GROUPING OBJECTS BASED ON THEIR APPEARANCE
|
|
- Edwin Oswald Merritt
- 5 years ago
- Views:
Transcription
1 Internatonal Journal of Artfcal Intellgence & Applcatons (IJAIA), Vol. 4, No. 4, July 203 GROUPING OBJECTS BASED ON THEIR APPEARANCE Altamrano Robles Lus Carlos, Zacarías Flores Fernando 2, De León Gómez Oscar Antono 3 and Sánchez López Abraham 4,2,3,4 ComputerScence, AutonomousUnversty of Puebla (BUAP), Puebla, Méxco fzflores@yahoo.com.mx, altamrano@cs.buap.mx, oleongomez@gmal.com asanchez@cs.buap.mx ABSTRACT The use of clusterng algorthms for partton to establsh a herarchcal structure n a lbrary of object models based on appearance s deployed. The man contrbuton corresponds to a novel and ntutve algorthm for clusterng of models based on ther appearance, closer to human behavor. Ths dvdes the complete set nto subclasses. Immedately, dvdes each of these n a number of predefned groups to complete the levels of herarchy that the user wants. Whose man purpose s to obtan a compettve classfcaton compared to what a human would perform. KEYWORDS Vson, Appearance, Models,Clusterng and Herarchy. INTRODUCTION At the begnnng of object recognton through dgtal mages, was used the well-known geometrc paradgm. Ths was manly to the wde avalablty of algorthms for detectng edges and corners. Despte the good results obtaned, s shown that usng only the way to recognze objects has ts lmtatons. For nstance, f we have two equal cubes n shape and sze but one red and the other blue, all systems based on geometrc characterstcswll say that t s the same object because both objects have the same shape. Ths happens because there s a substantal nformaton loss,nherent to the object n queston. Other sde, appearance-based approach has been proposed as an alternatve to geometrcal approaches. However, ths approach also has advantages (ncreases recognton rates, prncpally) and dsadvantages (depends on llumnaton condtons, pose and other parameters, y of course, requres a lot of mages n order to buld models). But, snce appearance models offer a promses alternatve to tradtonal models, they are been studyng wdely[0, 3, 4]. Because t s not easy to establsh a herarchy among models learnt by the machne, most authors prefer to works n dfferent drectons, such that: mprove recognton rate, mprove lost of nformaton, compresson methods (PCA, for example); one excepton are works made by Nelson and others [, 2]. However, they use just contour of objects for establsh any knd of herarchy no clear and depends of prmtves prevously selected. In ths paper, we present a novel and ntutve approach to establsh a herarchy whch does not depend of any prmtve or uses just contour of objects. Instead, we use all vsual nformaton about object appearance: of course contour, but also, color, texture, pose, etc. Our approach s ntutve because s closer to human behavor : our proposal uses a not supervsed approach n DOI : 0.52/jaa
2 Internatonal Journal of Artfcal Intellgence & Applcatons (IJAIA), Vol. 4, No. 4, July 203 order to buld several dfferent clusters and each one has an explanaton n terms of human behavor. They establsh a herarchy begnnng wth objects and fnalzng wth, 2, or the number of trees that we gve a pror, lke we want to dvde a group between men and women: we establsh the only parameter for our algorthm equal to 2 and we can obtan ths groupng wth 2 classes, but whch classes? Answer: several dfferent 2 classes, one par for each possble dualty: men/women, black/whte, short/tall, etc., lke human does. So, we developeda more natural object classfer that consders the followng aspects: Approprate ways to dfferentate between objects modeled by sets of dgtal mages (measures of smlarty/dssmlarty). Clusterng algorthms able to put objects models n class, Classfcaton based not only on ponts, as the tradtonal algorthms. Dfferent experments (shown below) shows that our approach exhbts results close to human results, n the sense above explaned. 2. PRELIMINARIES The followng subsectons present basc concepts about processng objects. 2.. Models based on appearance The object appearance s gven by: the combnaton of shape, ts reflectance propertes, and ts poston n front the camera n scene and the llumnaton present n ths [6]. Should be noted that the frst two are nherent to the object, however, n the thrd, the object s poston n front of the sensor and llumnaton may vary from scene to scene[6].lkewse, s clear that ths defnton mples that the appearance of an object at a partcular tme can be derved from a dgtal mage n whch that element s captured n a specfc scene. Smlarly, then may be used a set of mages to generate a model of the object under dfferent crcumstances. In fgure, you can see a graphcal representaton of an object's appearance[6]. Fgure. An object and ts model Usually, appearance-based approach works as follows: are taken a number of mages of the object you want to buld the model,an mage for each of the possble ways n whch the object s ntended that may appear n an mageand combned wth the varous lghtng condtons that mpnge upon t.next, these mages are stored n a format chosen (gf, tff, bmp, etc.),for later use; these mages themselves can form the model of the object or, they can extract some features that you thnk descrbe the object accurately, thus, these consttute the model. Fnally, the recognton process tself,s to compare the model wth a new mage that s lkely to contan the object (of course, ths latter process requres some extra processng on the new mage, such as object segmentaton of nterest n the mage, nterpolatng between mages, etc.) Creatng models usng Hotellng transform Due to the need for a large number of mages to a sngle object model and n practce t s common to have large quanttes of dfferent object, we need to fnd a compressed representaton 78
3 Internatonal Journal of Artfcal Intellgence & Applcatons (IJAIA), Vol. 4, No. 4, July 203 of all the mages. Normally, It s possble to use a common technque n pattern recognton named Hotellng transform or commonly known as prncpal component analyss [6]. Fgure. Egen-mage and a reconstructed mage The general procedure s as follows: [6], [4]:. Each mage s stacked by columns 2. Each mage s normalzed by dvdng by ther respectve norm 3. We obtan the average vector of all mages 4. Each mage s subtracted from the average 5. Matrx P s obtaned, where each column s one of the vectors prevously calculated 6. We obtan the covarance matrx mpled. T Q = P * P 7. Are obtaned the egenvalues and egenvectors of whch are obtaned correspondng to the covarance matrx, λ = λ e = λ /2 Pe Next, we obtan the set of egenvectors or egenmages, these vectors and egenvalues form a low- number of dmensonal egenspace. The dmenson of the egenspace correspondng to the prncpal components you want to use. 20 are normally suffcent as maxmum. We consder that ths should nclude all the mages of all objects. Immedately, we solve the followng system of equatons ea = P We obtan the matrx A contanng necessary coeffcents to reconstruct any mage of any pattern.ths compresson process shows nsgnfcant losses of nformaton for problem that concerns us. In fgure 3 the result of the manfolds extracton s shown for two objects and, n fgure 2 some of the egenmages can be observed obtaned by means of ths method Smlarty measures Typcally when you have a set of models and a test object but wth dfferent vews that contan some overlap n these, global data can dffer greatly the strategy s to fnd the common vew and measure the smlarty of those parts of the data. Then, we defne the dstance between manfolds as follows: d( M, M 2) = mn d( C, M 2) () 79
4 Internatonal Journal of Artfcal Intellgence & Applcatons (IJAIA), Vol. 4, No. 4, July 203 where = r, s the number of patches that make up the frst manfold. Thus, the dstance between manfolds s calculated as the best patch smlarty adjusted. For ths calculaton t s necessary to obtan other measures presented below Measure dstance between subspaces The measure of dstance between subspaces s denoted by d( S, S2) and currently there s not a unfed defnton. In ths paper we used the concept of prncpal angles for good performance [].Wth these concepts are defned the followng measures dstance. Man dea s that the nonlnear manfolds are composed of lnear subspaces and the dstance of thesee components defnes the smlarty between two manfolds. Formally denoted as: M = C, C2,, Cm where M s a manfold consstng of lnear subspaces. Once parttoned manfolds, t s only necessary to calculate the dstance among subspaces to proceed. A way to make t s to calculate the man angles that reflect the varatons among the subspaces wthout takng the dataa nto account []. Other methods take nto account the smlarty through the averages of each local pattern n a smlarty measure among exemplary. It s easy to understand that to use the coalton of both measures s better for the problem than we want to solve Measure dstance between subspaces The dstance from a subespacee to a manfold s denoted as d( S, M ) and s defned n the followng way d( S, M ) = mn d( S, C ) Ths means to fnd the mnmum dstance between a subspace and the lneal components of manfold Dstance from a manfold to a manfold Fgure 3. Examples calculated C M The dstance of a manfold to a manfold s denoted as d( M, M 2) and s defned n the followng way: d( M, M ) = mn d( C, M ) 2 2 C M = mn mn d( C, C ) C M C j M 2 j Clearly, the smlarty between M and M 2 s calculated as smlarty of local models that better they adapt. In [Wang2008] ths dstance s used n recognton of faces through sets of mages. In ths paper we wll use t to get groupng of models of objects based apparently. 80
5 Internatonal Journal of Artfcal Intellgence & Applcatons (IJAIA), Vol. 4, No. 4, July Dstance among manfolds As t was already defned prevously, Matrx A contans manfolds of the objects n the lbrary. k These defne a trajectory parameterzed n R where k s the number of selected man components. Man problem of groupng models of objects based on apparently s that t should be treated wth ths compact representaton. It s essental to fnd a measure or set of measures of smlarty are necessary to make a classfcaton of objects smlar to what a human would. In the lterature t s common to fnd smlarty measures between objects represented as a vector (column orrow) wthd characterstcs, stored n database of N objects (dataset) [2]. The strength of the relatonshp between two ponts of a dataset s ndcated by a smlarty coeffcent, the larger ths coeffcent, sad objects are related more. Next, we defne the dstance between manfolds as a measure of smlarty between models based on appearance Maxmal Lneal Patch After obtanng the Manfolds, as mentoned n prevous secton, s necessary that each manfold s dvded nto a seres of subspaces called maxmal lneal patches.the concept of maxmal lneal patche (MLP) s defned as a set of ponts defnng a maxmal lnear subspace, where the lnear perturbaton of these ponts s gven naturally as the deflecton between the Eucldean dstance and geodesc dstance. Fgure 4. Examples of Manfolds obtaned from COIL-20 The man dea s to splt the data set nto several MLPs such that sets of mages contanng very smlar to each other. These patches are calculated from a random pont of dataset untl the condton of lnearty s lost. Ths procedure guarantees the local lnearty of models and adaptvely controls the number of models calculated. Formally, [7] be X = x,..., xn a data set, n our case are ponts that make up manfolds, where x D R and Ns the number of ponts n X. we dvde X nto a dsjont collecton of MLPs X = N = C, C C = ( j,, j =,2,, m) j m ( ) ( ) ( ) 2 N = C = x, x,, x ( N = N) n where m s the number of patches and N s the number of ponts n the patch. U C such that: 8
6 Internatonal Journal of Artfcal Intellgence & Applcatons (IJAIA), Vol. 4, No. 4, July 203 Fgure 5. Parttonng of a model based on appearance For parttonng s necessary to calculate the Eucldean dstance matrx DE and the matrx of geodesc dstances D G. Calculaton of the MLPs s used as an ntermedate step to fnd the dstance between two manfolds. Ths dstance has been used n face recognton through mage sets [7]. As mentoned above, the dea s to splt the set of mages n several patches whch are embedded n a lnear subspace. The followng shows the algorthm developed. Intally =, C =, X T =, X T =, X R = X.. Whle XT 2. Randomly select a pont n X R as seed ( ) ( ) 2.. C = { x }, X = X { x } R R 2.2. For x C m x, Identfy each of k nearest neghbors xc as canddate ( ) ( ) If x c X R and ( DG ( xc, xn ) / DE ( xc, xn )) θ C = CU { xc} X R = X R { xc} EndIf EndFor 3 XT = UCj, XR = X XT; +, C = EndWhle j= ( ) After calculatng a lnear patch, ts average s calculated. Ths s taken as a representatve of set and summarzes nformaton of selected patch that s formed by numercally smlar mages. The mportance of these representatves s reflected n the reducton of the number of mages requred to represent the object appearance and s taken nto account for the calculaton of the smlarty of a complete model wth other Measure dstance between subspaces by varatons For two subspaces C and C j we defne a dstance measure by varatons as ther average canoncal correlatons as follows: d ( C, C ) = r / σ V j k k = r 82
7 Internatonal Journal of Artfcal Intellgence & Applcatons (IJAIA), Vol. 4, No. 4, July 203 Where σ σ r are the canoncal correlatons and are obtaned by the method descrbed n [] and r s the mnmum szes of lnear subspaces Measure dstance between copes To complement the above measure s defned as measured between copes of the duplcate correlaton e and e j : (, ) Smlarty between models based on appearance Fnally, the dstance between subspaces s obtaned as the weghted average of the two prevous measures as follows: d( C, C ) = ( α) d ( C, C ) + αd ( C, C ) j E j V j When comparng two sets of mages the frst shows how smlar the appearance of objects, whle the other s so close that common varaton trends of the two sets. Thus, we can easly defne the dstance between manfolds as n equaton () Clusterng algorthms The K-means algorthm s one of the most used due to ts smplcty and speed. The core dea s to group the data repeatedly around the centrods calculated from an ntal allocaton of the data n a fxed number of k groups. The class membershp changes accordng to the evaluaton of an functon error untl ths error does not change sgnfcantly or data belongng ether. The error functon s defned as follows [5]: Let D be a data set wth n nstances, and be C, C2,, Ck k cluster dsjont from D. The error functon s calculated as: k = = x C E d( x, µ ( C )) where µ ( C ) s the centrod of cluster C and d (.,.) s the Eucldean dstance typcally. The Keywords secton begns wth the word, Keywords n 3 pt. Tmes New Roman, bold talcs, Small Caps font wth a 6pt. spacng followng. There may be up to fve keywords (or short phrases) separated by commas and sx spaces, n 0 pt. Tmes New Roman talcs. An 8 pt. lne spacng follows Spectral clusterng Graph theory can be used to partton data. In these technques we use a measure of smlarty between the data to form a weghted adjacency matrx. Intutvely, we want fnd a partton of the graph where the edges of the nodes that are n dfferent groups have low weght and have hgh weghts when they are n the same group [3], [8]. Let G = ( V, E) be anundrected graph wth a set of vertcesv = v,, vn. We assume that each edge has an assocated postve weght. The heavy adjacency matrx W = w,, j =,, n, where f w j = 0, vertces and j are not connected. Snce G s not addressed must be w j = w j. Furthermore, degree of vertex s defned as follow: j 83
8 Internatonal Journal of Artfcal Intellgence & Applcatons (IJAIA), Vol. 4, No. 4, July 203 d n = j= The dagonal matrx havng the degrees of the vertces s called degree matrx and s denoted as D. Lkewse defne heavy adjacency matrx and degree matrx. Next, we defne the unnormalzed Laplacan matrx of a graph as: L = D W Ths matrx has several nterestng propertes and. s postve sem-defnte and the number of egenvectors of ths matrx s drectly related to the number of connected components of the graph [3]. Besdes, ths method s closely related to the method kmeans kernelzado and cuts of graphs normalzed [9] Normalzed Laplacan Matrces In the lterature there are two standard Laplacan matrces. Each one s closely related to each other and s defned as: /2 /2 L = D LD L sym rw = D L Matrx L sym s denoted so because t s symmetrcal and matrx L rw s ntmately related to random walks. As n the case of the non-normalzed matrx, the multplcty of the frst Laplacan egenvalue of the matrx s related to the number of connected components of the graph of heavy adjacences [8].In ths study we used the non-standard matrx and matrx L for the calculaton of the groupngs. 3. Our proposal The ntenton of usng dfferent clusterng methods for parttonng s to acheve an ntal dvson of the set of models and then apply the same algorthm agan, dvdng the classes obtaned n the prevous step n another herarchcal level. Besdes usng the smlarty measure manfolds also be transformed so that they can be manpulated as a sngle object and not as a set of ponts. Ths was acheved by calculatng the centrod of each manfold and calculatng the standard devaton.ths defnes a sphere that contans almost all ponts of the manfold. Ths defnes then a sphere that contans almost all ponts of the manfold. Immedately show the method generates a herarchy of a set of object models. The proposed method can be summarzed as follows:. It creates an ntal partton of the orgnal set. 2. Each class obtaned s dvded k nto groups, the parameter k s user defned for each level of the herarchy 3. Results are stored 4. Repeat from step 2 to complete the requred levels Fnally, we obtaned a tree structure that shows how have been dvded every class obtaned,n each of the teratons calculated. The rates of each object dsplayed n the tree nodes. 4. TEST AND RESULTS Tests were conducted on the classc-lookng set of models suppled by Columba Unversty, COIL-20. The results obtaned were compared wth those valdated by a user vsually, ths n w j sym 84
9 Internatonal Journal of Artfcal Intellgence & Applcatons (IJAIA), Vol. 4, No. 4, July 203 order to check that the classfcaton obtaned s smlar or not. The results obtaned are presented n a tree structure where nodes tags are nterpreted accordng to the followng scheme: Fgure 6. scheme of object ndex The system was executed several tmes for each of the algorthms and wth dfferent numbers of levels and Parttons by level, found emprcally that n the partcular case of the lbrary used, three levels and no more than three parttons per group are suffcent to acheve good rankngs. In fgure 7, you can see one of the groups calculated by the algorthm. Note that the classfcaton of objects 4.8 and 6 places n the same class and ths wll not change usng other methods proposed, change only the objects that are grouped nto hgher levels of the herarchy Ths s because the calculaton of the dstance between manfolds gves greater weght to objects appearance so those n whch the domnant gray tone s very smlar are grouped n the same class as evdenced n the case of the wood peces. Ths measure can also reflect some specal characterstcs as how objects n the case of fgure 7 the majorty shares (objects are hgher than wde, almost the same gray and have almost the same knd of symmetry) whch may be an ndcaton that the results can be mproved somewhat by mxng these results to those obtaned usng only borders. Make a change of models representaton calculatng the centrods allows us to use unmodfed exstng algorthms but at the same tme preserves all the nnate drawbacks of these same. To avod the partcular algorthms derved usng k-means, we wll use the spectral clusterng on the centrods. 4.. Spectral clusterng by manfolds centrods Below are shown some of the results obtaned n testng algorthm. For ths partcular test we used three levels and a vector (3,3,2) v =. Fgure 7. One group obtaned wth the proposal. 85
10 Internatonal Journal of Artfcal Intellgence & Applcatons (IJAIA), Vol. 4, No. 4, July CONCLUSIONS AND FUTURE WORK Although algorthm s the type of dvde and conquer, ts complexty s polynomal snce the algorthms used (kmeans, spectral) assure us that the groups formed are very smlar to each other and very dfferent from the other groups thereby avodng combnatoral check what s the best partton. However, when an object has been msclassfed, s mpossble to assgn another class and n the best case can be segregated from the group to the next level calculated. On the other hand,usng a measure of smlarty between objects, results were very smlar to those obtaned by other methods snce the model lbrary allows. In addton, t was found that the system s able to match the results obtaned by the user and also s able to detect smlartes not evdent at a glance.in fgure 8 we show some results obtaned wth our proposal. The results have demonstrated the usefulness of smlarty measures alternatve to classc and the data processng as a means to obtan better results. However, much work remans to be done n ths feld n whch ths research ams to poneer.some thngs to ncorporate nto future research are: Fgure 8. An unexpected group?. Use models based on appearance only for objects borders. Wth the am of usng as a complementary method. Select dfferent algorthms to apply to each level. The results may be mproved f dfferent algorthms are used at each level of the herarchy. Testng wth other models lbrares and compare results. It s essental to experment wth dfferent lbrares to delmt the scope of the proposal developed n ths work and and modfy t to make t applcable to more general problems. Automatcally analyze the results. To quanttatvely compare the classfcatons obtaned by the system wth those made by a human. Improvng the representaton of herarchcal structure. Explore the possblty of calculatng the tree dendrogram. Investgate possble applcatons of ths proposal. All these ponts can be taken as a bass for further work n order to fnd new applcatons, detect problems andand generate other proposals that serve to extend knowledge currently hasn the areas of computer vson and clusterng algorthms. ACKNOWLEDGEMENTS The authors would lke to thank to Autonomous Unversty of Puebla for ther fnancal support.óscar De León Gómez would lke to thank to CONACyT by the scholarshp No REFERENCES [] Å.BjörckandG.H.Golub.Numercal methods forcomputng angles between lnear subspaces. Mathematcs of Computaton, 27:579594,
11 Internatonal Journal of Artfcal Intellgence & Applcatons (IJAIA), Vol. 4, No. 4, July 203 [2] Clusterng. IEEE Press, [3] Ulrke Luxburg. A tutoral on spectral clusterng. Statstcs and Computng, 7(4):395 46, December [4] Taylor C.J. Cootes,T.F. Statstcal model of appearance for computer vson. Techncal report, Imagng Scence Bomedcal Engneerng, Unversty of Manchester,March, [5] Janhong Wu Guojun Gan, Chaoqun Ma. Data Clusterng: Theory, Algorthms, and applcatons. ASA-SIAM Seres on Statstcs and Appled Probablty, SIAM, [6] Nene,Nayar, Murase. Parametrc appearance representaton. Nayar, S. K., Poggo, T., (Eds.) Early Vsual Learnng,OxfordUnversty Press. NewYork, :3 60, 996. [7] Xln Chen WenGao Rupng Wang, Shguang Shan. Manfold-manfold dstance wth applcaton to face recognton based on mage set. IEEE,2008. [8] Francs R. Bach y Mchael I. Jordan.Learnng spectral clusterng.techncal Report UCB/CSD , EECS Department, Unverstyof Calforna, Berkeley, Jun [9] Inderjt S. Dhllon. Kernel k-means, spectral clusterng and normalzed cuts. In pp ACM Press, [0] Mundy, J., Lu, A., Pllow, N., et. al., An expermental comparson ofappearance and geometrc model based recognton, Object Representatonn Computer Vson II, n Proceedngs of ECCV96 Internatonal Workshop,Cambrdge U.K., pp Aprl, 996. [] Andrea Selnger and Randal C. Nelson, ``A Perceptual Groupng Herarchy for Appearance-Based 3D Object Recognton'', Computer Vson and Image Understandng, vol. 76, no., October 999, pp [2] Randal C. Nelson and Andrea Selnger ``A Cubst Approach to Object Recognton'', Internatonal Conference on Computer Vson (ICCV98), Bombay, Inda, January 998, [3] Zaman Khan and Adnan Ibraheem. Hand Gesture recognton: A lterature revew. Internatonal journal of artfcal Intellgence & Applcatons, Vol 3, No. 4, July 202. [4] Bm Jan, M.K. Gupta and JyotBhart.Effcent rs recognton algorthm usng method of moment.internatonal journal of artfcal Intellgence & Applcatons, Vol 3, No. 5, September 202. Authors Bography Dr. Fernando Zacaras s full tme professor n the Computer Scence Department at Autonomous Unversty of Puebla. He has drected and partcpated n research projects and development n ths area from 995, wth results that they have been reported n more than 50 publcatons of nternatonal level. And one of the most mportant s Answer set programmng and applcatons. Conacyt project, Reference number: A. He has been a member of several nternatonal commttees such as: The IEEE Latn Amerca Transacton, IAENG Internatonal journal of Computer Scence, Internatonal conference on Advances n Moble Computng and Multmeda, etc. Dr. Lus Carlos Altamrano receved hs Ph.D. degree at Natonal Polytechnc Insttute (Méxco) n Hs nterest areas nclude: Computer Vson, Image Processng, Remote Sensng and Artfcal Intellgence. He has partcpated n several dfferent projects ncludng: Ol projects and Computer Vson Projects. Currently, he s full tme professor n the Computer Scence Department at Autonomous Unversty of Puebla and head of Postgraduate Department at the same Insttuton. Dr. Abraham Sánchez López s full tme professor n the Computer Scence Department at Autonomous Unversty of Puebla. He was member of SNI (Natonal System of Researchers) from 2003 untl 202 (Level ). Hs man areas nclude: Artfcal Intellgence, Robot plannng and smulaton. He receved hs Ph. D. degree n M.C. Óscar Antono De León Gomez receved hs Master s degree at Computer Scence Department atautonomous Unversty of Puebla (202). He has been developng several dfferent projects on Computer Vson at INAOE (Natonal Insttute of Astrophyscs, Optcal and Electroncs), where currently he s workng 87
MULTISPECTRAL IMAGES CLASSIFICATION BASED ON KLT AND ATR AUTOMATIC TARGET RECOGNITION
MULTISPECTRAL IMAGES CLASSIFICATION BASED ON KLT AND ATR AUTOMATIC TARGET RECOGNITION Paulo Quntlano 1 & Antono Santa-Rosa 1 Federal Polce Department, Brasla, Brazl. E-mals: quntlano.pqs@dpf.gov.br and
More informationCluster Analysis of Electrical Behavior
Journal of Computer and Communcatons, 205, 3, 88-93 Publshed Onlne May 205 n ScRes. http://www.scrp.org/ournal/cc http://dx.do.org/0.4236/cc.205.350 Cluster Analyss of Electrcal Behavor Ln Lu Ln Lu, School
More informationContent Based Image Retrieval Using 2-D Discrete Wavelet with Texture Feature with Different Classifiers
IOSR Journal of Electroncs and Communcaton Engneerng (IOSR-JECE) e-issn: 78-834,p- ISSN: 78-8735.Volume 9, Issue, Ver. IV (Mar - Apr. 04), PP 0-07 Content Based Image Retreval Usng -D Dscrete Wavelet wth
More informationA Binarization Algorithm specialized on Document Images and Photos
A Bnarzaton Algorthm specalzed on Document mages and Photos Ergna Kavalleratou Dept. of nformaton and Communcaton Systems Engneerng Unversty of the Aegean kavalleratou@aegean.gr Abstract n ths paper, a
More informationMachine Learning: Algorithms and Applications
14/05/1 Machne Learnng: Algorthms and Applcatons Florano Zn Free Unversty of Bozen-Bolzano Faculty of Computer Scence Academc Year 011-01 Lecture 10: 14 May 01 Unsupervsed Learnng cont Sldes courtesy of
More informationLearning the Kernel Parameters in Kernel Minimum Distance Classifier
Learnng the Kernel Parameters n Kernel Mnmum Dstance Classfer Daoqang Zhang 1,, Songcan Chen and Zh-Hua Zhou 1* 1 Natonal Laboratory for Novel Software Technology Nanjng Unversty, Nanjng 193, Chna Department
More informationSubspace clustering. Clustering. Fundamental to all clustering techniques is the choice of distance measure between data points;
Subspace clusterng Clusterng Fundamental to all clusterng technques s the choce of dstance measure between data ponts; D q ( ) ( ) 2 x x = x x, j k = 1 k jk Squared Eucldean dstance Assumpton: All features
More informationA Fast Visual Tracking Algorithm Based on Circle Pixels Matching
A Fast Vsual Trackng Algorthm Based on Crcle Pxels Matchng Zhqang Hou hou_zhq@sohu.com Chongzhao Han czhan@mal.xjtu.edu.cn Ln Zheng Abstract: A fast vsual trackng algorthm based on crcle pxels matchng
More informationClassifier Selection Based on Data Complexity Measures *
Classfer Selecton Based on Data Complexty Measures * Edth Hernández-Reyes, J.A. Carrasco-Ochoa, and J.Fco. Martínez-Trndad Natonal Insttute for Astrophyscs, Optcs and Electroncs, Lus Enrque Erro No.1 Sta.
More informationParallelism for Nested Loops with Non-uniform and Flow Dependences
Parallelsm for Nested Loops wth Non-unform and Flow Dependences Sam-Jn Jeong Dept. of Informaton & Communcaton Engneerng, Cheonan Unversty, 5, Anseo-dong, Cheonan, Chungnam, 330-80, Korea. seong@cheonan.ac.kr
More informationSupport Vector Machines
/9/207 MIST.6060 Busness Intellgence and Data Mnng What are Support Vector Machnes? Support Vector Machnes Support Vector Machnes (SVMs) are supervsed learnng technques that analyze data and recognze patterns.
More informationA Fast Content-Based Multimedia Retrieval Technique Using Compressed Data
A Fast Content-Based Multmeda Retreval Technque Usng Compressed Data Borko Furht and Pornvt Saksobhavvat NSF Multmeda Laboratory Florda Atlantc Unversty, Boca Raton, Florda 3343 ABSTRACT In ths paper,
More informationEfficient Segmentation and Classification of Remote Sensing Image Using Local Self Similarity
ISSN(Onlne): 2320-9801 ISSN (Prnt): 2320-9798 Internatonal Journal of Innovatve Research n Computer and Communcaton Engneerng (An ISO 3297: 2007 Certfed Organzaton) Vol.2, Specal Issue 1, March 2014 Proceedngs
More informationHierarchical clustering for gene expression data analysis
Herarchcal clusterng for gene expresson data analyss Gorgo Valentn e-mal: valentn@ds.unm.t Clusterng of Mcroarray Data. Clusterng of gene expresson profles (rows) => dscovery of co-regulated and functonally
More informationOutline. Type of Machine Learning. Examples of Application. Unsupervised Learning
Outlne Artfcal Intellgence and ts applcatons Lecture 8 Unsupervsed Learnng Professor Danel Yeung danyeung@eee.org Dr. Patrck Chan patrckchan@eee.org South Chna Unversty of Technology, Chna Introducton
More informationFEATURE EXTRACTION. Dr. K.Vijayarekha. Associate Dean School of Electrical and Electronics Engineering SASTRA University, Thanjavur
FEATURE EXTRACTION Dr. K.Vjayarekha Assocate Dean School of Electrcal and Electroncs Engneerng SASTRA Unversty, Thanjavur613 41 Jont Intatve of IITs and IISc Funded by MHRD Page 1 of 8 Table of Contents
More informationDetection of an Object by using Principal Component Analysis
Detecton of an Object by usng Prncpal Component Analyss 1. G. Nagaven, 2. Dr. T. Sreenvasulu Reddy 1. M.Tech, Department of EEE, SVUCE, Trupath, Inda. 2. Assoc. Professor, Department of ECE, SVUCE, Trupath,
More informationTN348: Openlab Module - Colocalization
TN348: Openlab Module - Colocalzaton Topc The Colocalzaton module provdes the faclty to vsualze and quantfy colocalzaton between pars of mages. The Colocalzaton wndow contans a prevew of the two mages
More informationTsinghua University at TAC 2009: Summarizing Multi-documents by Information Distance
Tsnghua Unversty at TAC 2009: Summarzng Mult-documents by Informaton Dstance Chong Long, Mnle Huang, Xaoyan Zhu State Key Laboratory of Intellgent Technology and Systems, Tsnghua Natonal Laboratory for
More informationSkew Angle Estimation and Correction of Hand Written, Textual and Large areas of Non-Textual Document Images: A Novel Approach
Angle Estmaton and Correcton of Hand Wrtten, Textual and Large areas of Non-Textual Document Images: A Novel Approach D.R.Ramesh Babu Pyush M Kumat Mahesh D Dhannawat PES Insttute of Technology Research
More informationAn Optimal Algorithm for Prufer Codes *
J. Software Engneerng & Applcatons, 2009, 2: 111-115 do:10.4236/jsea.2009.22016 Publshed Onlne July 2009 (www.scrp.org/journal/jsea) An Optmal Algorthm for Prufer Codes * Xaodong Wang 1, 2, Le Wang 3,
More informationCorner-Based Image Alignment using Pyramid Structure with Gradient Vector Similarity
Journal of Sgnal and Informaton Processng, 013, 4, 114-119 do:10.436/jsp.013.43b00 Publshed Onlne August 013 (http://www.scrp.org/journal/jsp) Corner-Based Image Algnment usng Pyramd Structure wth Gradent
More informationModule Management Tool in Software Development Organizations
Journal of Computer Scence (5): 8-, 7 ISSN 59-66 7 Scence Publcatons Management Tool n Software Development Organzatons Ahmad A. Al-Rababah and Mohammad A. Al-Rababah Faculty of IT, Al-Ahlyyah Amman Unversty,
More informationMachine Learning. Topic 6: Clustering
Machne Learnng Topc 6: lusterng lusterng Groupng data nto (hopefully useful) sets. Thngs on the left Thngs on the rght Applcatons of lusterng Hypothess Generaton lusters mght suggest natural groups. Hypothess
More informationOutline. Self-Organizing Maps (SOM) US Hebbian Learning, Cntd. The learning rule is Hebbian like:
Self-Organzng Maps (SOM) Turgay İBRİKÇİ, PhD. Outlne Introducton Structures of SOM SOM Archtecture Neghborhoods SOM Algorthm Examples Summary 1 2 Unsupervsed Hebban Learnng US Hebban Learnng, Cntd 3 A
More informationGraph-based Clustering
Graphbased Clusterng Transform the data nto a graph representaton ertces are the data ponts to be clustered Edges are eghted based on smlarty beteen data ponts Graph parttonng Þ Each connected component
More informationSLAM Summer School 2006 Practical 2: SLAM using Monocular Vision
SLAM Summer School 2006 Practcal 2: SLAM usng Monocular Vson Javer Cvera, Unversty of Zaragoza Andrew J. Davson, Imperal College London J.M.M Montel, Unversty of Zaragoza. josemar@unzar.es, jcvera@unzar.es,
More informationFace Recognition University at Buffalo CSE666 Lecture Slides Resources:
Face Recognton Unversty at Buffalo CSE666 Lecture Sldes Resources: http://www.face-rec.org/algorthms/ Overvew of face recognton algorthms Correlaton - Pxel based correspondence between two face mages Structural
More informationUsing Fuzzy Logic to Enhance the Large Size Remote Sensing Images
Internatonal Journal of Informaton and Electroncs Engneerng Vol. 5 No. 6 November 015 Usng Fuzzy Logc to Enhance the Large Sze Remote Sensng Images Trung Nguyen Tu Huy Ngo Hoang and Thoa Vu Van Abstract
More informationMathematics 256 a course in differential equations for engineering students
Mathematcs 56 a course n dfferental equatons for engneerng students Chapter 5. More effcent methods of numercal soluton Euler s method s qute neffcent. Because the error s essentally proportonal to the
More informationRelated-Mode Attacks on CTR Encryption Mode
Internatonal Journal of Network Securty, Vol.4, No.3, PP.282 287, May 2007 282 Related-Mode Attacks on CTR Encrypton Mode Dayn Wang, Dongda Ln, and Wenlng Wu (Correspondng author: Dayn Wang) Key Laboratory
More informationLaplacian Eigenmap for Image Retrieval
Laplacan Egenmap for Image Retreval Xaofe He Partha Nyog Department of Computer Scence The Unversty of Chcago, 1100 E 58 th Street, Chcago, IL 60637 ABSTRACT Dmensonalty reducton has been receved much
More informationLECTURE : MANIFOLD LEARNING
LECTURE : MANIFOLD LEARNING Rta Osadchy Some sldes are due to L.Saul, V. C. Raykar, N. Verma Topcs PCA MDS IsoMap LLE EgenMaps Done! Dmensonalty Reducton Data representaton Inputs are real-valued vectors
More informationThe Greedy Method. Outline and Reading. Change Money Problem. Greedy Algorithms. Applications of the Greedy Strategy. The Greedy Method Technique
//00 :0 AM Outlne and Readng The Greedy Method The Greedy Method Technque (secton.) Fractonal Knapsack Problem (secton..) Task Schedulng (secton..) Mnmum Spannng Trees (secton.) Change Money Problem Greedy
More informationModular PCA Face Recognition Based on Weighted Average
odern Appled Scence odular PCA Face Recognton Based on Weghted Average Chengmao Han (Correspondng author) Department of athematcs, Lny Normal Unversty Lny 76005, Chna E-mal: hanchengmao@163.com Abstract
More informationAnalysis of Continuous Beams in General
Analyss of Contnuous Beams n General Contnuous beams consdered here are prsmatc, rgdly connected to each beam segment and supported at varous ponts along the beam. onts are selected at ponts of support,
More informationRange images. Range image registration. Examples of sampling patterns. Range images and range surfaces
Range mages For many structured lght scanners, the range data forms a hghly regular pattern known as a range mage. he samplng pattern s determned by the specfc scanner. Range mage regstraton 1 Examples
More informationAn Image Fusion Approach Based on Segmentation Region
Rong Wang, L-Qun Gao, Shu Yang, Yu-Hua Cha, and Yan-Chun Lu An Image Fuson Approach Based On Segmentaton Regon An Image Fuson Approach Based on Segmentaton Regon Rong Wang, L-Qun Gao, Shu Yang 3, Yu-Hua
More informationR s s f. m y s. SPH3UW Unit 7.3 Spherical Concave Mirrors Page 1 of 12. Notes
SPH3UW Unt 7.3 Sphercal Concave Mrrors Page 1 of 1 Notes Physcs Tool box Concave Mrror If the reflectng surface takes place on the nner surface of the sphercal shape so that the centre of the mrror bulges
More informationUnsupervised Learning
Pattern Recognton Lecture 8 Outlne Introducton Unsupervsed Learnng Parametrc VS Non-Parametrc Approach Mxture of Denstes Maxmum-Lkelhood Estmates Clusterng Prof. Danel Yeung School of Computer Scence and
More information12/2/2009. Announcements. Parametric / Non-parametric. Case-Based Reasoning. Nearest-Neighbor on Images. Nearest-Neighbor Classification
Introducton to Artfcal Intellgence V22.0472-001 Fall 2009 Lecture 24: Nearest-Neghbors & Support Vector Machnes Rob Fergus Dept of Computer Scence, Courant Insttute, NYU Sldes from Danel Yeung, John DeNero
More informationFor instance, ; the five basic number-sets are increasingly more n A B & B A A = B (1)
Secton 1.2 Subsets and the Boolean operatons on sets If every element of the set A s an element of the set B, we say that A s a subset of B, or that A s contaned n B, or that B contans A, and we wrte A
More informationLecture 5: Multilayer Perceptrons
Lecture 5: Multlayer Perceptrons Roger Grosse 1 Introducton So far, we ve only talked about lnear models: lnear regresson and lnear bnary classfers. We noted that there are functons that can t be represented
More informationEdge Detection in Noisy Images Using the Support Vector Machines
Edge Detecton n Nosy Images Usng the Support Vector Machnes Hlaro Gómez-Moreno, Saturnno Maldonado-Bascón, Francsco López-Ferreras Sgnal Theory and Communcatons Department. Unversty of Alcalá Crta. Madrd-Barcelona
More informationUnsupervised Learning and Clustering
Unsupervsed Learnng and Clusterng Supervsed vs. Unsupervsed Learnng Up to now we consdered supervsed learnng scenaro, where we are gven 1. samples 1,, n 2. class labels for all samples 1,, n Ths s also
More informationOptimizing Document Scoring for Query Retrieval
Optmzng Document Scorng for Query Retreval Brent Ellwen baellwe@cs.stanford.edu Abstract The goal of ths project was to automate the process of tunng a document query engne. Specfcally, I used machne learnng
More informationProblem Set 3 Solutions
Introducton to Algorthms October 4, 2002 Massachusetts Insttute of Technology 6046J/18410J Professors Erk Demane and Shaf Goldwasser Handout 14 Problem Set 3 Solutons (Exercses were not to be turned n,
More informationCollaboratively Regularized Nearest Points for Set Based Recognition
Academc Center for Computng and Meda Studes, Kyoto Unversty Collaboratvely Regularzed Nearest Ponts for Set Based Recognton Yang Wu, Mchhko Mnoh, Masayuk Mukunok Kyoto Unversty 9/1/013 BMVC 013 @ Brstol,
More informationLocal Quaternary Patterns and Feature Local Quaternary Patterns
Local Quaternary Patterns and Feature Local Quaternary Patterns Jayu Gu and Chengjun Lu The Department of Computer Scence, New Jersey Insttute of Technology, Newark, NJ 0102, USA Abstract - Ths paper presents
More informationBioTechnology. An Indian Journal FULL PAPER. Trade Science Inc.
[Type text] [Type text] [Type text] ISSN : 0974-74 Volume 0 Issue BoTechnology 04 An Indan Journal FULL PAPER BTAIJ 0() 04 [684-689] Revew on Chna s sports ndustry fnancng market based on market -orented
More informationLoad Balancing for Hex-Cell Interconnection Network
Int. J. Communcatons, Network and System Scences,,, - Publshed Onlne Aprl n ScRes. http://www.scrp.org/journal/jcns http://dx.do.org/./jcns.. Load Balancng for Hex-Cell Interconnecton Network Saher Manaseer,
More information2x x l. Module 3: Element Properties Lecture 4: Lagrange and Serendipity Elements
Module 3: Element Propertes Lecture : Lagrange and Serendpty Elements 5 In last lecture note, the nterpolaton functons are derved on the bass of assumed polynomal from Pascal s trangle for the fled varable.
More informationParallel matrix-vector multiplication
Appendx A Parallel matrx-vector multplcaton The reduced transton matrx of the three-dmensonal cage model for gel electrophoress, descrbed n secton 3.2, becomes excessvely large for polymer lengths more
More informationCompiler Design. Spring Register Allocation. Sample Exercises and Solutions. Prof. Pedro C. Diniz
Compler Desgn Sprng 2014 Regster Allocaton Sample Exercses and Solutons Prof. Pedro C. Dnz USC / Informaton Scences Insttute 4676 Admralty Way, Sute 1001 Marna del Rey, Calforna 90292 pedro@s.edu Regster
More informationImprovement of Spatial Resolution Using BlockMatching Based Motion Estimation and Frame. Integration
Improvement of Spatal Resoluton Usng BlockMatchng Based Moton Estmaton and Frame Integraton Danya Suga and Takayuk Hamamoto Graduate School of Engneerng, Tokyo Unversty of Scence, 6-3-1, Nuku, Katsuska-ku,
More informationAn Indian Journal FULL PAPER ABSTRACT KEYWORDS. Trade Science Inc.
[Type text] [Type text] [Type text] ISSN : 97-735 Volume Issue 9 BoTechnology An Indan Journal FULL PAPER BTAIJ, (9), [333-3] Matlab mult-dmensonal model-based - 3 Chnese football assocaton super league
More informationLecture 4: Principal components
/3/6 Lecture 4: Prncpal components 3..6 Multvarate lnear regresson MLR s optmal for the estmaton data...but poor for handlng collnear data Covarance matrx s not nvertble (large condton number) Robustness
More informationReading. 14. Subdivision curves. Recommended:
eadng ecommended: Stollntz, Deose, and Salesn. Wavelets for Computer Graphcs: heory and Applcatons, 996, secton 6.-6., A.5. 4. Subdvson curves Note: there s an error n Stollntz, et al., secton A.5. Equaton
More informationLobachevsky State University of Nizhni Novgorod. Polyhedron. Quick Start Guide
Lobachevsky State Unversty of Nzhn Novgorod Polyhedron Quck Start Gude Nzhn Novgorod 2016 Contents Specfcaton of Polyhedron software... 3 Theoretcal background... 4 1. Interface of Polyhedron... 6 1.1.
More informationProper Choice of Data Used for the Estimation of Datum Transformation Parameters
Proper Choce of Data Used for the Estmaton of Datum Transformaton Parameters Hakan S. KUTOGLU, Turkey Key words: Coordnate systems; transformaton; estmaton, relablty. SUMMARY Advances n technologes and
More informationClustering Algorithm of Similarity Segmentation based on Point Sorting
Internatonal onference on Logstcs Engneerng, Management and omputer Scence (LEMS 2015) lusterng Algorthm of Smlarty Segmentaton based on Pont Sortng Hanbng L, Yan Wang*, Lan Huang, Mngda L, Yng Sun, Hanyuan
More informationA New Feature of Uniformity of Image Texture Directions Coinciding with the Human Eyes Perception 1
A New Feature of Unformty of Image Texture Drectons Concdng wth the Human Eyes Percepton Xng-Jan He, De-Shuang Huang, Yue Zhang, Tat-Mng Lo 2, and Mchael R. Lyu 3 Intellgent Computng Lab, Insttute of Intellgent
More informationAn Improved Image Segmentation Algorithm Based on the Otsu Method
3th ACIS Internatonal Conference on Software Engneerng, Artfcal Intellgence, Networkng arallel/dstrbuted Computng An Improved Image Segmentaton Algorthm Based on the Otsu Method Mengxng Huang, enjao Yu,
More informationUnsupervised Learning and Clustering
Unsupervsed Learnng and Clusterng Why consder unlabeled samples?. Collectng and labelng large set of samples s costly Gettng recorded speech s free, labelng s tme consumng 2. Classfer could be desgned
More informationClassification of Face Images Based on Gender using Dimensionality Reduction Techniques and SVM
Classfcaton of Face Images Based on Gender usng Dmensonalty Reducton Technques and SVM Fahm Mannan 260 266 294 School of Computer Scence McGll Unversty Abstract Ths report presents gender classfcaton based
More informationS1 Note. Basis functions.
S1 Note. Bass functons. Contents Types of bass functons...1 The Fourer bass...2 B-splne bass...3 Power and type I error rates wth dfferent numbers of bass functons...4 Table S1. Smulaton results of type
More informationAssignment # 2. Farrukh Jabeen Algorithms 510 Assignment #2 Due Date: June 15, 2009.
Farrukh Jabeen Algorthms 51 Assgnment #2 Due Date: June 15, 29. Assgnment # 2 Chapter 3 Dscrete Fourer Transforms Implement the FFT for the DFT. Descrbed n sectons 3.1 and 3.2. Delverables: 1. Concse descrpton
More informationA PATTERN RECOGNITION APPROACH TO IMAGE SEGMENTATION
1 THE PUBLISHING HOUSE PROCEEDINGS OF THE ROMANIAN ACADEMY, Seres A, OF THE ROMANIAN ACADEMY Volume 4, Number 2/2003, pp.000-000 A PATTERN RECOGNITION APPROACH TO IMAGE SEGMENTATION Tudor BARBU Insttute
More informationQuery Clustering Using a Hybrid Query Similarity Measure
Query clusterng usng a hybrd query smlarty measure Fu. L., Goh, D.H., & Foo, S. (2004). WSEAS Transacton on Computers, 3(3), 700-705. Query Clusterng Usng a Hybrd Query Smlarty Measure Ln Fu, Don Hoe-Lan
More informationActive Contours/Snakes
Actve Contours/Snakes Erkut Erdem Acknowledgement: The sldes are adapted from the sldes prepared by K. Grauman of Unversty of Texas at Austn Fttng: Edges vs. boundares Edges useful sgnal to ndcate occludng
More informationFace Recognition Based on SVM and 2DPCA
Vol. 4, o. 3, September, 2011 Face Recognton Based on SVM and 2DPCA Tha Hoang Le, Len Bu Faculty of Informaton Technology, HCMC Unversty of Scence Faculty of Informaton Scences and Engneerng, Unversty
More informationA New Measure of Cluster Validity Using Line Symmetry *
JOURAL OF IFORMATIO SCIECE AD EGIEERIG 30, 443-461 (014) A ew Measure of Cluster Valdty Usng Lne Symmetry * CHIE-HSIG CHOU 1, YI-ZEG HSIEH AD MU-CHU SU 1 Department of Electrcal Engneerng Tamkang Unversty
More informationFuzzy C-Means Initialized by Fixed Threshold Clustering for Improving Image Retrieval
Fuzzy -Means Intalzed by Fxed Threshold lusterng for Improvng Image Retreval NAWARA HANSIRI, SIRIPORN SUPRATID,HOM KIMPAN 3 Faculty of Informaton Technology Rangst Unversty Muang-Ake, Paholyotn Road, Patumtan,
More informationMachine Learning 9. week
Machne Learnng 9. week Mappng Concept Radal Bass Functons (RBF) RBF Networks 1 Mappng It s probably the best scenaro for the classfcaton of two dataset s to separate them lnearly. As you see n the below
More informationAvailable online at Available online at Advanced in Control Engineering and Information Science
Avalable onlne at wwwscencedrectcom Avalable onlne at wwwscencedrectcom Proceda Proceda Engneerng Engneerng 00 (2011) 15000 000 (2011) 1642 1646 Proceda Engneerng wwwelsevercom/locate/proceda Advanced
More informationA Clustering Algorithm for Key Frame Extraction Based on Density Peak
Journal of Computer and Communcatons, 2018, 6, 118-128 http://www.scrp.org/ournal/cc ISSN Onlne: 2327-5227 ISSN Prnt: 2327-5219 A Clusterng Algorthm for Key Frame Extracton Based on Densty Peak Hong Zhao
More informationA New Approach For the Ranking of Fuzzy Sets With Different Heights
New pproach For the ankng of Fuzzy Sets Wth Dfferent Heghts Pushpnder Sngh School of Mathematcs Computer pplcatons Thapar Unversty, Patala-7 00 Inda pushpndersnl@gmalcom STCT ankng of fuzzy sets plays
More informationObject-Based Techniques for Image Retrieval
54 Zhang, Gao, & Luo Chapter VII Object-Based Technques for Image Retreval Y. J. Zhang, Tsnghua Unversty, Chna Y. Y. Gao, Tsnghua Unversty, Chna Y. Luo, Tsnghua Unversty, Chna ABSTRACT To overcome the
More informationCS434a/541a: Pattern Recognition Prof. Olga Veksler. Lecture 15
CS434a/541a: Pattern Recognton Prof. Olga Veksler Lecture 15 Today New Topc: Unsupervsed Learnng Supervsed vs. unsupervsed learnng Unsupervsed learnng Net Tme: parametrc unsupervsed learnng Today: nonparametrc
More information3D vector computer graphics
3D vector computer graphcs Paolo Varagnolo: freelance engneer Padova Aprl 2016 Prvate Practce ----------------------------------- 1. Introducton Vector 3D model representaton n computer graphcs requres
More informationX- Chart Using ANOM Approach
ISSN 1684-8403 Journal of Statstcs Volume 17, 010, pp. 3-3 Abstract X- Chart Usng ANOM Approach Gullapall Chakravarth 1 and Chaluvad Venkateswara Rao Control lmts for ndvdual measurements (X) chart are
More informationK-means and Hierarchical Clustering
Note to other teachers and users of these sldes. Andrew would be delghted f you found ths source materal useful n gvng your own lectures. Feel free to use these sldes verbatm, or to modfy them to ft your
More informationPictures at an Exhibition
1 Pctures at an Exhbton Stephane Kwan and Karen Zhu Department of Electrcal Engneerng Stanford Unversty, Stanford, CA 9405 Emal: {skwan1, kyzhu}@stanford.edu Abstract An mage processng algorthm s desgned
More informationDetermining the Optimal Bandwidth Based on Multi-criterion Fusion
Proceedngs of 01 4th Internatonal Conference on Machne Learnng and Computng IPCSIT vol. 5 (01) (01) IACSIT Press, Sngapore Determnng the Optmal Bandwdth Based on Mult-crteron Fuson Ha-L Lang 1+, Xan-Mn
More informationShape Representation Robust to the Sketching Order Using Distance Map and Direction Histogram
Shape Representaton Robust to the Sketchng Order Usng Dstance Map and Drecton Hstogram Department of Computer Scence Yonse Unversty Kwon Yun CONTENTS Revew Topc Proposed Method System Overvew Sketch Normalzaton
More informationFlatten a Curved Space by Kernel: From Einstein to Euclid
Flatten a Curved Space by Kernel: From Ensten to Eucld Quyuan Huang, Dapeng Olver Wu Ensten s general theory of relatvty fundamentally changed our vew about the physcal world. Dfferent from Newton s theory,
More informationVirtual Machine Migration based on Trust Measurement of Computer Node
Appled Mechancs and Materals Onlne: 2014-04-04 ISSN: 1662-7482, Vols. 536-537, pp 678-682 do:10.4028/www.scentfc.net/amm.536-537.678 2014 Trans Tech Publcatons, Swtzerland Vrtual Machne Mgraton based on
More informationIEEE TRANSACTIONS ON IMAGE PROCESSING, VOL. *, NO. *, Dictionary Pair Learning on Grassmann Manifolds for Image Denoising
IEEE TRANSACTIONS ON IMAGE PROCESSING, VOL. *, NO. *, 2015 1 Dctonary Par Learnng on Grassmann Manfolds for Image Denosng Xanhua Zeng, We Ban, We Lu, Jale Shen, Dacheng Tao, Fellow, IEEE Abstract Image
More informationOutline. Discriminative classifiers for image recognition. Where in the World? A nearest neighbor recognition example 4/14/2011. CS 376 Lecture 22 1
4/14/011 Outlne Dscrmnatve classfers for mage recognton Wednesday, Aprl 13 Krsten Grauman UT-Austn Last tme: wndow-based generc obect detecton basc ppelne face detecton wth boostng as case study Today:
More informationAppearance-based Statistical Methods for Face Recognition
47th Internatonal Symposum ELMAR-2005, 08-10 June 2005, Zadar, Croata Appearance-based Statstcal Methods for Face Recognton Kresmr Delac 1, Mslav Grgc 2, Panos Latss 3 1 Croatan elecom, Savsa 32, Zagreb,
More informationA Computer Vision System for Automated Container Code Recognition
A Computer Vson System for Automated Contaner Code Recognton Hsn-Chen Chen, Chh-Ka Chen, Fu-Yu Hsu, Yu-San Ln, Yu-Te Wu, Yung-Nen Sun * Abstract Contaner code examnaton s an essental step n the contaner
More informationThe Discriminate Analysis and Dimension Reduction Methods of High Dimension
Open Journal of Socal Scences, 015, 3, 7-13 Publshed Onlne March 015 n ScRes. http://www.scrp.org/journal/jss http://dx.do.org/10.436/jss.015.3300 The Dscrmnate Analyss and Dmenson Reducton Methods of
More informationA CALCULATION METHOD OF DEEP WEB ENTITIES RECOGNITION
A CALCULATION METHOD OF DEEP WEB ENTITIES RECOGNITION 1 FENG YONG, DANG XIAO-WAN, 3 XU HONG-YAN School of Informaton, Laonng Unversty, Shenyang Laonng E-mal: 1 fyxuhy@163.com, dangxaowan@163.com, 3 xuhongyan_lndx@163.com
More informationOn Some Entertaining Applications of the Concept of Set in Computer Science Course
On Some Entertanng Applcatons of the Concept of Set n Computer Scence Course Krasmr Yordzhev *, Hrstna Kostadnova ** * Assocate Professor Krasmr Yordzhev, Ph.D., Faculty of Mathematcs and Natural Scences,
More informationBAYESIAN MULTI-SOURCE DOMAIN ADAPTATION
BAYESIAN MULTI-SOURCE DOMAIN ADAPTATION SHI-LIANG SUN, HONG-LEI SHI Department of Computer Scence and Technology, East Chna Normal Unversty 500 Dongchuan Road, Shangha 200241, P. R. Chna E-MAIL: slsun@cs.ecnu.edu.cn,
More informationImage Alignment CSC 767
Image Algnment CSC 767 Image algnment Image from http://graphcs.cs.cmu.edu/courses/15-463/2010_fall/ Image algnment: Applcatons Panorama sttchng Image algnment: Applcatons Recognton of object nstances
More informationCorrelative features for the classification of textural images
Correlatve features for the classfcaton of textural mages M A Turkova 1 and A V Gadel 1, 1 Samara Natonal Research Unversty, Moskovskoe Shosse 34, Samara, Russa, 443086 Image Processng Systems Insttute
More informationClassifying Acoustic Transient Signals Using Artificial Intelligence
Classfyng Acoustc Transent Sgnals Usng Artfcal Intellgence Steve Sutton, Unversty of North Carolna At Wlmngton (suttons@charter.net) Greg Huff, Unversty of North Carolna At Wlmngton (jgh7476@uncwl.edu)
More informationHistogram of Template for Pedestrian Detection
PAPER IEICE TRANS. FUNDAMENTALS/COMMUN./ELECTRON./INF. & SYST., VOL. E85-A/B/C/D, No. xx JANUARY 20xx Hstogram of Template for Pedestran Detecton Shaopeng Tang, Non Member, Satosh Goto Fellow Summary In
More informationMachine Learning. Support Vector Machines. (contains material adapted from talks by Constantin F. Aliferis & Ioannis Tsamardinos, and Martin Law)
Machne Learnng Support Vector Machnes (contans materal adapted from talks by Constantn F. Alfers & Ioanns Tsamardnos, and Martn Law) Bryan Pardo, Machne Learnng: EECS 349 Fall 2014 Support Vector Machnes
More information