AUTOMATED personal identification using biometrics

Size: px
Start display at page:

Download "AUTOMATED personal identification using biometrics"

Transcription

1 A 3D Feature Descrptor Recovered from a Sngle 2D Palmprnt Image Qan Zheng,2, Ajay Kumar, and Gang Pan 2 Abstract Desgn and development of effcent and accurate feature descrptors s crtcal for the success of many computer vson applcatons. Ths paper proposes a new feature descrptor, referred to as DoN, for the 2D palmprnt matchng. The descrptor s extracted for each pont on the palmprnt. It s based on the ordnal measure whch partally descrbes the dfference of the neghborng ponts normal vectors. DoN has at least two advantages: ) t descrbes the 3D nformaton, whch s expected to be hghly stable under commonly occurrng llumnaton varatons durng contactless magng; 2) the sze of DoN for each pont s only one bt, whch s computatonally smple to extract, easy to match, and effcent to storage. We show that such 3D nformaton can be extracted from a sngle 2D palmprnt mage. The analyss for the effectveness of ordnal measure for palmprnt matchng s also provded. Four publcly avalable 2D palmprnt databases are used to evaluate the effectveness of DoN, both for dentfcaton and the verfcaton. Our method on all these databases acheves the state-of-the-art performance. INTRODUCTION AUTOMATED personal dentfcaton usng bometrcs characterstcs s one of the most crtcal and challengng tasks to meet growng demand for strngent securty. There s ever growng need to develop more accurate and effcent bometrcs matchng technologes. Therefore our goal n ths work has been to desgn, develop and evaluate more accurate, compact and faster matchng algorthms for the palmprnt dentfcaton. In the context of advancements n the matchng of 2D palmprnts, ths paper ntroduces new algorthm to further mprove feature extracton and matchng technques for the contactless palmprnt dentfcaton. Our key contrbutons are three-folds: ) we present a new palmprnt feature whch s effcent and effectve for 2D palmprnt matchng; 2) we show that such feature descrbes the 3D nformaton and t can be extracted from a sngle 2D palmprnt mage; 3) our feature s based on the ordnal measure and we show that the ordnal measure s powerful for palmprnt matchng. Our method acheves the state-of-the-art performance on four publcly avalable 2D palmprnt databases. Besdes, extra experments are well desgned to verfy our arguments on contrbuton 2) and 3). The paper s organzed as follows: we frst ntroduce the DoN feature and explan ts recovery from a sngle 2D mage n Sec. 2; the effectveness of the ordnal measure s ascertaned n Sec. 3; fnally, the expermental results are reported n Sec DON FOR PALMPRINT MATCHING In ths secton, the DoN feature s frstly ntroduced. We then llustrate that how such feature whch descrbes 3D nformaton can be recovered from a sngle 2D mage. Fnally, Department of Computng, The Hong Kong Polytechnc Unversty, Hung Hom, Kowloon, Hong Kong; 2 Department of Computer Scence, Zhejang Unversty, Hangzhou, Chna.. The mplementaton of proposed method s publcly made avalable [] the extracton of DoN from the 2D palmprnt mage and the feature matchng between two palmprnt templates are ntroduced. 2. DoN: Dfference of Vertex Normal Vectors For the better descrpton of the feature, we buld Cartesan coordnates for each acqured palmprnt mage, mage plane formng x-y plane, wth the normal of mage surface N representng z-axs, as shown n Fg.. Each pont/pxel p on the mage plane s one-to-one correspondng to a vertex v on palmprnt surface. Let n = (x, y, z ) denote the normal vector of vertex v. Suppose pont p has two neghborng regons R and R 2, ts DoN feature DoN() s DoN() = τ( z j z j ), () j R j R 2 where τ( ) s the sgn functon whch s defned as {, α <, τ(α) =, α. The DoN feature for p descrbes the dfference between the z-component of normal vectors from ts two neghborng regons. Accordng to (), two key ssues n recoverng the DoN feature of pont p are: ) how to obtan z value for each pont from ts neghborng regons (Sec. 2.2); 2) how to dentfy the neghborng regons of pont p (Sec. 2.3) D Feature from a Sngle 2D Image We address the frst problem by ntroducng the Lambertan Model n our analyss. The Lambertan Model has also been utlzed by Basr and Jacobs [2] to prove that the Lambertan objects appear close to a 9D lnear subspace. Ths concluson was evaluated for face recognton problem and excellent performance was acheved [2]. In ths paper, we further explore Lambertan Model for the contactless palmprnt matchng. (2)

2 2 Presented Palm Palmrpnt Image Fg. : The 2D palmprnt mage s essentally the projecton of the palmprnt surface. The drecton of llumnaton L s always from the front sde of the palmprnt magng. Accordng to the Lambertan Model, the pxel value I() of pont p on mage I s jontly determned by reflectance k d, llumnaton ntensty l d, llumnaton drecton vector L, and pont normal vector n, I() = k d l d Ln. (3) For dfferent ponts on the same palmprnt mage, k d, l d, and L can be consdered as the same. Gven a pont p, the dfference between the ponts ntensty from ts neghborng regon R and R 2 s j R I(j) I(j). We can substtute these two components n sgn functon τ and defne D() as follows, D() = τ( I(j) I(j)). (4) j R 2 j R Usng (3), (4) can be rewrtten as D() = τ(k d l d L( j R j R 2 n j n j )) j R 2 = τ(l( n j n j )). j R j R 2 D() s determned by the angle between llumnaton vector L and vector ( j R n j j R n 2 j ). For the smplcty, let X = x x, j R j R 2 Y = y y, j R j R 2 Z = z z, j R and let L = (a, b, c), we have j R 2 (5) (6) D() = τ( X a + Y b + Z c). (7) It may be noted that palmprnt mages are always acqured under frontal llumnaton. Therefore, under the Cartesan coordnate n Fg., we have c >. When Zc k > Xa k + Y b k s satsfed, D() s determned by the sgn of Z,.e., D() = τ( X a + Y b + Z c) = τ( Z c) = τ( Z ). (8) Accordng to the descrpton n Sec. 2., τ( Z ) represents the DoN feature DoN() we plan to recover. (8) llustrates that the 3D nformaton DoN() can be recovered from a sngle 2D mage. When constrant Zc k > Xa k + Y b k s satsfed, we can use avalable texture-level nformaton to recover the 3D shape nformaton. Wthout loss of generalty, for the sample case when X a > Y b, the constrant can be rewrtten as c a c a > 2 X Z. (9) s determned by the llumnaton durng the magng whch cannot be always controlled. One feasble way to meet the constrant n (9) s to ensure the rght sde 2 X Z as small as possble. We show that the proper partton for the neghborng regons can be used to acheve ths goal whch s ntroduced n next secton. For the smplcty, the partton for each pont on the palmprnt are set to be the same. Therefore the feature extracton usng DoN can be acheved by usng a convoluton operaton followed by a sgn functon. The partton of the neghborng regons on palmprnt mages s then replaced by the dvson of regons on the spatal flter. The spatal regons correspondng to the operatng flter s dvded nto two subsets, say R and R 2 n an M N sze spatal flter. All the entres n regon R are set as and the entres n regon R 2 are set as -. Extendng τ to matrx operaton, the DoN feature matrx or template F for mage I can be computed as follows: F = τ(f I), () () s a computatonally smpler operaton and s used to recover the 3D nformaton from a sngle 2D mage I n our approach. 2.3 Parttonng Local Regons The second problem outlned n Sec. 2. s to decde the neghborng regons. We acheve ths goal by parttonng the local regons on the spatal flter as dscussed above. Consderng the fact that bnarzed feature template s generated from the (contactless magng) nosy mages, the operator or the flter should be desgned n such a way that the postve and negatve flterng results from multple pxels are evened out. Ths mples that the sum of all the entres n the flter should be zero and the spatal dstrbuton of or - n the flter be symmetrc and orderly. On the other hand, as we dscussed above, such dvson should ensure the rght sde of the constrant n (9) be as small as possble. We now brefly revst the characterstcs of the palmprnt surface. Human palm surface s a complex 3D textured surface consstng of many rdges, wrnkles and lnes [3], [4]. The spatal extents of these key elements are dfferent from each other. Generally rdges represent the smallest element whle the wrnkles are larger than rdges, and the lnes are the most remarkable features n a typcal palmprnt database. All these key elements share the shape lke a valley. The valley typcally represents a symmetrcal shape whose symmetry axs

3 3 s the valley bottom. A 3D palm surface can be seen as a combnaton of several such symmetrcal unts. Such unts dstrbute rregularly on palmprnt surface. We utlze these propertes of such unts to make the rght sde n (9) as small as possble, that s, make X approxmate to zero and Z be a large value. If the neghborng regons partton s desgned to be contnuously connected and n patches, t s expected that j R x, j R x 2, thus ther dfference X. Ths s for two reasons: ) the symmetrc unts are over x-y plane and such partton ensures the surface symmetry for the regons. When the pont normal vectors are added, the azmuth components have large chance to be nearly cancelled out or elmnated (Fg. 2); 2) the x-y plane s almost a flat plane, whch ndcates that the x-component for most of ponts are excepted to be zero. The value of Z s excepted be large for two reasons, ) j R z tself s large value snce each z s a postve value, whch s manly determned by the sze and the length of valley; 2) j R z s excepted to be dfferent from j R z 2 due to the rregular dstrbuton of the valley unt on the palmprnt surface. 2 X Therefore, for the palmprnt mage, Z s more lkely to have small value and satsfy the constrant n (9). Fg. 2 llustrates the summaton result of normal vectors from the ponts on a typcal palm lne. Y Z X Palmprnt surface Small regon of palmprnt Cross sec on ofbluelne Normal vectors summa on Fg. 2: For a small regon on palmprnt, f we add the normal vector of each pxel pont, the summaton result wll almost be vertcal. Ths fgure llumnates the summaton of ponts on arbtrary palm lne. Fg. 3 llustrates some canddate flters. Fg. 3 -(c) show three such spatal dvsons wth the cross or number of parttons ncreasng from left to rght. Fg. 3 (d)-(f) llustrate spatal dstrbuton of values for the three dfferent drectons of a subset when the cross number s fxed to 2. The ncrease n the number of cross n parttons s expected to help n suppressng the photometrc dstortons by balancng the fltered results from the postve and negatve values. However, too many crosses wll make the flter rotatonally senstve. Ths wll also ntroduce asymmetry by reducng the symmetrcal property of the small regons on palmprnt, whch wll make the argument j R x, j R x 2 unrelable or nvald. Besdes, too many crosses wll also make Z. Ths can further pose lmtatons n meetng the constrant n (9). It may be noted that the symmetrcal unts (such as rdges, wrnkles and lnes) representng domnant palmprnt features are expected to have some wdth. The drecton of ntersecton (c) (d) (e) (f) Fg. 3: Dfferent colors represent dfferent subsets n a flter. In -(c) the crossng n the flter ncreases from to 3 whle n (d)-(f) the drectons of flters are vared when the crossng s fxed to 2. Fg. 4: The feature code of palmprnt. Orgnal gray level mage. The feature code. boundary (center or whte lne on flter n Fg. 3) also has wdth. In order to ensure the symmetry, the whte lne n the flter confguraton should be located orthogonal or parallel to the domnant symmetrcal unts. The expermental results n Sec. 4.2 also support these arguments. Accordng to prevous analyss, the flter of Fg. 3 s selected to construct the feature extractor or the flter. Ths flter can be defned as follows: > j f,j = < j otherwse. () where, j s the ndexes,, j [ B, B]. The flter sze s (2B + ) (2B + ). Fg. 4 shows a typcal palmprnt ROI mage and ts encoded features. 2.4 Template Denosng and Matchng Our analyss n prevous sectons suggests that the constrant n (9) may not be satsfed for some cases. Ths can be due to two key reasons: ) n some of the extreme locatons, there s stll a chance that Z, and 2) when porton of the valleys appears on the boundary or edges of the flter, the response from the small regons wll no longer be symmetrcal. In order to allevate the unrelable codes or features resultng from the nose or extreme cases as dscussed above, we ncorporate a denosng strategy durng matchng. The morphologcal operatons,.e., openng operaton and closng operaton, are performed on the feature templates, and the weghted sum of three scores s computed as the fnal matchng score. The morphologcal operatons are adopted for two reasons, ) the nosy perturbatons due to the lmtatons of the feature extractor are expected to be dscretely dstrbuted

4 4 n the feature templates and 2) the feature template s bnary and has spatal contnuty. It should be noted that ths step does not ncrease the feature template sze. Gven two feature template F and F 2, the matchng dstance s computed by the weghted sum of three scores, ds(f, F 2 ) = w S(F, F 2 ) + w 2 S( F, F 2 ) + w 3 S( F, F 2 ), (2) where F and F are the results after applyng closng and openng operatons on feature template F. S(F, F 2 ) s defned as S(F, F 2 ) = Γ(F F 2 &M(F )&M(F 2 )), (3) Γ(M(I )&M(I 2 )) where and & are XOR and AND operaton, Γ(F) computes the number of non-zero value n matrx F, M(I) s the mask matrx ndcatng the vald regon on palmprnt mage I. It s defne as {, background M(I)(, j) = (4), otherwse. We use horzontal and vertcal translatons between two matched templates to mprove algnment. In all our experments, the best match score among these algnments s used as the fnal match score. 3 ANALYSIS ON PALMPRINT MATCHING The DoN feature for each pont on the mage can have only two values. Therefore, t s necessary to analyss whether ths feature can be powerful enough for the accurate palmprnt matchng. The DoN feature s a knd of ordnal measure. In the lterature, Snha [5] beleved that ordnal measure can only be used for smple detecton or classfcaton task and t should be mpossble usng ordnal measure alone to solve complex object recognton problems. Ths concern arses from the nature of ordnal measure whch loses some numercal nformaton. Sun et al. [6], [7] demonstrated that the ordnal measure can play a defnng role for the complex rs recognton and palmprnt recognton problems. They acheved sgnfcant performance mprovement over competng methods. However, they essentally used three features to ncrease the dscrmnaton of the ordnal measure and the dscrmnaton usng only one feature based on ordnal measure was not nvestgated. Besdes, they acheved most of the success from desgnng the feature extractor,.e., usng dervaton of Gaussan flter, and dd not provde much analyss for the role played by the ordnal measure. In ths secton, we demonstrate that a sngle feature based on ordnal measure can be very powerful for the palmprnt matchng. In the palmprnt matchng, feature or template of a palmprnt mage s usually a feature matrx [6], [8], [9], [], [], [2]. Each entry on the matrx s an encoded feature code. Dstance between two templates s defned as the sum of dstances between such codes. Hammng dstance s usually employed to measure the codes dstance as the codes are often bnarzed. In ordnal measure, the number of encodng classes for each code s 2 whle n [9] or [8], the encodng classes s 6. For smplcty, the number of encodng classes s represented by λ n the followng analyss. Let us consder the smlarty between two palmprnt templates beng matched. If these two palmprnt mages belong to the same subject, we refer them as ntra-class matchng dstance or genune matchng dstance. Smlarly, f these templates belong to dfferent subjects, we refer them as nterclass matchng dstance or mposter matchng dstance. In order to smplfy the analyss, dstance between two codes s assumed to be zero f they are equal, otherwse one. For a palmprnt template, the dmenson of ts feature vector s Consderng the spatal dependence of palmprnt regon, t s qute reasonable to assume that the ntrnsc dmenson of a sngle palmprnt feature vector, let us denote here by M, s much smaller than 6384,.e., M For the convenence of llustratons, we can consder a certan value for M, say. Consderng the nter-class matchng attempts, whether the codes on the templates are matched or msmatched can be consdered as a random event. Ths s because two palmprnts are from dfferent subjects and are unknown. Therefore, we assume 2 that the dstrbuton of nter-class matchng dstance D nter follows Bnomal dstrbuton D nter B(n nter, p), (5) where n nter s the number of trals n the Bnomal dstrbuton whch s the same as M, p s the success probablty. The relaton between λ and p can be expressed as p = λ. (6) Fg. 5 shows the nter-class dstance dstrbutons when n ter =, λ s 6 and 2 respectvely. Ths fgure llustrates the nter-class palmprnt matchng scores for the deal cases,.e., when features are robust and free from magng varatons, wth the changes n λ. Probablty Densty λ=2 λ= Matchng Dstance Probablty Densty nter class ntra class, ω=.8 ntra class, ω=.9 ntra class, ω= Matchng Dstance Fg. 5: Typcal matchng dstance dstrbuton for dfferent λ. When p = 5, the nter-class dstance dstrbuton and three ntra-class 6 dstance dstrbuton wth dfferent ω. Consderng the ntra-class matchng attempts, among all the codes pars, we call the codes par as relable when both of partcpatng matchng codes are encoded n the rght class 3, otherwse the pars are consdered as unrelable. For a relable codes par, the matchng dstance should certanly be zero. For an unrelable par, the matchng dstance stll have chance to be 2. Analyss of rs codes usng mllons of matchng scores presented by Daugman [3] also justfes ths assumpton. 3. Rght class descrbes the actual depth nformaton.

5 5 zero. Therefore, the ntra-class matchng dstance between two templates s effectvely determned by the number of unrelable codes pars. For the unrelable pars, whether these respectve codes are matched or not can also be consdered as a random event. Ths s because the unrelable codes are manly caused by nose and msalgnment whose exact nfluence unknown. Representng the number of unrelable pars as ωm, where ω s the unrelable pars rate ( < ω < ), we assume ntra-class matchng dstance D ntra follows the Bnomal dstrbuton D ntra B(n ntra, p), (7) n ntra s number of trals, whch s equal to the unrelable pars number ωm, p s the success probablty. The relatonshp between p and λ s the same as for nter-class dstance dstrbuton. When λ = 6, M =, several ntra-class dstance dstrbutons wth dfferent ω as well as the nter-class dstance dstrbuton are shown n Fg. 5. From ths fgure, we can observe that when ω becomes larger, whch means the number of unrelable pars s large, the overlappng area between ntra-class and nter-class dstance dstrbutons also becomes larger. It s known that n order to acheve accurate dentfcaton, least overlap between the ntra-class and nterclass matchng dstance dstrbuton s desrable. The lkelhood of an unrelable par beng generated can be estmated as θ = ( p)( p). (8) It may be noted that the ω also represents the expected rate of unrelable code pars. Therefore t s qute reasonable to assume that ω s proportonal to θ,.e., ω θ. (9) Gven two λ, the relaton between ther correspondng p, p 2, ω, and ω 2 can be expressed as ω = ( p )( p ) ω 2 ( p 2 )( p 2 ). (2) When two λ are 6 and 2, p s 5 6 and p 2 s 2 respectvely, gven ω =.95,, accordng to (2), the correspondng ω 2 =.73,.76. Fg. 6 and show these dstrbuton wth ω =.95, respectvely. Probablty Densty nter class, λ=2 nter class, λ=6 ntra class, λ=2 ntra class, λ= Matchng Dstance Problty Densty nter class,λ = 2 nter class,λ = 6 ntra class,λ = 2 ntra class,λ = Matchng Dstance Fg. 6: If ω =.95, the nter-class and ntra-class dstance dstrbuton wth λ = 2 and λ = 6 respectvely. If ω =, the nter-class and ntra-class dstance dstrbuton wth λ = 2 and λ = 6 respectvely. We can observe from Fg. 6 that under the same condton, the overlappng area between nter-class and ntra-class dstance dstrbutons for λ = 2 s always smaller than that for λ = 6. No overlap means the nter-class dstance s always larger than ntra-class dstance, whch results n good performance. Note that λ = 2 means the feature template s bnary, t s expected that bnary representaton for feature s more effectve. The analyss presented n ths secton theoretcally argues the effectveness of ordnal measure for palmprnt matchng. Ths argument s further supported by our expermental results on 2D palmprnt databases n Sec EXPERIMENTAL VALIDATION AND RESULTS In ths secton, we frstly evaluate the performance from our method on four publcly avalable 2D palmprnt databases. Three competng methods, [8], [9], and Ordnal Code [6] are mplemented 4 for the comparatve performance. To facltate far comparson wth pror work, dfferent protocols are adopted for dfferent databases. The computaton complexty s also compared. In order provde detaled analyss on our method, the performance from our method usng dfferent flters (Fg. 3) as well as wthout denosng strategy s also reported. To demonstrate the robustness of our feature to the llumnaton changes, the experment from a face database wth sgnfcant llumnaton changes s also reported. Besdes, a fast verson of, namely Fast-, whch s based on the ordnal measure, s employed to support the arguments we made n Sec Expermental Results from Proposed Method 4.. PolyU Contactless 2D/3D Palmprnt Database Ths contactless palmprnt database [5] s acqured from 77 dfferent subjects (rght hand). There are 2 samples from each subject wth 2D mages and depth mages. It also provdes segmented palmprnt mages of pxels. Our experments are performed on full 2D part of ths database. The 2D mages n ths database are acqured under poor (ambent) llumnaton condtons. In our experment, the frst 5 samples of each subject are enrolled as tranng set. The rest 5 are as the test set. There are 885 samples for tranng/gallery and 885 samples for testng/probe. The ROC, CMC curves as well as the EER are used to evaluate the performance. The poor or ambent llumnaton n ths database, along wth contactless magng, makes t most challengng among other databases. Fg. 7 llustrates the ROC and CMC curves. Table provdes the EER and average rank-one recognton accuracy. It can be observed that the proposed method acheves outperformng results over the other methods for both the verfcaton and dentfcaton IITD Palmprnt Database The IITD touchless palmprnt database [6] provdes contactless palmprnt mages from the rght and left hands of 23 subjects. There are more than 5 samples for each rght hand or left hand. Ths database provdes 5 5 pxels segmented palmprnt mages. 4. The three methods are mplemented by us. All of the results usng the verson of our mplementaton acheve smlar or slghtly better than those reported n the authors paper [8], [9], [4], wth same protocol and same datasets. The mplementaton as well as the parameters can be found on [].

6 6 Genune Acceptance Rate.96 Ordnal Code 4 2 Recognton Rate Ordnal Code Rank Fg. 7: The ROC curves and CMC curves of dfferent methods from the PolyU 2D/3D contactless palmprnt database. TABLE : The EER and rank-one recognton rate (accuracy) of dfferent methods from PolyU 2D/3D palmprnt database. Method Compettve Code Ordnal Code EER (%) Accuracy (%) The protocol s exactly the same as n []. All the 3 rght hand palmprnt mages are used for our experments. For each subject, one mage for testng and the rest for tranng. The average performance s reported. The ROC, EER and CMC are used to ascertan the performance. Fg. 8 llustrates the ROC and CMC curves. Table 2 presents the EER and average rank-one recognton accuracy. The EER and rankone recognton rate from our method acheves outperformng results. TABLE 2: The EER and rank-one recognton rate (accuracy) of dfferent methods from the IITD palmprnt database. Method Compettve Code Ordnal Code EER (%) Accuracy (%) PolyU Palmprnt Database The PolyU palmprnt database [7] contans 7752 palmprnt mages from 386 dfferent palms. These mages were automatcally segmented to pxel. In ths database, there are several mages whch are poorly algned due to ther rotatonal varaton. In our experments, we used the same protocol as reported n [8]. Only the frst sample of each ndvdual s used to construct the tranng set. Then, the tranng set s enlarged by rotatng each mage n tranng set at -9, -6, -3, 3, 6 and 9 respectvely. Consequently, there are seven tranng samples for each of the palms from database. Fg. 9 llustrates the comparatve ROC and CMC curves. Table 3 summarzes the EER and rank-one recognton accuracy from dfferent methods. The results from our method and the Ordnal Code are superor than those from the other two methods. Note that the performance from our method s observed to be slghtly better than that of Ordnal Code. The reason for the performance mprovement not beng sgnfcant, for ths PolyU palmprnt database, s that the Ordnal Code s specal case of our feature. Ths palmprnt database s a contact-based database acqured under controlled or very good llumnaton condton. The drecton of llumnaton vector L s almost parallel to z-axs (Fg. ). In ths case, the left sde Genune Acceptance Rate.96 Ordnal Code 4 2 Recognton Rate Ordnal Code Rank Fg. 8: The ROC curves and CMC curves of dfferent methods from the IITD palmprnt database. Genune Acceptance Rate Ordnal Code Recognton Rate Rank Ordnal Code Fg. 9: The ROC curves and CMC curves of dfferent methods from the PolyU palmprnt database. of the constrant n (9) s large enough to make t s satsfed. Accordng to our prevous analyss, the Ordnal Code method s essentally the dfference between the weghted combnaton of the pont normal vectors. Even f the flter s not carefully desgned, the constrant n (9) s excepted to hold good. Note that the feature sze of Ordnal Code s three tmes larger than ours, whle our method outperforms Ordnal Code. TABLE 3: The EER and rank-one recognton rate (accuracy) of dfferent methods from PolyU palmprnt database. Method Compettve Code Ordnal Code EER (%) Accuracy (%) CASIA Palmprnt Database The CASIA palmprnt database contans 5239 palmprnt mages from 3 ndvduals. It s the largest publcly avalable palmprnt database n terms of the number of ndvduals. In ths database, the ndvdual s the same as the ndvdual 9 and therefore these two classes were merged nto one class. The th mage from the left hand of ndvdual 27 s also msplaced to the rght hand. The 3rd mage from left hand of ndvdual 76 s a dstorted sample whose qualty s very poor. These two samples can also be automatcally detected by our palmprnt segmentaton program []. We elmnated these two mages n our experment. Therefore all our experments wth ths database used 5237 mages belongng to 6 dfferent palms. We segmented and scaled the resultng mages n the database to pxels. In our experments, the total number of resultng matches s 3,692,466, whch ncludes 2,567 genune and 3,689,899 mposter matches. Fg. llustrates the ROC and CMC results from the proposed method,, compettve code and Ordnal Code.

7 7 Genune Acceptance Rate Ordnal Code 5 Fg. : The ROC curves from the CASIA palmprnt database. Table 4 summarzes EER acheved from these methods. Our method sgnfcantly outperforms competng three methods. TABLE 4: The EER(%) comparson from CASIA palmprnt database. Method Compettve Code Ordnal Code EER Computatonal Complexty Table 5 lsts the computatonal tme for our method,, and Ordnal Code. The matchng speed of our method s the fastest among other methods. The feature extracton speed of our method s much faster than the, Ordnal Code and margnally slower than for. However, the matchng speed of our method s more than 22 tmes faster than that for the. TABLE 5: Feature extracton and matchng tme (ms) of dfferent methods. Method Feature Extracton Matchng Compettve Code Ordnal Code Note: The expermental envronment s: Wndows 8 Professonal, Intel(R) Core(TM) 5-32M CPU@2.5GHz, 8G RAM, VS Addtonal Expermental Results In ths secton, we nvestgate the nfluence of usng dfferent flter confguratons as well as the denosng strategy for the palmprnt matchng performance. We use four publcly avalable databases employed n earler experments, wth ther descrbed protocols, to ascertan the performance from other flter confguratons. The weght rato and flter sze have been emprcally selected to acheve the llustrated performance. The EER results usng the dfferent flter confguratons n Fg. 3 are shown n Fg.. The expermental results valdate our analyss presented n Sec. 2.3, regardng the expected nfluence n the performance from the varatons n the number of cross and the drectons of the dfferent subset. The flter we selected acheves the best performance as llustrated. The results for the three competng methods are also shown n the fgure. All the other flter results are also not sgnfcantly nferor because they are ordnal measures and they can stll extract robust codes or feature n ther respectve templates. Ths fgure also llustrates the results wthout usng the denosng strategy,.e., employng ds(f, F 2 ) = S(F, F 2 ). nstead of (2), to compute the matchng dstance. As can be observed from the results, even wthout employng the denosng strategy, our feature can also acheve competng performance wth the state-of-the-art methods reported n the lterature. It should be noted that wthout denosng strategy, the matchng speed s three tmes faster. EER(%) PloyU 2D/3D IITD PolyU CASIA /(d),ours (c) (e) (f) w/o denosng Ordnal Code Fg. : Comparatve performance usng EER from dfferent flter confguratons for four palmprnt databases. 4.3 Experment on Yale Face Database In ths experment, Extended Yale Face Database B [8] s employed. These experments are not ntended to nvestgate our feature on face surface, but we just want to evaluate the effectveness of our feature to support the arguments that our feature s ndeed descrbng the 3D shape nformaton whch s nsenstve to llumnaton changes. We choose Extended Yale Face Database B to acheve the goal for three reasons: ) There are only few palmprnt mages acqured under extreme lghtng condtons. The face data, whose surface s almost flat, s smlar to palmprnt surface. 2) Ths database s acqured under extreme llumnaton varatons. 3) The face data n ths database doesn t have any vew or expresson varatons. The Extended Yale Face Database B contans 38 subjects and each subject s wth 64 llumnaton condtons. The cropped faces wth the resoluton of are also provded. In our experment, the most neutral lght sources (A+E+) mages are used as the gallery, and all the other frontal mages are used as probes (n summary, 38 mages consttute the tranng set and 2376 mages as test set). Ths database s used to evaluate the dentfcaton performance. We do not desgn addtonal flters accordng to the geometrcal propertes of the face surface but smply use the same dvson strategy as for the palmprnt. Besdes, the denosng matchng strategy s not employed to underlne our feature s robustness to the extremely varyng llumnaton condtons. The rank-one recognton result s 99.3% whch s the stateof-the-art performance. Table 6 provdes the comparson wth another two state-of-art methods on ths database. The results demonstrate that our feature s also robust to the llumnaton varatons.

8 8 TABLE 6: Identfcaton rate comparson wth the state-of-the-art methods from the Extended Yale Face Database B. Method PP+LTP/DT [9] G LDP [2] Rank- rate (%) Expermental Results from Fast- The experments n ths secton are used to verfy the effectveness of ordnal measure for the 2D palmprnt matchng. Frstly, we smplfy the to generate bnary representaton feature. As reported n [9], the uses sx flters to generate sx pre-templates. In the smplfed verson, two of the sx flters (the orentaton of these two flters are orthogonal) are used to generate two ntermedate pre-templates and subsequently bnary template s generated as the fnal feature. The rest of mplementaton of the s the same as n [9]. Ths method s referred [2] to as Fast-. Fg. 2 llustrates the ROC for the Fast- and on four 2D palmprnt databases usng the same protocols as reported above. It can be observed that the performance of smplfed/ordnal approach s superor to that of orgnal one. Therefore t s reasonable to conclude that ordnal measure s not only faster but also more accurate whch valdates our analyss n Sec. 3. Genune Acceptance Rate Genune Acceptance Rate Fast (c) Fast Genune Acceptance Rate Genune Acceptance Rate Fast 4 2 Fast 5 Fg. 2: The ROC curves for Fast- and from PolyU Contactless 2D/3D database, IITD palmprnt database, (c) PolyU 2D palmprnt database and (d) CASIA 2D palmprnt database. 5 CONCLUSIONS Ths paper has proposed a new feature whch s based on ordnal measure. Ths feature recovers 3D shape nformaton of the surface whle t s extracted from pxel level nformaton n the contactless mages. The feature extracton and matchng s very effcent and the mplementaton s smple whch emphaszes on ts practcalty. The template sze from ths feature s also very small whle t acheves excellent performances on many databases. (d) The proposed feature s sutable to be combned [3] wth other features to further mprove the performance. There are two key reasons: ) t has lower storage requrements whle beng effcent to recover/extract and match, and most mportantly t s effectve n achevng accurate performance, 2) most of 2D feature mentoned n prevous sectons extract the texture nformaton whle our feature s recoverng 3D shape nformaton, whch means t s lkely to have less redundancy than the other features. Despte promsng results, there are several aspects of proposed approach that requre further consderaton. The mpact of non-lambertan surface and non-lnear palm deformatons, n the feature recovery and matchng, requres further study and analyss. REFERENCES [] Weblnk for downloadng codes for all algorthms n ths paper, avalable from: csajaykr/2dto3d.htm. [2] R. Basr and D. W. Jacobs, Lambertan reflectance and lnear subspaces, Patt. Anal. Mach. Intell., IEEE Trans. on, vol. 25, no. 2, pp , 23. [3] A. Kumar and C. Kwong, Towards contactless, low-cost and accurate 3d fngerprnt dentfcaton, Pattern Analyss and Machne Intellgence, IEEE Transactons on, vol. 37, no. 3, pp , 25. [4] D. Malton, D. Mao, A. K. Jan, and S. Prabhakar, Handbook of fngerprnt recognton. Sprnger Scence & Busness Meda, 29. [5] P. Snha, Qualtatve representatons for recognton, n Bologcally Motvated Computer Vson. Sprnger, 22, pp [6] Z. Sun, T. Tan, Y. Wang, and S. Z. L, Ordnal palmprnt representon for personal dentfcaton [representon read representaton], n Comp. Vs. Patt. Recogn., 25. CVPR 25. IEEE Computer Socety Conf. on, vol.. IEEE, 25, pp [7] Z. Sun and T. Tan, Ordnal measures for rs recognton, Patt. Anal. Mach. Intell., IEEE Trans. on, vol. 3, no. 2, pp , 29. [8] W. Ja, D.-S. Huang, and D. Zhang, Palmprnt verfcaton based on robust lne orentaton code, Patt. Recogn., pp , 28. [9] A.-K. Kong and D. Zhang, Compettve codng scheme for palmprnt verfcaton, n Patt. Recogn., 24. ICPR 24. Proc. of the 7th Inter. Conf. on, vol.. IEEE, 24, pp [] A. Kumar and S. Shekhar, Personal dentfcaton usng multbometrcs rank-level fuson, Systems, Man, and Cybernetcs, Part C: Applcatons and Revews, IEEE Trans. on, vol. 4, no. 5, pp , 2. [] V. Kanhangad, A. Kumar, and D. Zhang, A unfed framework for contactless hand verfcaton, Info. Fore. Sec., IEEE Trans. on, vol. 6, no. 3, pp. 4 27, 2. [2] D. Zhang, W.-K. Kong, J. You, and M. Wong, Onlne palmprnt dentfcaton, Patt. Anal. Mach. Intell., IEEE Trans. on, vol. 25, no. 9, pp. 4 5, 23. [3] J. Daugman, The mportance of beng random: statstcal prncples of rs recognton, Patt. recogn., vol. 36, no. 2, pp , 23. [4] D. N. Bhat and S. K. Nayar, Ordnal measures for mage correspondence, Patt. Anal. Mach. Intell., IEEE Trans. on, vol. 2, no. 4, pp , 998. [5] The Hong Kong Polytechnc Unversty Contact-free 3D/2D Hand Images Database (Ver.), avalable onlne. [6] IIT Delh Touchless Palmprnt Database (Ver.). (25, Jun. ). [onlne], avalable from: ajaykr/database Palm.htm. [7] PolyU Palmprnt Palmprnt Database [onlne], The Hong Kong Polytechnc Unversty, bometrcs/. [8] K.-C. Lee, J. Ho, and D. J. Kregman, Acqurng lnear subspaces for face recognton under varable lghtng, Patt. Anal. Mach. Intell., IEEE Trans. on, vol. 27, no. 5, pp , 25. [9] X. Tan and B. Trggs, Enhanced local texture feature sets for face recognton under dffcult lghtng condtons, Image Proc., IEEE Trans. on, vol. 9, no. 6, pp , 2. [2] B. Zhang, Y. Gao, S. Zhao, and J. Lu, Local dervatve pattern versus local bnary pattern: face recognton wth hgh-order local pattern descrptor, Image Proc., IEEE Trans. on, vol. 9, pp , 2. [2] Q. Zheng, A. Kumar, and G. Pan, Suspectng less and achevng more: new nsghts on palmprnt dentfcaton for faster and more accurate matchng, Info. Fore. Sec., IEEE Trans. on, 25.

Sixth Indian Conference on Computer Vision, Graphics & Image Processing

Sixth Indian Conference on Computer Vision, Graphics & Image Processing Sxth Indan Conference on Computer Vson, Graphcs & Image Processng Incorporatng Cohort Informaton for Relable Palmprnt Authentcaton Ajay Kumar Bometrcs Research Laboratory, Department of Electrcal Engneerng

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

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

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

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

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

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

Palmprint Feature Extraction Using 2-D Gabor Filters

Palmprint Feature Extraction Using 2-D Gabor Filters Palmprnt Feature Extracton Usng 2-D Gabor Flters Wa Kn Kong Davd Zhang and Wenxn L Bometrcs Research Centre Department of Computng The Hong Kong Polytechnc Unversty Kowloon Hong Kong Correspondng author:

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

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

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

R s s f. m y s. SPH3UW Unit 7.3 Spherical Concave Mirrors Page 1 of 12. Notes SPH3UW Unt 7.3 Sphercal Concave Mrrors Page 1 of 1 Notes Physcs Tool box Concave Mrror If the reflectng surface takes place on the nner surface of the sphercal shape so that the centre of the mrror bulges

More information

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

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

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

CS 534: Computer Vision Model Fitting

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

More information

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

Recognizing Faces. Outline

Recognizing 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 information

Face Recognition using 3D Directional Corner Points

Face 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 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

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

Accounting for the Use of Different Length Scale Factors in x, y and z Directions

Accounting for the Use of Different Length Scale Factors in x, y and z Directions 1 Accountng for the Use of Dfferent Length Scale Factors n x, y and z Drectons Taha Soch (taha.soch@kcl.ac.uk) Imagng Scences & Bomedcal Engneerng, Kng s College London, The Rayne Insttute, St Thomas Hosptal,

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

User Authentication Based On Behavioral Mouse Dynamics Biometrics

User 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 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

Face Recognition by Fusing Binary Edge Feature and Second-order Mutual Information

Face 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 information

Electrical analysis of light-weight, triangular weave reflector antennas

Electrical analysis of light-weight, triangular weave reflector antennas Electrcal analyss of lght-weght, trangular weave reflector antennas Knud Pontoppdan TICRA Laederstraede 34 DK-121 Copenhagen K Denmark Emal: kp@tcra.com INTRODUCTION The new lght-weght reflector antenna

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

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

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

Proper Choice of Data Used for the Estimation of Datum Transformation Parameters

Proper Choice of Data Used for the Estimation of Datum Transformation Parameters Proper Choce of Data Used for the Estmaton of Datum Transformaton Parameters Hakan S. KUTOGLU, Turkey Key words: Coordnate systems; transformaton; estmaton, relablty. SUMMARY Advances n technologes and

More information

Reducing Frame Rate for Object Tracking

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

More information

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

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

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

Finger-Vein Verification Based on Multi-Features Fusion

Finger-Vein Verification Based on Multi-Features Fusion Sensors 203, 3, 5048-5067; do:0.3390/s35048 Artcle OPEN ACCESS sensors ISSN 424-8220 www.mdp.com/journal/sensors Fnger-Ven Verfcaton Based on Mult-Features Fuson Huafeng Qn,2, *, Lan Qn 2, Lan Xue 2, Xpng

More information

Brushlet Features for Texture Image Retrieval

Brushlet Features for Texture Image Retrieval DICTA00: Dgtal Image Computng Technques and Applcatons, 1 January 00, Melbourne, Australa 1 Brushlet Features for Texture Image Retreval Chbao Chen and Kap Luk Chan Informaton System Research Lab, School

More information

Analysis of Continuous Beams in General

Analysis of Continuous Beams in General Analyss of Contnuous Beams n General Contnuous beams consdered here are prsmatc, rgdly connected to each beam segment and supported at varous ponts along the beam. onts are selected at ponts of support,

More information

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

Distance Calculation from Single Optical Image

Distance Calculation from Single Optical Image 17 Internatonal Conference on Mathematcs, Modellng and Smulaton Technologes and Applcatons (MMSTA 17) ISBN: 978-1-6595-53-8 Dstance Calculaton from Sngle Optcal Image Xao-yng DUAN 1,, Yang-je WEI 1,,*

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

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

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

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

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

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

Palmprint Recognition Using Directional Representation and Compresses Sensing

Palmprint Recognition Using Directional Representation and Compresses Sensing Research Journal of Appled Scences, Engneerng and echnology 4(22): 4724-4728, 2012 ISSN: 2040-7467 Maxwell Scentfc Organzaton, 2012 Submtted: March 31, 2012 Accepted: Aprl 30, 2012 Publshed: November 15,

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

SVM-based Learning for Multiple Model Estimation

SVM-based Learning for Multiple Model Estimation SVM-based Learnng for Multple Model Estmaton Vladmr Cherkassky and Yunqan Ma Department of Electrcal and Computer Engneerng Unversty of Mnnesota Mnneapols, MN 55455 {cherkass,myq}@ece.umn.edu Abstract:

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

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

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

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

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

Hermite Splines in Lie Groups as Products of Geodesics

Hermite 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 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

Comparison Study of Textural Descriptors for Training Neural Network Classifiers

Comparison Study of Textural Descriptors for Training Neural Network Classifiers Comparson Study of Textural Descrptors for Tranng Neural Network Classfers G.D. MAGOULAS (1) S.A. KARKANIS (1) D.A. KARRAS () and M.N. VRAHATIS (3) (1) Department of Informatcs Unversty of Athens GR-157.84

More information

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

Lobachevsky State University of Nizhni Novgorod. Polyhedron. Quick Start Guide Lobachevsky State Unversty of Nzhn Novgorod Polyhedron Quck Start Gude Nzhn Novgorod 2016 Contents Specfcaton of Polyhedron software... 3 Theoretcal background... 4 1. Interface of Polyhedron... 6 1.1.

More information

Intra-Parametric Analysis of a Fuzzy MOLP

Intra-Parametric Analysis of a Fuzzy MOLP Intra-Parametrc Analyss of a Fuzzy MOLP a MIAO-LING WANG a Department of Industral Engneerng and Management a Mnghsn Insttute of Technology and Hsnchu Tawan, ROC b HSIAO-FAN WANG b Insttute of Industral

More information

Helsinki University Of Technology, Systems Analysis Laboratory Mat Independent research projects in applied mathematics (3 cr)

Helsinki University Of Technology, Systems Analysis Laboratory Mat Independent research projects in applied mathematics (3 cr) Helsnk Unversty Of Technology, Systems Analyss Laboratory Mat-2.08 Independent research projects n appled mathematcs (3 cr) "! #$&% Antt Laukkanen 506 R ajlaukka@cc.hut.f 2 Introducton...3 2 Multattrbute

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

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

A Novel Fingerprint Matching Method Combining Geometric and Texture Features

A Novel Fingerprint Matching Method Combining Geometric and Texture Features A Novel ngerprnt Matchng Method Combnng Geometrc and Texture eatures Me Xe, Chengpu Yu and Jn Q Unversty of Electronc Scence and Technology of Chna. Chengdu,P.R.Chna xeme@ee.uestc.edu.cn Post Code:6154

More information

Steps for Computing the Dissimilarity, Entropy, Herfindahl-Hirschman and. Accessibility (Gravity with Competition) Indices

Steps for Computing the Dissimilarity, Entropy, Herfindahl-Hirschman and. Accessibility (Gravity with Competition) Indices Steps for Computng the Dssmlarty, Entropy, Herfndahl-Hrschman and Accessblty (Gravty wth Competton) Indces I. Dssmlarty Index Measurement: The followng formula can be used to measure the evenness between

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

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

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

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

Face Detection with Deep Learning

Face 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 information

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

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

More information

MOTION PANORAMA CONSTRUCTION FROM STREAMING VIDEO FOR POWER- CONSTRAINED MOBILE MULTIMEDIA ENVIRONMENTS XUNYU PAN

MOTION PANORAMA CONSTRUCTION FROM STREAMING VIDEO FOR POWER- CONSTRAINED MOBILE MULTIMEDIA ENVIRONMENTS XUNYU PAN MOTION PANORAMA CONSTRUCTION FROM STREAMING VIDEO FOR POWER- CONSTRAINED MOBILE MULTIMEDIA ENVIRONMENTS by XUNYU PAN (Under the Drecton of Suchendra M. Bhandarkar) ABSTRACT In modern tmes, more and more

More information

Object Recognition Based on Photometric Alignment Using Random Sample Consensus

Object 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 information

Learning-based License Plate Detection on Edge Features

Learning-based License Plate Detection on Edge Features Learnng-based Lcense Plate Detecton on Edge Features Wng Teng Ho, Woo Hen Yap, Yong Haur Tay Computer Vson and Intellgent Systems (CVIS) Group Unverst Tunku Abdul Rahman, Malaysa wngteng_h@yahoo.com, woohen@yahoo.com,

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

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

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

Complex Numbers. Now we also saw that if a and b were both positive then ab = a b. For a second let s forget that restriction and do the following.

Complex Numbers. Now we also saw that if a and b were both positive then ab = a b. For a second let s forget that restriction and do the following. Complex Numbers The last topc n ths secton s not really related to most of what we ve done n ths chapter, although t s somewhat related to the radcals secton as we wll see. We also won t need the materal

More information

A fast algorithm for color image segmentation

A 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 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

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

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

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

Competitive Sparse Representation Classification for Face Recognition

Competitive Sparse Representation Classification for Face Recognition 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

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

2x x l. Module 3: Element Properties Lecture 4: Lagrange and Serendipity Elements

2x 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 information

For instance, ; the five basic number-sets are increasingly more n A B & B A A = B (1)

For instance, ; the five basic number-sets are increasingly more n A B & B A A = B (1) Secton 1.2 Subsets and the Boolean operatons on sets If every element of the set A s an element of the set B, we say that A s a subset of B, or that A s contaned n B, or that B contans A, and we wrte A

More information

Face Tracking Using Motion-Guided Dynamic Template Matching

Face Tracking Using Motion-Guided Dynamic Template Matching ACCV2002: The 5th Asan Conference on Computer Vson, 23--25 January 2002, Melbourne, Australa. Face Trackng Usng Moton-Guded Dynamc Template Matchng Lang Wang, Tenu Tan, Wemng Hu atonal Laboratory of Pattern

More information

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

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

More information

12/2/2009. Announcements. Parametric / Non-parametric. Case-Based Reasoning. Nearest-Neighbor on Images. Nearest-Neighbor Classification

12/2/2009. 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 information

Improved SIFT-Features Matching for Object Recognition

Improved SIFT-Features Matching for Object Recognition Improved SIFT-Features Matchng for Obect Recognton Fara Alhwarn, Chao Wang, Danela Rstć-Durrant, Axel Gräser Insttute of Automaton, Unversty of Bremen, FB / NW Otto-Hahn-Allee D-8359 Bremen Emals: {alhwarn,wang,rstc,ag}@at.un-bremen.de

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

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

Enhanced Face Detection Technique Based on Color Correction Approach and SMQT Features

Enhanced 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 information

2 ZHENG et al.: ASSOCIATING GROUPS OF PEOPLE (a) Ambgutes from person re dentfcaton n solaton (b) Assocatng groups of people may reduce ambgutes n mat

2 ZHENG et al.: ASSOCIATING GROUPS OF PEOPLE (a) Ambgutes from person re dentfcaton n solaton (b) Assocatng groups of people may reduce ambgutes n mat ZHENG et al.: ASSOCIATING GROUPS OF PEOPLE 1 Assocatng Groups of People We-Sh Zheng jason@dcs.qmul.ac.uk Shaogang Gong sgg@dcs.qmul.ac.uk Tao Xang txang@dcs.qmul.ac.uk School of EECS, Queen Mary Unversty

More information

Image Matching Algorithm based on Feature-point and DAISY Descriptor

Image Matching Algorithm based on Feature-point and DAISY Descriptor JOURNAL OF MULTIMEDIA, VOL. 9, NO. 6, JUNE 2014 829 Image Matchng Algorthm based on Feature-pont and DAISY Descrptor L L School of Busness, Schuan Agrcultural Unversty, Schuan Dujanyan 611830, Chna Abstract

More information

Scale Selective Extended Local Binary Pattern For Texture Classification

Scale 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 information

EYE CENTER LOCALIZATION ON A FACIAL IMAGE BASED ON MULTI-BLOCK LOCAL BINARY PATTERNS

EYE 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 information

Object Tracking Based on PISC Image and Template Matching

Object Tracking Based on PISC Image and Template Matching ect Trackng Based on PISC Image and Template Matchng Bud Sugand Electrcal Engneerng Department Batam State Polytechnc Batam Indonesa ud_sugand@polatam.ac.d Astract Ths paper proposed a method for oect

More information

Palmprint Minutia Point Matching Algorithmand GPU Application

Palmprint Minutia Point Matching Algorithmand GPU Application Palmprnt Mnuta Pont Matchng Algorthmand GPU Applcaton 1 Beng Crmnal Scence Insttuton, Beng, 100054,Chna E-mal: wucs@sccas.cn Zhgang Lu 2 Publc Securty Bureau of Beng s Dongcheng, Beng, 100061, Chna Cagang

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

Image Alignment CSC 767

Image Alignment CSC 767 Image Algnment CSC 767 Image algnment Image from http://graphcs.cs.cmu.edu/courses/15-463/2010_fall/ Image algnment: Applcatons Panorama sttchng Image algnment: Applcatons Recognton of object nstances

More information