We show the development of the original SMSE filter from the MMSE (minimum mean square error) filter and the PMSE (parametric mean square error) filter, both of which suffer from the oversmooth phenomena. The SMSE filter is more efficient than the PMSE filter in terms of noise removal as it does not take into account all the correlation factors used for image restoration. An adaptive SMSE filter is also presented. The adaptive SMSE filter uses a mask operation technique. A user- defined mask is moved across the image and the filtering parameters are computed based on the local image statistics of the region below the mask. The original and adaptive SMSE filters are implemented using a Hopfield neural network based algorithm. A number of experiments were performed to test the filter characteristics.
Perry, SW & Guan, L 1995, 'Restoration of images degraded by space-variant distortion using a neural network', IEEE International Conference on Neural Networks - Conference Proceedings, pp. 2067-2070.
This paper introduces a neural network algorithm to the restoration of images suffering a known form of space-variant distortion. Using multiple weighting matrices to represent space-variance, the algorithm provides high quality restorations. The algorithm will also be shown to be very computationally inexpensive due to the ability to minimize the neural network's energy in the most efficient manner.
Perry, SW DSTO Aeronautical and Maritime Research Laboratory, Department of Defence, Commonwealth of Australia 2000, Applications of Image Processing to Mine Warfare Sonar, no. DSTO-GD-0237, Sydney, Australia.
Information from various mine warfare sonar systems is often presented to the operator in a visual form. To obtain the optimum performance of these systems, it is desirable to apply intelligent processing techniques to the corresponding imagery. This report examines image processing techniques which may have the potential to improve either system or operator performance. The types of mine warfare sonar imagery examined in this report are sector-scan, side-scan, and the AMI project imagery. For each of these three types of imagery/applicable image processing concepts and techniques are examined with reference to techniques recorded in the literature.