TY - RPRT T1 - Sparsity-Inspired Recognition of Targets in Infrared Images Y1 - 2010 A1 - Chellapa, Rama KW - *FORWARD LOOKING INFRARED SYSTEMS KW - *IMAGE PROCESSING KW - *TARGET RECOGNITION KW - algorithms KW - ARMY RESEARCH KW - ATR(AUTOMATIC TARGET RECOGNITION) KW - AUTOMATIC KW - CNN(CONVENTIONAL NEURAL NETWORKS) KW - CS(COMPRESSIVE SENSING) KW - DISCRIMINANT ANALYSIS KW - FLIR(FORWARD LOOKING INFRARED) KW - INFRARED DETECTION AND DETECTORS KW - INFRARED IMAGES KW - LDA(LINEAR DISCRIMINANT ANALYSIS) KW - MILITARY OPERATIONS KW - MNN(MODULAR NEURAL NETWORKS) KW - neural nets KW - PCA(PRINCIPAL COMPONENT ANALYSIS) KW - PE611102 KW - SPARSITY KW - TARGET DIRECTION, RANGE AND POSITION FINDING AB - Sparsity-based methods have recently been suggested for tasks such as face and iris recognition. In this project, we evaluated the effectiveness of such methods for automatic target recognition in infrared images. We show how sparsity can be helpful for efficient utilization of data for target recognition. We evaluated the effectiveness of the proposed algorithm in terms of recognition rate and confusion matrices on the well known Comanche forward-looking infrared (FLIR) data set consisting of ten different military targets at different orientations. This work was done in collaboration with Dr. Nasser Nasrabadi, Chief Scientist, SEDD, Army research laboratory. This work will be presented at the International Conference on Image Processing being held in Hong Kong in September 2010. A journal paper reporting our work is under preparation. UR - http://stinet.dtic.mil/oai/oai?&verb=getRecord&metadataPrefix=html&identifier=ADA535424 ER - TY - CONF T1 - Salient Clustering for View-dependent Multiresolution Rendering T2 - Computer Graphics and Image Processing (SIBGRAPI), 2009 XXII Brazilian Symposium on Y1 - 2009 A1 - Barni,R. A1 - Comba,J. A1 - Varshney, Amitabh KW - (computer KW - algorithms;cluster KW - analysis;mesh KW - attention;mesh KW - AUTOMATIC KW - centred KW - clustering KW - clustering;rendering KW - clustering;user-centric KW - clusters;low-level KW - dependent KW - design; KW - framework;salient KW - graphics);user KW - graphics;face KW - human KW - information;propagative KW - mesh KW - multiresolution KW - rendering;image KW - representation;mesh KW - resolution;image KW - saliency;mesh KW - seed KW - segmentation KW - segmentation;pattern KW - segmentation;perceptual KW - selection;computer KW - system;view KW - visual AB - Perceptual information is quickly gaining importance in mesh representation, analysis and rendering. User studies, eye tracking and other techniques are able to provide ever more useful insights for many user-centric systems, which form the bulk of computer graphics applications. In this work we build upon the concept of Mesh Saliency - an automatic measure of visual importance for triangle meshes based on models of low-level human visual attention - applying it to the problem of mesh segmentation and view-dependent rendering. We introduce a technique for segmentation that partitions an object into a set of face clusters, each encompassing a group of locally interesting features; Mesh Saliency is incorporated in a propagative mesh clustering framework, guiding cluster seed selection and triangle propagation costs and leading to a convergence of face clusters around perceptually important features. We compare our technique with different fully automatic segmentation algorithms, showing that it provides similar or better segmentation without the need for user input. We illustrate application of our clustering results through a saliency-guided view-dependent rendering system, achieving significant frame rate increases with little loss of visual detail. JA - Computer Graphics and Image Processing (SIBGRAPI), 2009 XXII Brazilian Symposium on M3 - 10.1109/SIBGRAPI.2009.34 ER - TY - CONF T1 - A Logic Framework for Sports Video Summarization Using Text-Based Semantic Annotation T2 - Semantic Media Adaptation and Personalization, 2008. SMAP '08. Third International Workshop on Y1 - 2008 A1 - Refaey,M.A. A1 - Abd-Almageed, Wael A1 - Davis, Larry S. KW - (mathematics);video KW - analysis;trees KW - annotation;Internet;broadcasting;sport;text KW - AUTOMATIC KW - detection;logic KW - engine;parse KW - event KW - PROCESSING KW - processing; KW - semantic KW - signal KW - summarization;text KW - trees;sports KW - video KW - Webcasting;text-based AB - Detection of semantic events in sports videos is an essential step towards video summarization. A large volume of research has been conducted for automatic semantic event detection and summarization of sports videos. In this paper we present a novel sports video summarization framework using a combination of text, video and logic analysis. Parse trees are used to analyze structured and free-style text webcasting of sports games and extract the game¿s semantic events, such as goals and penalties in soccer games. Semantic events are then hierarchically arranged before being passed to a logic processing engine. The logic engine receives the summary preferences from the user and subsequently parses the event hierarchy to generate the game¿s summary according to the user¿s preferences. The proposed framework was applied to both soccer and basketball videos. We achieved an average accuracy of 98.6% and 100% on soccer and basketball videos, respectively. JA - Semantic Media Adaptation and Personalization, 2008. SMAP '08. Third International Workshop on M3 - 10.1109/SMAP.2008.25 ER - TY - JOUR T1 - Detection and analysis of hair JF - Pattern Analysis and Machine Intelligence, IEEE Transactions on Y1 - 2006 A1 - Yacoob,Yaser A1 - Davis, Larry S. KW - analysis;hair KW - appearance;image KW - AUTOMATIC KW - detection; KW - detection;hair KW - extraction;image KW - hair KW - indexing;multidimensional KW - recognition;face KW - recognition;feature KW - representation;object KW - representation;person AB - We develop computational models for measuring hair appearance for comparing different people. The models and methods developed have applications to person recognition and image indexing. An automatic hair detection algorithm is described and results reported. A multidimensional representation of hair appearance is presented and computational algorithms are described. Results on a data set of 524 subjects are reported. Identification of people using hair attributes is compared to eigenface-based recognition along with a joint, eigenface-hair-based identification VL - 28 SN - 0162-8828 CP - 7 M3 - 10.1109/TPAMI.2006.139 ER - TY - RPRT T1 - Physics-Based Detectors Applied to Long-Wave Infrared Hyperspectral Data Y1 - 2006 A1 - Broadwater,Joshua A1 - Chellapa, Rama KW - *HYPERSPECTRAL IMAGERY KW - algorithms KW - AUTOMATIC KW - DETECTION KW - FAR INFRARED RADIATION KW - LOW STRENGTH KW - Matched filters KW - MILITARY OPERATIONS KW - NIGHT KW - OPTICS KW - SIGNATURES KW - SOILS KW - SPECTROMETERS KW - TARGET DETECTION AB - Long-wave infrared (LWIR) hyperspectral image (HSI) data presents an interesting challenge for automatic target detection algorithms. LWIR HSI data is useful for both day and night operations, but weak signatures like disturbed soil can be problematic for standard matched-filter techniques. In this paper, we augment the standard matched-filter techniques with physics-based information particular to HSI data. Our results show that these physics-based detectors provide improved detection performance with quick processing times. PB - CENTER FOR AUTOMATION RESEARCH, UNIVERSITY OF MARYLAND COLLEGE PARK UR - http://stinet.dtic.mil/oai/oai?&verb=getRecord&metadataPrefix=html&identifier=ADA481339 ER - TY - RPRT T1 - Headline Generation for Written and Broadcast News Y1 - 2005 A1 - Zajic, David A1 - Dorr, Bonnie J A1 - Schwartz,Richard KW - *INFORMATION RETRIEVAL KW - *RADIO BROADCASTING KW - AUTOMATIC KW - AUTOMATIC SUMMARIZATION KW - HEDGE TRIMMER KW - INFORMATION SCIENCE KW - RADIO COMMUNICATIONS KW - STATISTICAL PROCESSES KW - TEST AND EVALUATION KW - WORDS(LANGUAGE) AB - This technical report is an overview of work done on Headline Generation for written and broadcast news. The report covers HMM Hedge, a statistical approach based on the noisy channel model, Hedge Trimmer, a parse-and-trim approach using linguistically motivated trimming rules, and Topiary, a combination of Trimmer and Unsupervised Topic Discovery. Automatic evaluation of summaries using ROUGE and BLEU is described and used to evaluate the Headline Generation systems. PB - Instititue for Advanced Computer Studies, Univ of Maryland, College Park UR - http://stinet.dtic.mil/oai/oai?&verb=getRecord&metadataPrefix=html&identifier=ADA454198 ER - TY - CONF T1 - Automatic position calibration of multiple microphones T2 - Acoustics, Speech, and Signal Processing, 2004. Proceedings. (ICASSP '04). IEEE International Conference on Y1 - 2004 A1 - Raykar,V.C. A1 - Duraiswami, Ramani KW - approximations; KW - array KW - audio KW - AUTOMATIC KW - calibration; KW - closed KW - covariance; KW - dimensional KW - estimation; KW - form KW - function KW - implicit KW - least KW - likelihood KW - loudspeakers; KW - maximum KW - microphone KW - microphones; KW - minimisation; KW - minimization; KW - multiple KW - nonlinear KW - position KW - positions; KW - problem; KW - processing; KW - signal KW - solution; KW - squares KW - theorem; KW - three AB - We describe a method to determine automatically the relative three dimensional positions of multiple microphones using at least five loudspeakers in unknown positions. The only assumption we make is that there is a microphone which is very close to a loudspeaker. In our experimental setup, we attach one microphone to each loudspeaker. We derive the maximum likelihood estimator and the solution turns out to be a non-linear least squares problem. A closed form solution which can be used as the initial guess for the minimization routine is derived. We also derive an approximate expression for the covariance of the estimator using the implicit function theorem. Using this, we analyze the performance of the estimator with respect to the positions of the loudspeakers. The algorithm is validated using both Monte-Carlo simulations and a real-time experimental setup. JA - Acoustics, Speech, and Signal Processing, 2004. Proceedings. (ICASSP '04). IEEE International Conference on VL - 4 M3 - 10.1109/ICASSP.2004.1326765 ER - TY - RPRT T1 - Dimensionality Estimation in Hyperspectral Imagery Using Minimum Description Length Y1 - 2004 A1 - Broadwater,Joshua B A1 - Meth,Reuven A1 - Chellapa, Rama KW - *ALGORITHMS KW - *HYPERSPECTRAL IMAGERY KW - *TARGET DETECTION KW - *TARGET RECOGNITION KW - AUTOMATIC KW - BURIED OBJECTS KW - COMPONENT REPORTS KW - MINE DETECTION KW - NUMERICAL MATHEMATICS KW - SURFACE TARGETS. KW - SYMPOSIA KW - TARGET DIRECTION, RANGE AND POSITION FINDING AB - Numerous algorithms have been developed for hyperspectral automatic target recognition (ATR) applications. Many of these algorithms require estimation of a background subspace. The estimation of the background subspace has been addressed using multiple methods. but most of these methods assume a-priori knowledge of the background dimensionality. In order to automate the estimation of the background subspace. we present an algorithm based on minimum description length (MDL) that can identify the background dimension. Results show that the MDL criterion estimates the proper dimension of the background for ATR applications. PB - MARYLAND UNIV COLLEGE PARK CENTER FOR AUTOMATION RESEARCH UR - http://stinet.dtic.mil/oai/oai?&verb=getRecord&metadataPrefix=html&identifier=ADA431643 ER - TY - CONF T1 - Uncalibrated stereo rectification for automatic 3D surveillance T2 - Image Processing, 2004. ICIP '04. 2004 International Conference on Y1 - 2004 A1 - Lim,S.-N. A1 - Mittal,A. A1 - Davis, Larry S. A1 - Paragios,N. KW - 3D KW - AUTOMATIC KW - conjugate KW - epipolar KW - image KW - lines; KW - matching; KW - method; KW - processing; KW - rectification KW - scene; KW - stereo KW - surveillance; KW - uncalibrated KW - urban AB - We describe a stereo rectification method suitable for automatic 3D surveillance. We take advantage of the fact that in a typical urban scene, there is ordinarily a small number of dominant planes. Given two views of the scene, we align a dominant plane in one view with the other. Conjugate epipolar lines between the reference view and plane-aligned image become geometrically identical and can be added to the rectified image pair line by line. Selecting conjugate epipolar lines to cover the whole image is simplified since they are geometrically identical. In addition, the polarities of conjugate epipolar lines are automatically preserved by plane alignment, which simplifies stereo matching. JA - Image Processing, 2004. ICIP '04. 2004 International Conference on VL - 2 M3 - 10.1109/ICIP.2004.1419753 ER - TY - RPRT T1 - Domain-Specific Term-List Expansion Using Existing Linguistic Resources Y1 - 2002 A1 - Dorr, Bonnie J A1 - Zhao,Tiejun KW - *LEXICOGRAPHY KW - *LINGUISTICS KW - AUTOMATIC KW - CHINA KW - COMPARISON KW - EXTRACTION KW - HANDS KW - knowledge based systems KW - linguistics KW - resources KW - WORDS(LANGUAGE) AB - This report describes a series of experiments involving expansion of a domain-specific human-generated "seed list" using available linguistic resources. The resources used for the expansion are intended to be general purpose: two large-scale Chinese-English dictionaries and a Chinese lexical knowledge base (HowNet). The methodology involves three steps: (1) hand extraction of head words from each entry in the human-generated seed list; (2) automatic comparison of these head words against entries in the linguistic resources-where an entry matches if the head word matches the entry exactly or is included in its the semantic definition; and (3) collection of any resulting matching entries into a larger term list. The terms extracted by this process were verified manually to confirm whether they were relevant to the topic of a specific domain. An important contribution of this work is the finding that the use of a bilingual term list for the expansion proces does not provide a significant improvement over the use of a simpler, more easily produced, monoligual term list. PB - Instititue for Advanced Computer Studies, Univ of Maryland, College Park UR - http://stinet.dtic.mil/oai/oai?&verb=getRecord&metadataPrefix=html&identifier=ADA525800 ER - TY - CONF T1 - The processing of form documents T2 - Document Analysis and Recognition, 1993., Proceedings of the Second International Conference on Y1 - 1993 A1 - David Doermann A1 - Rosenfeld, A. KW - AUTOMATIC KW - business KW - detectors; KW - document KW - documents; KW - extraction; KW - feature KW - form KW - forms; KW - generation; KW - generic KW - handling; KW - known KW - markings; KW - model KW - modeling; KW - non-form KW - optimal KW - properties; KW - reconstruction; KW - set; KW - specialized KW - stroke KW - width AB - An overview of an approach to the generic modeling and processing of known forms is presented. The system provides a methodology by which models are generated from regions in the document based on their usage. Automatic extraction of an optimal set of features to be used for registration is proposed, and it is shown how specialized detectors can be designed for each feature based on their position, orientation and width properties. Registration of the form with the model is accomplished using probing to establish correspondence. Form components which are corrupted by markings are detected and isolated, the intersections are interpreted and the properties of the non-form markings are used to reconstruct the strokes through the intersections. The feasibility of these ideas is demonstrated through an implementation of key components of the system JA - Document Analysis and Recognition, 1993., Proceedings of the Second International Conference on M3 - 10.1109/ICDAR.1993.395687 ER -