REDUCED-REFERENCE ASSESSMENT OF PERCEIVED QUALITY BY EXPLOITING COLOR INFORMATION
|
|
- Aubrie Morton
- 5 years ago
- Views:
Transcription
1 REDUCED-REFERENCE ASSESSMENT OF PERCEIVED QUALITY BY EXPLOITING COLOR INFORMATION Jdith Redi, Paolo Gastaldo, Rodolfo Znino, and Ingrid Heyndericx (+) University of Genoa, DIBE, Via Opera Pia a Genova Italy (+) Philips Research Laboratories Prof. Holstlaan AA Eindhoven NL and Delft Technical University, Meelweg 4 68 CD Delft NL { dith.redi, paolo.gastaldo, ingrid.heyndericx@philips.com ABSTRACT Image qality assessment systems based on a redcedreference paradigm rely on the extraction of salient featres both from a target image and the corresponding original, ndistorted image. Ths, they can spport realtime modeling of perceived qality. This paper shows that the color correlogram and its featres can be effective descriptors for a redced-reference system, embedding relevant information on the change in color distribtion in the images. In the proposed framewor, the featre-based descriptions of the images feed a doble-layer system: first, the artifact affecting the image is identified by exploiting a classifier based on Spport Vector Machines (SVMs); secondly, the nmerical representation of the image is mapped onto a qality score by a dedicated predictor exploiting the MltiLayer Perceptron (MPL) paradigm, which is specifically trained to assess image qality for a given artifact. Experimental reslts based on sbective qality data confirm the general validity of the approach.. INTRODUCTION Modeling perceived qality is a tas of paramont interest in the field of digital commnication technologies. Accrate qality assessment sally relies on data derived from experiments with hman sbects, which are ased to evalate the overall qality of a representative set of stimli. Collecting data throgh sbective experiments can provide ratings reflecting coherently hman perception, bt is also time consming, and often difficlt to model in a deterministic way []. Hence, when aiming at bilding a real-time system that optimizes the qality of any inpt signal atomatically, qality assessment that does not reqire any hman evalation, is needed. To this prpose, obective methods can be exploited [, 3]. These methods rely on the extraction of salient featres from images and on the processing of these featres towards a consistent estimation of the images qality. Obective methods sally adopt a fll reference paradigm [, 3], reqiring the nowledge of both the target image and the corresponding original, ndistorted image. In addition, some fll-reference methods exploit analytical models of the hman visal system (HVS) to improve their accracy and reliability. When woring in real time, thogh, the fll information on the ndistorted image is sally not available. Redced reference methods circmvent this lac by sing in the processing phase only a limited nmber of featres extracted from the original signal [, 4]. This paper presents a qality assessment system based on a redced-reference obective method. Some specific characteristics are extracted from the original, ndistorted image, as well as from the corresponding distorted (e.g. compressed) image. The non-linear mapping between the extracted featres and the qality rating is determined sing comptational intelligent methods, sch as a neral networ. Actally, by training the neral networ to mimic perceived image qality, the design of an explicit model of the hman visal system is avoided. In this research, color information is sed to constrct a compact bt informative featre-based description of the image. Artifacts introdced in images by digital systems seriosly affect the original color distribtion. Moreover, often the changes cased by a particlar artifact are very pecliar, so that it is liely that both the distortion and its amont can be identified by comparing the statistics of the original and distorted image. Color statistics have been widely sed for image indexing [6] and textre recognition [5], bt not for their potential contribtion to obective qality assessment. Ths, based on previos wor [7, 8], this research adopts a set of featres derived from the color correlogram [6], a second-order histogram of color information in an image. In particlar, the model extracts significant statistics from the he layer of an image, both at a local and at a global level.
2 A doble-layer system based on connectionist paradigms is designed to map this featre-based description of the image into qality scores. First, a classification algorithm is exploited to determine the artifact affecting the image. Then, a regression machine trained on solving the problem of assessing qality for images affected by that particlar distortion is sed to obtain a reliable estimation of the qality gain/loss of the sample with respect to its original version. The second release of the LIVE database [9] is sed for the performance evalation of or proposed approach. In particlar, three inds of distortion are addressed: White noise, Gassian Blr and JPEG compression. The reslts confirm the general validity of the connectionist paradigm, as well as of the se of color statistics to efficiently predict image qality.. FRAMEWORK FOR REDUCED REFERENCE IMAGE QUALITY ESTIMATION In formal terms, a redced reference qality prediction problem, in which a few data extracted from the original, ndistorted image are sed to predict the qality of a target image, can be expressed as follows. Let I be a reference image, and I the image obtained by applying some artifact to I, being r the distortion level. Let also x and ( n x r denote a nmerical representation of I and I, respectively. Finally, let q and q be the sbective qality ratings for I and ( n I r, respectively. Then, the adopted paradigm aims at estimating the distance ( n) d S ( q, q ) between the sbective scores associated with the two images by comparing the obective descriptors, x and ( n x r. It shold be noticed that the information extraction from the original image can be performed before transmission and the reslting descriptor ( x ) can be sent along with the image throgh the broadcast channel as metadata. The descriptor x is then extracted at the otpt side of the channel, as visalized in Figre. Figre also illstrates that the proposed redcedreference framewor decoples the prediction of image qality into two tass. The rationale behind sch an approach is that every ind of distortion modifies the image in a different way, ths reqiring a specific obective metric to feed the qality predictor. So, having determined the set A={a, a,, a n } of artifacts of interest, implementing a dedicated qality predictor for each of them can be a sccessfl strategy. As no a-priori nowledge can be assmed on the natre of the distortion affecting the inpt signal, the framewor operates as follows. First, an artifact detector exploits the descriptors x and ( n x r to identify the distortion (a i A) applied to Figre - Overall scheme of the proposed system for obective qality assessment the image I. This modle then forwards the image descriptors to the qality predictor P(a i ), designed to qantify effects on perceived qality cased by the artifact a i. Finally, the specialized qality predictor provides a estimation for the difference in qality between the reference and distorted images, ( n) ( q, q ) d S. 3. COLOR INFORMATION EXTRACTION FOR QUALITY EVALUATION For the proposed architectre to be effective, an efficient nmerical representation of the image is needed. De to limitations in the transmission bandwidth, data representing the original image has to be small-sized, bt still very informative. The choice of a proper nmerical descriptor for an image is therefore crcial. The present stdy proposes to represent an image by the second order statistics of its color information. Recent wor [5, 6] showed how second order histograms can be sccessflly employed in different applications. Gastaldo et al. showed that co-occrrence matrices [7, 8] are sitable for a featre-based representation of an image able to convey information on the visal qality. The color correlogram [6] belongs to the family of the second order histograms, as well as the co-occrrence matrix. Aiming at maintaining in the final description both local and global information, the color correlogram is compted on eqally sized sb-regions of the image, and a featre representation is constrcted for each of them. Global information is then obtained by
3 assembling local information in a single descriptive vector by sing a statistical approach. 3.. Local information extraction A color correlogram Η b of a sb-region b (inclding W b H b pixels) for a predefined distance, is a threedimensional matrix that describes the spatial distribtion in lminance, color or any other image property that can be represented by qantized vales. More precisely, the color correlogram describes how the spatial correlation of pairs of colors changes with distance. Formally, the entry Η ( i, of the correlogram matrix is defined as: b ( m, n), m < Wb, n < Hb, s.t. Ηb ( i, = m, n] = Ci; p, q] = C ; dist( m, n], p, q]) = () Each matrix element Η b ( i, specifies the probability of finding a pixel of vale C at a distance from a pixel of vale C i. The dist() operator denotes the measre of distance between pixels selected to calclate the color correlogram. In or present wor, the L -norm is embedded for the operator dist(). In practical terms, choosing the L -norm means that only those pairs of pixels with a distance in horizontal and vertical direction are considered in the comptation. The proposed system preserves local information by compting the color correlogram on sb-regions of an image. To this end, images are split into non-overlapping sqare regions, each holding N a N a pixels. For each bloc, the set of featres Φ = {f 0, f, f 5 }, as defined in table I, is extracted from the correlogram. 3.. Global-level nmerical representation The local information extraction phase otpts as many vales for each featre of Φ as the nmber of blocs. This is far too mch data for a descriptor that shold be sent in real time throgh a commnication channel. Moreover, sbective scores sally express the qality of a whole image, and not of a set of blocs. So, from a modeling perspective, the featre-based image description shold consist of a single vector, to be associated with the single qality score. Toward this end, bloc-based information is combined sing statistical descriptors, namely percentiles, to represent the distribtion of a featre f over the image. Recalling the notations sed in the previos paragraphs, the following psedo-code can be applied to both the reference and the distorted image for constrcting the global descriptors. Inpts: a pictre I, a descriptive featre f Φ and a vale for distance. Bloc-level featre extraction. a. Split I into N b non-overlapping sqare blocs, and obtain the set Β = { b ; m =,..., } m N b Table I Featres calclated from the color correlogram for describing the images Featre name b. For each bloc b m B: compte the associate color correlogram: Η ( i,. c. For each matrix Η : compte the vale x b m, m of featre f, and obtain the set: Χ = b m { x ; m,.., N }, m =. Global level nmerical representation. a. Compte a percentile-based description of Χ ; let p α be the α th percentile: ϕ α, = α ( ) p Χ b. Assemble the obective descriptor vector, x, for the featre f on the image ( n, I Definition Energy f 0 = [ Η b ( i, ] Eventally, the extraction process reslts in a descriptor of an image, Ī, being a global pattern, x. This otpt of the extraction process is forwarded to the doble-layer assessment system for qality estimation. 4. COMPUTATIONAL INTELLIGENCE FOR OBJECTIVE QUALITY ASSESSMENT The system described in section consists of two steps. First, the artifact affecting the image has to be identified; secondly, the nmerical representation of the image has to be mapped onto a qality score by a dedicated predictor, which is specifically trained to assess image qality for a given artifact. To complete these two steps, two very different problems have to be solved. The first layer consists of a i, Diagonal Energy [ ] f = Η b ( i, i) b b ( i, i, Entropy f = Η ( i, log Η Contrast f 3 = z P( z) Homogeneity f 4 = Η b [ + ( i ] i, Energy Ratio f 5 = f0 f x P ( z) = Ηb ( i, i = z { ; [ 0,00] } ϕ α, = α z b ()
4 classification problem. Therefore, a system able to identify, given a set of classes, to which class an inpt pattern belongs to, needs to be designed. When aiming to detect the artifact a i affecting the sample I, the inpt pattern is the global descriptor (), and the m classes represent the artifacts of interest a, a, a m. The mapping fnction relates the inpt vector to a discrete vale, identifying the artifact. The second layer, instead, mst otpt a continos vale, being a qality score. Therefore, this modle has to be designed to solve a regression problem. This research adopts Spport Vector Machines (SVM) for the classification tas and a feed forward neral networ, namely the Circlar Bac-Propagation (CBP) networ for the regression tas. 4.. Spport Vector Machines In binary classification problems, the system is reqired to discriminate between only two classes. Spport vector machines are widely proven to be powerfl tools for solving these tass. Given a set of np patterns Z ={ (x l, y l ); l=,..,np; y l {-,+} }, a SVM relies on the soltion of the following Qadratic Programming problem, to find the optimal hyper-plane w separating the two classes: np np min αlα m yl ymk( x l, xm ) αl (3) α l, m= l= 0 α l C, l np sbect to: = ylα l 0 l= where α l are the SVM parameters setting the classseparating srface and the scalar qantity C is a fixed reglarization term that rles the trade-off between accracy and complexity. The ernel fnction Κ() allows inner prodcts of patterns in a higher dimensional, transformed space, yet disregarding the specific mapping of each single pattern. If Φ(x ) and Φ(x ) are the points in the featre space that are associated with x and x, respectively, then their dot prodct can be written as Φ ( x ), Φ( x ) = K( x, x). A common choice for the ernel Κ() that has been proven to be effective, is the Radial Basis Fnction (RBF) ernel, formlated as: K ( x x = e x /σ x Problem setting (3) has the crcial advantage of involving a qadratic-optimization problem with linear constraints, ensring that the soltion is niqe. The evental generalization performance of a SVM depends on the specific choice for the ernel parameters {C, σ}. Both theoretical [0] and empirical [] approaches can be adopted to determine the generalization (4) limits. In or present research, the parameters {C, σ} are tned following an empirical approach involving -fold cross validation []. 4.. Circlar Bac-Propagation In or proposed system feed-forward neral networs map featre-based image descriptions into the associated estimates of perceived qality, which, in the present formlation, are scalar vales. Theory proves that feed-forward networs embedding a sigmoidal nonlinearity can spport arbitrary mappings []. The MltiLayer Perceptron (MLP) model [] belongs to this class of networs, and has been proved to perform effectively for a wide variety of problems. The Circlar Bac Propagation (CBP) networ [3] extends the conventional MLP with an additional inpt, being the sm of the sqared vales of all the networ inpts. The qadratic term boosts the networ s representation ability withot affecting the fritfl properties of an MLP strctre. The CBP architectre can be formally described as follows. The inpt layer connects the n i inpt vales (featres) to each of the n h nerons of a hidden layer. The -th hidden neron performs a non-linear transformation on the weighted combination of the inpt vales, with coefficients w,i ( =,, n h ; i=,, n i ): ni ni a = sigm w,0 + w, i xi + w, n + x i i i= i= (5) where sigm(z)=(+e -z ) -, and a is the neron activation. Liewise, the otpt layer provides the actal networ responses, y, ( =,, n o ): n h y = sigm w,0 + w, a = (6) The strctral CBP enhancement still allows to adopt conventional bac-propagation algorithms [] for weight adstment, yielding an effective training. A qadratic cost fnction measres the distortion between the actal system otpt and the expected reference otpt for a sample of training patterns. The cost is expressed as: n p no ( l) ( l) ( t y ) E = non p l= = (7) where n p is the nmber of training patterns, and t is the desired reference otpt. In the present application, = and the expected otpt is given by the qality score obtained from sbective panel tests.
5 5. PRACTICAL IMPLEMENTATION The present section describes a possible implementation of the proposed redced-reference framewor, consisting of an SVM-based modle to solve the artifact identification problem, and a CBP architectre, mapping featres into qality scores. 5.. Nmerical description The first step in the neral-based qality estimator is the featre extraction, in which both the reference image I and the target image r I ) are processed to yield the vectors ( n x ) and x, compted according to the procedre reported in Section 3. The inpt image is divided into sqared sb-regions of 3x3 pixels, and for each bloc, all pixels are converted from the RGB color space to the HSV color space. Then, the color correlogram is compted on the he component (H-component) of the blocs. From these vales, the set of featres Φ defined in table xx is calclated. Finally, the global descriptor is assembled by compting 6 percentiles of the distribtion of each featre f, and combining them in the vector, r x n = ; α = 0,0,40,60,80,00; = 0,,...,5. ( ) { } ϕ α, For each featre, the obective representation of the stimls is obtained simply by combining the two vectors: [ ( n) ( ] n, z = x, x (8) The reslting -dimensional vectors r z ) = 0,...,5 feed the doble-layer qality estimator. It shold be noticed that not always the whole featre based information vector is needed, as will be shown later when discssing or experimental reslts. 5.. Neral-Based Qality Loss/Gain Qantification The role of the artifact identifier modle in the system is to solve a mlticlass problem, associating each z to the appropriate artifact a i A. To implement this mlticlass classifier, several binary predictors are connected in series. In this stdy three possible artifacts are considered: White noise, Gassian Blr and JPEG compression. The classification modle is implemented sing a first SVM to select images distorted by White Noise, and a second SVM to separate blrred images from JPEG compressed images. Both SVMs receive as inpt a vector based on one featre, namely Energy Ratio (see f 5 in table I). This featre seemed sfficiently effective for the artifact classification. The second layer, i.e. the qality level predictor is composed of as many sbsystems as the nmber of considered artifacts; in or case three. The three prediction modles have the same architectre; in particlar, to improve robstness, an ensemble strategy is applied [4]. To this end, a coordinate-partitioning approach is sed; in particlar, each single predictor involved in the ensemble fed with a single, distinct featre, satisfying the reqired hypothesis of disoint inpt sb-spaces. Being independent from one another, each artifact-dedicated sb-modle maes se of a particlar pair of featres, designed to be as mch informative as possible for the specific artifact: to qantify the effect of white noise on qality, the pair {Contrast, Total/Diagonal Energy} is sed; the pair {Diagonal Energy, Homogeneity} is sed to describe blrred images and the cople {Diagonal Energy, Entropy} is sed to characterize JPEG stimli. 6. EXPERIMENTAL RESULTS The second release of the LIVE database [9] was sed as a testbed for the performance evalation of or proposed model. In particlar, the three datasets inclding samples distorted with White noise, Gassian Blr and JPEG compression were considered. To evalate the overall performance of the doble-layer qality estimator, a - fold-lie test method was applied. From the LIVE database, which incldes several distorted versions of 9 original images, each artifact specific sbset was divided into 5 different folds, each containing all the distorted versions of a few original images, in sch a way that none of the 9 contents belonged to more than one grop. Both layers were trained independently, nonetheless for each of them the -fold strategy was adopted, performing 5 rns. In each rn 4 of the 5 folds were sed as training data, and the remaining one was sed as test data. In this way, the system was proven to be able to generalize independently of the specific image content sed for the training. For all experiments, the color correlogram was compted for a distance = sing the L norm. For the artifact classification system, it was necessary to tne the ernel parameters to improve the machine s generalization ability. To this prpose, the -fold crossvalidation techniqe was applied. First, the three datasets provided by the LIVE database were merged in a single one. The first SVM was trained to recognize noisy images. For this SVM, a linear ernel was employed, and the parameter C was finally set to 0 3. The second SVM was trained on a sbset of the first dataset, inclding only blrred and compressed images. In this case, the se of a RBF ernel was necessary, confirming that the problem of distingishing blr from compression artifacts (which is intitively more difficlt than the previos tas, since the visal effects prodced by the two distortions often overlap) is indeed of increased complexity. Based on the cross-validation otpt, the parameter C was set to 0 5 and σ =.
6 Table II - Performance of each SVM machine for artifact identification in terms of % of misclassified patterns White Noise recognition Table II reports the percentages of misclassified patterns for each rn and for both SVM classifiers. It clearly illstrates that the first SVM has a perfect performance, since none of the blrred or compressed images is misclassified as noisy image. The performance of the second SVM is worse, de to its more complex tas. Nonetheless, in the worst case still only abot 0% of the images is misclassified. For the qality estimation layer the three ensembles of two CBP neral networs were each trained on a different dataset. The datasets for white noise and blrred images contained 45 patterns, while the remaining testbed inclded 59 JPEG compressed images. For each image a qality score was provided in the LIVE database. The Difference Mean Opinion Score (DMOS), originally ranging between [0,00] was remapped for comptational reasons into the range [-, +]. De to the flexibility of the system, it was possible to design the networ architectre on prpose for each tas. For coherence, all the 6 neral networs shared the same inpt layer with the dimensionality of the global descriptor z. The networs for the qality estimation of noisy and blrred images were eqipped with a 5-nerons hidden layer, while, to maximize the generalization ability, the qality estimator for the JPEG compressed images conted 7 hidden nerons. For the three sb-systems, two different indicators were sed to evalate their accracy, defined as the discrepancy between the obectively predicted qality ˆ ( n) d q, q and the vale provided by the LIVE change S ( ) ( n) database d ( q, q ) S Blr/Compression recognition Rn # 0.00% 0.00% Rn # 0.00% 4.76% Rn #3 0.00% 0.00% Rn #4 0.00%.9% Rn #5 0.00% 8.0% Average 0.00% 4.80% : the mean vale of the absolte prediction error µ err between d s and dˆ S (in percentage), and the root mean sqare (RMS) error. Table III shows how the second layer in or system is able to prodce satisfactory qality predictions. Assessing the qality gain/loss of noisy images, the system s mean absolte prediction error, averaged over the five folds, is 5.8%. Dealing with compressed images, the mean absolte prediction error is 6.03%, which improves the accracy obtained in past stdies [8]. Finally, the qality estimator for the blrred images achieves an accracy of 93.5%, which represents an improvement of.90% with respect to previos reslts [8]. Table III - Performance of qality estimators in terms of percentage absolte error and RMS between predicted and sbective qality scores White Noise JPEG Compression 7. REFERENCES Gassian Blr %µ err RMS %µ err RMS %µ err RMS Rn # 4.87% % % 0.9 Rn # 5.06% % % 0.06 Rn # % % % 0.08 Rn #4 5.8% % % 0.09 Rn #5 5.64% % % 0.5 Average 5.77% % % 0.75 [] Wang, Z., Sheih, H.R., and Bovi, A.C.: Obective video qality assessment. In: Frth, B., and Marqes, O.: The Handboo of Video Databases: Design and Applications. CRC Press, Boca Raton, FL (003) [] International Telecommnication Union, Methodology for the sbective assessment of the qality of television pictres, ITU-R BT.500 (995). [3] Sheih, H.R., Sabir, M.F., and Bovi, A.C.: A statistical evalation of recent fll reference image qality assessment algorithm. IEEE Trans. Image Processing 5 (006) [4] Z. Wang and E. P. Simoncelli, Redced-reference image qality assessment sing a wavelet-domain natral image statistic model, in Proc. SPIE Hman Vision and Electronic Imaging X, vol (005). [5] Haralic, R.M., Shanmgam, K., and Dinstein, I.: Textral Featres for Image Classification. IEEE Trans. On Systems, Man and Cybernetics SMC-3 (973) 60- [6] 8. Hang, J., Ravi Kmar, S., Mitra, M., Zh, W.-J., and Zabih, R.: Image indexing sing color correlograms. In Proc. IEEE CVPR 97 (997) [7] Gastaldo P. and Znino Z.: Neral networs for the noreference assessment of perceived qality. Jornal of Electronic Imaging 4 (005) [8] Redi J., Gastaldo P., Znino R. and Heyndericx I.: Cooccrrence matrixes for the qality assessment of coded images. In Proc. ICANN 08 (008) [9] Sheih, H.R., Wang, Z., Cormac, L., and Bovi, A.C.: LIVE Image Qality Assessment Database at [0] V. Vapni, Statistical Learning Theory. New Yor: Wiley, 998. [] Bartlett P, Bocheron S, Lgosi G Model selection and error estimation. Machine Learning, 00, Volme 48, Nmber -3 pp.85-3 [] Rmelhart, D.E., and McClelland, J.L.: Parallel distribted processing. MIT Press, Cambridge, MA (986). [3] Ridella, S., Rovetta, S., and Znino, R.: Circlar bacpropagation networs for classification. IEEE Trans. on Neral Networs 8 (997) [4] Kittler, J., Hatef, M., Din, R. P. W., and Matas, J.: On combining classifiers. IEEE Trans. Pattern Analysis and Machine Intelligence. 0 (998) 6 39
Image Compression Compression Fundamentals
Compression Fndamentals Data compression refers to the process of redcing the amont of data reqired to represent given qantity of information. Note that data and information are not the same. Data refers
More informationTu P7 15 First-arrival Traveltime Tomography with Modified Total Variation Regularization
T P7 15 First-arrival Traveltime Tomography with Modified Total Variation Reglarization W. Jiang* (University of Science and Technology of China) & J. Zhang (University of Science and Technology of China)
More informationReal-time mean-shift based tracker for thermal vision systems
9 th International Conference on Qantitative InfraRed Thermography Jly -5, 008, Krakow - Poland Real-time mean-shift based tracker for thermal vision systems G. Bieszczad* T. Sosnowski** * Military University
More informationA sufficient condition for spiral cone beam long object imaging via backprojection
A sfficient condition for spiral cone beam long object imaging via backprojection K. C. Tam Siemens Corporate Research, Inc., Princeton, NJ, USA Abstract The response of a point object in cone beam spiral
More informationSTABILITY OF SIMULTANEOUS RECURRENT NEURAL NETWORK DYNAMICS FOR STATIC OPTIMIZATION
STABILITY OF SIMULTANEOUS RECURRENT NEURAL NETWOR DYNAMICS FOR STATIC OPTIMIZATION Grsel Serpen and Yifeng X Electrical Engineering and Compter Science Department, The University of Toledo, Toledo, OH
More informationAUTOMATIC REGISTRATION FOR REPEAT-TRACK INSAR DATA PROCESSING
AUTOMATIC REGISTRATION FOR REPEAT-TRACK INSAR DATA PROCESSING Mingsheng LIAO, Li ZHANG, Zxn ZHANG, Jiangqing ZHANG Whan Technical University of Srveying and Mapping, Natinal Lab. for Information Eng. in
More informationOn the Computational Complexity and Effectiveness of N-hub Shortest-Path Routing
1 On the Comptational Complexity and Effectiveness of N-hb Shortest-Path Roting Reven Cohen Gabi Nakibli Dept. of Compter Sciences Technion Israel Abstract In this paper we stdy the comptational complexity
More informationSZ-1.4: Significantly Improving Lossy Compression for Scientific Data Sets Based on Multidimensional Prediction and Error- Controlled Quantization
SZ-1.4: Significantly Improving Lossy Compression for Scientific Data Sets Based on Mltidimensional Prediction and Error- Controlled Qantization Dingwen Tao (University of California, Riverside) Sheng
More informationBias of Higher Order Predictive Interpolation for Sub-pixel Registration
Bias of Higher Order Predictive Interpolation for Sb-pixel Registration Donald G Bailey Institte of Information Sciences and Technology Massey University Palmerston North, New Zealand D.G.Bailey@massey.ac.nz
More informationFast Obstacle Detection using Flow/Depth Constraint
Fast Obstacle etection sing Flow/epth Constraint S. Heinrich aimlerchrylser AG P.O.Box 2360, -89013 Ulm, Germany Stefan.Heinrich@aimlerChrysler.com Abstract The early recognition of potentially harmfl
More informationAn Optimization of Granular Network by Evolutionary Methods
An Optimization of Granlar Networ by Evoltionary Methods YUN-HEE HAN, KEUN-CHANG KWAK* Dept. of Control, Instrmentation, and Robot Engineering Chosn University 375 Seos-dong, Dong-g, Gwangj, 50-759 Soth
More informationStatistical Methods in functional MRI. Standard Analysis. Data Processing Pipeline. Multiple Comparisons Problem. Multiple Comparisons Problem
Statistical Methods in fnctional MRI Lectre 7: Mltiple Comparisons 04/3/13 Martin Lindqist Department of Biostatistics Johns Hopkins University Data Processing Pipeline Standard Analysis Data Acqisition
More informationPARAMETER OPTIMIZATION FOR TAKAGI-SUGENO FUZZY MODELS LESSONS LEARNT
PAAMETE OPTIMIZATION FO TAKAGI-SUGENO FUZZY MODELS LESSONS LEANT Manfred Männle Inst. for Compter Design and Falt Tolerance Univ. of Karlsrhe, 768 Karlsrhe, Germany maennle@compter.org Brokat Technologies
More informationImage Denoising Algorithms
Image Denoising Algorithms Xiang Hao School of Compting, University of Utah, USA, hao@cs.tah.ed Abstract. This is a report of an assignment of the class Mathematics of Imaging. In this assignment, we first
More informationUncertainty Determination for Dimensional Measurements with Computed Tomography
Uncertainty Determination for Dimensional Measrements with Compted Tomography Kim Kiekens 1,, Tan Ye 1,, Frank Welkenhyzen, Jean-Pierre Krth, Wim Dewlf 1, 1 Grop T even University College, KU even Association
More informationOptimal Sampling in Compressed Sensing
Optimal Sampling in Compressed Sensing Joyita Dtta Introdction Compressed sensing allows s to recover objects reasonably well from highly ndersampled data, in spite of violating the Nyqist criterion. In
More informationA RECOGNITION METHOD FOR AIRPLANE TARGETS USING 3D POINT CLOUD DATA
A RECOGNITION METHOD FOR AIRPLANE TARGETS USING 3D POINT CLOUD DATA Mei Zho*, Ling-li Tang, Chan-rong Li, Zhi Peng, Jing-mei Li Academy of Opto-Electronics, Chinese Academy of Sciences, No.9, Dengzhang
More informationIMPLEMENTATION OF OBJECT ORIENTED APPROACH TO MODIFIED ANT ALGORITHM FOR TASK SCHEDULING IN GRID COMPUTING
International Jornal of Modern Engineering Research (IJMER) www.imer.com Vol.1, Isse1, pp-134-139 ISSN: 2249-6645 IMPLEMENTATION OF OBJECT ORIENTED APPROACH TO MODIFIED ANT ALGORITHM FOR TASK SCHEDULING
More informationIDENTIFICATION OF THE AEROELASTIC MODEL OF A LARGE TRANSPORT CIVIL AIRCRAFT FOR CONTROL LAW DESIGN AND VALIDATION
ICAS 2 CONGRESS IDENTIFICATION OF THE AEROELASTIC MODEL OF A LARGE TRANSPORT CIVIL AIRCRAFT FOR CONTROL LAW DESIGN AND VALIDATION Christophe Le Garrec, Marc Hmbert, Michel Lacabanne Aérospatiale Matra
More informationREPLICATION IN BANDWIDTH-SYMMETRIC BITTORRENT NETWORKS. M. Meulpolder, D.H.J. Epema, H.J. Sips
REPLICATION IN BANDWIDTH-SYMMETRIC BITTORRENT NETWORKS M. Melpolder, D.H.J. Epema, H.J. Sips Parallel and Distribted Systems Grop Department of Compter Science, Delft University of Technology, the Netherlands
More informationPicking and Curves Week 6
CS 48/68 INTERACTIVE COMPUTER GRAPHICS Picking and Crves Week 6 David Breen Department of Compter Science Drexel University Based on material from Ed Angel, University of New Mexico Objectives Picking
More informationFactors Influencing Performance of Firefly and Particle Swarm Optimization Algorithms
Factors Inflencing Performance of Firefly and Particle Swarm Optimization Algorithms Damanjeet Kar, UIET, Panjab University, Chandigarh Abstract In this paper, two natre inspired meta heristic approaches
More informationCOMPOSITION OF STABLE SET POLYHEDRA
COMPOSITION OF STABLE SET POLYHEDRA Benjamin McClosky and Illya V. Hicks Department of Comptational and Applied Mathematics Rice University November 30, 2007 Abstract Barahona and Mahjob fond a defining
More informationMethod to build an initial adaptive Neuro-Fuzzy controller for joints control of a legged robot
Method to bild an initial adaptive Nero-Fzzy controller for joints control of a legged robot J-C Habmremyi, P. ool and Y. Badoin Royal Military Academy-Free University of Brssels 08 Hobbema str, box:mrm,
More informationEvaluating Influence Diagrams
Evalating Inflence Diagrams Where we ve been and where we re going Mark Crowley Department of Compter Science University of British Colmbia crowley@cs.bc.ca Agst 31, 2004 Abstract In this paper we will
More informationDigital Image Processing Chapter 5: Image Restoration
Digital Image Processing Chapter 5: Image Restoration Concept of Image Restoration Image restoration is to restore a degraded image back to the original image while image enhancement is to maniplate the
More informationTdb: A Source-level Debugger for Dynamically Translated Programs
Tdb: A Sorce-level Debgger for Dynamically Translated Programs Naveen Kmar, Brce R. Childers, and Mary Lo Soffa Department of Compter Science University of Pittsbrgh Pittsbrgh, Pennsylvania 15260 {naveen,
More informationSwitched state-feedback controllers with multi-estimators for MIMO systems
Proceedings of the th WEA Int Conf on COMPUTATIONAL INTELLIGENCE MAN-MACHINE YTEM AND CYBERNETIC Venice Ital November - 6 89 witched state-feedback controllers with mlti-estimators for MIMO sstems LIBOR
More informationFINITE ELEMENT APPROXIMATION OF CONVECTION DIFFUSION PROBLEMS USING GRADED MESHES
FINITE ELEMENT APPROXIMATION OF CONVECTION DIFFUSION PROBLEMS USING GRADED MESHES RICARDO G. DURÁN AND ARIEL L. LOMBARDI Abstract. We consider the nmerical approximation of a model convection-diffsion
More informationFault Tolerance in Hypercubes
Falt Tolerance in Hypercbes Shobana Balakrishnan, Füsn Özgüner, and Baback A. Izadi Department of Electrical Engineering, The Ohio State University, Colmbs, OH 40, USA Abstract: This paper describes different
More informationWhat s New in AppSense Management Suite Version 7.0?
What s New in AMS V7.0 What s New in AppSense Management Site Version 7.0? AppSense Management Site Version 7.0 is the latest version of the AppSense prodct range and comprises three prodct components,
More informationMulti-lingual Multi-media Information Retrieval System
Mlti-lingal Mlti-media Information Retrieval System Shoji Mizobchi, Sankon Lee, Fmihiko Kawano, Tsyoshi Kobayashi, Takahiro Komats Gradate School of Engineering, University of Tokshima 2-1 Minamijosanjima,
More informationResolving Linkage Anomalies in Extracted Software System Models
Resolving Linkage Anomalies in Extracted Software System Models Jingwei W and Richard C. Holt School of Compter Science University of Waterloo Waterloo, Canada j25w, holt @plg.waterloo.ca Abstract Program
More informationApplication of Gaussian Curvature Method in Development of Hull Plate Surface
Marine Engineering Frontiers (MEF) Volme 2, 204 Application of Gassian Crvatre Method in Development of Hll Plate Srface Xili Zh, Yjn Li 2 Dept. of Infor., Dalian Univ., Dalian 6622, China 2 Dept. of Naval
More informationHardware-Accelerated Free-Form Deformation
Hardware-Accelerated Free-Form Deformation Clint Cha and Ulrich Nemann Compter Science Department Integrated Media Systems Center University of Sothern California Abstract Hardware-acceleration for geometric
More informationObject Pose from a Single Image
Object Pose from a Single Image How Do We See Objects in Depth? Stereo Use differences between images in or left and right eye How mch is this difference for a car at 00 m? Moe or head sideways Or, the
More informationSketch-Based Aesthetic Product Form Exploration from Existing Images Using Piecewise Clothoid Curves
Sketch-Based Aesthetic Prodct Form Exploration from Existing Images Using Piecewise Clothoid Crves Günay Orbay, Mehmet Ersin Yümer, Levent Brak Kara* Mechanical Engineering Department Carnegie Mellon University
More informationBlended Deformable Models
Blended Deformable Models (In IEEE Trans. Pattern Analysis and Machine Intelligence, April 996, 8:4, pp. 443-448) Doglas DeCarlo and Dimitri Metaxas Department of Compter & Information Science University
More informationComputer-Aided Mechanical Design Using Configuration Spaces
Compter-Aided Mechanical Design Using Configration Spaces Leo Joskowicz Institte of Compter Science The Hebrew University Jersalem 91904, Israel E-mail: josko@cs.hji.ac.il Elisha Sacks (corresponding athor)
More informationThe LS-STAG Method : A new Immersed Boundary / Level-Set Method for the Computation of Incompressible Viscous Flows in Complex Geometries
The LS-STAG Method : A new Immersed Bondary / Level-Set Method for the Comptation of Incompressible Viscos Flows in Complex Geometries Yoann Cheny & Olivier Botella Nancy Universités LEMTA - UMR 7563 (CNRS-INPL-UHP)
More informationAAA CENTER FOR DRIVING SAFETY & TECHNOLOGY
AAA CENTER FOR DRIVING SAFETY & TECHNOLOGY 2017 FORD MUSTANG PREMIUM CONVERTIBLE INFOTAINMENT SYSTEM* DEMAND RATING Very High Demand The Ford Mstang Convertible s SYNC 3 (version 2.20) infotainment system
More informationMETAMODEL FOR SOFTWARE SOLUTIONS IN COMPUTED TOMOGRAPHY
VOL. 10, NO 22, DECEBER, 2015 ISSN 1819-6608 ETAODEL FOR SOFTWARE SOLUTIONS IN COPUTED TOOGRAPHY Vitaliy ezhyev Faclty of Compter Systems and Software Engineering, Universiti alaysia Pahang, Gambang, alaysia
More informationImage Restoration Image Degradation and Restoration
Image Degradation and Restoration hxy Image Degradation Model: Spatial domain representation can be modeled by: g x y h x y f x y x y Freqency domain representation can be modeled by: G F N Prepared By:
More informationCS 4204 Computer Graphics
CS 424 Compter Graphics Crves and Srfaces Yong Cao Virginia Tech Reference: Ed Angle, Interactive Compter Graphics, University of New Mexico, class notes Crve and Srface Modeling Objectives Introdce types
More informationarxiv: v1 [cs.si] 12 Dec 2016
Connection Discovery sing Shared Images by Gassian Relational Topic Model Xiaopeng Li, Ming Cheng, James She HKUST-NIE Social Media Lab, Hong Kong University of Science & Technology, Hong Kong xlibo@connect.st.hk,
More informationDIRECT AND PROGRESSIVE RECONSTRUCTION OF DUAL PHOTOGRAPHY IMAGES
DIRECT AND PROGRESSIVE RECONSTRUCTION OF DUAL PHOTOGRAPHY IMAGES Binh-Son Ha 1 Imari Sato 2 Kok-Lim Low 1 1 National University of Singapore 2 National Institte of Informatics, Tokyo, Japan ABSTRACT Dal
More informationMaster for Co-Simulation Using FMI
Master for Co-Simlation Using FMI Jens Bastian Christoph Claß Ssann Wolf Peter Schneider Franhofer Institte for Integrated Circits IIS / Design Atomation Division EAS Zenerstraße 38, 69 Dresden, Germany
More informationTopic Continuity for Web Document Categorization and Ranking
Topic Continity for Web ocment Categorization and Ranking B. L. Narayan, C. A. Mrthy and Sankar. Pal Machine Intelligence Unit, Indian Statistical Institte, 03, B. T. Road, olkata - 70008, India. E-mail:
More informationNetworks An introduction to microcomputer networking concepts
Behavior Research Methods& Instrmentation 1978, Vol 10 (4),522-526 Networks An introdction to microcompter networking concepts RALPH WALLACE and RICHARD N. JOHNSON GA TX, Chicago, Illinois60648 and JAMES
More informationh-vectors of PS ear-decomposable graphs
h-vectors of PS ear-decomposable graphs Nima Imani 2, Lee Johnson 1, Mckenzie Keeling-Garcia 1, Steven Klee 1 and Casey Pinckney 1 1 Seattle University Department of Mathematics, 901 12th Avene, Seattle,
More informationIntroduction to Computational Manifolds and Applications
IMPA - Institto de Matemática Pra e Aplicada, Rio de Janeiro, RJ, Brazil Introdction to Comptational Manifolds and Applications Part 1 - Constrctions Prof. Marcelo Ferreira Siqeira mfsiqeira@dimap.frn.br
More informationSeismic trace interpolation with approximate message passing Navid Ghadermarzy and Felix Herrmann and Özgür Yılmaz, University of British Columbia
Seismic trace interpolation with approximate message passing Navid Ghadermarzy and Felix Herrmann and Özgür Yılmaz, University of British Colmbia SUMMARY Approximate message passing (AMP) is a comptationally
More informationMultiple-Choice Test Chapter Golden Section Search Method Optimization COMPLETE SOLUTION SET
Mltiple-Choice Test Chapter 09.0 Golden Section Search Method Optimization COMPLETE SOLUTION SET. Which o the ollowing statements is incorrect regarding the Eqal Interval Search and Golden Section Search
More informationMinimal Edge Addition for Network Controllability
This article has been accepted for pblication in a ftre isse of this jornal, bt has not been flly edited. Content may change prior to final pblication. Citation information: DOI 10.1109/TCNS.2018.2814841,
More informationFunctions of Combinational Logic
CHPTER 6 Fnctions of Combinational Logic CHPTER OUTLINE 6 6 6 6 6 5 6 6 6 7 6 8 6 9 6 6 Half and Fll dders Parallel inary dders Ripple Carry and Look-head Carry dders Comparators Decoders Encoders Code
More informationPipelined van Emde Boas Tree: Algorithms, Analysis, and Applications
This fll text paper was peer reviewed at the direction of IEEE Commnications Society sbject matter experts for pblication in the IEEE INFOCOM 007 proceedings Pipelined van Emde Boas Tree: Algorithms, Analysis,
More informationEMC ViPR. User Guide. Version
EMC ViPR Version 1.1.0 User Gide 302-000-481 01 Copyright 2013-2014 EMC Corporation. All rights reserved. Pblished in USA. Pblished Febrary, 2014 EMC believes the information in this pblication is accrate
More informationTOWARD AN UNCERTAINTY PRINCIPLE FOR WEIGHTED GRAPHS
TOWARD AN UNCERTAINTY PRINCIPLE FOR WEIGHTED GRAPHS Bastien Pasdelop, Réda Alami, Vincent Gripon Telecom Bretagne UMR CNRS Lab-STICC name.srname@telecom-bretagne.e Michael Rabbat McGill University ECE
More informationA Hybrid Weight-Based Clustering Algorithm for Wireless Sensor Networks
Open Access Library Jornal A Hybrid Weight-Based Clstering Algorithm for Wireless Sensor Networks Cheikh Sidy Mohamed Cisse, Cheikh Sarr * Faclty of Science and Technology, University of Thies, Thies,
More informationToday s Lecture. Software Architecture. Lecture 27: Introduction to Software Architecture. Introduction and Background of
Today s Lectre Lectre 27: Introdction to Software Architectre Kenneth M. Anderson Fondations of Software Engineering CSCI 5828 - Spring Semester, 1999 Introdction and Backgrond of Software Architectre
More informationA choice relation framework for supporting category-partition test case generation
Title A choice relation framework for spporting category-partition test case generation Athor(s) Chen, TY; Poon, PL; Tse, TH Citation Ieee Transactions On Software Engineering, 2003, v. 29 n. 7, p. 577-593
More informationRequirements Engineering. Objectives. System requirements. Types of requirements. FAQS about requirements. Requirements problems
Reqirements Engineering Objectives An introdction to reqirements Gerald Kotonya and Ian Sommerville To introdce the notion of system reqirements and the reqirements process. To explain how reqirements
More informationarxiv: v1 [cs.cr] 3 Jun 2017
Spectrm-based deep neral networks for frad detection arxiv:1706.00891v1 [cs.cr] 3 Jn 2017 ABSTRACT Shhan Yan Tongji University 4e66@tongji.ed.cn Jn Li University of Oregon lijn@cs.oregon.ed In this paper,
More informationAppearance Based Tracking with Background Subtraction
The 8th International Conference on Compter Science & Edcation (ICCSE 213) April 26-28, 213. Colombo, Sri Lanka SD1.4 Appearance Based Tracking with Backgrond Sbtraction Dileepa Joseph Jayamanne Electronic
More informationSemantic segmentation is the task of labeling every pixel in
Deep Learning for Visal Understanding: Part Anrag Arnab, Shai Zheng, Sadeep Jayasmana, Bernardino Romera-Paredes, Måns Larsson, Alexander Kirillov, Bogdan Savchynskyy, Carsten Rother, Fredrik Kahl, and
More informationConstructing and Comparing User Mobility Profiles for Location-based Services
Constrcting and Comparing User Mobility Profiles for Location-based Services Xihi Chen Interdisciplinary Centre for Secrity, Reliability and Trst, University of Lxemborg Jn Pang Compter Science and Commnications,
More information3-D SURFACE ROUGHNESS PROFILE OF 316-STAINLESS STEEL USING VERTICAL SCANNING INTERFEROMETRY WITH A SUPERLUMINESCENT DIODE
IMEKO 1 TC3, TC5 and TC Conferences Metrology in Modern Context November 5, 1, Pattaya, Chonbri, Thailand 3-D SURFACE ROUGHNESS PROFILE OF 316-STAINLESS STEEL USING VERTICAL SCANNING INTERFEROMETRY WITH
More informationPavlin and Daniel D. Corkill. Department of Computer and Information Science University of Massachusetts Amherst, Massachusetts 01003
From: AAAI-84 Proceedings. Copyright 1984, AAAI (www.aaai.org). All rights reserved. SELECTIVE ABSTRACTION OF AI SYSTEM ACTIVITY Jasmina Pavlin and Daniel D. Corkill Department of Compter and Information
More informationPlenoPatch: Patch-based Plenoptic Image Manipulation
1 PlenoPatch: Patch-based Plenoptic Image Maniplation Fang-Le Zhang, Member, IEEE, Je Wang, Senior Member, IEEE, Eli Shechtman, Member, IEEE, Zi-Ye Zho, Jia-Xin Shi, and Shi-Min H, Member, IEEE Abstract
More informationCurves and Surfaces. CS 537 Interactive Computer Graphics Prof. David E. Breen Department of Computer Science
Crves and Srfaces CS 57 Interactive Compter Graphics Prof. David E. Breen Department of Compter Science E. Angel and D. Shreiner: Interactive Compter Graphics 6E Addison-Wesley 22 Objectives Introdce types
More informationMultiView: Improving Trust in Group Video Conferencing Through Spatial Faithfulness David T. Nguyen, John F. Canny
MltiView: Improving Trst in Grop Video Conferencing Throgh Spatial Faithflness David T. Ngyen, John F. Canny CHI 2007, April 28 May 3, 2007, San Jose, California, USA Presented by: Stefan Stojanoski 1529445
More informationPlenoPatch: Patch-based Plenoptic Image Manipulation
1 PlenoPatch: Patch-based Plenoptic Image Maniplation Fang-Le Zhang, Member, IEEE, Je Wang, Senior Member, IEEE, Eli Shechtman, Member, IEEE, Zi-Ye Zho, Jia-Xin Shi, and Shi-Min H, Member, IEEE Abstract
More informationAlliances and Bisection Width for Planar Graphs
Alliances and Bisection Width for Planar Graphs Martin Olsen 1 and Morten Revsbæk 1 AU Herning Aarhs University, Denmark. martino@hih.a.dk MADAGO, Department of Compter Science Aarhs University, Denmark.
More informationDiscrete Cost Multicommodity Network Optimization Problems and Exact Solution Methods
Annals of Operations Research 106, 19 46, 2001 2002 Klwer Academic Pblishers. Manfactred in The Netherlands. Discrete Cost Mlticommodity Network Optimization Problems and Exact Soltion Methods MICHEL MINOUX
More informationImproved Estimation of Transmission Distortion for Error-resilient Video Coding
1 Improved Estimation of Transmission Distortion for Error-resilient Video Coding Zhifeng Chen 1, Peshala Pahalawatta 2, Alexis Michael Torapis 2, and Dapeng W 1,* 1 Department of Electrical and Compter
More informationSmoothing Low SNR Molecular Images Via Anisotropic Median-Diffusion
Smoothing Low SNR Moleclar Images Via Anisotropic Median-Diffsion Jian Ling 1, Member, IEEE, and Alan C. Bovik 2, Fellow, IEEE 1 Sothwest Research Institte Bioengineering Department 6220 Clebra Road San
More informationTowards applications based on measuring the orbital angular momentum of light
CHAPTER 8 Towards applications based on measring the orbital anglar momentm of light Efficient measrement of the orbital anglar momentm (OAM) of light has been a longstanding problem in both classical
More informationAn Adaptive Strategy for Maximizing Throughput in MAC layer Wireless Multicast
University of Pennsylvania ScholarlyCommons Departmental Papers (ESE) Department of Electrical & Systems Engineering May 24 An Adaptive Strategy for Maximizing Throghpt in MAC layer Wireless Mlticast Prasanna
More informationDVR 630/650 Series. Video DVR 630/650 Series. 8/16-Channel real-time recording with CIF resolution. Flexible viewing with two monitor outputs
Video DVR 630/650 Series DVR 630/650 Series 8/16-Channel real-time recording with resoltion Flexible viewing with two monitor otpts Remote viewing, playback, control, and configration Easy Pan/Tilt/Zoom
More informationStereo Matching and 3D Visualization for Gamma-Ray Cargo Inspection
Stereo Matching and 3D Visalization for Gamma-Ray Cargo Inspection Zhigang Zh *ab, Y-Chi H bc a Department of Compter Science, The City College of New York, New York, NY 3 b Department of Compter Science,
More informationNew-Sum: A Novel Online ABFT Scheme For General Iterative Methods
New-Sm: A Novel Online ABFT Scheme For General Iterative Methods Dingwen Tao (University of California, Riverside) Shaiwen Leon Song (Pacific Northwest National Laboratory) Sriram Krishnamoorthy (Pacific
More informationA GENERIC MODEL OF A BASE-ISOLATED BUILDING
Chapter 5 A GENERIC MODEL OF A BASE-ISOLATED BUILDING This chapter draws together the work o Chapters 3 and 4 and describes the assembly o a generic model o a base-isolated bilding. The irst section describes
More informationStaCo: Stackelberg-based Coverage Approach in Robotic Swarms
Maastricht University Department of Knowledge Engineering Technical Report No.:... : Stackelberg-based Coverage Approach in Robotic Swarms Kateřina Staňková, Bijan Ranjbar-Sahraei, Gerhard Weiss, Karl
More informationFeatures. ICMS Integrated Corrosion Management System
ICMS Integrated Corrosion System Featres Total Corrosion Data Data Exhange with DCS/PCS/SCADA Systems Correlate Corrosion & Process Data Enables Highly Cost-Effective Asset Designed Specifically for Corrosion
More information5.0 Curve and Surface Theory
5. Cre and Srface Theor 5.1 arametric Representation of Cres Consider the parametric representation of a cre as a ector t: t [t t t] 5.1 The deriatie of sch a ector ealated at t t is gien b t [ t t t ]
More informationA Frequency-optimized Discontinuous Formulation for Wave Propagation Problems
th Flid Dynamics Conference and Ehibit 8 Jne - Jly, Chicago, Illinois AIAA -55 A Freqency-optimized Discontinos Formlation for Wave Propagation Problems Yi Li and Z. J. Wang Department of Aerospace Engineering
More informationTemporal Light Field Reconstruction for Rendering Distribution Effects
Temporal Light Field Reconstrction for Rendering Distribtion Effects Jaakko Lehtinen NVIDIA Research PBRT, 16 spp, 403 s Timo Aila NVIDIA Research Jiawen Chen MIT CSAIL PBRT, 256 spp, 6426 s Samli Laine
More informationIdentification of the Symmetries of Linear Systems with Polytopic Constraints
2014 American Control Conference (ACC) Jne 4-6, 2014. Portland, Oregon, USA Identification of the Symmetries of Linear Systems with Polytopic Constraints Clas Danielson and Francesco Borrelli Abstract
More informationSECOND order computational methods commonly employed in production codes, while sufficient for many applications,
th AIAA Comptational Flid Dynamics Conference 7-3 Jne, Honoll, Hawaii AIAA -384 Active Fl Schemes for Systems Timothy A. Eymann DoD HPCMP/CREATE Kestrel Team, Eglin AFB, FL 354 Philip L. Roe Department
More informationNormalized averaging using adaptive applicability functions with application in image reconstruction from sparsely and randomly sampled data
Normalized averaging sing adaptive applicability fnctions with application in image reconstrction from sparsely and randomly sampled data Tan Q. Pham, Lcas J. van Vliet Pattern Recognition Grop, Faclty
More informationData/Metadata Data and Data Transformations
A Framework for Classifying Scientic Metadata Helena Galhardas, Eric Simon and Anthony Tomasic INRIA Domaine de Volcea - Rocqencort 7853 Le Chesnay France email: First-Name.Last-Name@inria.fr Abstract
More informationCohesive Subgraph Mining on Attributed Graph
Cohesive Sbgraph Mining on Attribted Graph Fan Zhang, Ying Zhang, L Qin, Wenjie Zhang, Xemin Lin QCIS, University of Technology, Sydney, University of New Soth Wales fanzhang.cs@gmail.com, {Ying.Zhang,
More informationTowards Understanding Bilevel Multi-objective Optimization with Deterministic Lower Level Decisions
Towards Understanding Bilevel Mlti-objective Optimization with Deterministic Lower Level Decisions Ankr Sinha Ankr.Sinha@aalto.fi Department of Information and Service Economy, Aalto University School
More informationA personalized search using a semantic distance measure in a graph-based ranking model
Article A personalized search sing a semantic distance measre in a graph-based ranking model Jornal of Information Science XX (X) pp. 1-23 The Athor(s) 2011 Reprints and Permissions: sagepb.co.k/jornalspermissions.nav
More informationLayered Light Field Reconstruction for Defocus Blur
Layered Light Field Reconstrction for Defocs Blr KARTHIK VAIDYANATHAN, JACOB MUNKBERG, PETRIK CLARBERG and MARCO SALVI Intel Corporation We present a novel algorithm for reconstrcting high qality defocs
More informationInteligencia Artificial. Revista Iberoamericana de Inteligencia Artificial ISSN:
Inteligencia Artificial. Revista Iberoamericana de Inteligencia Artificial ISSN: 1137-3601 revista@aepia.org Asociación Española para la Inteligencia Artificial España Zaballos, Lis J.; Henning, Gabriela
More informationScientific Computing on Bulk Synchronous Parallel Architectures
Scientific Compting on Blk Synchronos Parallel Architectres R. H. Bisseling Department of Mathematics, Utrecht University and W. F. McColl y Programming Research Grop, Oxford University April 27, 1994
More informationEUCLIDEAN SKELETONS USING CLOSEST POINTS. Songting Luo. Leonidas J. Guibas. Hong-Kai Zhao. (Communicated by the associate editor name)
Volme X, No. 0X, 200X, X XX Web site: http://www.aimsciences.org EUCLIDEAN SKELETONS USING CLOSEST POINTS Songting Lo Department of Mathematics, University of California, Irvine Irvine, CA 92697-3875,
More informationarxiv: v1 [cs.cg] 27 Nov 2015
On Visibility Representations of Non-planar Graphs Therese Biedl 1, Giseppe Liotta 2, Fabrizio Montecchiani 2 David R. Cheriton School of Compter Science, University of Waterloo, Canada biedl@waterloo.ca
More informationCombined cine- and tagged-mri for tracking landmarks on the tongue surface
INTERSPEECH 2015 Combined cine- and tagged-mri for tracking landmarks on the tonge srface Honghao Bao 1, Wenhan L 1, Kiyoshi Honda 2,*, Jiango Wei 1, Qiang Fang 3, Jianw Dang 2, 4 1 School of Compter Software,
More information