th International Conference on Document Analysis and Recognition

Size: px
Start display at page:

Download "th International Conference on Document Analysis and Recognition"

Transcription

1 th Internatonal Conference on Document Analyss and Recognton Onlne Handwrtten Cursve Word Recognton Usng Segmentaton-free n Combnaton wth P2DBM-MQDF Blan Zhu 1, Art Shvram 2, Srrangaraj Setlur 2, Venu Govndaraju 2 and Masak akagawa 1 1 Department of Computer and Informaton Scences, Tokyo Unversty Agrculture and Technology, Tokyo, Japan 2 Center for Unfed Bometrcs and Sensors, Unversty at Buffalo, Buffalo, US 1 {zhublan, nakagawa}@cc.tuat.ac.jp, 2 {ashvram, setlur, govnd}@buffalo.edu Abstract Ths paper descrbes an onlne handwrtten Englsh cursve word recognton method usng a segmentaton-free Markov random feld () model n combnaton wth an offlne recognton method whch uses pseudo 2D b-moment normalzaton (P2DBM) and modfed quadratc dscrmnant functon (MQDF). It extracts ture ponts along the pen-tp trace from pen-down to pen-up and uses the ture pont coordnates as unary tures and the dfferences n coordnates between the neghborng ture ponts as bnary tures. Each character s modeled as a and word s are constructed by concatenatng character s accordng to a tre lexcon of words durng recognton. Our method expands the search space usng a character-synchronous beam search strategy to search the segmentaton and recognton paths. Ths method restrcts the search paths from the tre lexcon of words and precedng paths, as well as the lengths of ture ponts durng path search. We also combne t wth a P2DBM-MQDF recognzer that s wdely used for Chnese and Japanese character recognton. Keywords Word Recognton, Segmentaton-free,, MQDF, Tre Lexcon, Beam Search I. Introducton Onlne handwrtten character recognton s recevng ncreased attenton due to the development and prolferaton of pen-based or touch-based nput devces such as tablet termnals, smart phones, electronc whteboards, and dgtal pens (e.g., the Anoto pen). Realzng onlne handwrtten character recognton wth hgh performance s vtal, especally for applcatons such as the natural nput of text on smart phones, to provde a satsfactory user experence. Hdden Markov models (HMMs) have been the domnant technque for onlne Englsh word recognton [1-2]. HMMs probablstcally treat a sequence of ture vectors n wrtng or poston order. Whle they can use the neghborhood relatonshps between the successvely adjacent ture vectors n wrtng or poston order (the so-called one-dmensonal neghborhood relatonshps), two-dmensonal neghborhood relatonshps, such as those among the geometrcally neghborng ture vectors are not explctly expressed. Even wth one-dmensonal neghborhood relatonshps, HMMs only use the state transton probabltes and unary tures, and bnary tures are not well utlzed. An addtonal drawback s that the neghborhood relatonshps among more than two neghborng ture vectors, such as ternary tures, cannot be used. Although some HMMs apply bnary tures, they only merge the bnary tures nto unary tures and use a vector of larger dmenson because HMMs do not take a new vew of the bnary tures, whch lmts recognton accuracy [3]. The model s descrbed usng an undrected graph n whch a set of random varables have the Markov property, and s can be used to effectvely ntegrate nformaton among neghborng ture vectors, such as bnary and ternary tures, and two-dmensonal neghborhood relatonshps [4]. Therefore, s have been effectvely appled to strokeanalyss-based structural offlne character recognton [5] and mage processng [4]. However, s had not been appled to onlne character recognton untl our pror work on Japanese character recognton [6]. Current onlne handwrtten character recognton approaches tend to use HMMs (note that HMMs can be vewed as specfc cases of s). s have more degrees of freedom than HMMs for explctly expressng relatons among multple ture vectors. Character segmentaton of cursve handwrtng s dffcult due to the fact that segmentaton ponts between characters are not obvous. Wthout character recognton cues and lngustc context, characters cannot be segmented unambguously. A sble way to overcome the ambguty of segmentaton s called ntegrated segmentaton and recognton, whch s classfed nto segmentaton-free and over-segmentaton-based methods [3]. The segmentaton-free method s also called as the word-based method [7], and a global ture vector s extracted from an nput word pattern and matched aganst a stored dctonary of prototype words rather than recognzng ndvdual characters. Ths word recognton method avods problems assocated wth segmentaton. HMM-based word recognton tends to use the segmentaton-free method [1-2] explotng the advantages that one-dmensonal structural methods (one-dmensonal neghborhood relatonshps) such as HMMs have to offer lke the ablty to concatenate character HMMs to construct word HMMs based on the provded lexcon of words durng recognton. On the other hand, the over-segmentaton-based method attempts to splt cursve words nto character patterns at ther true boundares and label the splt character patterns [8-9]. A word pattern s over-segmented nto prmtve segments such that each segment comprses of a sngle character or part of a character. The segments are combned to generate canddate character patterns (formng a canddate lattce), whch are evaluated usng character recognton, ncorporatng geometrc and lngustc contexts. Onlne handwrtten word recognton usng a Tme-Delay eural etwork [7] or a recurrent neural network [10] /13 $ IEEE DOI /ICDAR

2 performs word recognton by contnuously movng the nput wndow of the network across the frame sequence representaton of a word, thus generatng actvaton traces at the output of the network. These output traces are subsequently examned to determne the ASCII strng(s) best representng the word mage. Snce words have obvous lngustc context, lexcondrven methods can decrease recognton errors due to smlar characters, resultng n the wdespread use of lexcon-drven methods for word recognton. However, many prevous studes usng lexcon-drven methods have focused only on recognton models or ture extracton and have not placed much emphass on the matchng schemes and memory utlzaton whch are crtcal factors for handwrtng recognton systems runnng on personal hand-held devces such as tablets and moble phones. On these relatvely small devces, t s vtal for a handwrtng recognton system to have as small a memory footprnt as possble whle mantanng hgh accuracy. Therefore, we need to refne the matchng schemes and memory storage methods to mprove the effectveness for word recognton. Lexcon matchng schemes can be roughly dvded nto two classes: matchng the nput mage wth one lexcon entry n ts entrety each tme [8] and matchng wth all lexcon entres smultaneously [1, 9]. Although t s sad that the former scheme s sutable for a small vocabulary, we thnk that t s not effectve, even for a small vocabulary, and requres more recognton tme and memory storage. The latter often uses a tre lexcon, whch s effectve n reducng processng tme and memory storage sze. We have proposed an onlne handwrtten Japanese character recognton method usng a model and demonstrated that the model results n a hgher recognton accuracy compared to usng HMM [6]. In ths paper, we apply a segmentaton-free model of one-dmensonal structure for onlne handwrtten Englsh cursve word recognton. It extracts ture ponts along the pen-tp trace from pen-down to pen-up. It uses the ture pont coordnates as unary tures and the dfferences n coordnates between the neghborng ture ponts as bnary tures. Each character s modeled as a, and word s are constructed by concatenatng character s accordng to a tre lexcon of words durng recognton. It expands the search space usng a character-synchronous beam search strategy to search the segmentaton and recognton paths. Ths method restrcts the search paths from the tre lexcon of words and precedng paths, as well as the lengths of ture ponts durng path search. We then combne ths approach wth a P2DBM-MQDF recognzer [11-13] that s wdely used for Chnese and Japanese character recogntons. Expermental results from the new IBM_UB_1 dataset demonstrate the superorty of our method. The rest of ths paper s organzed as follows: Secton 2 begns wth the descrpton of the preprocessng steps. Secton 3 descrbes our recognton method, Secton 4 presents the expermental results, and Secton 5 presents our conclusons. Preprocessng II. We pre-process each onlne word pattern as follow. (1) Extractng ture ponts. To compute baselnes and normalze slant effcently, we frst extract ture ponts usng the method developed by Ramner [14]. The start and end ponts of every stroke are pcked up as ture ponts. Then, the most dstant pont from the straght lne between adjacent ture ponts s selected as a ture pont f the dstance to the straght lne s greater than a threshold value that s decded from the word heght. Ths selecton s done recursvely untl no more ture ponts are selected. Ths ture pont extractng process s shown n Fg. 1(a). We also reserve the orgnal nput ponts for recognton usng the P2DBM-MQDF recognzer, and use the extracted ture ponts for computng baselnes, normalzng slant and for recognton by the model. (a) (c) Fg. 1. Examples showng preprocessng of word pattern (2) Computng baselnes. We compute baselnes usng lnear regresson lnes that approxmate the local mnma (baselne) or the local maxma (corpus lne) of the trajectory as shown n Fg.1 (b). (3) ormalzng rotatons. Rotatons of words are corrected by rotatng wth an angle computed from the two lnes as shown n Fg.1 (b). (4) ormalzng slant. We normalze the dfferent slants of words, by shearng every word accordng to ts slant as shown n Fg.1 (c). The slant s determned by a hstogram over all angles subtended by the lnes connectng two successve ture ponts of the trajectory and the horzontal lne. The computed angles are weghted wth the dstance of every par of successve ponts. A smple search for the maxmum entry n the hstogram provdes us wth the slant of the word. (5) Removng delayed strokes. Delayed strokes, such as the crossng of t or the dot of, are troublesome n onlne handwrtng recognton. These strokes ntroduce addtonal temporal varaton and complcate onlne recognton. We detect such strokes and remove them before word recognton. A delayed stroke s usually a short sequence wrtten n the upper regon of the wrtng pad, above already wrtten parts of (b) (d) 350

3 a word, and accompaned by a pen movement to the left. We use some smple threshold values for characterzng these tures of delayed strokes as shown n Fg. 1 (d). (6) ormalzng sze. To ensure that the same characters have approxmately the same heght and the same Y coordnate poston for every handwrtten word, we transform every word to a gven corpus heght, where the corpus heght s the dstance between the baselne and the corpus lne, and set the baselne and the corpus lne to fxed Y coordnate postons. Recognton Method III. In ths secton, we present our segmentaton-free recognton method. A. Constructon of Tre Lexcon Frst, we construct a tre lexcon from a word database, as shown n Fg. 2. When tranng character s, we can count the range of the number (length) of ture ponts for each character class. For nstance, the length range of ture ponts for the character O s from 5 to 50 as shown n Fg. 2. Then, the length range of ture ponts to the termnal for each of the tre can be calculated accordng to the length range of ture ponts for each character class, as shown n Fg. 2, where the numbers shown n parentheses of each box are the length range of ture ponts to the termnal for the. We can restrct the searched paths by the length ranges resultng n mproved recognton accuracy. Fg. 2. Porton of tre lexcon of words B. Constructons of Character Recognzers We create two character recognzers: a recognzer and a P2DBM-MQDF recognzer. For the recognzer, we extract character patterns from true segmentaton ponts of preprocessed tranng word patterns and shft the X coordnates of each character pattern so that the mnmum of X coordnate s 0 whle the Y coordnates are unchanged. Here, we employ ture ponts extracted n the preprocessng. We use the ture ponts of obtaned character patterns to tran the model of each character category [6]. For the P2DBM-MQDF recognzer, we extract character patterns from the orgnal nput ponts of tranng word patterns and normalze each character pattern to a standard sze when extractng the tures. Ths s because the character recognton-based approach depends on the performance of each character recognton and latest normalzaton mproves t sgnfcantly. From each character pattern (a sequence of stroke coordnates), we extract drectonal tures: hstograms of normalzed stroke drecton [11]. For coordnate normalzaton, we apply P2DBM [12]. The local stroke drecton s decomposed nto eght drectons, and from the ture map of each drecton, 8x8 values are extracted by Gaussan blurrng so that the dmensonalty of ture vectors s 512. To mprove the Gaussanty of ture dstrbuton, each value of the 512 tures s transformed by the Box-Cox transformaton (also called varable transformaton). The nput ture vector s reduced from 512D to nd by the Fsher lnear dscrmnant analyss (FLDA) [3]. Then we use the nd ture vectors to create a MQDF model [13]. Fg. 3. Extracted ture ponts of word Offer C. Path Search We expand the search space usng a character-synchronous method and apply the beam search strategy to match the ture ponts of each nput pattern wth the states of word models, and obtan a smlarty for each word. Fg. 3 shows an example of the extracted ture ponts of a word pattern. We match the ture ponts F={f 1, f 2, f 3,,f 56 } wth the states of word models. We show an example of a search usng a character-synchronous lexcon-drven method n Fg. 4 to descrbe our process flow. We conduct the search and expanson from the expanson depth d 1 to d 5. We frst search the tre lexcon from ts start s. In Fg. 2, the start s are [O], [p], and [], and based on these we expand the root and set ts chldren s 1-1 wth a character category [O], 1-2 wth a character category [p] and 1-3 wth a character category [], where each chld has a character category C and a start pont of a ture sequence. In ths, they share the start pont f 1. For each, the length from the start pont f 1 to the termnal pont f 56 s 56 so that the 1-3 s erased because the length range to the termnal of the tre [] s 9-44 and does not satsfy the length from the start pont to the termnal pont. For each of 1-1 and 1-2, we match the ture ponts from each start pont f to the pont f +M wth the states S={s 1, s 2, s 3,,s J } of the character model of the correspondng C, where M s the maxmum number of the ture ponts of C and t s calculated from tranng patterns. Fg. 5 shows an example of the matchng process. Before matchng, we shft the X coordnates of the ture ponts from f to f +M so that the mnmum of X coordnate s 0. We apply the Vterb search to match. Then we can get paths at the [End] state whch correspond to some end ponts such as f 16, f 17 and f 18 for 1-1.We sort the scores of the paths (note that each score S s normalzed by the number of the ture ponts from the start pont to the end pont of the path to S / ), and select top paths. In Fg. 4, s set as three and three paths (from f 1 to f 16, from f 1 to f 17, from f 1 to f 18 )of 1-1, and three paths (from f 1 to f 12, from f 1 to f 15, from f 1 to f 18 )of 1-2 are selected, and we call them sub-s. ode 1-1 has three sub-s ( , and ) whle 1-2 has three sub-s ( , and ). 351

4 Fg. 4. Search and Recognton Fg. 5. Matchng between ture ponts and character For all the sub-s up to the depth d 1, we evaluate all search paths accordng to the path evaluaton crteron, sort them, then only select several top paths and erase others. The number of the selected top paths s called the beam band. In Fg. 4, the beam band s set as two, and two paths endng at and are selected for d 1. We try the followng three path evaluaton crtera: E E 1 S P2DBM MQDF 1 S 1 P2DBM-MQDF 1 E P2DBM-MQDF S SP2DBM-MQDF P2DBM-MQDF 1 1 where E, E P2DBM-MQDF and E +P2DBM-MQDF stand for the evaluaton crteron usng the score of the recognzer alone, that of usng the score of the P2DBM-MQDF recognzer alone and that of usng both scores respectvely, S, S P2DM-MQDF,,, σ and σ P2DBM-MQDF are the score of the recognzer for sub- sub,the score of the P2DBM-MQDF recognzer for sub, the number of the ture ponts from the start pont to the end pont of sub,the number of sub-s n the path, the varance of the recognzer scores, and the varance of the P2DBM-MQDF recognzer scores, respectvely. We use the number of the ture ponts n each path to normalze the path evaluaton crteron of the path. Scnce S has been normalzed by when matchng ture ponts wth a character, we need to multply S by. We consder that the score of the P2DBM-MQDF recognzer reflects the shape of the combnaton of the all ture ponts of, the more the ture ponts are, the lower the score. Therefore, we do not multply t by. After selectng the end ponts and decdng the sub-s for each k-j by matchng as shown n Fg. 5, we can extract the orgnal nput ponts from the start pont to the end pont of each sub- sub, and nput them to (1) the P2DBM-MQDF recognzer to obtan the score S P2DM- MQDF. To combne the two scores S, S P2DM-MQDF, we normalze them by ther respectve score varances. Then, we apply the same method to process the next depth. Fnally, the expanson reaches to the depth d 5. For all s up to the depth d 5, we evaluate all paths accordng to the path evaluaton crteron, sort them, and then select the optmal path as the recognton result. For each k-j wth a character category C and a start pont f, we need to execute matchng to decde ts end ponts and sub-s. After that we need to extract tures to recognze each sub- by the P2DBM-MQDF recognzer. Dfferent s may be requred for the same par of C and f. Therefore, for each par of C and f, once ts end ponts and sub-s are decded and the scores of ts sub-s are calculated, we store them and use them for other s from the second tme. For each par of start and end ponts, we also store the tures for the P2DBM-MQDF recognzer, and use them for the second pass. Ths can greatly mprove recognton tme. We call ths storage strategy of scores and tures (SSSF). Experments IV. We evaluate the word recognzers usng the new IBM_UB_1 dataset orgnally collected by IBM and beng released by the Unversty at Buffalo (CUBS). IBM_UB_1 contans cursve onlne handwrtng n Englsh wrtten by 43 wrters. A set of 10 topc scrpts were generated each of whch conssted of 100 or more than 100 documents relatng to the correspondng topc. For each document wrtten by a specfc wrter, there s a summary text and a correspondng query text. The summary text contans one or two pages of wrtng on a partcular topc, whle the query text contans approxmately 25 words that encapsulate the summary text and thus may be used to retreve the correspondng summares. Our experments use the query text from 43 wrters. ether IBM_UB_1 nor UIPE database [15] has word patterns wth both word labels and character-level segmentaton labels. We need both the labels to tran our character recognzers. Fortunately, we found that n tran r01 v07 of the UIPE database, Benchmark#3 (solated characters) and Benchmark#6 (solated words) share four common sets ( art data wth 6 wrters, cea data wth 6 wrters, ceb data wth 4 wrters and, ka data wth 28 wrters). We combned these sets to obtan a set wth the word labels and character-level segmentaton labels. The data set contans 14,691 characters and 2,127 words for 44 wrters. At the begnnng, we used the data set to tran the and the P2DBM-MQDF recognzers. Then we set the tre lexcon to have only one word (the word strng label of the recognzed word pattern n IBM_UB_1), and used the traned recognzer to recognze each word pattern of IBM_UB_1 and set the recognzed segmentaton result as the character segmentaton labels of the pattern. After that, we obtaned character-level segmentaton labels for the word patterns of IBM_UB_1. We selected four pages at random for each of 20 wrters as testng data, and used the remanng data as tranng data. We 352

5 used the tranng data to tran the and the P2DBM- MQDF recognzers. Then, we tested the recognton rate for the testng data. The tranng data has 61,105 words and 355,895 characters of 62 categores (dgt, uppercase and lowercase Latn alphabet), whle the testng data has 1,795 words and 10,987 characters of 62 categores. The constructed tre lexcon contans 5,000 words. The experments were mplemented on an Intel(R) Core(TM) 2 Quad CPU 2.83GHz 2.83 GHz wth 4.00 GB memory. The ture vector s reduced from 512D to 61D by the FLDA for the P2DBM-MQDF recognzer. s set as 100 and the beam band s set as 1,000. Table 1 shows the results. TABLE I. RECOGITIO RESULTS Method Performance E EP2DBM- MQDF E+P2DBM- MQDF E+P2DBM-MQDF wthout length range restrctng Word rec. rate (%) From these results, we can see that E +P2DBM-MQDF sgnfcantly mproves the recognton accuracy by combnng model wth a P2DBM-MQDF recognzer. The model apples a structural method that s weak at collectng global character nformaton, but robust aganst character shape varatons. The 2DBM-MQDF recognzer uses an unstructural method that s robust aganst noses but weak aganst character shape varatons. By combnng a structural method wth an un-structural method, the recognton accuracy mproves snce they compensate for ther respectve dsadvantages. Restrctng the searched paths by the length range can brng hgher recognton accuracy. The recognton rate for E P2DBM-MQDF s very low and we beleve ths s because the P2DBM-MQDF recognzer renormalzes each character pattern so that t loses the nformaton of character szes and character postons. We have realzed a word recognton rate of 76.89% whch s comparable to the performance reported usng a recurrent neural network [10]. The memory consumpton of our recognzer s about 8.5MB. We also nvestgated the results of dfferent beam bands. Table 2 shows the results, where the recognton tme s the average for recognzng a word. TABLE II. RECOGITIO RESULTS FOR BEAM BADS Performance Beam band Word rec. rate (%) Recognton tme (s) We can see that the larger the beam band, the hgher the recognton rate and the longer the recognton tme. The recognton speeds are stll low and we need to ncorporate our system wth a real-tme recognton approach [16] to mprove the recognton speed. We also evaluate the SSSF and got a result that t can mprove recognton speed about 89 tmes when not usng SSSF for beam band 300. V. Concluson Ths paper presented a method for onlne handwrtten Englsh cursve word recognton usng segmentaton-free model. We restrcted the search paths from the tre lexcon of words and precedng paths, as well as the lengths of ture ponts durng path search by the character-synchronous beam search strategy. We combned t wth a P2DBM- MQDF recognzer beng wdely used for Chnese and Japanese character recognton. Expermental results demonstrate a sgnfcant mprovement n recognton accuracy by combnng our segmentaton-free model wth a P2DBM-MQDF recognzer, and by restrctng the searched paths consderng the length range. We have already shown that combnng model wth P2DBM-MQDF acheves very hgh recognton rate for onlne handwrtten Japanese text [17]. In ths paper, we have shown ths combnaton s also very effectve and successful for onlne Englsh word recognton. Acknowledgment Ths work s beng supported by the Grant-n-Ad for Scentfc Research (B) References [1] S. Jaeger, S. Manke, J. Rechert and A. Wabel, Onlne handwrtng recognton: the pen++ recognzer, IJDAR, 3(1), pp , [2] M. Lwck and H. Bunke, HMM-based on-lne recognton of handwrtten whteboard notes, Proc. 10th IWFHR, pp , [3] M. Cheret,. Kharma, C.-L. Lu and C. Y. Suen, Character recognton systems, A Gude for Students and Practoners, John Wley & Sons, Inc., Hoboken, ew Jersey, [4] S. Z. L, Markov random feld modelng n mage analyss, Sprnger, Tokyo, [5] J. Zeng and Z.-Q. Lu, Markov random feld-based statstcal character structure modelng for handwrtten Chnese character recognton, IEEE Trans. PAMI, 30(5), [6] B. Zhu and M. akagawa, On-lne handwrtten Japanese characters recognton usng a model wth parameter optmzaton by CRF, Proc. 11th ICADR, pp , [7] G. Sen., Large vocabulary recognton of on-lne handwrtten cursve words, PhD thess, Department of Computer Scence of the State Unversty of ew York at Buffalo,.Y., USA, [8] G. Km and V. Govndaraju, A lexcon drven approach to handwrtten word recognton for real-tme applcatons, IEEE Trans. PAMI, 19(4), pp , [9] C.-L. Lu, M. Koga and H. Fujsawa, Lexcon-drven segmentaton and recognton of handwrtten character strngs for Japanese address readng, IEEE Trans. PAMI, 24(11), pp , [10] M. Lwck, A. Graves, S. Fern andez, H. Bunke, and J. Schmdhuber, A novel approach to on-lne handwrtng recognton based on bdrectonal long short-term memory networks, In Proc. 9th ICDAR, pp , [11] C.-L. Lu and X.-D. Zhou, Onlne Japanese character recognton usng trajectory-based normalzaton and drecton ture extracton, Proc. 10th IWFHR, pp , [12] C.-L. Lu and K. Marukawa, Pseudo two-dmensonal shape normalzaton methods for handwrtten Chnese character recognton, Pattern Recognton, 38(12), pp , [13] F. Kmura, Modfed duadratc dscrmnant functon and the applcaton to Chnese characters, IEEE Trans.PAMI, 9 (1), pp , [14] U. Ramer, An teratve procedure for the polygonal approxmaton of plan closed curves, Computer Graphcs and Image Processng, 1(3), pp , [15] I. Guyon, L. Schomaker, R. Plamondon,M. Lberman, and S. Janet. Unpen project of on-lne data exchange and recognzer benchmarks, Proc. 12 th ICPR, pp , [16] D.-H. Wang, C.-L. Lu, and X.-D. Zhou, An approach for real-tme recognton of onlne Chnese handwrtten sentences, Pattern Recognton, (45), pp , [17] B. Zhu, J. Gao and M. akagawa, Objectve functon desgn for MCEbased combnaton of on-lne and off-lne character recognzers for onlne handwrtten Japanese text recognton, Proc. 11th ICDAR, pp ,

Online Handwritten Cursive Word Recognition Using Segmentation-free and Segmentation-based Methods

Online Handwritten Cursive Word Recognition Using Segmentation-free and Segmentation-based Methods 3th Asan Conference on Pattern Recognton (ACPR2015), November 3-6, 2015, Kuala Lumpur, Malasa. Onlne Handwrtten Cursve Word Recognton Usng Segmentaton-free and Segmentaton-based Methods Blan Zhua, Art

More information

Segmentation Based Online Word Recognition: A Conditional Random Field Driven Beam Search Strategy

Segmentation Based Online Word Recognition: A Conditional Random Field Driven Beam Search Strategy 2013 12th Internatonal Conference on Document Analyss and Recognton Segmentaton Based Onlne Word Recognton: A Condtonal Random Feld Drven Beam Search Strategy Art Shvram 1, Blan Zhu 2, Srrangaraj Setlur

More information

An Approach to Real-Time Recognition of Chinese Handwritten Sentences

An Approach to Real-Time Recognition of Chinese Handwritten Sentences An Approach to Real-Tme Recognton of Chnese Handwrtten Sentences Da-Han Wang, Cheng-Ln Lu Natonal Laboratory of Pattern Recognton, Insttute of Automaton of Chnese Academy of Scences, Bejng 100190, P.R.

More information

Accumulated-Recognition-Rate Normalization for Combining Multiple On/Off-Line Japanese Character Classifiers Tested on a Large Database

Accumulated-Recognition-Rate Normalization for Combining Multiple On/Off-Line Japanese Character Classifiers Tested on a Large Database 4 th Internatonal Workshop on Multple Classfer Systems (MCS23) Guldford, UK Accumulated-Recognton-Rate Normalzaton for Combnng Multple On/Off-Lne Japanese Character Classfers Tested on a Large Database

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

Incremental MQDF Learning for Writer Adaptive Handwriting Recognition 1

Incremental MQDF Learning for Writer Adaptive Handwriting Recognition 1 200 2th Internatonal Conference on Fronters n Handwrtng Recognton Incremental MQDF Learnng for Wrter Adaptve Handwrtng Recognton Ka Dng, Lanwen Jn * School of Electronc and Informaton Engneerng, South

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

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

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

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

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

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

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

Skew Angle Estimation and Correction of Hand Written, Textual and Large areas of Non-Textual Document Images: A Novel Approach

Skew 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 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

An Empirical Comparative Study of Online Handwriting Chinese Character Recognition:Simplified v.s.traditional

An Empirical Comparative Study of Online Handwriting Chinese Character Recognition:Simplified v.s.traditional 2013 12th Internatonal Conference on Document Analyss and Recognton An Emprcal Comparatve Study of Onlne Handwrtng Chnese Recognton:Smplfed v.s.tradtonal Yan Gao, Lanwen Jn +, Wexn Yang School of Electronc

More information

Using Fuzzy Logic to Enhance the Large Size Remote Sensing Images

Using 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 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

S1 Note. Basis functions.

S1 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 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

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

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

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

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

Foreground and Background Information in an HMM-based Method for Recognition of Isolated Characters and Numeral Strings

Foreground and Background Information in an HMM-based Method for Recognition of Isolated Characters and Numeral Strings Foreground and Background Informaton n an HMM-based Method for Recognton of Isolated Characters and Numeral Strngs Alceu de S. Brtto Jr a,b, Robert Sabourn c,d, Flavo Bortolozz a,chng Y. Suen d a Pontfíca

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

Description of NTU Approach to NTCIR3 Multilingual Information Retrieval

Description of NTU Approach to NTCIR3 Multilingual Information Retrieval Proceedngs of the Thrd NTCIR Workshop Descrpton of NTU Approach to NTCIR3 Multlngual Informaton Retreval Wen-Cheng Ln and Hsn-Hs Chen Department of Computer Scence and Informaton Engneerng Natonal Tawan

More information

A Novel Adaptive Descriptor Algorithm for Ternary Pattern Textures

A Novel Adaptive Descriptor Algorithm for Ternary Pattern Textures A Novel Adaptve Descrptor Algorthm for Ternary Pattern Textures Fahuan Hu 1,2, Guopng Lu 1 *, Zengwen Dong 1 1.School of Mechancal & Electrcal Engneerng, Nanchang Unversty, Nanchang, 330031, Chna; 2. School

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

Signature and Lexicon Pruning Techniques

Signature and Lexicon Pruning Techniques Sgnature and Lexcon Prunng Technques Srnvas Palla, Hansheng Le, Venu Govndaraju Centre for Unfed Bometrcs and Sensors Unversty at Buffalo {spalla2, hle, govnd}@cedar.buffalo.edu Abstract Handwrtten word

More information

Recognition of Handwritten Numerals Using a Combined Classifier with Hybrid Features

Recognition of Handwritten Numerals Using a Combined Classifier with Hybrid Features Recognton of Handwrtten Numerals Usng a Combned Classfer wth Hybrd Features Kyoung Mn Km 1,4, Joong Jo Park 2, Young G Song 3, In Cheol Km 1, and Chng Y. Suen 1 1 Centre for Pattern Recognton and Machne

More information

Edge Detection in Noisy Images Using the Support Vector Machines

Edge 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 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

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

Detection of an Object by using Principal Component Analysis

Detection 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 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

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

An Iterative Solution Approach to Process Plant Layout using Mixed Integer Optimisation

An Iterative Solution Approach to Process Plant Layout using Mixed Integer Optimisation 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 An Iteratve Soluton Approach to Process Plant Layout usng Mxed

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

Pictures at an Exhibition

Pictures 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 information

Smoothing Spline ANOVA for variable screening

Smoothing Spline ANOVA for variable screening Smoothng Splne ANOVA for varable screenng a useful tool for metamodels tranng and mult-objectve optmzaton L. Rcco, E. Rgon, A. Turco Outlne RSM Introducton Possble couplng Test case MOO MOO wth Game Theory

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

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

HYBRID OFF-LINE OCR FOR ISOLATED HANDWRITTEN GREEK CHARACTERS

HYBRID OFF-LINE OCR FOR ISOLATED HANDWRITTEN GREEK CHARACTERS HYBRID OFF-LINE OCR FOR ISOLATED HANDWRITTEN GREEK CHARACTERS G. amvakas,2, B. Gatos, I. ratkaks, N. Stamatopoulos,2, A. Ronots 2, S.J. erantons Computatonal Intellgence Laboratory Insttute of Informatcs

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

Local Quaternary Patterns and Feature Local Quaternary Patterns

Local 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 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

Collaboratively Regularized Nearest Points for Set Based Recognition

Collaboratively 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 information

Face Recognition University at Buffalo CSE666 Lecture Slides Resources:

Face 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 information

BOOSTING CLASSIFICATION ACCURACY WITH SAMPLES CHOSEN FROM A VALIDATION SET

BOOSTING CLASSIFICATION ACCURACY WITH SAMPLES CHOSEN FROM A VALIDATION SET 1 BOOSTING CLASSIFICATION ACCURACY WITH SAMPLES CHOSEN FROM A VALIDATION SET TZU-CHENG CHUANG School of Electrcal and Computer Engneerng, Purdue Unversty, West Lafayette, Indana 47907 SAUL B. GELFAND School

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

VISUAL SELECTION OF SURFACE FEATURES DURING THEIR GEOMETRIC SIMULATION WITH THE HELP OF COMPUTER TECHNOLOGIES

VISUAL SELECTION OF SURFACE FEATURES DURING THEIR GEOMETRIC SIMULATION WITH THE HELP OF COMPUTER TECHNOLOGIES UbCC 2011, Volume 6, 5002981-x manuscrpts OPEN ACCES UbCC Journal ISSN 1992-8424 www.ubcc.org VISUAL SELECTION OF SURFACE FEATURES DURING THEIR GEOMETRIC SIMULATION WITH THE HELP OF COMPUTER TECHNOLOGIES

More information

Type-2 Fuzzy Non-uniform Rational B-spline Model with Type-2 Fuzzy Data

Type-2 Fuzzy Non-uniform Rational B-spline Model with Type-2 Fuzzy Data Malaysan Journal of Mathematcal Scences 11(S) Aprl : 35 46 (2017) Specal Issue: The 2nd Internatonal Conference and Workshop on Mathematcal Analyss (ICWOMA 2016) MALAYSIAN JOURNAL OF MATHEMATICAL SCIENCES

More information

PRÉSENTATIONS DE PROJETS

PRÉSENTATIONS DE PROJETS PRÉSENTATIONS DE PROJETS Rex Onlne (V. Atanasu) What s Rex? Rex s an onlne browser for collectons of wrtten documents [1]. Asde ths core functon t has however many other applcatons that make t nterestng

More information

High-Boost Mesh Filtering for 3-D Shape Enhancement

High-Boost Mesh Filtering for 3-D Shape Enhancement Hgh-Boost Mesh Flterng for 3-D Shape Enhancement Hrokazu Yagou Λ Alexander Belyaev y Damng We z Λ y z ; ; Shape Modelng Laboratory, Unversty of Azu, Azu-Wakamatsu 965-8580 Japan y Computer Graphcs Group,

More information

Online Character Segmentation Method for Unconstrained Handwriting Strings Using Off-stroke Features

Online Character Segmentation Method for Unconstrained Handwriting Strings Using Off-stroke Features Onlne Character Segmentaton Method for Unconstraned Handwrtng Strngs Usng Off-stroke Features Naohro Furukawa, Junko Tokuno, Hsash Ikeda To cte ths verson: Naohro Furukawa, Junko Tokuno, Hsash Ikeda. Onlne

More information

Unsupervised Learning

Unsupervised 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 information

The Research of Support Vector Machine in Agricultural Data Classification

The Research of Support Vector Machine in Agricultural Data Classification The Research of Support Vector Machne n Agrcultural Data Classfcaton Le Sh, Qguo Duan, Xnmng Ma, Me Weng College of Informaton and Management Scence, HeNan Agrcultural Unversty, Zhengzhou 45000 Chna Zhengzhou

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

Discriminative Dictionary Learning with Pairwise Constraints

Discriminative Dictionary Learning with Pairwise Constraints Dscrmnatve Dctonary Learnng wth Parwse Constrants Humn Guo Zhuoln Jang LARRY S. DAVIS UNIVERSITY OF MARYLAND Nov. 6 th, Outlne Introducton/motvaton Dctonary Learnng Dscrmnatve Dctonary Learnng wth Parwse

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

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

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

Improving online handwritten mathematical expressions recognition with contextual modeling

Improving online handwritten mathematical expressions recognition with contextual modeling Improvng onlne handwrtten mathematcal expressons recognton wth contextual modelng Ahmad-Montaser Awal, Harold Mouchère, Chrstan Vard-Gaudn IRCCyN/IVC UMR CNRS 6597 Ecole polytechnque de l unversté de Nantes

More information

MOTION BLUR ESTIMATION AT CORNERS

MOTION BLUR ESTIMATION AT CORNERS Gacomo Boracch and Vncenzo Caglot Dpartmento d Elettronca e Informazone, Poltecnco d Mlano, Va Ponzo, 34/5-20133 MILANO boracch@elet.polm.t, caglot@elet.polm.t Keywords: Abstract: Pont Spread Functon Parameter

More information

Sum of Linear and Fractional Multiobjective Programming Problem under Fuzzy Rules Constraints

Sum of Linear and Fractional Multiobjective Programming Problem under Fuzzy Rules Constraints Australan Journal of Basc and Appled Scences, 2(4): 1204-1208, 2008 ISSN 1991-8178 Sum of Lnear and Fractonal Multobjectve Programmng Problem under Fuzzy Rules Constrants 1 2 Sanjay Jan and Kalash Lachhwan

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

Human Face Recognition Using Generalized. Kernel Fisher Discriminant

Human Face Recognition Using Generalized. Kernel Fisher Discriminant Human Face Recognton Usng Generalzed Kernel Fsher Dscrmnant ng-yu Sun,2 De-Shuang Huang Ln Guo. Insttute of Intellgent Machnes, Chnese Academy of Scences, P.O.ox 30, Hefe, Anhu, Chna. 2. Department of

More information

Writer Identification using a Deep Neural Network

Writer Identification using a Deep Neural Network Wrter Identfcaton usng a Deep Neural Network Jun Chu and Sargur Srhar Department of Computer Scence and Engneerng Unversty at Buffalo, The State Unversty of New York Buffalo, NY 1469, USA {jchu6, srhar}@buffalo.edu

More information

Outline. Self-Organizing Maps (SOM) US Hebbian Learning, Cntd. The learning rule is Hebbian like:

Outline. 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 information

A MOVING MESH APPROACH FOR SIMULATION BUDGET ALLOCATION ON CONTINUOUS DOMAINS

A MOVING MESH APPROACH FOR SIMULATION BUDGET ALLOCATION ON CONTINUOUS DOMAINS Proceedngs of the Wnter Smulaton Conference M E Kuhl, N M Steger, F B Armstrong, and J A Jones, eds A MOVING MESH APPROACH FOR SIMULATION BUDGET ALLOCATION ON CONTINUOUS DOMAINS Mark W Brantley Chun-Hung

More information

An efficient method to build panoramic image mosaics

An efficient method to build panoramic image mosaics An effcent method to buld panoramc mage mosacs Pattern Recognton Letters vol. 4 003 Dae-Hyun Km Yong-In Yoon Jong-Soo Cho School of Electrcal Engneerng and Computer Scence Kyungpook Natonal Unv. Abstract

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

A Computer Vision System for Automated Container Code Recognition

A 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 information

3D Virtual Eyeglass Frames Modeling from Multiple Camera Image Data Based on the GFFD Deformation Method

3D Virtual Eyeglass Frames Modeling from Multiple Camera Image Data Based on the GFFD Deformation Method NICOGRAPH Internatonal 2012, pp. 114-119 3D Vrtual Eyeglass Frames Modelng from Multple Camera Image Data Based on the GFFD Deformaton Method Norak Tamura, Somsangouane Sngthemphone and Katsuhro Ktama

More information

Lecture 5: Multilayer Perceptrons

Lecture 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 information

Bootstrapping Structured Page Segmentation

Bootstrapping Structured Page Segmentation Bootstrappng Structured Page Segmentaton Huanfeng Ma and Davd Doermann Laboratory for Language and Meda Processng Insttute for Advanced Computer Studes (UMIACS) Unversty of Maryland, College Park, MD {hfma,

More information

Histogram of Template for Pedestrian Detection

Histogram 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 information

A Background Subtraction for a Vision-based User Interface *

A Background Subtraction for a Vision-based User Interface * A Background Subtracton for a Vson-based User Interface * Dongpyo Hong and Woontack Woo KJIST U-VR Lab. {dhon wwoo}@kjst.ac.kr Abstract In ths paper, we propose a robust and effcent background subtracton

More information

Image Representation & Visualization Basic Imaging Algorithms Shape Representation and Analysis. outline

Image Representation & Visualization Basic Imaging Algorithms Shape Representation and Analysis. outline mage Vsualzaton mage Vsualzaton mage Representaton & Vsualzaton Basc magng Algorthms Shape Representaton and Analyss outlne mage Representaton & Vsualzaton Basc magng Algorthms Shape Representaton and

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

Snakes-based approach for extraction of building roof contours from digital aerial images

Snakes-based approach for extraction of building roof contours from digital aerial images Snakes-based approach for extracton of buldng roof contours from dgtal aeral mages Alur P. Dal Poz and Antono J. Fazan São Paulo State Unversty Dept. of Cartography, R. Roberto Smonsen 305 19060-900 Presdente

More information

X- Chart Using ANOM Approach

X- 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 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

Solving two-person zero-sum game by Matlab

Solving two-person zero-sum game by Matlab Appled Mechancs and Materals Onlne: 2011-02-02 ISSN: 1662-7482, Vols. 50-51, pp 262-265 do:10.4028/www.scentfc.net/amm.50-51.262 2011 Trans Tech Publcatons, Swtzerland Solvng two-person zero-sum game by

More information

Fast Feature Value Searching for Face Detection

Fast Feature Value Searching for Face Detection Vol., No. 2 Computer and Informaton Scence Fast Feature Value Searchng for Face Detecton Yunyang Yan Department of Computer Engneerng Huayn Insttute of Technology Hua an 22300, Chna E-mal: areyyyke@63.com

More information

Outline. Discriminative classifiers for image recognition. Where in the World? A nearest neighbor recognition example 4/14/2011. CS 376 Lecture 22 1

Outline. 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 information

Dynamic wetting property investigation of AFM tips in micro/nanoscale

Dynamic wetting property investigation of AFM tips in micro/nanoscale Dynamc wettng property nvestgaton of AFM tps n mcro/nanoscale The wettng propertes of AFM probe tps are of concern n AFM tp related force measurement, fabrcaton, and manpulaton technques, such as dp-pen

More information

Gender Classification using Interlaced Derivative Patterns

Gender Classification using Interlaced Derivative Patterns Gender Classfcaton usng Interlaced Dervatve Patterns Author Shobernejad, Ameneh, Gao, Yongsheng Publshed 2 Conference Ttle Proceedngs of the 2th Internatonal Conference on Pattern Recognton (ICPR 2) DOI

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

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

A Study on the Application of Spatial-Knowledge-Tags using Human Motion in Intelligent Space

A Study on the Application of Spatial-Knowledge-Tags using Human Motion in Intelligent Space A Study on the Applcaton of Spatal-Knowledge-Tags usng Human Moton n Intellgent Space Tae-Seok Jn*, Kazuyuk Moroka**, Mhoko Ntsuma*, Takesh Sasak*, and Hdek Hashmoto * * Insttute of Industral Scence, the

More information

Face Recognition Based on SVM and 2DPCA

Face 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 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

Research and Application of Fingerprint Recognition Based on MATLAB

Research and Application of Fingerprint Recognition Based on MATLAB Send Orders for Reprnts to reprnts@benthamscence.ae The Open Automaton and Control Systems Journal, 205, 7, 07-07 Open Access Research and Applcaton of Fngerprnt Recognton Based on MATLAB Nng Lu* Department

More information

Meta-heuristics for Multidimensional Knapsack Problems

Meta-heuristics for Multidimensional Knapsack Problems 2012 4th Internatonal Conference on Computer Research and Development IPCSIT vol.39 (2012) (2012) IACSIT Press, Sngapore Meta-heurstcs for Multdmensonal Knapsack Problems Zhbao Man + Computer Scence Department,

More information

y and the total sum of

y and the total sum of Lnear regresson Testng for non-lnearty In analytcal chemstry, lnear regresson s commonly used n the constructon of calbraton functons requred for analytcal technques such as gas chromatography, atomc absorpton

More information

A PATTERN RECOGNITION APPROACH TO IMAGE SEGMENTATION

A 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 information

Reading. 14. Subdivision curves. Recommended:

Reading. 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 information