Structured neural networks for pattern recognition

This paper proposes a novel approach for the design of structures of neural networks for pattern recognition. The basic idea lies in subdividing the whole classification problem in smaller and simpler problems at different levels, each managed by appropriate components of a complex neural architecture. Three neural structures are presented and applied in a surveillance system aimed at monitoring a railway waiting room classifying potential dangerous situations. Each architecture is composed by nodes, which are actual multilayer perceptrons trained to discriminate between subsets of classes until a complete separation among the classes is achieved. This approach showed better performances with respect to a classical statistical classification procedures and to a single neural network.     

Geo-located image categorization and location recognition

Image categorization is undoubtedly one of the most recent and challenging problems faced in Computer Vision. The scientific literature is plenty of methods more or less efficient and dedicated to a specific class of images; further, commercial systems are also going to be advertised in the market. Nowadays, additional data can also be attached to the images, enriching its semantic interpretation beyond the pure appearance. This is the case of geo-location data that contain information about the geographical place where an image has been acquired. This data allow, if not require, a different management of the images, for instance, to the purpose of easy retrieval from a repository, or of identifying the geographical place of an unknown picture, given a geo-referenced image repository. This paper constitutes a first step in this sense, presenting a method for geo-referenced image categorization, and for the recognition of the geographical location of an image without such information available. The solutions presented are based on robust pattern recognition techniques, such as the probabilistic Latent Semantic Analysis, the Mean Shift clustering and the Support Vector Machines. Experiments have been carried out on a couple of geographical image databases: results are actually very promising, opening new interesting challenges and applications in this research field. The article is published in the original. Marco Cristani received the Laurea degree in 2002 and the Ph.D. degree in 2006, both in Computer Science from the University of Verona, Verona, Italy. He was a visiting Ph.D. student at the Computer Vision Lab, Institute for Robotics and Intelligent Systems School of Engineering (IRIS), University of Southern California, Los Angeles, in 2004–2005. He is now an Assistant Professor with the Department of Computer Science, University of Verona, working with the Vision, Image Processing and Sounds (VIPS) Lab. His main research interests include statistical pattern recognition, generative modeling via graphical models, and non-parametric data fusion techniques, with applications on surveillance, segmentation and image and video retrieval. He is the author of several papers in the above subjects and a reviewer for several international conferences and journals. Alessandro Perina received the BD and MS degrees in Information Technologies and Intelligent and Multimedia Systems from the University of Verona, Verona, Italy, in 2004 and 2006, respectively. He is currently a Ph.D. candidate in the Computer Science Department at the University of Verona. His research interests include computer vision, machine learning and pattern recognition. He is a student member of the IEEE. Umberto Castellani is Ricercatore (i.e., Research Assistant) of Department of Computer Science at University of Verona. He received his Dottorato di Ricerca (Ph.D.) in Computer Science from the University of Verona in 2003 working on 3D data modelling and reconstruction. During his Ph.D., he had been Visiting Research Fellow at the Machine Vision Unit of the Edinburgh University, in 2001. In 2007 he has been an Invited Professor for two months at the LASMEA laboratory in Clermont-Ferrand, France. In 2008, he has been Visiting Researcher for two months at the PRIP laboratory of the Michigan State University (USA). His main research interests concern the processing of 3D data coming from different acquisition systems such as 3D models from 3D scanners, acoustic images for the vision in underwater environment, and MRI scans for biomedical applications. The addressed methodologies are focused on the intersections among Machine Learning, Computer Vision and Computer Graphics. Vittorio Murino received the Laurea degree in electronic engineering in 1989 and the Ph.D. degree in electronic engineering and computer science in 1993, both from the University of Genoa, Genoa, Italy. He is a Full Professor with the Department of Computer Science, University of Verona. From 1993 to 1995, he was a Postdoctoral Fellow in the Signal Processing and Understanding Group, Department of Biophysical and electronic Engineering, University of Genoa, where he supervised of research activities on image processing for object recognition and pattern classification in underwater environments. From 1995 to 1998, he was an Assistant Professor of the Department of Mathematics and Computer Science, University of Udine, Udine, Italy. Since 1998, he has been with the University of Verona, where he founded and is responsible for the Vision, Image processing, and Sound (VIPS) Laboratory. He is scientifically responsible for several national and European projects and is an Evaluator for the European Commission of research project proposals related to different scientific programmes and frameworks. His main research interests include computer vision and pattern recognition, probabilistic techniques for image and video processing, and methods for integrating graphics and vision. He is author or co-author of more than 150 papers published in refereed journals and international conferences. Dr. Murino is a referee for several international journals, a member of the technical committees for several conferences (ECCV, ICPR, ICIP), and a member of the editorial board of Pattern Recognition, IEEE Transactions on Systems, Man, and Cybernetics, Pattern Analysis and Applications and Electronic Letters on Computer Vision and Image Analysis (ELCVIA). He was the promotor and Guest Editor off our special issues of Pattern Recognition and is a Fellow of the IAPR.     

Learning natural scene categories by selective multi-scale feature extraction

Natural scene categorization from images represents a very useful task for automatic image analysis systems. In the literature, several methods have been proposed facing this issue with excellent results. Typically, features of several types are clustered so as to generate a vocabulary able to describe in a multi-faceted way the considered image collection. This vocabulary is formed by a discrete set of visual codewords whose co-occurrence and/or composition allows to classify the scene category. A common drawback of these methods is that features are usually extracted from the whole image, actually disregarding whether they derive properly from the natural scene to be classified or from foreground objects, possibly present in it, which are not peculiar for the scene. As quoted by perceptual studies, objects present in an image are not useful to natural scene categorization, indeed bringing an important source of clutter, in dependence of their size.     

A real-time versatile roadway path extraction and tracking on an FPGA platform

This paper presents an algorithm for roadway path extraction and tracking and its implementation in a Field Programmable Gate Array (FPGA) device. The implementation is particularly suitable for use as a core component of a Lane Departure Warning (LDW) system, which requires high-performance digital image processing as well as low-cost semiconductor devices, appropriate for the high volume production of the automotive market. The FPGA technology proved to be a proper platform to meet these two contrasting requirements. The proposed algorithm is specifically designed to be completely embedded in FPGA hardware to process wide VGA resolution video sequences at 30 frames per second. The main contributions of this work lie in (i) the proper selection, customization and integration of the main functions for road extraction and tracking to cope with the addressed application, and (ii) the subsequent FPGA hardware implementation as a modular architecture of specialized blocks. Experiments on real road scenario video sequences running on the FPGA device illustrate the good performance of the proposed system prototype and its ability to adapt to varying common roadway conditions, without the need for a per-installation calibration procedure.     

A complete system for on-line 3D modelling from acoustic images

This paper presents a system for the three-dimensional (3D) reconstruction of an underwater environment on the basis of multiple range views from an acoustical camera. The challenge is to provide the reconstruction on-line, as the range views are obtained from the sensor. The final target of the work is to improve the understanding of a human operator driving an underwater Remotely Operated Vehicle. The acoustic camera provides a sequence of 3D images in real time. Data must be registered and fused to generate a unique 3D mosaic in the form of a triangle mesh, which is rendered through a graphical interface. Available technologies for registration and meshing have been modified and extended to match time constraints. Some experiments on real data are reported.     

Markov-based methodology for the restoration of underwater acoustic images

This article describes a probabilistic technique for the restoration of underwater acoustic images that is based on the Markov random fields (MRFs) methodology. The beamforming is applied to rough acoustic data that derive from multibeam systems or acoustic cameras to build a three-dimensional (3D) map, that is associated point by point with the estimates of the reliability of such measures. Specifically, backscattered echoes that are received by a 2D array antenna are arranged to generate two images in which each pixel represents the distance (range) from the sensor plane and the confidence of the measures, respectively. Unfortunately, this kind of image is affected by several problems due to the nature of the signal and the related sensing system. In the proposed algorithm, the range and the confidence images are modeled as separate MRFs whose associated probability distributions embed knowledge of the acoustic system, of the considered scene, and of the noise affecting the measures. In particular, the confidence image is first restored and the result is used to reconstruct the 3D image to allow an active integration of the reliability information. Optimal (in the maximum a posteriori probability sense) estimates of the reconstructed 3D map and the restored confidence image are obtained by minimizing the energy functionals, using simulated annealing. Experimental results on synthetic and real images show the performance of the proposed approach. © 1997 John Wiley & Sons, Inc. Int J Imaging Syst Technol, 8, 386–395, 1997     

Beyond classical consensus clustering: The least squares approach to multiple solutions

Clustering is one of the most important unsupervised learning problems and it consists of finding a common structure in a collection of unlabeled data. However, due to the ill-posed nature of the problem, different runs of the same clustering algorithm applied to the same data-set usually produce different solutions. In this scenario choosing a single solution is quite arbitrary. On the other hand, in many applications the problem of multiple solutions becomes intractable, hence it is often more desirable to provide a limited group of “good” clusterings rather than a single solution. In the present paper we propose the least squares consensus clustering. This technique allows to extrapolate a small number of different clustering solutions from an initial (large) ensemble obtained by applying any clustering algorithm to a given data-set. We also define a measure of quality and present a graphical visualization of each consensus clustering to make immediately interpretable the strength of the consensus. We have carried out several numerical experiments both on synthetic and real data-sets to illustrate the proposed methodology.     

Social interactions by visual focus of attention in a three‐dimensional environment

In human behaviour analysis, the visual focus of attention (VFOA) of a person is a very important cue. VFOA detection is difficult, though, especially in a unconstrained and crowded environment, typical of video surveillance scenarios. In this paper, we estimate the VFOA by defining the Subjective View Frustum, which approximates the visual field of a person in a three‐dimensional representation of the scene. This opens up to several intriguing behavioural investigations. In particular, we propose the Inter‐Relation Pattern Matrix, which suggests possible social interactions between the people present in a scene. Theoretical justifications and experimental results substantiate the validity and the goodness of the analysis performed.     

A probabilistic approach to the coupled reconstruction andrestoration of underwater acoustic images

Describes a probabilistic technique for the coupled reconstruction and restoration of underwater acoustic images. The technique is founded on the physics of the image-formation process. Beamforming, a method widely applied in acoustic imaging, is used to build a range image from backscattered echoes, associated point by point with another type of information representing the reliability (or confidence) of such an image. Unfortunately, this kind of images is plagued by problems due to the nature of the signal and to the related sensing system. In the proposed algorithm, the range and confidence images are modeled as Markov random fields whose associated probability distributions are specified by a single energy function. This function has been designed to fully embed the physics of the acoustic image-formation process by modeling a priori knowledge of the acoustic system, the considered scene, and the noise-affecting measures and also by integrating reliability information to allow the coupled and simultaneous reconstruction and restoration of both images. Optimal (in the maximum a posteriori probability sense) estimates of the reconstructed range image map and the restored confidence image are obtained by minimizing the energy function using simulated annealing. Experimental results show the improvement of the processed images over those obtained by other methods performing separate reconstruction and restoration processes that disregard reliability information