Competitive Sparse Representation Classification for Face Recognition
|
|
- Dominick Thomas
- 5 years ago
- Views:
Transcription
1 Vol. 6, No. 8, 05 Compettve Sparse Representaton Classfcaton for Face Recognton Yng Lu Chongqng Key Laboratory of Computatonal Intellgence Chongqng Unversty of Posts and elecommuncatons Chongqng, Chna Jan-Xun M Chongqng Key Laboratory of Computatonal Intellgence Chongqng Unversty of Posts and elecommuncatons Chongqng, Chna Cong L Chongqng Key Laboratory of Computatonal Intellgence Chongqng Unversty of Posts and elecommuncatons Chongqng, Chna Chao L Chongqng Key Laboratory of Computatonal Intellgence Chongqng Unversty of Posts and elecommuncatons Chongqng, Chna Abstract A method, named compettve sparse representaton classfcaton (CSRC), s proposed for face recognton n ths paper. CSRC ntroduces a lowest compettve deleton mechansm whch removes the lowest compettve sample based on the compettve ablty of tranng samples for representng a probe n multple rounds collaboratve lnear representaton. In other words, n each round of competng, whether a tranng sample s retaned or not n the next round depends on the ablty of representng the nput probe. Because of the number of tranng samples used for representng the probe decreases n CSRC, the codng vector s transformed nto a low dmensonal space comparng wth the ntal codng vector. hen the sparse representaton makes CSRC dscrmnatve for classfyng the probe. In addton, due to the fast algorthm, the FR system has less computatonal cost. o verfy the valdty of CSRC, we conduct a seres of experments on AR, Extended YB, and ORL databases respectvely. Keywords face recognton; collaboratve representaton sparse representaton; and compettve representaton I. INRODUCION Face Recognton (FR) has become to a hot research area for ts convenence n daly lfe. Recently, lnear representaton methods are very popular whch represent the probe wth tranng samples from gallery set. Collaboratve representaton (CR) method has acheved good performance for FR [-4],n whch a gven testng mage y can be represented by a tranng set A wth a codng vector x,.e. y=ax. he tranng set A ncludng all samples from all subjects s an over-complete dctonary. It s known that face mages from a specfc class le n a lnear subspace and a probe can be represented by mages whch have the same label as the probe. Comparng to the sngle-representaton method, collaboratve representaton has more ablty to compensate the pxels of probe. In order to make the codng vector more dscrmnatve, the sparse constrant was ntroduced n regular term. Inspred by compressve sensng, the nduced sparse constrant on codng vector uses l 0 -norm so that the representaton problem s formulzed as: mn x st.. y Ax () 0 Where. denotes the 0 l -norm, whch counts the 0 number of nonzero entres of the codng vector and s a small error tolerance. he l -norm s wdely dscussed and used n many 0 researches. However, the problem of fnd the sparsest soluton of an underdetermned system suffers the ssue of NP-hard. Researchers put forward dfferent solutons to l 0 -norm [5-7]. Now, the sparse constrant methods n CR can be dvded nto two categores: frst one uses l -norm constrant nstead of l 0 - norm, and second one employs supervsed sparse scheme. he frst CR method uses l -norm n FR s Sparse Representaton Classfcaton (SRC) []. In SRC, the sparse representaton s solved usng Lasso formulaton n whch the sparse degree of the codng vector can be adjusted by the norm constrant ntensty. However, the l -norm constrant could not brng good performance when the tranng samples have hgh correlaton [8]. In addton, l -norm based sparse representaton problem s qute tme consumng. A two-phase sparse representaton (PSR) was proposed by Xu. et.al usng supervsed sparse constrant scheme [9]. he M nearest neghbors of a probe are selected based on the frst phase of representaton and then used as the new tranng set n the second phase of representaton n PSR. However, the onetme deleton n PSR may lead to all samples from the classes as probe are removed when the probe s serously dstorted. In addton, PSR s senstve to llumnaton because of the samples wth negatve coeffcents are lkely removed. In ths paper, we propose a supervsed sparse constrant method named as compettve sparse representaton classfcaton (CSRC). Based on the compettve ablty of each tranng sample for representng a probe, the proposed CSRC ntroduces a lowest compettve deleton mechansm whch removes the lowest compettve samples based on the compettve ablty of each tranng sample n collaboratve P a g e
2 Vol. 6, No. 8, 05 lnear representaton. Only those samples wth hgh compettve can be used n the next collaboratve representaton. hen the dmensonalty of the codng vector n the next representaton s bgger than the current one. he mult-phase deleton s more useful for classfcaton than two-phases deleton [0]. Accordng to that the probe s represented based on the mult-phase deleton and untl the condton s satsfed. Meanwhle, the dmensonalty of the fnal codng vector s much smaller than the frst one s,.e. the codng vector s sparse. In CSRC, the compettve ablty of samples from correct class as probe s ncreasng as the lowest compettve ones are removed. In addton, the fast algorthm of CSRC enhances the effcency of the FR system, because the algorthm avods the procedure of fndng nverse matrxes n each collaboratve representaton. One advantage of CSRC s the mult-phases deleton lets the compettve ablty of the correct class s strengthened gradually and avods all samples form correct class of probe are removed n the one-tme collaboratve representaton. he other one s that comparng wth l -norm sparse constrant, CSRC has lower computatonal complexty wth the fast algorthm. hs paper s organzed as followed: n secton, three parts are descrbed: the ntroducton about a basc general framework for classfcaton usng compettve sparse representaton, descrpton the optmzaton method, and the analyss of the computatonal cost of CSRC. he features of CSRC are descrbed n secton 3.We conduct a seres of experments to verfy the good performance of our method n secton 4 and the conclusons are demonstrated n secton 5. II. COMPEIIVE SPARSE REPRESENAION CLASSIFICAION Face mages from a same class le n a lnear subspace, a query mage can be represented by wthn-class samples [, ]. But face recognton s a lack of samples problem n general. When the number of tranng samples of each class s not bg enough, t s hard to obtan a good representaton of a query mage by a small part of tranng samples that from a sngle-subject. hat s to say, the representaton has large dstance wth the query mage. hus, the recognton result s unstable. However, the query mage can be represented fathfully by collaboratve lnear representaton whch each tranng samples are put n the dctonary (sometmes the dctonary s over-completed). Because more tranng samples partcpate n representng the query mage n collaboratve lnear representaton, the compettve of tranng samples that from the class as the query decreases. As a consequence, the query mage s lkely classfed nto the wrong class. However, n the collaboratve representaton, not every sample has hgh compettveness (hgh coeffcent value). So removng these less compettve mages from the dctonary can ncrease the compettve of samples from correct class. he lowest compettve deleton mechansm n CSRC, whch delete the lowest compettve tranng samples from the dctonary n mult-phases. hs mechansm can ncrease the compettve of the correct samples through removng the lowest compettve samples. In the meantme, as the samples are removed n CSRC, the dmensonal of representaton coeffcents are smaller. Compared wth the space of the ntal representaton coeffcents, the fnal representaton coeffcents le n a subspace of t. In other words, the representaton coeffcents are sparse. he sparse codng vector (representaton coeffcents) has more dscrmnate nformaton. A. Compettve Sparse Representaton Classfcaton Gven suffcent tranng samples of the th object class, m m n A =a (,,..., c ), where n denotes the n, R R number of tranng samples and each class contans n tranng samples. Meanwhle each tranng sample s an m - m dmensonalty feature vector. A probe mage y R s represented by collaboratve lnear representaton over the dctonary. Snce CSRC s a mult-phase deleton method, the t t symbol A and x ( t,,... ) denote the tranng samples and codng vector n t th collaboratve representaton respectvely. he representaton framework s wrtten as followng: mn x s.t. y A x () Where s the nosy term. he Eq. () can be wrtten as rdge regresson form: x =arg mn y A x x (3) x he codng vector can be computed as: - n x =( A A + I ) A y x R (4) n n Where the unt matrx I R and s a Lagrangan coeffcent. Snce CSRC deletes only one tranng sample n each phase, CSRC removes the least compettve samples based on the correspondng entres of the codng vector,.e., t fnds the mnmum absolute value x amongst the representaton coeffcents xj mn{ x, x... x n }, and remove the correspondng tranng sample. So the dctonary s dvded nto two subsets: the frst subset ncludes the deleted mage, and the second subset ncludes the retaned samples. Here the samples n the second part wll be used as a r dctonary n the second phase. Let A and A denote the removed sample after the frst representaton and the tranng samples n th representaton respectvely. So the above two r m A a a and subsets can be descrbed as j j R r m (n ) A A A, A R j a j. hen the test y can be represented over the new dctonary A. In the same way, repeatedly conduct the above operaton n the next representaton phase. Assume the k th collaboratve representaton reaches the maxmum number of the deleton phases. he fnal codng k vector x can be represented as followng: k k k - k k n k x =( A A + I) A y x R (5) Snce many samples are removed from the dctonary A P a g e
3 Vol. 6, No. 8, 05 k that s used n the frst representaton, t s very lkely that A excludes all the samples of some classes. herefore, the orgn classfcaton problem s weakened to a smpler problem whch contans fewer classes. he coeffcents of the deleted samples as zero and then select the coeffcents assocated wth the th class and mark t as δ,,,..., c. hen the Eucldean dstance s used for measurng the dstance between each class and the test mage y. he rule of classfcaton s n favor of the class wth mnmum dstance. he formula can be expressed as followng: ID() arg mn d ( y) arg mn y - Aδ,,..., c (6) he detaled algorthm s gven as followng: Algorthm: Compettve Sparse Representaton Classfcaton (CSRC). Input: an undentfed mage n y R, the ntal dctonary A R nm.. Intal value: t 3. Repeat 4. Compute the codng vector accordng to (3) 5. Removng the tranng sample a j from the dctonary t A 6. t Update the dctonary : A, t t 7. Untl satsfy termnaton condtons 8. Identfy : ID arg mn( yaδ ) 9. Output: the dentty of y as. B. Optmzaton As t known to all, the analytcal soluton of the above lnear model s x ( A A I) A y. Due to the deleton operaton, the dctonary s updated n each phase. However, the new dctonary s the subset of the last dctonary. CSRC mplements a fast algorthm to avodng the repeated matrx nverson calculaton. Now let a new symbol A expresses the elementary transformaton of A,.e. A AE, where E s an elementary matrx. A can be treated that the matrx contans two matrces s A and r s r r s A A, A. he two matrces A and A A,.e. denote the deleted sample and the new tranng samples respectvely. Snce the matrx A s gven, so the nverse matrx ( A A I ) s avalable. hen the matrx ( A A I) can be derved from ( A A I ). he detaled dervaton processes are wrtten as follows: ( A A I) (( AE) ( AE) I) ( E A AE I) (7) ( E A AE E IE) ( E ( A A I) E) ( ( ) ) ( ) E A A I E E A A I E he key problem for solvng the codng vector s nverse matrx. In order to have a convenence expresson n (7), the equaton can be represented by the four matrces (O, P, C, and V ) as: ( A A ) s s ss s r sr I r s rs r r rr A A I A A I O P A A I A A I C V (8) Accordng the elementary transformaton of matrx, the nverse matrx wll be transformed as followng:. O P Is 0 O P Is C V 0 Ir 0 V CO P CO Ir hen ( A A I) A 0 Is + P( V CO P) CO - P( V CO P) = V CO P CO Ir O 0 Is + P( V CO P) CO - P( V CO P) = Ir ( V CO P) CO ( V CO P) Is 0 O + O P( V CO P) CO O P( V CO P) = Ir ( V CO P) CO ( V CO P) (9) O P O + O P( V CO P) CO - O P( V CO P) C V ( V CO P) CO ( V CO P) (0) In fact, the ultmate goal s to obtan the soluton nverse s matrx about A, whch s used as the new dctonary n the next teraton, then the nverse matrx can be wrtten as s s s s A A I. Not hard to fnd that s s s s ( A A I ) O O + O P( V CO P) CO () Where the delghted thngs are that O, P, C and V are already known n the matrx A. A group of new symbols are ntroduced to express the four block of ( A A I ),.e., Q O O P V CO P CO () Q O P V CO P (3) Q V CO P CO (4) Q V CO P (5) Combnng () and (3), (4), and (5) respectvely, the four block matrces can be descrpted as followng Q O O PQ Q Q Q V CO P O PQ - -QCO - - (6) 3 P a g e
4 Vol. 6, No. 8, 05 After a few tmes terate replacements, the expresses by other blocks,.e., O can be O Q Q Q Q (7) Here Q s a nvertble square matrx n general case. Snce only one the lowest tranng sample s removed n each phase, so Q s a scalar. In other words, the analytcal soluton about the nverse matrx s s s s A A I s obtaned,.e. It s easy to fnd that CSRC reduces the nterference of the wrong classes. s s s s ( A A I ) Q Q Q Q (8) - C. Complexty analyss We only analyze the tme complexty of (5) n ths secton, snce ths process s the most tme consumng n algorthm of CSRC. It s tme consumng way to solve (5) drectly n a certan collaboratve representaton and the tme complexty s 3 O( n m+n ) However, the tme complexty that CSRC obtans the codng vector s much less than (9). Snce the matrx nverson n (5) s replaced by (8). Moreover, the matrx Q n (8) s only one element, t save more calculaton. herefore, the complexty of CSRC s denoted as O( n m ). In addton, t s much less than the complexty that SRC obtans the spares.5 codng vector,.e. O( n m ). III. ANALYSIS OF CSRC he method Collaboratve representaton based on Classfcaton (CRC) uses l -norm constrant codng vector n the collaboratve representaton [3]. Although the codng vector s not sparse, t fully embodes the compettve level of each tranng sample n CRC. However, CRC obtans the regresson model uses only one collaboratve representaton. When the number of the tranng samples s small, t may lead to regresson model over-ftted. he compettve representaton, adopted n CSRC, removes the lowest compettve tranng sample from the tranng set n the current round and the rest tranng samples wll be used n the nest round. After several rounds of competng, all samples of the subject whch has a low correlaton wth the query may be removed. So CSRC reduces the scale of the FR problem. he Fg., depcts the resduals between the probe and predcton of each class whch are calculated by CRC and CSRC respectvely, llustrates ths phenomenon on ORL database. he upper one n the fgure s obtaned by CRC and the bottom one s obtaned by CSRC. he dstance between the probe and the predctons from over 0 classes s, whch means the tranng samples from these classes are removed and the coeffcents respect to them are zeros. In addton, n CSRC method, the lowest compettve deleton mechansm reduces resdual the correct class of the probe and enlarges the resduals of the wrong classes. Accordng to the fgure t s easy to calculate that the rato of two smallest resduals by CRC and CSRC are.579 and.77 respectvely. he rato of two smallest resduals s enlarged by CSRC, whch means CSRC has better dscrmnatve than CRC. Fg.. he resdual mages by CRC and CSRC for a clear testng face on the ORL database. We select the ffth mage of the frst man as a probe and the frst fve mages as tranng samples. In above two hstograms, the horzontal axs denotes the number of the class and the vertcal axs denotes the resduals between the probe and each class. he top one: the two smallest resdual are 0.49 and by CRC and the rato of them s.579. he bottom one: two smallest resdual are and by CSRC and the rato of them s.77 IV. EXPERIMENAL RESULS o evaluate the proposed CSRC algorthm, we conduct a serous of experments on mages from AR database, Extended YB database and ORL database respectvely, as well as comparng wth state-of-the-art methods ncludng CRC, SRC (wthout extended matrx), and PSR (the canddate set s 0%). We also assess the recognton rate of CSRC(the canddate set s 0%) on the occluded testng faces. All experments are performed n MALAB on 04b on desktop wth 4GHZ CPU and 8G RAM. A. Face Recognton wthout Occluson AR database More than 4000 color face mages of 6 people (70 men and 56 women) consst n AR database [4]. Each people has 6 mages nclude frontal vews of face wth dfferent facal expresson, llumnaton and occluson. he pctures of each ndvdual were taken n two sessons (separated by two weeks). Each secton contans 3 color mages and 0 ndvduals (65 men and 55 women) partcpated n both sessons. he mages of these 00 ndvduals (50 women and 50 men) were selected and used n our experment. Faces that are used to test these methods are gray and then normalzed t to pxels. 4 P a g e
5 Vol. 6, No. 8, 05 We select the frst seven faces n sesson one as tranng samples and the frst seven faces n the sesson two as the testng samples for each class and a specfc class faces are shown n Fg.. he recognton rates of the four methods are shown n the ab.. Snce the testng samples and tranng samples are collected n dfferent tme, none of all methods have a 00% recognton rate. Snce the tranng samples have hgh correlaton, the sparse representaton by SRC could not obtan a good performance. However, snce CRC could not ncrease the compettve ablty of the samples from the correct class as probe, so CRC has lower result than PSR and CSRC. Furthermore, PSR has lower recognton rate than CSRC, because of that PSR s lkely to delete all mages from the correct class n the frst phase. From the experment we can see that the lowest compettve deleton mechansm n CSRC makes the codng vector has more dscrmnant nformaton ndeed. Extended Yale B database he extended Yale B face database contans 38 persons under 64 llumnaton condtons [5,6]. A subset (contans 3 ndvduals) s used n ths experment. he 64 mages of a person n a partcular pose are acqured at camera frame rate of 30 frames/ second, so there s only small change n head pose and facal expresson for those 64 mages. Each mage s reszed to pxels n our experment. Several frontal faces of one person are shown n Fg. 3. As s known to all, llumnaton s another bg challenge for face recognton. Faces were captured under carous laboratory-controlled lghtng condtons. Samples n subset one (seven mages per person) under nomnal lghtng condton was used as the gallery. Snce the recognton rate for test subset and 3 (characterze slght-to-moderate lumnance varatons) are by all methods. Here we select faces n subset 4 are used for verfy CSRC method. Due to the ncreasng llumnaton condton, the recognton results are not very hgh n subset 4. From the ab., CSRC s better than CRC for testng the llumnated mages, whch means the deleton mechansm makes the classfcaton more dscrmnatve. In addton, compared wth PSR, CSRC has about % hgher recognton result than t. he reason for whch s that the tranng samples from the correct class as probe s easy removed n the two-phases deleton, as well as the samples wth negatve coeffcent are not deleted. o the contrary CSRC reduces the rsk that the correct tranng samples wll be removed n one tme through the mult-phase compettve deleton. ORL database ORL database, created by A& lab n Cambrdge Unversty, contans 400 face mages of 40 subjects,.e. each ndvdual provdng 0 face mages, ncludng expresson varants, multple drectons of posture change wthn 0% of the scale of the change. Dmensonalty of each face s reduced to All face mages are show n Fg. 4. For Fg.. Frontal faces wth emoton and llumnaton changes on AR database. he top seven faces are from sesson one and the down seven samples are from sesson two ABLE I. RECOGNIION RAES (%) ON AR DSABASE Methods CRC PSR SRC CSRC Recognton rate Fg. 3. Some sample faces of a subject from Extended Yale B database. he top row: seven mages wth moderate llumnance varatons from subset. he down row: a part mages wth large llumnaton varatons from subset 4 5 P a g e
6 Vol. 6, No. 8, 05 ABLE II. RECOGNIION RAES (%) ON EXENDED YALE B DAABASE methods CRC PSR SRC CSRC Subset each subject, we choose the frst fve mages as the tranng mages and the rest mages are used for testng. From the ab. 3, the recognton results of these four methods are close. Snce the deleton operaton n CSRC, the sparse codng vector has more dscrmnaton than CRC, so CSRC has.5% hgher recognton rate than CRC. ABLE IV. RECOGNIION RAES (%) ON AR DAABASE FOR SUNGLASSES AND SCARF Methods PSR CSRC Sunglasses 56 6 Scarf Fg. 4. en mages of a specfc class from the ORL database. he top row represents the tranng samples and the mages n the bottom row are testng samples ABLE III. RECOGNIION RAES (%) ON ORL DAABASE methods CRC PSR SRC CSRC Recognton rate B. Recognton wth sunglasses and scarf In ths secton we test CSRC s ablty to cope wth real possbly malcous occlusons usng a subset of AR database. he chosen subset conssts of 00 mages of 00 subjects, 50 male and 50 female. For each subject, eght frontal faces (half face are from sesson one and another half are from sesson two) wthout occlusons are used as tranng samples. We select the testng face mages wth sunglasses (two samples for each subject and each sample wth about 0 percent occluson) and scarf (two samples for each subject and each sample wth approxmately 40 percent occluson on the faces) respectvely and the testng samples of a specfc subject are show n Fg. 5. he recognton rates by PSR and CSRC are shown n ab. 4. CSRC has lttle better than PSR for testng samples wth sunglasses. In the scarf case, the recognton rate by CSRC s 6% hgher than PSR s. Because that the proporton of the scarf almost reaches to 40%, t s lkely that the mages of correct class as probe are deleted n the frst collaboratve representaton n PSR. On the contrary, CSRC makes sure the mages of correct class of probe have hgh compettve. Fg. 5. Face mages wth sunglasses and scarf respectvely on AR database V. CONCLUSIONS In ths paper, a compettve representaton framework s proposed to solve the sparse representaton problem. he lowest compettve deleton mechansm ensures the compettve ablty for representng a probe decrease and enhances the compettve ablty of the correct class as the probe. What s more the fast algorthm makes the FR system more effcency. Accordng to the experments, the multple rounds of compettve representaton has better performance n general than the two-phase deleton. In addton, SCRC adoptvely reduces the over-fttng ssue of the regresson model. However, CSRC also has some dsadvantages, such as has not enough robustness to deal wth occlusons, dsguses, and corrupton. In the further, we wll pay more attenton on these dsadvantages. ACKNOWLEDGMEN hs work was supported partally by the Natonal Nature Scence Commttee of Chna under Grant Nos and and partally by Chongqng basc and fronter research project under Grant Nos. cstc04jcyja4008. REFERENCES [] J. Wrght et al., Robust Face Recognton va Sparse Representaton, Ieee ransactons on Pattern Analyss And Machne Intellgence, vol. 3, no., pp. 0-7, Feb, 009. [] M. Yang et al., Regularzed robust codng for face recognton, IEEE rans Image Process, vol., no. 5, pp , May, 03. [3] M. Yang et al., Robust Sparse Codng for Face Recognton, 0 Ieee Conference on Computer Vson And Pattern Recognton (Cvpr), pp , 0, 0. [4] L. Zhang et al., Collaboratve representaton based classfcaton for face recognton, arxv preprnt arxv:04.358, 0. [5] A. Y. Yang et al., "Fast l -mnmzaton algorthms and an applcaton n robust face recognton: A revew." pp [6] M. Yang, and L. Zhang, "Gabor feature based sparse representaton for face recognton wth gabor occluson dctonary," Computer Vson ECCV 00, pp : Sprnger, 00. [7] N. Kwak, Prncpal component analyss based on l-norm maxmzaton, IEEE rans Pattern Anal Mach Intell, vol. 30, no. 9, pp , Sep, 008. [8] J. Wang et al., Robust face recognton va adaptve sparse representaton, IEEE rans Cybern, vol. 44, no., pp , Dec, 04. [9] Y. Xu et al., A wo-phase est Sample Sparse Representaton Method for Use Wth Face Recognton, Ieee ransactons on Crcuts And Systems for Vdeo echnology, vol., no. 9, pp. 55-6, Sep, 0. 6 P a g e
7 Vol. 6, No. 8, 05 [0] J.-X. M, Face mage recognton va collaboratve representaton on selected tranng samples, Optk - Internatonal Journal for Lght and Electron Optcs, vol. 4, no. 8, pp , 9//, 03. [] I. Naseem, R. ogner, and M. Bennamoun, Robust regresson for face recognton, Pattern Recognton, vol. 45, no., pp. 04-8, Jan, 0. [] I. Naseem, R. ogner, and M. Bennamoun, Lnear Regresson for Face Recognton, Ieee ransactons on Pattern Analyss And Machne Intellgence, vol. 3, no., pp. 06-, Nov, 00. [3] L. Zhang, M. Yang, and X. Feng, "Sparse representaton or collaboratve representaton: Whch helps face recognton?." pp [4] A. M. Martnez, he AR face database, CVC echncal Report, vol. 4, 998. [5] A. S. Georghades, P. N. Belhumeur, and D. Kregman, From few to many: Illumnaton cone models for face recognton under varable lghtng and pose, Pattern Analyss and Machne Intellgence, IEEE ransactons on, vol. 3, no. 6, pp , 00. [6] K. C. Lee, J. Ho, and D. J. Kregman, Acqurng lnear subspaces for face recognton under varable lghtng, IEEE rans Pattern Anal Mach Intell, vol. 7, no. 5, pp , May, P a g e
Learning a Class-Specific Dictionary for Facial Expression Recognition
BULGARIAN ACADEMY OF SCIENCES CYBERNETICS AND INFORMATION TECHNOLOGIES Volume 16, No 4 Sofa 016 Prnt ISSN: 1311-970; Onlne ISSN: 1314-4081 DOI: 10.1515/cat-016-0067 Learnng a Class-Specfc Dctonary for
More informationGeneral Regression and Representation Model for Face Recognition
013 IEEE Conference on Computer Vson and Pattern Recognton Workshops General Regresson and Representaton Model for Face Recognton Janjun Qan, Jan Yang School of Computer Scence and Engneerng Nanjng Unversty
More informationKernel Collaborative Representation Classification Based on Adaptive Dictionary Learning
Internatonal Journal of Intellgent Informaton Systems 2018; 7(2): 15-22 http://www.scencepublshnggroup.com/j/js do: 10.11648/j.js.20180702.11 ISSN: 2328-7675 (Prnt); ISSN: 2328-7683 (Onlne) Kernel Collaboratve
More informationCollaboratively Regularized Nearest Points for Set Based Recognition
Academc Center for Computng and Meda Studes, Kyoto Unversty Collaboratvely Regularzed Nearest Ponts for Set Based Recognton Yang Wu, Mchhko Mnoh, Masayuk Mukunok Kyoto Unversty 9/1/013 BMVC 013 @ Brstol,
More informationRobust Low-Rank Regularized Regression for Face Recognition with Occlusion
Robust Low-Rank Regularzed Regresson for ace Recognton wth Occluson Janjun Qan, Jan Yang, anlong Zhang and Zhouchen Ln School of Computer Scence and ngneerng, Nanjng Unversty of Scence and echnology Key
More informationRobust Kernel Representation with Statistical Local Features. for Face Recognition
Robust Kernel Representaton wth Statstcal Local Features for Face Recognton Meng Yang, Student Member, IEEE, Le Zhang 1, Member, IEEE Smon C. K. Shu, Member, IEEE, and Davd Zhang, Fellow, IEEE Dept. of
More informationSupport Vector Machines
/9/207 MIST.6060 Busness Intellgence and Data Mnng What are Support Vector Machnes? Support Vector Machnes Support Vector Machnes (SVMs) are supervsed learnng technques that analyze data and recognze patterns.
More informationRecognizing Faces. Outline
Recognzng Faces Drk Colbry Outlne Introducton and Motvaton Defnng a feature vector Prncpal Component Analyss Lnear Dscrmnate Analyss !"" #$""% http://www.nfotech.oulu.f/annual/2004 + &'()*) '+)* 2 ! &
More informationCluster Analysis of Electrical Behavior
Journal of Computer and Communcatons, 205, 3, 88-93 Publshed Onlne May 205 n ScRes. http://www.scrp.org/ournal/cc http://dx.do.org/0.4236/cc.205.350 Cluster Analyss of Electrcal Behavor Ln Lu Ln Lu, School
More informationComputer Aided Drafting, Design and Manufacturing Volume 25, Number 2, June 2015, Page 14
Computer Aded Draftng, Desgn and Manufacturng Volume 5, Number, June 015, Page 14 CADDM Face Recognton Algorthm Fusng Monogenc Bnary Codng and Collaboratve Representaton FU Yu-xan, PENG Lang-yu College
More informationFace Recognition Method Based on Within-class Clustering SVM
Face Recognton Method Based on Wthn-class Clusterng SVM Yan Wu, Xao Yao and Yng Xa Department of Computer Scence and Engneerng Tong Unversty Shangha, Chna Abstract - A face recognton method based on Wthn-class
More informationLearning the Kernel Parameters in Kernel Minimum Distance Classifier
Learnng the Kernel Parameters n Kernel Mnmum Dstance Classfer Daoqang Zhang 1,, Songcan Chen and Zh-Hua Zhou 1* 1 Natonal Laboratory for Novel Software Technology Nanjng Unversty, Nanjng 193, Chna Department
More informationSubspace clustering. Clustering. Fundamental to all clustering techniques is the choice of distance measure between data points;
Subspace clusterng Clusterng Fundamental to all clusterng technques s the choce of dstance measure between data ponts; D q ( ) ( ) 2 x x = x x, j k = 1 k jk Squared Eucldean dstance Assumpton: All features
More informationRange images. Range image registration. Examples of sampling patterns. Range images and range surfaces
Range mages For many structured lght scanners, the range data forms a hghly regular pattern known as a range mage. he samplng pattern s determned by the specfc scanner. Range mage regstraton 1 Examples
More informationDiscriminative 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 informationA Binarization Algorithm specialized on Document Images and Photos
A Bnarzaton Algorthm specalzed on Document mages and Photos Ergna Kavalleratou Dept. of nformaton and Communcaton Systems Engneerng Unversty of the Aegean kavalleratou@aegean.gr Abstract n ths paper, a
More informationMULTISPECTRAL 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 informationHuman 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 informationClassifier Selection Based on Data Complexity Measures *
Classfer Selecton Based on Data Complexty Measures * Edth Hernández-Reyes, J.A. Carrasco-Ochoa, and J.Fco. Martínez-Trndad Natonal Insttute for Astrophyscs, Optcs and Electroncs, Lus Enrque Erro No.1 Sta.
More informationModular PCA Face Recognition Based on Weighted Average
odern Appled Scence odular PCA Face Recognton Based on Weghted Average Chengmao Han (Correspondng author) Department of athematcs, Lny Normal Unversty Lny 76005, Chna E-mal: hanchengmao@163.com Abstract
More information3D Face Recognition Fusing Spherical Depth Map and Spherical Texture Map
Journal of Computer and Communcatons, 14, *, ** Publshed Onlne **** 14 n ScRes. http://www.scrp.org/journal/jcc http://dx.do.org/1.436/jcc.14.***** 3D Face Recognton Fusng Sphercal Depth Map and Sphercal
More informationFeature Extraction Based on Maximum Nearest Subspace Margin Criterion
Neural Process Lett DOI 10.7/s11063-012-9252-y Feature Extracton Based on Maxmum Nearest Subspace Margn Crteron Y Chen Zhenzhen L Zhong Jn Sprnger Scence+Busness Meda New York 2012 Abstract Based on the
More informationLocal Quaternary Patterns and Feature Local Quaternary Patterns
Local Quaternary Patterns and Feature Local Quaternary Patterns Jayu Gu and Chengjun Lu The Department of Computer Scence, New Jersey Insttute of Technology, Newark, NJ 0102, USA Abstract - Ths paper presents
More informationHermite Splines in Lie Groups as Products of Geodesics
Hermte Splnes n Le Groups as Products of Geodescs Ethan Eade Updated May 28, 2017 1 Introducton 1.1 Goal Ths document defnes a curve n the Le group G parametrzed by tme and by structural parameters n the
More informationParallelism for Nested Loops with Non-uniform and Flow Dependences
Parallelsm for Nested Loops wth Non-unform and Flow Dependences Sam-Jn Jeong Dept. of Informaton & Communcaton Engneerng, Cheonan Unversty, 5, Anseo-dong, Cheonan, Chungnam, 330-80, Korea. seong@cheonan.ac.kr
More informationTwo-Dimensional Supervised Discriminant Projection Method For Feature Extraction
Appl. Math. Inf. c. 6 No. pp. 8-85 (0) Appled Mathematcs & Informaton cences An Internatonal Journal @ 0 NP Natural cences Publshng Cor. wo-dmensonal upervsed Dscrmnant Proecton Method For Feature Extracton
More informationEYE CENTER LOCALIZATION ON A FACIAL IMAGE BASED ON MULTI-BLOCK LOCAL BINARY PATTERNS
P.G. Demdov Yaroslavl State Unversty Anatoly Ntn, Vladmr Khryashchev, Olga Stepanova, Igor Kostern EYE CENTER LOCALIZATION ON A FACIAL IMAGE BASED ON MULTI-BLOCK LOCAL BINARY PATTERNS Yaroslavl, 2015 Eye
More informationNeurocomputing 101 (2013) Contents lists available at SciVerse ScienceDirect. Neurocomputing
Neurocomputng (23) 4 5 Contents lsts avalable at ScVerse ScenceDrect Neurocomputng journal homepage: www.elsever.com/locate/neucom Localty constraned representaton based classfcaton wth spatal pyramd patches
More informationContent Based Image Retrieval Using 2-D Discrete Wavelet with Texture Feature with Different Classifiers
IOSR Journal of Electroncs and Communcaton Engneerng (IOSR-JECE) e-issn: 78-834,p- ISSN: 78-8735.Volume 9, Issue, Ver. IV (Mar - Apr. 04), PP 0-07 Content Based Image Retreval Usng -D Dscrete Wavelet wth
More informationCombination of Local Multiple Patterns and Exponential Discriminant Analysis for Facial Recognition
Sensors & ransducers 203 by IFSA http://.sensorsportal.com Combnaton of Local Multple Patterns and Exponental Dscrmnant Analyss for Facal Recognton, 2 Lfang Zhou, 2 Bn Fang, 3 Wesheng L, 3 Ldou Wang College
More informationDetection of an Object by using Principal Component Analysis
Detecton of an Object by usng Prncpal Component Analyss 1. G. Nagaven, 2. Dr. T. Sreenvasulu Reddy 1. M.Tech, Department of EEE, SVUCE, Trupath, Inda. 2. Assoc. Professor, Department of ECE, SVUCE, Trupath,
More information6.854 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 informationFeature 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 informationRobust Dictionary Learning with Capped l 1 -Norm
Proceedngs of the Twenty-Fourth Internatonal Jont Conference on Artfcal Intellgence (IJCAI 205) Robust Dctonary Learnng wth Capped l -Norm Wenhao Jang, Fepng Ne, Heng Huang Unversty of Texas at Arlngton
More informationTone-Aware Sparse Representation for Face Recognition
Tone-Aware Sparse Representaton for Face Recognton Lngfeng Wang, Huayu Wu and Chunhong Pan Abstract It s stll a very challengng task to recognze a face n a real world scenaro, snce the face may be corrupted
More informationFace Recognition University at Buffalo CSE666 Lecture Slides Resources:
Face Recognton Unversty at Buffalo CSE666 Lecture Sldes Resources: http://www.face-rec.org/algorthms/ Overvew of face recognton algorthms Correlaton - Pxel based correspondence between two face mages Structural
More informationEdge Detection in Noisy Images Using the Support Vector Machines
Edge Detecton n Nosy Images Usng the Support Vector Machnes Hlaro Gómez-Moreno, Saturnno Maldonado-Bascón, Francsco López-Ferreras Sgnal Theory and Communcatons Department. Unversty of Alcalá Crta. Madrd-Barcelona
More informationFace Recognition Based on SVM and 2DPCA
Vol. 4, o. 3, September, 2011 Face Recognton Based on SVM and 2DPCA Tha Hoang Le, Len Bu Faculty of Informaton Technology, HCMC Unversty of Scence Faculty of Informaton Scences and Engneerng, Unversty
More informationSkew Angle Estimation and Correction of Hand Written, Textual and Large areas of Non-Textual Document Images: A Novel Approach
Angle Estmaton and Correcton of Hand Wrtten, Textual and Large areas of Non-Textual Document Images: A Novel Approach D.R.Ramesh Babu Pyush M Kumat Mahesh D Dhannawat PES Insttute of Technology Research
More informationThe 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 informationAn Improved Image Segmentation Algorithm Based on the Otsu Method
3th ACIS Internatonal Conference on Software Engneerng, Artfcal Intellgence, Networkng arallel/dstrbuted Computng An Improved Image Segmentaton Algorthm Based on the Otsu Method Mengxng Huang, enjao Yu,
More informationCHAPTER 3 SEQUENTIAL MINIMAL OPTIMIZATION TRAINED SUPPORT VECTOR CLASSIFIER FOR CANCER PREDICTION
48 CHAPTER 3 SEQUENTIAL MINIMAL OPTIMIZATION TRAINED SUPPORT VECTOR CLASSIFIER FOR CANCER PREDICTION 3.1 INTRODUCTION The raw mcroarray data s bascally an mage wth dfferent colors ndcatng hybrdzaton (Xue
More informationSLAM Summer School 2006 Practical 2: SLAM using Monocular Vision
SLAM Summer School 2006 Practcal 2: SLAM usng Monocular Vson Javer Cvera, Unversty of Zaragoza Andrew J. Davson, Imperal College London J.M.M Montel, Unversty of Zaragoza. josemar@unzar.es, jcvera@unzar.es,
More informationShape Representation Robust to the Sketching Order Using Distance Map and Direction Histogram
Shape Representaton Robust to the Sketchng Order Usng Dstance Map and Drecton Hstogram Department of Computer Scence Yonse Unversty Kwon Yun CONTENTS Revew Topc Proposed Method System Overvew Sketch Normalzaton
More informationFace Recognition by Fusing Binary Edge Feature and Second-order Mutual Information
Face Recognton by Fusng Bnary Edge Feature and Second-order Mutual Informaton Jatao Song, Bejng Chen, We Wang, Xaobo Ren School of Electronc and Informaton Engneerng, Nngbo Unversty of Technology Nngbo,
More informationProblem 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 informationFace Detection with Deep Learning
Face Detecton wth Deep Learnng Yu Shen Yus122@ucsd.edu A13227146 Kuan-We Chen kuc010@ucsd.edu A99045121 Yzhou Hao y3hao@ucsd.edu A98017773 Mn Hsuan Wu mhwu@ucsd.edu A92424998 Abstract The project here
More informationNearest. Points between Image Sets. Meng Yang. Electrical Engineering/IBBT, K.U. Leuven, Belgium
Face Recognton based on Regularzed Ponts between Image Sets Nearest Meng Yang, Pengfe Zhu, Luc Van Gool,3, Le Zhang Department of Informatonn echnology and Electrcal Engneerng, EH Zurch, Swtzerland Department
More informationFast 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 informationCS 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 informationCorrelative features for the classification of textural images
Correlatve features for the classfcaton of textural mages M A Turkova 1 and A V Gadel 1, 1 Samara Natonal Research Unversty, Moskovskoe Shosse 34, Samara, Russa, 443086 Image Processng Systems Insttute
More informationOutline. Self-Organizing Maps (SOM) US Hebbian Learning, Cntd. The learning rule is Hebbian like:
Self-Organzng Maps (SOM) Turgay İBRİKÇİ, PhD. Outlne Introducton Structures of SOM SOM Archtecture Neghborhoods SOM Algorthm Examples Summary 1 2 Unsupervsed Hebban Learnng US Hebban Learnng, Cntd 3 A
More informationCorner-Based Image Alignment using Pyramid Structure with Gradient Vector Similarity
Journal of Sgnal and Informaton Processng, 013, 4, 114-119 do:10.436/jsp.013.43b00 Publshed Onlne August 013 (http://www.scrp.org/journal/jsp) Corner-Based Image Algnment usng Pyramd Structure wth Gradent
More informationAn Image Fusion Approach Based on Segmentation Region
Rong Wang, L-Qun Gao, Shu Yang, Yu-Hua Cha, and Yan-Chun Lu An Image Fuson Approach Based On Segmentaton Regon An Image Fuson Approach Based on Segmentaton Regon Rong Wang, L-Qun Gao, Shu Yang 3, Yu-Hua
More informationCompiler Design. Spring Register Allocation. Sample Exercises and Solutions. Prof. Pedro C. Diniz
Compler Desgn Sprng 2014 Regster Allocaton Sample Exercses and Solutons Prof. Pedro C. Dnz USC / Informaton Scences Insttute 4676 Admralty Way, Sute 1001 Marna del Rey, Calforna 90292 pedro@s.edu Regster
More informationTsinghua University at TAC 2009: Summarizing Multi-documents by Information Distance
Tsnghua Unversty at TAC 2009: Summarzng Mult-documents by Informaton Dstance Chong Long, Mnle Huang, Xaoyan Zhu State Key Laboratory of Intellgent Technology and Systems, Tsnghua Natonal Laboratory for
More informationIntegrated Expression-Invariant Face Recognition with Constrained Optical Flow
Integrated Expresson-Invarant Face Recognton wth Constraned Optcal Flow Chao-Kue Hseh, Shang-Hong La 2, and Yung-Chang Chen Department of Electrcal Engneerng, Natonal Tsng Hua Unversty, Tawan 2 Department
More information12/2/2009. Announcements. Parametric / Non-parametric. Case-Based Reasoning. Nearest-Neighbor on Images. Nearest-Neighbor Classification
Introducton to Artfcal Intellgence V22.0472-001 Fall 2009 Lecture 24: Nearest-Neghbors & Support Vector Machnes Rob Fergus Dept of Computer Scence, Courant Insttute, NYU Sldes from Danel Yeung, John DeNero
More information2x x l. Module 3: Element Properties Lecture 4: Lagrange and Serendipity Elements
Module 3: Element Propertes Lecture : Lagrange and Serendpty Elements 5 In last lecture note, the nterpolaton functons are derved on the bass of assumed polynomal from Pascal s trangle for the fled varable.
More informationJoint Example-based Depth Map Super-Resolution
Jont Example-based Depth Map Super-Resoluton Yanje L 1, Tanfan Xue,3, Lfeng Sun 1, Janzhuang Lu,3,4 1 Informaton Scence and Technology Department, Tsnghua Unversty, Bejng, Chna Department of Informaton
More informationPCA Based Gait Segmentation
Honggu L, Cupng Sh & Xngguo L PCA Based Gat Segmentaton PCA Based Gat Segmentaton Honggu L, Cupng Sh, and Xngguo L 2 Electronc Department, Physcs College, Yangzhou Unversty, 225002 Yangzhou, Chna 2 Department
More informationWIRELESS CAPSULE ENDOSCOPY IMAGE CLASSIFICATION BASED ON VECTOR SPARSE CODING.
WIRELESS CAPSULE ENDOSCOPY IMAGE CLASSIFICATION BASED ON VECTOR SPARSE CODING Tao Ma 1, Yuexan Zou 1 *, Zhqang Xang 1, Le L 1 and Y L 1 ADSPLAB/ELIP, School of ECE, Pekng Unversty, Shenzhen 518055, Chna
More informationCollaboratively Regularized Nearest Points for Set Based Recognition
WU, MINOH, MUKUNOKI: COLLABORATIVELY REGULARIZED NEAREST POINTS 1 Collaboratvely Regularzed Nearest Ponts for Set Based Recognton Yang Wu mm.meda.kyoto-u.ac.jp/members/yangwu Mchhko Mnoh mm.meda.kyoto-u.ac.jp/members/mnoh
More informationR s s f. m y s. SPH3UW Unit 7.3 Spherical Concave Mirrors Page 1 of 12. Notes
SPH3UW Unt 7.3 Sphercal Concave Mrrors Page 1 of 1 Notes Physcs Tool box Concave Mrror If the reflectng surface takes place on the nner surface of the sphercal shape so that the centre of the mrror bulges
More informationAn Efficient Illumination Normalization Method with Fuzzy LDA Feature Extractor for Face Recognition
www.mer.com Vol.2, Issue.1, pp-060-065 ISS: 2249-6645 An Effcent Illumnaton ormalzaton Meod w Fuzzy LDA Feature Extractor for Face Recognton Behzad Bozorgtabar 1, Hamed Azam 2 (Department of Electrcal
More informationDependence of the Color Rendering Index on the Luminance of Light Sources and Munsell Samples
Australan Journal of Basc and Appled Scences, 4(10): 4609-4613, 2010 ISSN 1991-8178 Dependence of the Color Renderng Index on the Lumnance of Lght Sources and Munsell Samples 1 A. EL-Bally (Physcs Department),
More informationOutline. Discriminative classifiers for image recognition. Where in the World? A nearest neighbor recognition example 4/14/2011. CS 376 Lecture 22 1
4/14/011 Outlne Dscrmnatve classfers for mage recognton Wednesday, Aprl 13 Krsten Grauman UT-Austn Last tme: wndow-based generc obect detecton basc ppelne face detecton wth boostng as case study Today:
More informationClassifying Acoustic Transient Signals Using Artificial Intelligence
Classfyng Acoustc Transent Sgnals Usng Artfcal Intellgence Steve Sutton, Unversty of North Carolna At Wlmngton (suttons@charter.net) Greg Huff, Unversty of North Carolna At Wlmngton (jgh7476@uncwl.edu)
More informationA 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 informationLOCAL FEATURE EXTRACTION AND MATCHING METHOD FOR REAL-TIME FACE RECOGNITION SYSTEM. Ho-Chul Shin, Hae Chul Choi and Seong-Dae Kim
LOCAL FEATURE EXTRACTIO AD MATCHIG METHOD FOR REAL-TIME FACE RECOGITIO SYSTEM Ho-Chul Shn, Hae Chul Cho and Seong-Dae Km Vsual Communcatons Lab., Department of EECS Korea Advanced Insttute of Scence and
More informationProgramming 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 informationParallel matrix-vector multiplication
Appendx A Parallel matrx-vector multplcaton The reduced transton matrx of the three-dmensonal cage model for gel electrophoress, descrbed n secton 3.2, becomes excessvely large for polymer lengths more
More informationPictures at an Exhibition
1 Pctures at an Exhbton Stephane Kwan and Karen Zhu Department of Electrcal Engneerng Stanford Unversty, Stanford, CA 9405 Emal: {skwan1, kyzhu}@stanford.edu Abstract An mage processng algorthm s desgned
More informationUser Authentication Based On Behavioral Mouse Dynamics Biometrics
User Authentcaton Based On Behavoral Mouse Dynamcs Bometrcs Chee-Hyung Yoon Danel Donghyun Km Department of Computer Scence Department of Computer Scence Stanford Unversty Stanford Unversty Stanford, CA
More informationTerm 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 informationObject Recognition Based on Photometric Alignment Using Random Sample Consensus
Vol. 44 No. SIG 9(CVIM 7) July 2003 3 attached shadow photometrc algnment RANSAC RANdom SAmple Consensus Yale Face Database B RANSAC Object Recognton Based on Photometrc Algnment Usng Random Sample Consensus
More informationScale Selective Extended Local Binary Pattern For Texture Classification
Scale Selectve Extended Local Bnary Pattern For Texture Classfcaton Yutng Hu, Zhlng Long, and Ghassan AlRegb Multmeda & Sensors Lab (MSL) Georga Insttute of Technology 03/09/017 Outlne Texture Representaton
More informationComputer Science Technical Report
Computer Scence echncal Report NLYSIS OF PCSED ND FISHER DISCRIMINNSED IMGE RECOGNIION LGORIHMS Wendy S. Yambor July echncal Report CS3 Computer Scence Department Colorado State Unversty Fort Collns, CO
More informationUsing Fuzzy Logic to Enhance the Large Size Remote Sensing Images
Internatonal Journal of Informaton and Electroncs Engneerng Vol. 5 No. 6 November 015 Usng Fuzzy Logc to Enhance the Large Sze Remote Sensng Images Trung Nguyen Tu Huy Ngo Hoang and Thoa Vu Van Abstract
More informationA fast algorithm for color image segmentation
Unersty of Wollongong Research Onlne Faculty of Informatcs - Papers (Arche) Faculty of Engneerng and Informaton Scences 006 A fast algorthm for color mage segmentaton L. Dong Unersty of Wollongong, lju@uow.edu.au
More informationMachine Learning: Algorithms and Applications
14/05/1 Machne Learnng: Algorthms and Applcatons Florano Zn Free Unversty of Bozen-Bolzano Faculty of Computer Scence Academc Year 011-01 Lecture 10: 14 May 01 Unsupervsed Learnng cont Sldes courtesy of
More informationFace Recognition using 3D Directional Corner Points
2014 22nd Internatonal Conference on Pattern Recognton Face Recognton usng 3D Drectonal Corner Ponts Xun Yu, Yongsheng Gao School of Engneerng Grffth Unversty Nathan, QLD, Australa xun.yu@grffthun.edu.au,
More informationFinite Element Analysis of Rubber Sealing Ring Resilience Behavior Qu Jia 1,a, Chen Geng 1,b and Yang Yuwei 2,c
Advanced Materals Research Onlne: 03-06-3 ISSN: 66-8985, Vol. 705, pp 40-44 do:0.408/www.scentfc.net/amr.705.40 03 Trans Tech Publcatons, Swtzerland Fnte Element Analyss of Rubber Sealng Rng Reslence Behavor
More informationEnhanced Face Detection Technique Based on Color Correction Approach and SMQT Features
Journal of Software Engneerng and Applcatons, 2013, 6, 519-525 http://dx.do.org/10.4236/jsea.2013.610062 Publshed Onlne October 2013 (http://www.scrp.org/journal/jsea) 519 Enhanced Face Detecton Technque
More informationMachine Learning 9. week
Machne Learnng 9. week Mappng Concept Radal Bass Functons (RBF) RBF Networks 1 Mappng It s probably the best scenaro for the classfcaton of two dataset s to separate them lnearly. As you see n the below
More informationQuality Improvement Algorithm for Tetrahedral Mesh Based on Optimal Delaunay Triangulation
Intellgent Informaton Management, 013, 5, 191-195 Publshed Onlne November 013 (http://www.scrp.org/journal/m) http://dx.do.org/10.36/m.013.5601 Qualty Improvement Algorthm for Tetrahedral Mesh Based on
More informationHigh resolution 3D Tau-p transform by matching pursuit Weiping Cao* and Warren S. Ross, Shearwater GeoServices
Hgh resoluton 3D Tau-p transform by matchng pursut Wepng Cao* and Warren S. Ross, Shearwater GeoServces Summary The 3D Tau-p transform s of vtal sgnfcance for processng sesmc data acqured wth modern wde
More informationHistogram of Template for Pedestrian Detection
PAPER IEICE TRANS. FUNDAMENTALS/COMMUN./ELECTRON./INF. & SYST., VOL. E85-A/B/C/D, No. xx JANUARY 20xx Hstogram of Template for Pedestran Detecton Shaopeng Tang, Non Member, Satosh Goto Fellow Summary In
More informationFast Computation of Shortest Path for Visiting Segments in the Plane
Send Orders for Reprnts to reprnts@benthamscence.ae 4 The Open Cybernetcs & Systemcs Journal, 04, 8, 4-9 Open Access Fast Computaton of Shortest Path for Vstng Segments n the Plane Ljuan Wang,, Bo Jang
More informationSolving 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 informationSENSITIVITY ANALYSIS IN LINEAR PROGRAMMING USING A CALCULATOR
SENSITIVITY ANALYSIS IN LINEAR PROGRAMMING USING A CALCULATOR Judth Aronow Rchard Jarvnen Independent Consultant Dept of Math/Stat 559 Frost Wnona State Unversty Beaumont, TX 7776 Wnona, MN 55987 aronowju@hal.lamar.edu
More informationA Fast Visual Tracking Algorithm Based on Circle Pixels Matching
A Fast Vsual Trackng Algorthm Based on Crcle Pxels Matchng Zhqang Hou hou_zhq@sohu.com Chongzhao Han czhan@mal.xjtu.edu.cn Ln Zheng Abstract: A fast vsual trackng algorthm based on crcle pxels matchng
More informationShape-adaptive DCT and Its Application in Region-based Image Coding
Internatonal Journal of Sgnal Processng, Image Processng and Pattern Recognton, pp.99-108 http://dx.do.org/10.14257/sp.2014.7.1.10 Shape-adaptve DCT and Its Applcaton n Regon-based Image Codng Yamn Zheng,
More informationRECOGNIZING GENDER THROUGH FACIAL IMAGE USING SUPPORT VECTOR MACHINE
Journal of Theoretcal and Appled Informaton Technology 30 th June 06. Vol.88. No.3 005-06 JATIT & LLS. All rghts reserved. ISSN: 99-8645 www.jatt.org E-ISSN: 87-395 RECOGNIZING GENDER THROUGH FACIAL IMAGE
More informationDecision Strategies for Rating Objects in Knowledge-Shared Research Networks
Decson Strateges for Ratng Objects n Knowledge-Shared Research etwors ALEXADRA GRACHAROVA *, HAS-JOACHM ER **, HASSA OUR ELD ** OM SUUROE ***, HARR ARAKSE *** * nsttute of Control and System Research,
More informationAngle-Independent 3D Reconstruction. Ji Zhang Mireille Boutin Daniel Aliaga
Angle-Independent 3D Reconstructon J Zhang Mrelle Boutn Danel Alaga Goal: Structure from Moton To reconstruct the 3D geometry of a scene from a set of pctures (e.g. a move of the scene pont reconstructon
More informationOn Some Entertaining Applications of the Concept of Set in Computer Science Course
On Some Entertanng Applcatons of the Concept of Set n Computer Scence Course Krasmr Yordzhev *, Hrstna Kostadnova ** * Assocate Professor Krasmr Yordzhev, Ph.D., Faculty of Mathematcs and Natural Scences,
More informationKernel principal component analysis network for image classification 1
Kernel prncpal component analyss network for mage classfcaton Wu Dan 4 Wu Jasong 34 Zeng Ru 4 Jang Longyu 4 Lotf Senhadj 34 Shu Huazhong 4 ( Key Laboratory of Computer Network and Informaton Integraton
More informationLecture 13: High-dimensional Images
Lec : Hgh-dmensonal Images Grayscale Images Lecture : Hgh-dmensonal Images Math 90 Prof. Todd Wttman The Ctadel A grayscale mage s an nteger-valued D matrx. An 8-bt mage takes on values between 0 and 55.
More informationThe 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