Structural Analysis of Musical Signals for Indexing and Thumbnailing

Size: px
Start display at page:

Download "Structural Analysis of Musical Signals for Indexing and Thumbnailing"

Transcription

1 Structural Analyss of Muscal Sgnals for Indexng and Thumbnalng We Cha Barry Vercoe MIT Meda Laboratory {chawe, Abstract A muscal pece typcally has a repettve structure. Analyss of ths structure wll be useful for musc segmentaton, ndexng and thumbnalng. Ths paper presents an algorthm that can automatcally analyze the repettve structure of muscal sgnals. Frst, the algorthm detects the repettons of each segment of fxed length n a pece usng dynamc programmng. Second, the algorthm summarzes ths repetton nformaton and nfers the structure based on heurstc rules. The performance of the approach s demonstrated vsually usng fgures for qualtatve evaluaton, and by two structural smlarty measures for quanttatve evaluaton. Based on the structural analyss result, ths paper also proposes a method for musc thumbnalng. The prelmnary results obtaned usng a corpus of Beatles songs show that automatc structural analyss and thumbnalng of musc are possble. 1. Introducton A muscal pece typcally has a repettve structure. For example, a song may have a structure of ABA, ndcatng a three-part compostonal form n whch the second secton contrasts wth the frst secton, and the thrd secton s a restatement of the frst. Methods for automatcally detectng the repettve structure of a muscal pece from acoustcal sgnals s valuable for nformaton retreval systems and dgtal lbrares; for example, the result can be used for ndexng the muscal content or for musc thumbnalng. There has been some recent research on ths topc. Dannenberg and Hu presented a method to automatcally detect the repettve structure of muscal sgnals [6]. The process conssts of searchng for smlar segments n a muscal pece, formng clusters of smlar segments, and explanng the muscal structure n terms of these clusters. Three representatons were nvestgated: monophonc ptch estmaton, chroma representaton, and polyphonc transcrpton followed by harmonc analyss. Although the promse of ths method was demonstrated n several examples, there was no quanttatve evaluaton of the method n ther paper. Two topcs closely related to structural analyss of musc have also been nvestgated. One s musc thumbnalng (or musc summarzaton), whch ams at fndng the most representatve part (normally assumed to be the most repeated secton) of a song. Some research on musc thumbnalng deals wth symbolc muscal data (e.g., MIDI fles and scores) [10]. There have also been studes on thumbnalng of muscal sgnals. Logan and Chu attempted to use a clusterng technque or Hdden Markov Models to fnd key phrases of songs. Mel Cepstral features were used to characterze each song [11]. The other related topc s musc segmentaton. Most prevous research n ths area attempted to segment muscal peces by detectng the locatons where a sgnfcant change of statstcal propertes occurs [1]. Ths method s more approprate for segmentng dfferent local events rather than segmentng the semantc components of the global structure. Addtonally, Foote proposed a representaton called a smlarty matrx for vsualzng and analyzng the structure of audo, ncludng symbolc musc, acoustc muscal sgnals or more general audo [7][8]. One attempt usng ths representaton was to locate ponts of sgnfcant change n musc (e.g., score analyss) or audo (e.g., speech/musc segmentaton) [8]. Bartsch and Wakefeld used the smlarty matrx and chroma-based features for musc thumbnalng [4]. A varaton of the smlarty matrx was also proposed for musc thumbnalng [12]. Ths paper descrbes research nto automatc dentfcaton of the repettve structure of muscal peces from acoustc sgnals. Specfcally, an algorthm s presented that wll output structural nformaton, ncludng both the form (e.g., AABABA) and the boundares ndcatng the begnnng and the end of each secton. It s assumed that no pror knowledge about muscal forms or the length of each secton s provded, and the restatement of a secton may have varatons. Ths assumpton requres both robustness and effcency of the algorthm. Two novel structural smlarty measures are also proposed n ths paper to quanttatvely evaluate the performance of the algorthm, n addton to the qualtatve evaluaton presented by fgures.

2 The remander of ths paper s organzed as follows. Secton 2 llustrates the structural analyss approach. Secton 3 presents the expermental results. Secton 4 explans how the structural analyss result can be used for musc thumbnalng. Secton 5 gves conclusons and proposes future work. 2. Approach Ths secton llustrates the structural analyss method, whch follows fve steps and s also llustrated n Fgure 1: 1) Segment the sgnal nto frames and compute the feature of each frame; 2) Segment the feature sequence nto overlapped segments of fxed length and compute the repetton property of each segment usng dynamc programmng; 3) Detect the repettons of each segment by fndng the local mnma n the dynamc programmng result; 4) Merge consecutve segments that have the same repettve property nto sectons and generate pars of smlar sectons; 5) Segment and label the repettve structure. Segment the muscal sgnal nto overlaped frames Frame 1 Frame 2 Frame 3 Frame n Compute the feature vector of each frame Feature vector 1 Feature vector 2 Feature vector 3 Feature vector n Segment the feature vector sequence nto overlapped segments Segment 1 Segment 2 Segment 3 Segment m Match each segment aganst the feature vector sequence usng dynamc programmng Detect the repettons of each segment Merge consecutve segments that have the same repettve property nto sectons Label the repettve structure Fgure 1. Overvew of the approach. The followng fve sectons explan each step n detal. All the parameter confguratons are tuned based on the expermental corpus, whch s descrbed n Secton Feature extracton The algorthm frst segments the sgnal nto overlapped frames (e.g., 1024-sample wndow length wth 512- sample overlap) and computes the feature of each frame. Two representatons are nvestgated n ths paper. One s the ptch representaton, whch uses autocorrelaton [13] to estmate the man frequency component of each frame. Although all the test data n the experment are polyphonc, t turns out that, for muscal sgnals wth a leadng vocal, ths feature can stll capture much nformaton. The other representaton explored s the spectral representaton,.e., FFT magntude coeffcents. The dstance between two ptch features v 1 and v2 s defned as v1 v2 d p ( v1, v2 ) = (1) normalzaton factor The dstance between two spectral features v 1 and v 2 s defned as v v v v 1 2 d f ( 1, 2 ) = (2) v v 1 2 In both cases, a dstance value ranges between 0 and Pattern matchng After computng the feature vector v (onedmensonal vector for the ptch representaton and N- j dmensonal vector for the spectral representaton) for each frame, the algorthm segments the feature vector sequence V[ 1, n] = { v j j = 1,, n} (n s the number of frames) nto overlapped segments of fxed length l (e.g., 200 consecutve vectors wth 150 vectors overlap). Snce prevous research has shown that dynamc programmng s effectve for musc pattern matchng [9][14], here dynamc programmng s used to match each segment (.e., s = V [ j, j + l 1] ) wth the feature vector sequence startng from ths segment (.e., V [ j, n] ). The dynamc programmng algorthm wll fll n a matrx M (.e., the dynamc programmng matrx of the th segment) as shown n Fgure 2 based on Equaton 3. M[ p 1, q] + e M[ p, q 1] + e M[ p, q] = mn M[ p 1, q 1] + c 0 ( p 1) ( q 1) ( p, q 1) o. w. (3)

3 where e s the nserton or deleton cost, c s the dstance between the two correspondng feature vectors, whch has been defned n Secton 2.1. The last row of matrx M s defned as functon d [r] (shown as the shaded area n Fgure 2). In addton, the trace-back step of dynamc programmng determnes the actual algnments (.e., the locatons n V [ j, n] matchng the begnnng of s ) that result n d [r]. The trace-back result s denoted as t [r]. (3) No two repettons are closer than d, the algorthm adds the mnmum nto the detected repetton set. In our experment, we set w=400, d=5, and p=0.1. Fgure 3 shows the repetton detecton result of one segment n the song Yesterday repetton detecton: Yesterday 60 vj V(j+1) V(j+2) vn d [k] vj V(j+1) V(j+l-1) e 2e le Fgure 2. Dynamc programmng matrx M. Ths step s the most tme consumng one n the structural analyss algorthm; ts tme complexty s O ( n 2 ) Repetton detecton Ths step of the algorthm detects the repetton of each segment. To acheve ths, the algorthm detects the local mnma n the functon d [r] for each, because normally a repetton of segment wll correspond to a local mnmum n ths functon. There are four predefned parameters n the algorthm of detectng the local mnma: the wdth parameter w, the dstance parameter d, the heght parameter h, and the shape parameter p. To detect local mnma of d [r], the algorthm sldes the wndow of wdth w over d [r]. Assume the ndex of the mnmum wthn the wndow s r 0 wth value d [ r 0 ], the ndex of the maxmum wthn the wndow but left to r 0 s r 1 (.e., r 1 <r 0 ) wth value d [ r 1 ], and the ndex of the maxmum wthn the wndow but rght to r 0 s r 2 (.e., r 2 >r 0 ) wth value d [ r 2 ]. If (1) d [ r1 ] d [ r0 ] > h and d [ r2 ] d [ r0 ] > h (.e., the local mnmum s deep enough); and (2) d [ r1 ] d [ r0 ] d [ r2 ] d [ r0 ] > p or > p (.e., the r1 r0 r2 r0 local mnmum s sharp enough); and k Fgure 3. One-segment repetton detecton result of Yesterday. The local mnma ndcated by crcles correspond to detected repettons of the segment. The repettons detected may have add or drop errors, meanng a repetton s falsely detected or mssed. For example, n Fgure 3, the frst, the second, the fourth and the ffth detected local mnma correspond to the four restatements of the same melodc segment n the song ( here to stay, over me, hde away, hde away ). However, there s an add error occurrng at the thrd detected local mnmum. The number of add errors and that of the drop errors are balanced by the predefned parameter h; whenever the local mnmum s deeper than heght h, the algorthm reports a detecton of repetton. Thus, when h ncreases, there are more drop errors but fewer add errors, and vse versa. For balancng between these two knds of errors, the algorthm searches wthn a range for the best value of h (e.g., decreasng from 10 to 5 wth step -1 for the ptch representaton, and decreasng from 12 to 8 wth step -1 for the spectral representaton), so that the number of detected repettons of the whole song s reasonable (e.g., # detected repetton s / n 2 ). For each detected mnmum d [ r * ] for s = V [ j, j + l 1], let k * = t [ r * ] ; thus, t s detected that the segment startng at v s repeated at j v. Please * j + k note that by the nature of dynamc programmng, the matchng part may not be of length l due to the varatons n the repetton.

4 2.4. Segment mergng The algorthm merges consecutve segments that have the same repettve property nto sectons and generates pars of smlar sectons n ths step. k repetton detecton (whole song): Yesterday j Fgure 4. Whole-song repetton detecton result of Yesterday. A crcle or a square at locaton (j, k) ndcates that the segment startng at v j s detected to repeat at v j+k. Fgure 4 shows the repetton detecton result of the song Yesterday after ths step. In ths fgure, a crcle or a square at (j, k) corresponds to a repetton detected n the last step (.e., the segment startng at v j s repeated at v j + ). Snce typcally one muscal phrase conssts of k multple segments, based on the confguratons n prevous steps, f one segment n a phrase s repeated by a shft of k, all the segments n ths phrase are repeated by shfts roughly equal to k. Ths phenomenon can be seen from Fgure 4, where the squares have the horzontal patterns ndcatng consecutve segments have roughly the same shfts. By detectng these horzontal patterns (denoted by squares n Fgure 4) and dscardng other detected repettons (denoted by crcles n Fgure 4) obtaned from the thrd step, the effects of add/drop errors are further reduced. The output of ths step s a set of merged sectons n terms of tuples < j 1, j2, shft >, ndcatng that the segment startng at v and endng at j 1 v repeats roughly from j 2 v j + to 1 shft v j +. Each tuple corresponds to one 2 shft horzontal pattern n the whole-song repetton detecton result. For example, the tuple correspondng to the leftbottom horzontal pattern n Fgure 4 s <100, 450, 370>. Snce the shfts of repettons may not be exactly the same for segments n the merged secton, the shft of the whole secton s the average value Structure labelng Based on the tuples obtaned from the fourth step, the last step of the algorthm segments the whole pece nto sectons and labels each secton accordng to the repettve relaton (.e., gves each secton a symbol such as A, B, etc.). Thus, ths step wll output the structural nformaton, ncludng both the form (e.g., AABABA) and the boundares ndcatng the begnnng and the end of each secton. To solve conflcts that mght occur, the rule for labelng s always labelng the most frequently repeated secton frst. Specfcally, the algorthm fnds the most frequently repeated secton based on the frst two columns n the tuples, and labels t and ts shfted versons as secton A. Then the algorthm deletes the tuples already labeled, repeats the same procedure for the remanng tuples, and labels sectons produced n each step as B, C, D and so on. If conflcts occur (e.g., a later labeled secton has overlap wth the prevous labeled sectons), the prevous labeled sectons wll always reman ntact and the current secton wll be truncated. 3. Experment and evaluaton Ths secton presents the expermental results and evaluatons of the structural analyss approach Data set The expermental corpus conssts of the 26 Beatles songs n the two CDs The Beatles ( ). All these songs have clear repettve structures and leadng vocal. The data were mxed to 8-bt mono and down-sampled to 11kHz Measures of structural smlarty Fgure 5. Comparson of the computed structure (above) usng the ptch representaton and the deal structure (below) of Yesterday. Sectons n the same color ndcate restatements of the secton. Sectons n the lghtest grey correspond to the sectons wth no repetton. To qualtatvely evaluate the results, fgures as shown n Fgure 5 are used to compare the structure obtaned from the algorthm wth the deal structure obtaned by manually labelng the repetton. Ths paper also proposes two measures of structural smlarty to quanttatvely

5 evaluate the result. Both of the measures need to be as small as possble, deally equal to zero. Measure 1 (structural measure) s defned as the edt dstance between the strngs representng dfferent forms. For the example n Fgure 5, the dstance between the deal structure AABABA and the computed structure AABBABA s 1, ndcatng one nserton. Here how the algorthm labels each secton s not mportant as long as the repettve relaton s the same; thus, ths deal structure s deemed as equvalent (0-dstance) to structure BBABAB, or structure AACACA. Measure 2 (boundary measure) s manly used to evaluate how accurate the boundares of each secton are. It s defned as BM = ( 1 r) / s (4) where r s the rato of the length of parts where both structures have the same labelng to the whole length, and s s the number of the repettve sectons n the deal structure. BM song d Boundary Measures of 26 Beatles' Songs (ptch representaton) Boundary Measures of 26 Beatles' Songs (FFT representaton) 3.3. Results BM Fgure 6 and Fgure 7 show the structural and boundary measures of the expermental results usng both the ptch representaton and the spectral representaton. In Fgure 7, the baselne results correspondng to labelng the whole song as a sngle secton are also plotted for a comparson. SM Structural Measures of 26 Beatles' songs song d Fgure 7. Boundary measures of the 26 Beatles songs. The sold lnes wth crcle markers correspond to the computed results (above: ptch representaton; below: spectral representaton). The dotted lnes wth x markers correspond to the baselne. It s easly seen from the above fgures that the performance of the thrd, the eghth and the nnth song usng the ptch representaton are the best (the structural measures are 0 and the boundary measures are low). For example, the result of the thrd song From me to you usng the ptch representaton s shown n Fgure song d Fgure 6. Structural measures of the 26 Beatles songs. The sold lne wth crcle markers corresponds to the ptch representaton results. The dashed lne wth square markers corresponds to the spectral representaton results. Fgure 8. Comparson of the computed structure (above) usng the ptch representaton and the deal structure (below) of From me to you. The one of the worst performance s the seventeenth song Day trpper usng the ptch representaton, whose result s shown n Fgure 9.

6 Fgure 9. Comparson of the computed structure (above) usng the ptch representaton and the deal structure (below) of Day trpper. Some nterestng results also occur. For example, for the twelfth song Tcket to rde, although the computed structure usng the spectral representaton s dfferent from the deal structure as shown n Fgure 10, t also looks reasonable by seeng secton A n the computed structure as the combnaton of secton A and secton B n the deal structure. Fgure 10. Comparson of the computed structure (above) usng the spectral representaton and the deal structure (below) of Tcket to rde Dscussons The expermental result shows that, by ether the ptch representaton or the spectral representaton, the performance of 15 out of 26 songs have structural measures less than or equal to 2 (Fgure 6) and the results of all the songs have boundary measures better than the baselne (Fgure 11). Ths demonstrates the promse of the method. BM Boundary Measures of 26 Beatles' Songs (mn) song d Fgure 11. Boundary measures of the 26 Beatles songs. The sold lne wth crcle markers corresponds to the best computed result for each song usng ether the ptch representaton or the spectral representaton. The dotted lne wth x markers corresponds to the baselne. The result does not show one representaton s sgnfcantly superor to the other. However, for each song, the result of one representaton s often better than the result of the other. Ths ndcates that the representaton does play an mportant role n performance and nvestgatng other feature representatons mght help mprove the accuracy of the algorthm. One can notce that, even for the song wth the best performance, the computed boundares of each secton were slghtly shfted from the deal boundares. Ths was manly caused by the naccuracy of the approxmate pattern matchng. To tackle ths problem, other muscal features (e.g., chord progressons, change n dynamcs, etc.) can be used to detect local events so as to locate the boundares accurately. In fact, ths problem suggests that computng only the repettve relaton mght not be suffcent for fndng the semantc structure. The poston of phrase boundares n tonal melodes relates to a number of nteractng muscal factors. The most obvous determnants of muscal phrases are the standard chord progressons known as cadence. Other factors nclude surface features such as relatvely large nterval leaps, change n dynamcs, and mcropauses ( groupng preference rules ), and repeated muscal patterns n terms of harmony, rhythm and melodc contour. [3] 4. Musc thumbnalng va structural analyss The problem of musc thumbnalng ams at fndng the most representatve part of a song, whch can be used for musc browsng and searchng. It would be helpful f the song has been segmented nto semantcally meanngful sectons before summarzaton, because, although what makes a part of a song most memorable s not clear, ths part often appears at partcular locatons wthn the structure, e.g., the begnnng or the end part of each secton. For example, among the 26 Beatles songs, 6 songs have the song ttles n the frst sentence of a secton; 9 songs have them n the last sentence of a secton; and 10 songs have them n both the frst and the last sentences of a secton. For many pop/rock songs, ttles are contaned n the hook sentences. Ths nformaton s very useful for musc thumbnalng: once we have the structure of a song, a straghtforward strategy s to choose the begnnng or the end part of the most repeated secton as the summary of the musc. For example, f ten-second summares are wanted, Table 1 shows the performance of ths strategy based on the crtera proposed by Logan and Chu [11]. The four columns n Table 1 ndcate the percentage of summares that contan a vocal porton, contan the song s ttle, are the begnnng of a secton, and are the begnnng of a phrase. The algorthm frst fnd the most repeated sectons, take the frst secton among these and truncate the begnnng ten seconds of t as the summary of the

7 song. The thumbnalng result usng ths method hghly depends on the accuracy of the structural analyss result. For example, the summary of the song From me to you usng the ptch representaton s If there's anythng that you want; If there's anythng I can do; Just call ; the summary of the song Yesterday usng the ptch representaton s Yesterday, all my troubles seemed so far away; Now t looks as though. Both of the summares start rght at the begnnng of the most repettve secton. However, the summary of song Day trpper usng the ptch representaton does not contan any vocal porton due to the poor structural analyss result of ths song. Table 1: Thumbnalng results of the 26 Beatles songs Vocal Ttle Begnnng of a secton Begnnng of a phrase Ptch 85% 23% 50% 58% FFT 88% 35% 46% 58% 5. Conclusons and future work Ths paper presents an algorthm for automatcally analyzng the repettve structure of musc from acoustc sgnals. Prelmnary results were evaluated both qualtatvely and quanttatvely, whch demonstrate the promse of the proposed method. To mprove the accuracy, more representatons need to be nvestgated. The possblty of generalzng ths method to other musc genres should also be explored. Addtonally, nferrng the herarchcal repettve structures of musc and dentfyng the functonalty of each secton wthn the structure would be a more complcated yet nterestng topc. Musc segmentaton, thumbnalng and structural analyss are three coupled problems. Dscovery of effectve methods for solvng any one of the three problems wll beneft the other two. Furthermore, the soluton to any of them depends on the study of humans percepton of musc, for example, what makes part of musc sounds lke a complete phrase and what makes t memorable or dstngushable. Human experments are always necessary for explorng such questons. Fnally, whle most prevous research on musc dgtal lbrares was based on symbolc representatons of musc (e.g., MIDI, scores), ths paper attempts to address the structural analyss problem of acoustc muscal data. We beleve that current automatc musc transcrpton technques are stll far from robust and effcent, and thus analyzng acoustc muscal data drectly wthout transcrpton s of great applcaton value for ndexng the dgtal musc repostory, segmentng musc at transtons, and summarzng the thumbnals of musc, all of whch wll beneft the users browsng and searchng n a musc dgtal lbrary. 6. References [1] J.J. Aucouturer and M. Sandler. Segmentaton of Muscal Sgnals usng Hdden Markov Models, In Proc. AES 110 th Conventon, May [2] J.J. Aucouturer and M. Sandler. Usng Long-Term Structure to Retreve Musc: Representaton and Matchng, In Proc. Internatonal Symposum on Musc Informaton Retreval, Bloomngton, IN, [3] M. Balaban, K. Ebcoglu, and O. Laske, Understandng Musc wth AI: Perspectves on Musc Cognton, Cambrdge: MIT Press; Menlo Park: AAAI Press, [4] M.A. Bartsch and G.H. Wakefeld, To Catch a Chorus: Usng Chroma-based Representatons for Audo Thumbnalng, In Proc. Workshop on Applcatons of Sgnal Processng to Audo and Acoustcs, [5] W.P. Brmngham, R.B. Dannenberg, G.H. Wakefeld, M. Bartsch, D. Bykowsk, D. Mazzon, C. Meek, M. Mellody, and W. Rand, MUSART: Musc Retreval va Aural Queres, In Proc. Internatonal Symposum on Musc Informaton Retreval, Bloomngton, IN, [6] R.B. Dannenberg and N. Hu, Pattern Dscovery Technques for Musc Audo, In Proc. Internatonal Conference on Musc Informaton Retreval, October [7] J. Foote, "Vsualzng Musc and Audo usng Self- Smlarty," In Proc. ACM Multmeda Conference, Orlando, FL, pp , [8] J. Foote, "Automatc Audo Segmentaton usng a Measure of Audo Novelty." In Proc. of IEEE Internatonal Conference on Multmeda and Expo, vol. I, pp , [9] J. Foote, ARTHUR: Retrevng Orchestral Musc by Long-Term Structure, In Proc. Internatonal Symposum on Musc Informaton Retreval, October [10] J.L. Hsu, C.C. Lu, and L.P. Chen, Dscoverng Nontrval Repeatng Patterns n Musc Data, IEEE Transactons on Multmeda, Vol. 3, No. 3, pp , September [11] B. Logan and S. Chu, Musc Summarzaton usng Key Phrases, In Proc. Internatonal Conference on Acoustcs, Speech and Sgnal Processng, [12] G. Peeters, A. L. Burthe and X. Rodet, Toward Automatc Musc Audo Summary Generaton from Sgnal Analyss,

8 In Proc. Internatonal Conference on Musc Informaton Retreval, October [13] C. Roads, The Computer Musc Tutoral, MIT Press, [14] C. Yang, Musc Database Retreval Based on Spectral Smlarty, In Proc. Internatonal Symposum on Musc Informaton Retreval, 2001.

A Fast Content-Based Multimedia Retrieval Technique Using Compressed Data

A 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 information

Music/Voice Separation using the Similarity Matrix. Zafar Rafii & Bryan Pardo

Music/Voice Separation using the Similarity Matrix. Zafar Rafii & Bryan Pardo Musc/Voce Separaton usng the Smlarty Matrx Zafar Raf & Bryan Pardo Introducton Muscal peces are often characterzed by an underlyng repeatng structure over whch varyng elements are supermposed Propellerheads

More information

UB at GeoCLEF Department of Geography Abstract

UB at GeoCLEF Department of Geography   Abstract UB at GeoCLEF 2006 Mguel E. Ruz (1), Stuart Shapro (2), June Abbas (1), Slva B. Southwck (1) and Davd Mark (3) State Unversty of New York at Buffalo (1) Department of Lbrary and Informaton Studes (2) Department

More information

Load Balancing for Hex-Cell Interconnection Network

Load 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 information

A Fast Visual Tracking Algorithm Based on Circle Pixels Matching

A 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 information

A Binarization Algorithm specialized on Document Images and Photos

A 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 information

Content Based Image Retrieval Using 2-D Discrete Wavelet with Texture Feature with Different Classifiers

Content 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 information

Parallelism for Nested Loops with Non-uniform and Flow Dependences

Parallelism 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 information

Query Clustering Using a Hybrid Query Similarity Measure

Query 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 information

An Optimal Algorithm for Prufer Codes *

An 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 information

Improvement of Spatial Resolution Using BlockMatching Based Motion Estimation and Frame. Integration

Improvement 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 information

Corner-Based Image Alignment using Pyramid Structure with Gradient Vector Similarity

Corner-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 information

Shape Representation Robust to the Sketching Order Using Distance Map and Direction Histogram

Shape 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 information

Term Weighting Classification System Using the Chi-square Statistic for the Classification Subtask at NTCIR-6 Patent Retrieval Task

Term Weighting Classification System Using the Chi-square Statistic for the Classification Subtask at NTCIR-6 Patent Retrieval Task Proceedngs of NTCIR-6 Workshop Meetng, May 15-18, 2007, Tokyo, Japan Term Weghtng Classfcaton System Usng the Ch-square Statstc for the Classfcaton Subtask at NTCIR-6 Patent Retreval Task Kotaro Hashmoto

More information

Fuzzy C-Means Initialized by Fixed Threshold Clustering for Improving Image Retrieval

Fuzzy 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 information

R s s f. m y s. SPH3UW Unit 7.3 Spherical Concave Mirrors Page 1 of 12. Notes

R 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 information

Tsinghua University at TAC 2009: Summarizing Multi-documents by Information Distance

Tsinghua 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 information

Compiler Design. Spring Register Allocation. Sample Exercises and Solutions. Prof. Pedro C. Diniz

Compiler 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 information

TN348: Openlab Module - Colocalization

TN348: 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 information

Course Introduction. Algorithm 8/31/2017. COSC 320 Advanced Data Structures and Algorithms. COSC 320 Advanced Data Structures and Algorithms

Course Introduction. Algorithm 8/31/2017. COSC 320 Advanced Data Structures and Algorithms. COSC 320 Advanced Data Structures and Algorithms Course Introducton Course Topcs Exams, abs, Proects A quc loo at a few algorthms 1 Advanced Data Structures and Algorthms Descrpton: We are gong to dscuss algorthm complexty analyss, algorthm desgn technques

More information

3D vector computer graphics

3D 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 information

Reducing Frame Rate for Object Tracking

Reducing Frame Rate for Object Tracking Reducng Frame Rate for Object Trackng Pavel Korshunov 1 and We Tsang Oo 2 1 Natonal Unversty of Sngapore, Sngapore 11977, pavelkor@comp.nus.edu.sg 2 Natonal Unversty of Sngapore, Sngapore 11977, oowt@comp.nus.edu.sg

More information

USING GRAPHING SKILLS

USING GRAPHING SKILLS Name: BOLOGY: Date: _ Class: USNG GRAPHNG SKLLS NTRODUCTON: Recorded data can be plotted on a graph. A graph s a pctoral representaton of nformaton recorded n a data table. t s used to show a relatonshp

More information

Problem Definitions and Evaluation Criteria for Computational Expensive Optimization

Problem Definitions and Evaluation Criteria for Computational Expensive Optimization Problem efntons and Evaluaton Crtera for Computatonal Expensve Optmzaton B. Lu 1, Q. Chen and Q. Zhang 3, J. J. Lang 4, P. N. Suganthan, B. Y. Qu 6 1 epartment of Computng, Glyndwr Unversty, UK Faclty

More information

MULTISPECTRAL IMAGES CLASSIFICATION BASED ON KLT AND ATR AUTOMATIC TARGET RECOGNITION

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 information

Efficient Content Representation in MPEG Video Databases

Efficient Content Representation in MPEG Video Databases Effcent Content Representaton n MPEG Vdeo Databases Yanns S. Avrths, Nkolaos D. Doulams, Anastasos D. Doulams and Stefanos D. Kollas Department of Electrcal and Computer Engneerng Natonal Techncal Unversty

More information

Subspace clustering. Clustering. Fundamental to all clustering techniques is the choice of distance measure between data points;

Subspace 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 information

Hierarchical clustering for gene expression data analysis

Hierarchical 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 information

A Deflected Grid-based Algorithm for Clustering Analysis

A Deflected Grid-based Algorithm for Clustering Analysis A Deflected Grd-based Algorthm for Clusterng Analyss NANCY P. LIN, CHUNG-I CHANG, HAO-EN CHUEH, HUNG-JEN CHEN, WEI-HUA HAO Department of Computer Scence and Informaton Engneerng Tamkang Unversty 5 Yng-chuan

More information

Array transposition in CUDA shared memory

Array transposition in CUDA shared memory Array transposton n CUDA shared memory Mke Gles February 19, 2014 Abstract Ths short note s nspred by some code wrtten by Jeremy Appleyard for the transposton of data through shared memory. I had some

More information

A Unified Framework for Semantics and Feature Based Relevance Feedback in Image Retrieval Systems

A Unified Framework for Semantics and Feature Based Relevance Feedback in Image Retrieval Systems A Unfed Framework for Semantcs and Feature Based Relevance Feedback n Image Retreval Systems Ye Lu *, Chunhu Hu 2, Xngquan Zhu 3*, HongJang Zhang 2, Qang Yang * School of Computng Scence Smon Fraser Unversty

More information

A Gradient Difference based Technique for Video Text Detection

A Gradient Difference based Technique for Video Text Detection A Gradent Dfference based Technque for Vdeo Text Detecton Palaahnakote Shvakumara, Trung Quy Phan and Chew Lm Tan School of Computng, Natonal Unversty of Sngapore {shva, phanquyt, tancl }@comp.nus.edu.sg

More information

Analysis of Continuous Beams in General

Analysis 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 information

Efficient Mean Shift Algorithm based Color Images Categorization and Searching

Efficient Mean Shift Algorithm based Color Images Categorization and Searching 152 Effcent Mean Shft Algorthm based Color Images Categorzaton and Searchng 1 Dr S K Vay, 2 Sanay Rathore, 3 Abhshek Verma and 4 Hemra Sngh Thakur 1 Professor, Head of Dept Physcs, Govt Geetanal Grl s

More information

An Image Fusion Approach Based on Segmentation Region

An 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 information

Cluster Analysis of Electrical Behavior

Cluster 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 information

A Gradient Difference based Technique for Video Text Detection

A Gradient Difference based Technique for Video Text Detection 2009 10th Internatonal Conference on Document Analyss and Recognton A Gradent Dfference based Technque for Vdeo Text Detecton Palaahnakote Shvakumara, Trung Quy Phan and Chew Lm Tan School of Computng,

More information

Machine Learning: Algorithms and Applications

Machine 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 information

CSCI 104 Sorting Algorithms. Mark Redekopp David Kempe

CSCI 104 Sorting Algorithms. Mark Redekopp David Kempe CSCI 104 Sortng Algorthms Mark Redekopp Davd Kempe Algorthm Effcency SORTING 2 Sortng If we have an unordered lst, sequental search becomes our only choce If we wll perform a lot of searches t may be benefcal

More information

SLAM Summer School 2006 Practical 2: SLAM using Monocular Vision

SLAM 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 information

Unsupervised Content Discovery in Composite Audio Rui Cai Department of Computer Science and Technology, Tsinghua Univ. Beijing, , China

Unsupervised Content Discovery in Composite Audio Rui Cai Department of Computer Science and Technology, Tsinghua Univ. Beijing, , China Unsupervsed Content Dscovery n Composte Audo Ru Ca Department of Computer Scence and Technology, Tsnghua Unv. Bejng, 100084, Chna caru01@mals.tsnghua.edu.cn Le Lu Mcrosoft Research Asa No. 49 Zhchun Road

More information

Intrinsic Plagiarism Detection Using Character n-gram Profiles

Intrinsic Plagiarism Detection Using Character n-gram Profiles Intrnsc Plagarsm Detecton Usng Character n-gram Profles Efstathos Stamatatos Unversty of the Aegean 83200 - Karlovass, Samos, Greece stamatatos@aegean.gr Abstract: The task of ntrnsc plagarsm detecton

More information

Mathematics 256 a course in differential equations for engineering students

Mathematics 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 information

Querying by sketch geographical databases. Yu Han 1, a *

Querying by sketch geographical databases. Yu Han 1, a * 4th Internatonal Conference on Sensors, Measurement and Intellgent Materals (ICSMIM 2015) Queryng by sketch geographcal databases Yu Han 1, a * 1 Department of Basc Courses, Shenyang Insttute of Artllery,

More information

A Clustering Algorithm for Key Frame Extraction Based on Density Peak

A 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 information

A mathematical programming approach to the analysis, design and scheduling of offshore oilfields

A mathematical programming approach to the analysis, design and scheduling of offshore oilfields 17 th European Symposum on Computer Aded Process Engneerng ESCAPE17 V. Plesu and P.S. Agach (Edtors) 2007 Elsever B.V. All rghts reserved. 1 A mathematcal programmng approach to the analyss, desgn and

More information

The Codesign Challenge

The Codesign Challenge ECE 4530 Codesgn Challenge Fall 2007 Hardware/Software Codesgn The Codesgn Challenge Objectves In the codesgn challenge, your task s to accelerate a gven software reference mplementaton as fast as possble.

More information

Performance Evaluation of Information Retrieval Systems

Performance Evaluation of Information Retrieval Systems Why System Evaluaton? Performance Evaluaton of Informaton Retreval Systems Many sldes n ths secton are adapted from Prof. Joydeep Ghosh (UT ECE) who n turn adapted them from Prof. Dk Lee (Unv. of Scence

More information

Positive Semi-definite Programming Localization in Wireless Sensor Networks

Positive Semi-definite Programming Localization in Wireless Sensor Networks Postve Sem-defnte Programmng Localzaton n Wreless Sensor etworks Shengdong Xe 1,, Jn Wang, Aqun Hu 1, Yunl Gu, Jang Xu, 1 School of Informaton Scence and Engneerng, Southeast Unversty, 10096, anjng Computer

More information

Chapter 6 Programmng the fnte element method Inow turn to the man subject of ths book: The mplementaton of the fnte element algorthm n computer programs. In order to make my dscusson as straghtforward

More information

Support Vector Machines

Support 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 information

Combining Multiple Resources, Evidence and Criteria for Genomic Information Retrieval

Combining Multiple Resources, Evidence and Criteria for Genomic Information Retrieval Combnng Multple Resources, Evdence and Crtera for Genomc Informaton Retreval Luo S 1, Je Lu 2 and Jame Callan 2 1 Department of Computer Scence, Purdue Unversty, West Lafayette, IN 47907, USA ls@cs.purdue.edu

More information

Classifying Acoustic Transient Signals Using Artificial Intelligence

Classifying 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 information

Object-Based Techniques for Image Retrieval

Object-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 information

COMPLEX WAVELET TRANSFORM-BASED COLOR INDEXING FOR CONTENT-BASED IMAGE RETRIEVAL

COMPLEX WAVELET TRANSFORM-BASED COLOR INDEXING FOR CONTENT-BASED IMAGE RETRIEVAL COMPLEX WAVELET TRANSFORM-BASED COLOR INDEXING FOR CONTENT-BASED IMAGE RETRIEVAL Nader Safavan and Shohreh Kasae Department of Computer Engneerng Sharf Unversty of Technology Tehran, Iran skasae@sharf.edu

More information

Programming in Fortran 90 : 2017/2018

Programming in Fortran 90 : 2017/2018 Programmng n Fortran 90 : 2017/2018 Programmng n Fortran 90 : 2017/2018 Exercse 1 : Evaluaton of functon dependng on nput Wrte a program who evaluate the functon f (x,y) for any two user specfed values

More information

Semantic Image Retrieval Using Region Based Inverted File

Semantic Image Retrieval Using Region Based Inverted File Semantc Image Retreval Usng Regon Based Inverted Fle Dengsheng Zhang, Md Monrul Islam, Guoun Lu and Jn Hou 2 Gppsland School of Informaton Technology, Monash Unversty Churchll, VIC 3842, Australa E-mal:

More information

Parallel matrix-vector multiplication

Parallel 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 information

Load-Balanced Anycast Routing

Load-Balanced Anycast Routing Load-Balanced Anycast Routng Chng-Yu Ln, Jung-Hua Lo, and Sy-Yen Kuo Department of Electrcal Engneerng atonal Tawan Unversty, Tape, Tawan sykuo@cc.ee.ntu.edu.tw Abstract For fault-tolerance and load-balance

More information

Determining Fuzzy Sets for Quantitative Attributes in Data Mining Problems

Determining Fuzzy Sets for Quantitative Attributes in Data Mining Problems Determnng Fuzzy Sets for Quanttatve Attrbutes n Data Mnng Problems ATTILA GYENESEI Turku Centre for Computer Scence (TUCS) Unversty of Turku, Department of Computer Scence Lemmnkäsenkatu 4A, FIN-5 Turku

More information

Learning-Based Top-N Selection Query Evaluation over Relational Databases

Learning-Based Top-N Selection Query Evaluation over Relational Databases Learnng-Based Top-N Selecton Query Evaluaton over Relatonal Databases Lang Zhu *, Wey Meng ** * School of Mathematcs and Computer Scence, Hebe Unversty, Baodng, Hebe 071002, Chna, zhu@mal.hbu.edu.cn **

More information

Visual Thesaurus for Color Image Retrieval using Self-Organizing Maps

Visual Thesaurus for Color Image Retrieval using Self-Organizing Maps Vsual Thesaurus for Color Image Retreval usng Self-Organzng Maps Chrstopher C. Yang and Mlo K. Yp Department of System Engneerng and Engneerng Management The Chnese Unversty of Hong Kong, Hong Kong ABSTRACT

More information

Real-Time View Recognition and Event Detection for Sports Video

Real-Time View Recognition and Event Detection for Sports Video Real-Tme Vew Recognton and Event Detecton for Sports Vdeo Authors: D Zhong and Shh-Fu Chang {dzhong, sfchang@ee.columba.edu} Department of Electrcal Engneerng, Columba Unversty For specal ssue on Multmeda

More information

Optimizing Document Scoring for Query Retrieval

Optimizing 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 information

Learning the Kernel Parameters in Kernel Minimum Distance Classifier

Learning 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 information

Lobachevsky State University of Nizhni Novgorod. Polyhedron. Quick Start Guide

Lobachevsky 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 information

Classifier Selection Based on Data Complexity Measures *

Classifier 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 information

Lecture #15 Lecture Notes

Lecture #15 Lecture Notes Lecture #15 Lecture Notes The ocean water column s very much a 3-D spatal entt and we need to represent that structure n an economcal way to deal wth t n calculatons. We wll dscuss one way to do so, emprcal

More information

FEATURE EXTRACTION. Dr. K.Vijayarekha. Associate Dean School of Electrical and Electronics Engineering SASTRA University, Thanjavur

FEATURE 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 information

AP PHYSICS B 2008 SCORING GUIDELINES

AP PHYSICS B 2008 SCORING GUIDELINES AP PHYSICS B 2008 SCORING GUIDELINES General Notes About 2008 AP Physcs Scorng Gudelnes 1. The solutons contan the most common method of solvng the free-response questons and the allocaton of ponts for

More information

The Greedy Method. Outline and Reading. Change Money Problem. Greedy Algorithms. Applications of the Greedy Strategy. The Greedy Method Technique

The 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 information

A Multi-step Strategy for Shape Similarity Search In Kamon Image Database

A Multi-step Strategy for Shape Similarity Search In Kamon Image Database A Mult-step Strategy for Shape Smlarty Search In Kamon Image Database Paul W.H. Kwan, Kazuo Torach 2, Kesuke Kameyama 2, Junbn Gao 3, Nobuyuk Otsu 4 School of Mathematcs, Statstcs and Computer Scence,

More information

TPL-Aware Displacement-driven Detailed Placement Refinement with Coloring Constraints

TPL-Aware Displacement-driven Detailed Placement Refinement with Coloring Constraints TPL-ware Dsplacement-drven Detaled Placement Refnement wth Colorng Constrants Tao Ln Iowa State Unversty tln@astate.edu Chrs Chu Iowa State Unversty cnchu@astate.edu BSTRCT To mnmze the effect of process

More information

Related-Mode Attacks on CTR Encryption Mode

Related-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 information

6.854 Advanced Algorithms Petar Maymounkov Problem Set 11 (November 23, 2005) With: Benjamin Rossman, Oren Weimann, and Pouya Kheradpour

6.854 Advanced Algorithms Petar Maymounkov Problem Set 11 (November 23, 2005) With: Benjamin Rossman, Oren Weimann, and Pouya Kheradpour 6.854 Advanced Algorthms Petar Maymounkov Problem Set 11 (November 23, 2005) Wth: Benjamn Rossman, Oren Wemann, and Pouya Kheradpour Problem 1. We reduce vertex cover to MAX-SAT wth weghts, such that the

More information

Content Based Retrieval

Content Based Retrieval 2010 2nd Internatonal Conference on Sgnal rocessng Systems (ICSS) A Novel Commercal Break Detecton and Automatc Annotaton of TV rograms for Abstract In ths paper, we present a novel approach for automatc

More information

Machine Learning. Topic 6: Clustering

Machine 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 information

Fitting: Deformable contours April 26 th, 2018

Fitting: Deformable contours April 26 th, 2018 4/6/08 Fttng: Deformable contours Aprl 6 th, 08 Yong Jae Lee UC Davs Recap so far: Groupng and Fttng Goal: move from array of pxel values (or flter outputs) to a collecton of regons, objects, and shapes.

More information

K-means and Hierarchical Clustering

K-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 information

CS 534: Computer Vision Model Fitting

CS 534: Computer Vision Model Fitting CS 534: Computer Vson Model Fttng Sprng 004 Ahmed Elgammal Dept of Computer Scence CS 534 Model Fttng - 1 Outlnes Model fttng s mportant Least-squares fttng Maxmum lkelhood estmaton MAP estmaton Robust

More information

Can We Beat the Prefix Filtering? An Adaptive Framework for Similarity Join and Search

Can We Beat the Prefix Filtering? An Adaptive Framework for Similarity Join and Search Can We Beat the Prefx Flterng? An Adaptve Framework for Smlarty Jon and Search Jannan Wang Guolang L Janhua Feng Department of Computer Scence and Technology, Tsnghua Natonal Laboratory for Informaton

More information

Determining the Optimal Bandwidth Based on Multi-criterion Fusion

Determining 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 information

Fitting & Matching. Lecture 4 Prof. Bregler. Slides from: S. Lazebnik, S. Seitz, M. Pollefeys, A. Effros.

Fitting & Matching. Lecture 4 Prof. Bregler. Slides from: S. Lazebnik, S. Seitz, M. Pollefeys, A. Effros. Fttng & Matchng Lecture 4 Prof. Bregler Sldes from: S. Lazebnk, S. Setz, M. Pollefeys, A. Effros. How do we buld panorama? We need to match (algn) mages Matchng wth Features Detect feature ponts n both

More information

A Webpage Similarity Measure for Web Sessions Clustering Using Sequence Alignment

A Webpage Similarity Measure for Web Sessions Clustering Using Sequence Alignment A Webpage Smlarty Measure for Web Sessons Clusterng Usng Sequence Algnment Mozhgan Azmpour-Kv School of Engneerng and Scence Sharf Unversty of Technology, Internatonal Campus Ksh Island, Iran mogan_az@ksh.sharf.edu

More information

Active Contours/Snakes

Active 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 information

An Entropy-Based Approach to Integrated Information Needs Assessment

An Entropy-Based Approach to Integrated Information Needs Assessment Dstrbuton Statement A: Approved for publc release; dstrbuton s unlmted. An Entropy-Based Approach to ntegrated nformaton Needs Assessment June 8, 2004 Wllam J. Farrell Lockheed Martn Advanced Technology

More information

Outline. Type of Machine Learning. Examples of Application. Unsupervised Learning

Outline. 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 information

KIDS Lab at ImageCLEF 2012 Personal Photo Retrieval

KIDS Lab at ImageCLEF 2012 Personal Photo Retrieval KD Lab at mageclef 2012 Personal Photo Retreval Cha-We Ku, Been-Chan Chen, Guan-Bn Chen, L-J Gaou, Rong-ng Huang, and ao-en Wang Knowledge, nformaton, and Database ystem Laboratory Department of Computer

More information

Video Content Representation using Optimal Extraction of Frames and Scenes

Video Content Representation using Optimal Extraction of Frames and Scenes Vdeo Content Representaton usng Optmal Etracton of rames and Scenes Nkolaos D. Doulam Anastasos D. Doulam Yanns S. Avrths and Stefanos D. ollas Natonal Techncal Unversty of Athens Department of Electrcal

More information

Research of Dynamic Access to Cloud Database Based on Improved Pheromone Algorithm

Research of Dynamic Access to Cloud Database Based on Improved Pheromone Algorithm , pp.197-202 http://dx.do.org/10.14257/dta.2016.9.5.20 Research of Dynamc Access to Cloud Database Based on Improved Pheromone Algorthm Yongqang L 1 and Jn Pan 2 1 (Software Technology Vocatonal College,

More information

BioTechnology. An Indian Journal FULL PAPER. Trade Science Inc.

BioTechnology. 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 information

SURFACE PROFILE EVALUATION BY FRACTAL DIMENSION AND STATISTIC TOOLS USING MATLAB

SURFACE PROFILE EVALUATION BY FRACTAL DIMENSION AND STATISTIC TOOLS USING MATLAB SURFACE PROFILE EVALUATION BY FRACTAL DIMENSION AND STATISTIC TOOLS USING MATLAB V. Hotař, A. Hotař Techncal Unversty of Lberec, Department of Glass Producng Machnes and Robotcs, Department of Materal

More information

A MODEL-BASED BOOK BOUNDARY DETECTION TECHNIQUE FOR BOOKSHELF IMAGE ANALYSIS

A MODEL-BASED BOOK BOUNDARY DETECTION TECHNIQUE FOR BOOKSHELF IMAGE ANALYSIS A MODEL-BASED BOOK BOUNDARY DETECTION TECHNIQUE FOR BOOKSHELF IMAGE ANALYSIS Ej TAIRA, Sech UCHIDA, Hroak SAKOE Graduate School of Informaton Scence and Electrcal Engneerng, Kyushu Unversty Faculty of

More information

A Comparison and Evaluation of Three Different Pose Estimation Algorithms In Detecting Low Texture Manufactured Objects

A Comparison and Evaluation of Three Different Pose Estimation Algorithms In Detecting Low Texture Manufactured Objects Clemson Unversty TgerPrnts All Theses Theses 12-2011 A Comparson and Evaluaton of Three Dfferent Pose Estmaton Algorthms In Detectng Low Texture Manufactured Objects Robert Krener Clemson Unversty, rkrene@clemson.edu

More information

A New Approach For the Ranking of Fuzzy Sets With Different Heights

A 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 information

CMPS 10 Introduction to Computer Science Lecture Notes

CMPS 10 Introduction to Computer Science Lecture Notes CPS 0 Introducton to Computer Scence Lecture Notes Chapter : Algorthm Desgn How should we present algorthms? Natural languages lke Englsh, Spansh, or French whch are rch n nterpretaton and meanng are not

More information

A Robust Method for Estimating the Fundamental Matrix

A Robust Method for Estimating the Fundamental Matrix Proc. VIIth Dgtal Image Computng: Technques and Applcatons, Sun C., Talbot H., Ourseln S. and Adraansen T. (Eds.), 0- Dec. 003, Sydney A Robust Method for Estmatng the Fundamental Matrx C.L. Feng and Y.S.

More information

Finding Repetitive Patterns in 3D Human Motion Captured Data

Finding Repetitive Patterns in 3D Human Motion Captured Data Fndng Repettve Patterns n 3D Human Moton Captured Data Ka-Ta Tang Department of Computer Scence, Cty Unversty of Hong Kong, Hong Kong tjeff@ctyu.edu.hk Howard Leung Department of Computer Scence, Cty Unversty

More information

Sequential search. Building Java Programs Chapter 13. Sequential search. Sequential search

Sequential search. Building Java Programs Chapter 13. Sequential search. Sequential search Sequental search Buldng Java Programs Chapter 13 Searchng and Sortng sequental search: Locates a target value n an array/lst by examnng each element from start to fnsh. How many elements wll t need to

More information

Feature Reduction and Selection

Feature Reduction and Selection Feature Reducton and Selecton Dr. Shuang LIANG School of Software Engneerng TongJ Unversty Fall, 2012 Today s Topcs Introducton Problems of Dmensonalty Feature Reducton Statstc methods Prncpal Components

More information