Image Segmentation with Simultaneous Illumination and Reflectance Estimation: An Energy Minimization Approach

Size: px
Start display at page:

Download "Image Segmentation with Simultaneous Illumination and Reflectance Estimation: An Energy Minimization Approach"

Transcription

1 Image Segmentaton wth Smultaneous Illumnaton and Reflectance Estmaton: An Energy Mnmzaton Approach Chunmng L 1, Fang L 2, Chu-Yen Kao 3, Chenyang Xu 4 1 Vanderblt Unversty Insttute of Imagng Scence, ashvlle, T 37232, USA 2 Dept. of Mathematcs, East Chna ormal Unversty, Shangha , Chna 3 Dept. of Mathematcs, The Oho State Unversty, Columbus, OH 43210, USA 4 Semens Corporate Research, Prnceton, J 08540, USA lchunmng@gmal.com Abstract Spatal ntensty varatons caused by llumnaton changes have been a challenge for mage segmentaton and many other computer vson tasks. Ths paper presents a novel method for mage segmentaton wth smultaneous estmaton of llumnaton and reflectance mages. The proposed method s based on the composton of an observed scene mage wth an llumnaton component and a reflectance component, known as ntrnsc mages. We defne an energy functonal n terms of an llumnaton mage, the membershp functons of the regons, and the correspondng reflectance constants of the regons n the scene. Ths energy s convex n each of ts varables. By mnmzng the energy, mage segmentaton result s obtaned n the form of the membershp functons of the regons. The llumnaton and reflectance components of the observed mage are estmated smultaneously as the result of energy mnmzaton. Wth llumnaton taken nto account, the proposed method s able to segment mages wth non-unform ntenstes caused by spatal varatons n llumnaton. Comparsons wth the state-of-the-art pecewse smooth model demonstrate the superor performance of our method. 1. Introducton An observed mage of a scene s a composton of llumnaton and reflectance components, whch have been termed as ntrnsc mages n [1]. The reflectance component characterzes a unque physcal property of the object surfaces n the scene the albedo of the object surfaces. The llumnaton component depends on the lght drecton and the orentaton of the object surfaces, and can be used to nfer the geometry of the objects beng vewed. However, spatal varatons n llumnaton cause changes of the ntenstes n the observed mages, whch has been a challenge for many computer vson tasks, such as segmentaton, trackng, and object recognton. In mage segmentaton, a consderable dffculty arses from spatal varatons n llumnaton. The observed mage ntenstes n a sngle object may not be unform due to spatal varatons n llumnaton. In ths case, t s dffcult to defne a regon descrptor n regon-based methods for mage segmentaton. Therefore, many wdely used regonbased methods [15, 5] rely on the unformty of the ntenstes n the regons to be segmented, as t s easy to model the ntenstes wthn the regons of nterest and to defne a regon descrptor. These methods are not applcable to real-world mages wth non-unform ntenstes caused by spatal varatons n llumnaton. The well-known Mumford-Shah model n [11] has provded a theoretcal framework of mage segmentaton, whch can be appled to a wder range of mages, ncludng those wth non-unform ntenstes. The Mumford-Shah model assumes that an mage can be approxmated by a pecewse smooth functon. Based on ths assumpton, the Mumford-Shah model seeks a contour and a number of smooth functons that approxmate the mage ntenstes n dsjont regons separated by the contour. The smoothness of these functons s ensured by mposng a smoothng term n the Mumford-Shah model, whch leads to a set of partal dfferental equatons that have to be solved at each teraton for the evoluton of the contour. Therefore, the computaton n these methods s rather expensve, whch lmts ther practcal applcatons. In addton, due to the non-convexty of the underlyng energy functonals, the correspondng energy mnmzaton algorthms may converge to false local mnma. The pecewse smooth characterzaton of mages n the Mumford-Shah model s too general for real mages. A 2009 IEEE 12th Internatonal Conference on Computer Vson (ICCV) /09/$ IEEE 702

2 more specfc characterzaton of real mages would provde more useful nformaton, whch can be taken nto account n mage segmentaton. In fact, there s a more specfc characterzaton of real mages based on the theory of lghtness percepton: an observed mage I, ts llumnaton mage component S, and reflectance mage component R are related by the followng multplcatve model: I = R S. (1) Furthermore, the llumnaton mage S s assumed to be smooth [6, 8, 2]. The reflectance mage R can be approxmated by a pecewse constant map, as t descrbes a physcal property of the object surfaces n the scene. Therefore, the observed mage can be approxmated by the product of a smooth functon and a pecewse constant functon. In ths paper, we propose a novel energy mnmzaton method for mage segmentaton wth smultaneous estmaton of llumnaton and reflectance mages, based on the above mage composton model n Eq. (1) and the propertes of the llumnaton and reflectance mages. We defne an energy functonal n terms of an llumnaton mage, the membershp functons of the regons, and the correspondng reflectance constants of the regons n the scene. By mnmzng the energy, mage segmentaton result s obtaned n the form of the membershp functons of the regons. Meanwhle, the llumnaton and reflectance components of the observed mage are estmated smultaneously as the result of energy mnmzaton. The proposed method s related to the technques for llumnaton and reflectance estmaton (IRE) from an observed mage, such as those n [7, 14, 9, 12]. However, our method s fundamentally dfferent from these methods for llumnaton and reflectance estmaton. Our method s manly targeted for mage segmentaton. In the formulaton of our method, we ntroduce membershp functons of the object surfaces n the mage, whch drectly represent an mage segmentaton result. The exstng methods for IRE do not provde mage segmentaton results. 2. Problem Formulaton Based on Intrnsc Image Composton For theoretcal completeness, we slghtly extend the mage model (1) wth addtve nose as below: I = R S + n (2) where n s assumed to be zero-mean Gaussan nose. The reflectance mage R s approxmately a constant r n a regon Ω, whch corresponds to an object surface n the scene. The property of the llumnaton mage S and the reflectance mage R mentoned n Secton 1 can be made more specfc as below: (A1) R(x) r for x Ω for =1,,, wth Ω Ω j = for any j and Ω =Ω. (A2) The llumnaton mage S s smooth. The above dsjont regons Ω 1,, Ω form a partton of the mage doman Ω. The goal of mage segmentaton s to determne such a partton. From the above mage model (2), we propose to seek a pecewse constant map ˆR and a smooth functon Ŝ to estmate the ntrnsc mages R and S, respectvely. In general, a pecewse constant map R can be expressed as a lnear combnaton: R = r χ Ω (3) where χ Ω s the characterstc functon of the regon Ω. Each characterstc functon χ Ω s a membershp functon of the regon Ω. Thus, estmaton of the reflectance mage R from an observed mage I can be acheved by seekng membershp functons u assocated wth the regons Ω and the correspondng reflectance constants r,for =1,,. Thus, mage segmentaton result s gven by the obtaned membershp functons. For convenence, the membershp functons u 1,,u and the constants r 1,,r are denoted by a vector valued membershp functon U =(u 1,,u ), and a vector r =(r 1,,r ), respectvely. Thus, the problem of estmatng the ntrnsc mages R and S can be formulated as one that seeks optmal vector membershp functon U, vector r, and a smooth functon S accordng to an optmalty crteron. We wll propose a varatonal formulaton, n whch the optmal U, r, and S are obtaned by mnmzng an energy functonal F(U, r,s). 3. Energy Formulaton 3.1. Statstcal Property of Local Intenstes The varaton n appearance caused by the llumnaton changes has been a challengng problem for vsual tasks, such as segmentaton. However, the smoothness of llumnaton mage, along wth the pecewse constant property of the reflectance mage, mples a useful property of the observed mage ntenstes n local area, whch wll be used n the formulaton of the proposed energy mnmzaton framework for segmentaton wth smultaneous llumnaton and reflectance estmaton. In fact, the observed mage ntenstes n a small local area follow a smple dstrbuton as descrbed below. We consder a crcular neghborhood of each pont x Ω, defned by O x {y : y x ρ}, where ρ s the radus of the neghborhood. The partton {Ω } of the entre doman Ω nduces a partton of the neghborhood O x nto subsets O x Ω, whch form a partton of O x. For a smooth 703

3 functon S, the values S(y) for all y n the crcular neghborhood O x can be well approxmated by S(x). Therefore, the ntenstes S(y)R(y) n each subregon O x Ω are approxmately the constant S(x)r. Thus, we have the followng approxmaton S(y)R(y) S(x)r for y O x Ω (4) From the mage model (2), we have I(y) S(x)r + n(y) for y O x Ω where n(y) s addtve zero-mean Gaussan nose. The above arguments show that the ntenstes wthn the neghborhood O x are samples from Gaussan dstrbutons wth dstnct means S(x)r, = 1,,. In the other word, the ntenstes n the neghborhood O x form clusters: {I(y) :y O x Ω }, each wth a cluster center m S(x)r. As shown above, the cluster center m can be approxmated by the product of a regon dependent reflectance constants r and an llumnaton factor S(x), whch approxmates the llumnaton wthn the neghborhood O x. Ths property of local ntenstes s fully exploted n the proposed method for mage segmentaton and estmaton of ntrnsc mages Clusterng Crteron for Local Intenstes In vew of the above clusterng characterzaton of local ntenstes n terms of llumnaton and reflectance, we ntroduce a clusterng-based approach for mage segmentaton wth smultaneous llumnaton and reflectance estmaton. Lets frst revew a basc clusterng crteron that has been used to derve the well known fuzzy C-means and K-means algorthms. A standard clusterng algorthm seeks the membershp functons u and the cluster centers c, such that the followng clusterng crteron functon s mnmzed: J FCM = u q (x) I(x) c 2 dx (5) where q 1 s the fuzzfer, I s the nput mage, u (x) s the membershp value at pont x for class such that u (x) =1, and c s the cluster center of class. Ths clusterng crteron has been used to derve a smple yet powerful algorthm to classfy data that conssts of separable clusters wth dstnct cluster centers and determne the cluster centers at the same tme. The correspondng mnmzaton process s the well-known fuzzy C-means (FCM) algorthm. As a specal case when the fuzzfer q s 1, the mnmzaton of the above clusterng crteron s acheved by the K-means algorthm. The above FCM clusterng crteron s n general not vald for mage I wth a spatally varyng llumnaton mage component S. However, from the statstcs of local ntenstes n Secton 3.1, the above standard clusterng crteron can be appled for local ntensty classfcaton. The analyss n Secton 3.1 has shown that the ntenstes n the neghborhood O x form clusters, each wth a dstnct cluster center m S(x)r. Therefore, we propose to mnmze the followng clusterng crteron for local ntenstes n the neghborhood O x : J loc x (U, r,s(x)) O x u q (y)k(x y) (6) I(y) S(x)r 2 dy where u s the membershp functon of the regon Ω, and K(x y) s a weghtng functon, whch decreases as the dstance from y to the neghborhood center x ncreases, and becomes zero when x y >ρ,.e., y / O x. In ths paper, the notaton a represents the Eucld norm for a vector a or the absolute value for a scalar a. The above crteron functon can be used for both gray level mages and color mages. The above clusterng crteron functon Jx loc can be wrtten as J loc x (U, r,s(x)) = u q (y)k(x y) (7) I(y) S(x)r 2 dy due to the fact that K(x y) =0for y / O x Integraton of Local Clusterng Crteron The optmal U, r, and S are defned as those that mnmze the clusterng crteron functon Jx loc (U, r, S(x)) for all the neghborhood O x. Mnmzaton of Jx loc for all x Ω can be acheved by mnmzng the ntegral of Jx loc over Ω. Therefore, we defne an energy J (u, r,s) J loc x (u, r,s(x))dx,.e. J (U, r,s) u q (y)k(x y) (8) I(y) S(x)r 2 dydx By changng the order of ntegraton, the above energy J (u, r,s) can be wrtten n the form: J (U, r,s)= u q (y)d (I(y))dy (9) where d (I(y)) K(x y) I(y) S(x)r 2 dx (10) 704

4 Image segmentaton and the estmaton of llumnaton and reflectance can be acheved by solvng the followng constraned energy mnmzaton problem: mn J (U, r, S) subject to U U, (11) where U s the space of all the membershp functons,.e. U {(u 1,,u ) T :0 u (x) 1,,,, and u (x) =1, for all x Ω} (12) 3.4. Energy Formulaton wth Membershp Regularzaton For mages wth hgh level nose, t s necessary to regularze membershp functon by addng a regularzaton term R(U) to the above clusterng energy J (U, r,s). Therefore, we defne J reg (U, r,s)=j (U, r,s)+γr(u) (13) In ths work, we use total varaton of the membershp functons as the membershp regularzaton term,.e. Then we defne R(U) = u (x) dx (14) J reg (U, r,s) J (U, r,s)+γr(u), (15) and solve the followng constraned energy mnmzaton problem: mn J reg (U, r,s) subject to U U, (16) It s worth notng that ths energy s convex n the varables r and S, whch mples a unque global mnmum n each of these varables gven the others. The mnmzaton of the energy functonals J (U, r,s) and J reg (U, r,s) s descrbed n the next secton. 4. Energy Mnmzaton We have defned two constraned energy mnmzaton problems n Eqs. (11) and (16) n the prevous secton. The energy mnmzaton for the varables r and S are the same for both cases, as the addtonal membershp regularzaton term γr(u) n (16) s fxed as a constant when mnmzng the energy functonals J (U, r,s) and J reg (U, r,s)wth respect to r and S Mnmzaton of J (U, r,s) Case q>1 Mnmzaton of the energy J (U, r,s) can be performed by nterleavng the mnmzaton wth respect to the varables U, r, and S. ote that the energy J (U, r,s) s convex n each varable, and t s a quadratc form n terms of the varables r and S. The mnmzaton wth respect to each varable, gven the other two fxed, can be acheved by a closed form soluton as below: For fxed U and S, there s a unque mnmzer of the energy J (U, r,s) wth respect to r, whch s n the form denoted by ˆr =(ˆr 1,, ˆr ), s gven by (S K)Iu q ˆr = dx (S2 K)u q, =1,,. (17) dx where s the convoluton operaton. For fxed r and S, there s a unque mnmzer of the energy J (U, r,s) wth respect to U, whch s gven by û (y) = 1 k=1 ( d(i(y)) d k (I(y)) ) 1 q 1 (18) Gven U and r, there s a unque mnmzer of the energy J (U, r,s) wth respect to S, whch s gven by Ŝ = (IR(1) ) K (19) R (2) K where R (1) = r u q and R(2) = r2 uq. ote that the convolutons wth a kernel K n the expresson of Ŝ n Eq. (19) confrms the smoothness of the derved optmal llumnaton mage Ŝ, whch has been explaned from the defnton of the clusterng crteron functon Jx loc n the prevous secton Case q =1 For the case of q =1, the optmzaton of the varables r and S n the energy J (U, r,s) s the same as (17) and (19) wth q =1. However, the optmal membershp functon U cannot be computed by (18) n ths case. The optmzaton of r for the case of q =1s gven below. Let u 1,,u be postve functons such that u (x) =1for all x Ω, then the energy d (I(x))u (x)dx (20) s mnmzed f for each x Ω, the values of u 1 (x),,u (x) are gven by { 1, = mn (x); û (x) = 0, mn (x). (21) 705

5 where mn (x) =argmn{d (I(x))} Mnmzaton of J reg (U, r,s) wth Respect to U Usng the same technque as n [10]. The frst constrant can be relaxed by addng exact penalty convex functons ν(u )=max (0, 2u 1 1). Usng Lagrange multpler method, the second constrant can be removed by addng a term α ( u 1) 2 dx n (15) where α s a postve Lagrange multpler. Then the equvalent unconstraned problem s to mnmze J reg (U, r,s)= ( γ u dx + u d dx + ν(u ) ) + α 2 ( u 1) 2 dx. (22) Observng that the energy (22) s convex n varables u,we choose to follow [3] and take use of the fast dualty projecton method n [4]. So we add auxlary varables v and mnmze the followng approxmate energy J reg (U, V, r,s)= (γ v dx + 1 2θ (v u ) 2 dx + u d dx + ν(u )) + α 2 ( u 1) 2 dx (23) where θ>0s chosen to be small enough so that u and v are almost dentcal wth respect to the L 2 norm. Snce (23) s convex n u and v, we can use alternate mnmzaton method to fnd the mnmzer of (23). Decompose the problem nto four sub-problems as follows For fxed U, r,s,solvev by mnmzng γ v dx + 1 (v u ) 2 dx 2θ Ths problem can be effcently solved by fast dualty projecton algorthm. The soluton s gven by v = u θdv p,,..., K where the vector p can be solved by fxed pont method: Intalzng p 0 =0and teratng p n+1 = pn + τ (dv pn u /θ) 1+τ (dv p n u /θ). wth τ 1/8 to ensure convergence. For fxed V,r,S,wesolveu by mnmzng 1 2θ (v u ) 2 dx + u d dx +ν(u )+ α 2 ( u 1) 2 dx. Remark that the presence of the ν(u ) term n the energy s equvalent to cuttng off each u at 0 and at 1. On the other hand, f u [0, 1], then ν(u )=0, and the Euler-Lagrange equaton of ths problem s 1 θ (u v )+d + α( u 1) = 0 Then the mnmzaton s gven by u = mn(max( v θd θα( j u j 1), 0), 1). 1+θα For fxed U and V, r and S are gven by Eqs. (17) and (19) respectvely. 5. Implementaton and Expermental Results 5.1. Implementaton and Parameter Settng The energy mnmzaton s performed by an nterleaved optmzaton of each varable of the energy J (U, r,s) n an teratve process. In every teraton, we mnmze the energy wth respect to one of the three varables gven the other two obtaned from the prevous teraton. Before the teraton, the three varables U, r, and S are ntalzed as follows. In our mplementaton, we always ntalze the llumnaton mage S as a constant map for smplcty. To reduce the teraton number and save computaton tme, we apply the effcent K-means algorthm as a prelmnary segmentaton for the ntalzaton of the membershp functon U and the reflectance constant r. After the optmal Û, ˆr, and Ŝ are obtaned as the result of energy mnmzaton, the segmentaton result s gven by the membershp functons n Û =(û 1,, û ). The llumnaton and reflectance mages are gven by Ŝ and ˆR(x) = ˆr û (x) (24) To estmate the llumnaton mage more accurately, t s preferable to choose a small scale parameter σ n the Gaussan kernel K. In ths case, we use σ =1for the estmaton of llumnaton and reflectance mages. A larger scale parameter σ would blur the resultng llumnaton mage. If one s only nterested n the segmentaton result, a large scale parameter σ can be used n our method. In addton, the fuzzfer q s set to q =1.2 n most of our experments Expermental Results We frst demonstrate the effectveness of the proposed method for three mages n Column 1 of Fg. 1. The spatal llumnaton varaton s obvous n these mages. The segmentaton results and llumnaton mages are shown n Columns 2 and 3, respectvely. The segmentaton results 706

6 are hghly consstent wth the orgnal mages. The estmated reflectance mages exhbt desrable homogenety n each regon, whle ntensty varaton due to the llumnaton changes s extracted to the estmated llumnaton mages n Column 3, whch characterze the shapes of the surfaces n the scenes and the lghtng condton. Fgure 2. Result for two face mages. Column 1: Orgnal mage; Column 2: Segmentaton result; Column 3: Estmated llumnaton mage. Fgure 1. Results for three color mages. Column 1: Orgnal mage; Column 2: Segmentaton result; Column 3: Estmated llumnaton mage. We have also appled our method to face mages, as shown n Fg. 2. The varaton n appearance caused by spatal llumnaton changes often poses addtonal challenge n face recognton. Therefore t s n general desrable to obtan an llumnaton free mage,.e., a pure reflectance mage, as the nput to face recognton systems to mprove robustness and recognton rate. Fg. 2 shows the results of our method for two face mages n Column 1. The segmentaton results are shown n Column 2. ote that the skns on both faces and necks are correctly dentfed as a sngle regon. As a result, the correspondng estmaton of the reflectance mages exhbt such homogenety n the faces and necks. The spatal varaton n llumnaton, characterzng the geometry of the face, s extracted n the llumnaton mages, as shown n Column 3. The proposed method s by nature able to deal wth ntensty nhomogenetes n mage segmentaton. Fg. 3 shows the segmentaton results of our method for four mages. There are fast local ntensty varatons, such as the ntensty varatons wthn the grass or the squrrel n the frst mage. For such mages, t would be helpful to smooth the mage to sgnfcantly suppress fast local ntensty varatons. Our method s then appled to the smoothed mages, whch gves desrable segmentaton results as shown n Fg. 3. As mentoned n Secton 1, the Mumford-Shah model s able to deal wth ntensty n mage segmentaton. aturally, we compare our method wth the pecewse smooth (PS) model proposed n [13], whch s a well-known method that Fgure 3. Applcaton of our method for mage segmentaton. Column 1: The nput mages; Columns 2 and 3 show the segmentaton results n the forms of regons and boundares, respectvely. mplements the Mumford-Shah model usng a level set approach. We have mplemented the PS model for gray level mages n a two-phase formulaton. For a far comparson, we chose a synthetc gray level mage n Fg. 4 that can be segmented by the PS model n a two-phase formulaton. For ths mage, we appled our model wth the membershp regularzaton term. We used dfferent ntalzatons to test the robustness of the two methods. The results show that our method s robust to ntalzaton. Ths can be seen from the results n Fgs. 4(b) and 4(c) for two dfferent ntalzatons. The values of the ntal membershp functons are generated as random numbers for the frst ntalzaton (for the result n Fg. 4(b)), and a crcular regon s used to defne the membershp functons for the second ntalzaton (for the result n Fg. 4(c)). It s clear that these two results are almost the same, whch demonstrates the robustness of our method to ntalzaton. The PS model often produces qute dfferent results for even slghtly dfferent ntalzatons n our mplementaton. The results of the PS model wth two 707

7 method s able to segment mages wth non-unform ntenstes caused by spatal varatons n llumnaton. Comparsons wth the state-of-the-art pecewse smooth model demonstrate the superor performance of our method. (a) Orgnal mage. (b) Result 1 of our (c) Result 2 of our method. method. (d) Result 1 of the PS (e) Result 2 of the PS model. model. Fgure 4. Comparson of our method wth the pecewse smooth model. ntalzatons are shown n Fgs. 4(d) and 4(e). The ntalzaton of the level set functons n the PS model are obtaned by two smlar crcular regons of dfferent szes. We also compared the CPU tmes of the two methods wth the same crcular regon for the defnton of the ntal membershp functon n our method (for the result n Fg. 4(c))) and the level set functon n the PS model (for the result n Fg. 4(e)). The CPU tmed consumed by our method and the PS model are 7.83 and seconds respectvely. These CPU tmes are recorded n runnng our Matlab programs on a Lenovo ThnkPad notebook wth Intel (R) Core (TM)2 Duo CPU, 2.40 GHz, 2 GB RAM, wth Matlab 7.4 on Wndows Vsta. Ths experment demonstrates the advantage of our method n terms robustness and computatonal effcency over the PS model. 6. Concluson We have presented a novel method for mage segmentaton wth smultaneous estmaton of llumnaton and reflectance mages. The proposed method s based on the composton of an observed scene mage as the product of an llumnaton component and a reflectance component, known as ntrnsc mages. We defne an energy functonal n terms of an llumnaton mage, the membershp functons of the regons, and the correspondng reflectance constants of the regons n the scene. Ths energy s convex n each of ts varables. By mnmzng the energy, mage segmentaton result s obtaned n the form of the membershp functons of the regons. Meanwhle, the llumnaton and reflectance components of the observed mage are estmated smultaneously as the result of energy mnmzaton. Wth llumnaton taken nto account, the proposed References [1] H. Barrow and J. Tenenbaum. Recoverng ntrnsc scene characterstcs from mages. In A. Hanson and E. Rseman, edtors, Computer Vson Systems, pages Academc Press, [2] A. Blake. Boundary condtons for lghtness computaton n Mondran world. Computer Vson, Graphcs and Image Processng, 32(3): , December [3] X. Bresson, S. Esedoglu, P. Vandergheynst, J.-P. Thran, and S. Osher. Fast global mnmzaton of the actve contour/snake model. J. Math. Imagng Vs., 28(2): , [4] A. Chambolle. An algorthm for total varaton mnmzaton and applcatons. J. Math. Imagng Vs., 20(1-2):89 97, January [5] T. Chan and L. Vese. Actve contours wthout edges. IEEE Trans. Imag. Proc., 10(2): , February [6] B. Horn. Determnng lghtness from an mage. Comput. Graphcs, Image Processng, 3(4): , December [7] R. Kmmel, M. Elad, D. Shaked, R. Keshet, and I. Sobel. A varatonal framework for retnex. Int l J. Comp. Vs., 52(1):7 23, Aprl [8] E. Land and J. McCann. Lghtness and retnex theory. J. Opt. Soc. of Amerca, 61(1):1 11, [9] Y. Matsushta, K. shno, K. Ikeuch, and M. Sakauch. Illumnaton normalzaton wth tme-dependent ntrnsc mages for vdeo survellance. IEEE Trans. Pattern Anal. Mach. Intell., 26(10): , October [10] S. E. Mla kolova and T. F. Chan. Algorthms for fndng global mnmzers of mage segmentaton and denosng models. SIAM J. Appl. Math., 66(5): , [11] D. Mumford and J. Shah. Optmal approxmatons by pecewse smooth functons and assocated varatonal problems. Commun. Pure Appl. Math., 42(5): , [12] M. Tappen, E. Adelson, and W. Freeman. Estmatng ntrnsc component mages usng non-lnear regresson. In Proceedngs of IEEE Conference on Computer Vson and Pattern Recognton (CVPR), volume 2, pages , [13] L. Vese and T. Chan. A multphase level set framework for mage segmentaton usng the Mumford and Shah model. Int l J. Comp. Vs., 50(3): , December [14] Y. Wess. Dervng ntrnsc mages from mage sequences. In Proceedngs of the Eghth Internatonal Conference on Computer Vson (ICCV), volume II, pages 68 75, [15] S.-C. Zhu and A. Yulle. Regon competton: Unfyng snakes, regon growng, and Bayes/MDL for multband mage segmentaton. IEEE Trans. Patt. Anal. Mach. Intell., 18(9): , September

Nonlocal Mumford-Shah Model for Image Segmentation

Nonlocal Mumford-Shah Model for Image Segmentation for Image Segmentaton 1 College of Informaton Engneerng, Qngdao Unversty, Qngdao, 266000,Chna E-mal:ccluxaoq@163.com ebo e 23 College of Informaton Engneerng, Qngdao Unversty, Qngdao, 266000,Chna E-mal:

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

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

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

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

Learning the Kernel Parameters in Kernel Minimum Distance Classifier

Learning the Kernel Parameters in Kernel Minimum Distance Classifier Learnng the Kernel Parameters n Kernel Mnmum Dstance Classfer Daoqang Zhang 1,, Songcan Chen and Zh-Hua Zhou 1* 1 Natonal Laboratory for Novel Software Technology Nanjng Unversty, Nanjng 193, Chna Department

More information

GSLM Operations Research II Fall 13/14

GSLM Operations Research II Fall 13/14 GSLM 58 Operatons Research II Fall /4 6. Separable Programmng Consder a general NLP mn f(x) s.t. g j (x) b j j =. m. Defnton 6.. The NLP s a separable program f ts objectve functon and all constrants are

More information

NUMERICAL SOLVING OPTIMAL CONTROL PROBLEMS BY THE METHOD OF VARIATIONS

NUMERICAL SOLVING OPTIMAL CONTROL PROBLEMS BY THE METHOD OF VARIATIONS ARPN Journal of Engneerng and Appled Scences 006-017 Asan Research Publshng Network (ARPN). All rghts reserved. NUMERICAL SOLVING OPTIMAL CONTROL PROBLEMS BY THE METHOD OF VARIATIONS Igor Grgoryev, Svetlana

More information

Determining the Optimal Bandwidth Based on Multi-criterion Fusion

Determining the Optimal Bandwidth Based on Multi-criterion Fusion Proceedngs of 01 4th Internatonal Conference on Machne Learnng and Computng IPCSIT vol. 5 (01) (01) IACSIT Press, Sngapore Determnng the Optmal Bandwdth Based on Mult-crteron Fuson Ha-L Lang 1+, Xan-Mn

More information

A Robust Parametric Method for Bias Field Estimation and Segmentation of MR Images

A Robust Parametric Method for Bias Field Estimation and Segmentation of MR Images A Robust Parametrc Method for Bas Feld Estmaton and Segmentaton of MR Images Chunmng L, Chrs Gatenby,LWang 2, John C. Gore Vanderblt Unversty of Imagng Scence, Nashvlle, TN 37232, USA 2 Nanjng Unversty

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

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

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

Support Vector Machines

Support Vector Machines Support Vector Machnes Decson surface s a hyperplane (lne n 2D) n feature space (smlar to the Perceptron) Arguably, the most mportant recent dscovery n machne learnng In a nutshell: map the data to a predetermned

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

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

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

Solving two-person zero-sum game by Matlab

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

More information

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

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

More information

Boundary Detection Using Open Spline Curve Based on Mumford-Shah Model

Boundary Detection Using Open Spline Curve Based on Mumford-Shah Model Vol. 35, o. 2 ACTA AUTOMATICA SIICA February, 2009 Boundary Detecton Usng Open Splne Curve Based on Mumford-Shah Model LI ao-mao 1, 2 ZHU Ln-Ln 1, 2 TAG Yan-Dong 1 Abstract Inspred by Cremers s work, ths

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

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

Skew Angle Estimation and Correction of Hand Written, Textual and Large areas of Non-Textual Document Images: A Novel Approach Angle Estmaton and Correcton of Hand Wrtten, Textual and Large areas of Non-Textual Document Images: A Novel Approach D.R.Ramesh Babu Pyush M Kumat Mahesh D Dhannawat PES Insttute of Technology Research

More information

Optimized Region Competition Algorithm Applied to the Segmentation of Artificial Muscles in Stereoscopic Images

Optimized Region Competition Algorithm Applied to the Segmentation of Artificial Muscles in Stereoscopic Images Vol. 2, No. 3, Page 185-195 Copyrght 2008, TSI Press Prnted n the USA. All rghts reserved Optmzed Regon Competton Algorthm Appled to the Segmentaton of Artfcal Muscles n Stereoscopc Images Rafael Verdú-Monedero,

More information

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

A New Approach For the Ranking of Fuzzy Sets With Different Heights New pproach For the ankng of Fuzzy Sets Wth Dfferent Heghts Pushpnder Sngh School of Mathematcs Computer pplcatons Thapar Unversty, Patala-7 00 Inda pushpndersnl@gmalcom STCT ankng of fuzzy sets plays

More information

An Image Segmentation Method Based on Partial Differential Equation Models

An Image Segmentation Method Based on Partial Differential Equation Models An Image Segmentaton Method Based on Partal Dfferental Equaton Models Jang We, Lu Chan* College of Informaton Engneerng, Tarm Unversty, Alar, Chna *Correspondng Author Emal: 76356718@qq.com Abstract In

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

A Region based Active Contour Approach for Liver CT Image Analysis Driven by Local likelihood Image Fitting Energy

A Region based Active Contour Approach for Liver CT Image Analysis Driven by Local likelihood Image Fitting Energy Internatonal Journal of Engneerng and Advanced Technology (IJEAT) ISSN: 49 8958, Volume-6 Issue-5, June 07 A Regon based Actve Contour Approach for Lver CT Image Analyss Drven by Local lkelhood Image Fttng

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

Quality Improvement Algorithm for Tetrahedral Mesh Based on Optimal Delaunay Triangulation

Quality 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 information

Positive Semi-definite Programming Localization in Wireless Sensor Networks

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

More information

Smoothing Spline ANOVA for variable screening

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

More information

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

Feature Reduction and Selection

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

More information

An Accurate Evaluation of Integrals in Convex and Non convex Polygonal Domain by Twelve Node Quadrilateral Finite Element Method

An Accurate Evaluation of Integrals in Convex and Non convex Polygonal Domain by Twelve Node Quadrilateral Finite Element Method Internatonal Journal of Computatonal and Appled Mathematcs. ISSN 89-4966 Volume, Number (07), pp. 33-4 Research Inda Publcatons http://www.rpublcaton.com An Accurate Evaluaton of Integrals n Convex and

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

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

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

NAG Fortran Library Chapter Introduction. G10 Smoothing in Statistics

NAG Fortran Library Chapter Introduction. G10 Smoothing in Statistics Introducton G10 NAG Fortran Lbrary Chapter Introducton G10 Smoothng n Statstcs Contents 1 Scope of the Chapter... 2 2 Background to the Problems... 2 2.1 Smoothng Methods... 2 2.2 Smoothng Splnes and Regresson

More information

Fitting: Deformable contours April 26 th, 2018

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

More information

A Bilinear Model for Sparse Coding

A Bilinear Model for Sparse Coding A Blnear Model for Sparse Codng Davd B. Grmes and Rajesh P. N. Rao Department of Computer Scence and Engneerng Unversty of Washngton Seattle, WA 98195-2350, U.S.A. grmes,rao @cs.washngton.edu Abstract

More information

An Efficient Genetic Algorithm with Fuzzy c-means Clustering for Traveling Salesman Problem

An Efficient Genetic Algorithm with Fuzzy c-means Clustering for Traveling Salesman Problem An Effcent Genetc Algorthm wth Fuzzy c-means Clusterng for Travelng Salesman Problem Jong-Won Yoon and Sung-Bae Cho Dept. of Computer Scence Yonse Unversty Seoul, Korea jwyoon@sclab.yonse.ac.r, sbcho@cs.yonse.ac.r

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

Novel Fuzzy logic Based Edge Detection Technique

Novel Fuzzy logic Based Edge Detection Technique Novel Fuzzy logc Based Edge Detecton Technque Aborsade, D.O Department of Electroncs Engneerng, adoke Akntola Unversty of Tech., Ogbomoso. Oyo-state. doaborsade@yahoo.com Abstract Ths paper s based on

More information

Fuzzy Filtering Algorithms for Image Processing: Performance Evaluation of Various Approaches

Fuzzy Filtering Algorithms for Image Processing: Performance Evaluation of Various Approaches Proceedngs of the Internatonal Conference on Cognton and Recognton Fuzzy Flterng Algorthms for Image Processng: Performance Evaluaton of Varous Approaches Rajoo Pandey and Umesh Ghanekar Department of

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

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

The Shortest Path of Touring Lines given in the Plane

The Shortest Path of Touring Lines given in the Plane Send Orders for Reprnts to reprnts@benthamscence.ae 262 The Open Cybernetcs & Systemcs Journal, 2015, 9, 262-267 The Shortest Path of Tourng Lnes gven n the Plane Open Access Ljuan Wang 1,2, Dandan He

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

An Application of the Dulmage-Mendelsohn Decomposition to Sparse Null Space Bases of Full Row Rank Matrices

An Application of the Dulmage-Mendelsohn Decomposition to Sparse Null Space Bases of Full Row Rank Matrices Internatonal Mathematcal Forum, Vol 7, 2012, no 52, 2549-2554 An Applcaton of the Dulmage-Mendelsohn Decomposton to Sparse Null Space Bases of Full Row Rank Matrces Mostafa Khorramzadeh Department of Mathematcal

More information

Problem Definitions and Evaluation Criteria for Computational Expensive Optimization

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

More information

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

Unsupervised Learning

Unsupervised Learning Pattern Recognton Lecture 8 Outlne Introducton Unsupervsed Learnng Parametrc VS Non-Parametrc Approach Mxture of Denstes Maxmum-Lkelhood Estmates Clusterng Prof. Danel Yeung School of Computer Scence and

More information

A Robust Method for Estimating the Fundamental Matrix

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

More information

LECTURE : MANIFOLD LEARNING

LECTURE : MANIFOLD LEARNING LECTURE : MANIFOLD LEARNING Rta Osadchy Some sldes are due to L.Saul, V. C. Raykar, N. Verma Topcs PCA MDS IsoMap LLE EgenMaps Done! Dmensonalty Reducton Data representaton Inputs are real-valued vectors

More 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

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

Optimal Scheduling of Capture Times in a Multiple Capture Imaging System

Optimal Scheduling of Capture Times in a Multiple Capture Imaging System Optmal Schedulng of Capture Tmes n a Multple Capture Imagng System Tng Chen and Abbas El Gamal Informaton Systems Laboratory Department of Electrcal Engneerng Stanford Unversty Stanford, Calforna 9435,

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

A NEW FUZZY C-MEANS BASED SEGMENTATION STRATEGY. APPLICATIONS TO LIP REGION IDENTIFICATION

A NEW FUZZY C-MEANS BASED SEGMENTATION STRATEGY. APPLICATIONS TO LIP REGION IDENTIFICATION A NEW FUZZY C-MEANS BASED SEGMENTATION STRATEGY. APPLICATIONS TO LIP REGION IDENTIFICATION Mhaela Gordan *, Constantne Kotropoulos **, Apostolos Georgaks **, Ioanns Ptas ** * Bass of Electroncs Department,

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

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

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

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

Video Object Tracking Based On Extended Active Shape Models With Color Information

Video Object Tracking Based On Extended Active Shape Models With Color Information CGIV'2002: he Frst Frst European Conference Colour on Colour n Graphcs, Imagng, and Vson Vdeo Object rackng Based On Extended Actve Shape Models Wth Color Informaton A. Koschan, S.K. Kang, J.K. Pak, B.

More information

Programming in Fortran 90 : 2017/2018

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

More information

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

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

Geodesic Active Regions for Supervised Texture Segmentation

Geodesic Active Regions for Supervised Texture Segmentation Geodesc Actve egons for Supervsed Texture Segmentaton Nkos Paragos achd Derche INIA BP 9, 00, oute des Lucoles 0690 Sopha Antpols Cedex, France e-mal: {nparago,der}@sophanrafr Abstract Ths paper presents

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

Categories and Subject Descriptors B.7.2 [Integrated Circuits]: Design Aids Verification. General Terms Algorithms

Categories and Subject Descriptors B.7.2 [Integrated Circuits]: Design Aids Verification. General Terms Algorithms 3. Fndng Determnstc Soluton from Underdetermned Equaton: Large-Scale Performance Modelng by Least Angle Regresson Xn L ECE Department, Carnege Mellon Unversty Forbs Avenue, Pttsburgh, PA 3 xnl@ece.cmu.edu

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

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

Learning a Class-Specific Dictionary for Facial Expression Recognition

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 information

Range images. Range image registration. Examples of sampling patterns. Range images and range surfaces

Range 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 information

The Research of Ellipse Parameter Fitting Algorithm of Ultrasonic Imaging Logging in the Casing Hole

The Research of Ellipse Parameter Fitting Algorithm of Ultrasonic Imaging Logging in the Casing Hole Appled Mathematcs, 04, 5, 37-3 Publshed Onlne May 04 n ScRes. http://www.scrp.org/journal/am http://dx.do.org/0.436/am.04.584 The Research of Ellpse Parameter Fttng Algorthm of Ultrasonc Imagng Loggng

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

Vanishing Hull. Jinhui Hu, Suya You, Ulrich Neumann University of Southern California {jinhuihu,suyay,

Vanishing Hull. Jinhui Hu, Suya You, Ulrich Neumann University of Southern California {jinhuihu,suyay, Vanshng Hull Jnhu Hu Suya You Ulrch Neumann Unversty of Southern Calforna {jnhuhusuyay uneumann}@graphcs.usc.edu Abstract Vanshng ponts are valuable n many vson tasks such as orentaton estmaton pose recovery

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

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

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

More information

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

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

More information

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

Robust Mean Shift Tracking with Corrected Background-Weighted Histogram

Robust Mean Shift Tracking with Corrected Background-Weighted Histogram Robust Mean Shft Trackng wth Corrected Background-Weghted Hstogram Jfeng Nng, Le Zhang, Davd Zhang and Chengke Wu Abstract: The background-weghted hstogram (BWH) algorthm proposed n [] attempts to reduce

More information

Active Contours/Snakes

Active Contours/Snakes Actve Contours/Snakes Erkut Erdem Acknowledgement: The sldes are adapted from the sldes prepared by K. Grauman of Unversty of Texas at Austn Fttng: Edges vs. boundares Edges useful sgnal to ndcate occludng

More information

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

Multi-stable Perception. Necker Cube

Multi-stable Perception. Necker Cube Mult-stable Percepton Necker Cube Spnnng dancer lluson, Nobuuk Kaahara Fttng and Algnment Computer Vson Szelsk 6.1 James Has Acknowledgment: Man sldes from Derek Hoem, Lana Lazebnk, and Grauman&Lebe 2008

More information

A PATTERN RECOGNITION APPROACH TO IMAGE SEGMENTATION

A PATTERN RECOGNITION APPROACH TO IMAGE SEGMENTATION 1 THE PUBLISHING HOUSE PROCEEDINGS OF THE ROMANIAN ACADEMY, Seres A, OF THE ROMANIAN ACADEMY Volume 4, Number 2/2003, pp.000-000 A PATTERN RECOGNITION APPROACH TO IMAGE SEGMENTATION Tudor BARBU Insttute

More information

Face Recognition Method Based on Within-class Clustering SVM

Face 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 information

Model-Based Bundle Adjustment to Face Modeling

Model-Based Bundle Adjustment to Face Modeling Model-Based Bundle Adjustment to Face Modelng Oscar K. Au Ivor W. sang Shrley Y. Wong oscarau@cs.ust.hk vor@cs.ust.hk shrleyw@cs.ust.hk he Hong Kong Unversty of Scence and echnology Realstc facal synthess

More information

Using Fuzzy Logic to Enhance the Large Size Remote Sensing Images

Using Fuzzy Logic to Enhance the Large Size Remote Sensing Images Internatonal Journal of Informaton and Electroncs Engneerng Vol. 5 No. 6 November 015 Usng Fuzzy Logc to Enhance the Large Sze Remote Sensng Images Trung Nguyen Tu Huy Ngo Hoang and Thoa Vu Van Abstract

More information

Fixing Max-Product: Convergent Message Passing Algorithms for MAP LP-Relaxations

Fixing Max-Product: Convergent Message Passing Algorithms for MAP LP-Relaxations Fxng Max-Product: Convergent Message Passng Algorthms for MAP LP-Relaxatons Amr Globerson Tomm Jaakkola Computer Scence and Artfcal Intellgence Laboratory Massachusetts Insttute of Technology Cambrdge,

More information

Classification / Regression Support Vector Machines

Classification / Regression Support Vector Machines Classfcaton / Regresson Support Vector Machnes Jeff Howbert Introducton to Machne Learnng Wnter 04 Topcs SVM classfers for lnearly separable classes SVM classfers for non-lnearly separable classes SVM

More information

Ecient Computation of the Most Probable Motion from Fuzzy. Moshe Ben-Ezra Shmuel Peleg Michael Werman. The Hebrew University of Jerusalem

Ecient Computation of the Most Probable Motion from Fuzzy. Moshe Ben-Ezra Shmuel Peleg Michael Werman. The Hebrew University of Jerusalem Ecent Computaton of the Most Probable Moton from Fuzzy Correspondences Moshe Ben-Ezra Shmuel Peleg Mchael Werman Insttute of Computer Scence The Hebrew Unversty of Jerusalem 91904 Jerusalem, Israel Emal:

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

New Appearance Models for Natural Image Matting

New Appearance Models for Natural Image Matting New Appearance Models for Natural Image Mattng Dheeraj Sngaraju Johns Hopkns Unversty Baltmore, MD, USA. dheeraj@cs.jhu.edu Carsten Rother Mcrosoft Research Cambrdge, UK. carrot@mcrosoft.com Chrstoph Rhemann

More information

Maximum Variance Combined with Adaptive Genetic Algorithm for Infrared Image Segmentation

Maximum Variance Combined with Adaptive Genetic Algorithm for Infrared Image Segmentation Internatonal Conference on Logstcs Engneerng, Management and Computer Scence (LEMCS 5) Maxmum Varance Combned wth Adaptve Genetc Algorthm for Infrared Image Segmentaton Huxuan Fu College of Automaton Harbn

More information

Resolving Ambiguity in Depth Extraction for Motion Capture using Genetic Algorithm

Resolving Ambiguity in Depth Extraction for Motion Capture using Genetic Algorithm Resolvng Ambguty n Depth Extracton for Moton Capture usng Genetc Algorthm Yn Yee Wa, Ch Kn Chow, Tong Lee Computer Vson and Image Processng Laboratory Dept. of Electronc Engneerng The Chnese Unversty of

More information

Wishing you all a Total Quality New Year!

Wishing you all a Total Quality New Year! Total Qualty Management and Sx Sgma Post Graduate Program 214-15 Sesson 4 Vnay Kumar Kalakband Assstant Professor Operatons & Systems Area 1 Wshng you all a Total Qualty New Year! Hope you acheve Sx sgma

More information

Mathematics 256 a course in differential equations for engineering students

Mathematics 256 a course in differential equations for engineering students Mathematcs 56 a course n dfferental equatons for engneerng students Chapter 5. More effcent methods of numercal soluton Euler s method s qute neffcent. Because the error s essentally proportonal to the

More information

Support Vector Machines. CS534 - Machine Learning

Support Vector Machines. CS534 - Machine Learning Support Vector Machnes CS534 - Machne Learnng Perceptron Revsted: Lnear Separators Bnar classfcaton can be veed as the task of separatng classes n feature space: b > 0 b 0 b < 0 f() sgn( b) Lnear Separators

More information

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

Fuzzy C-Means Initialized by Fixed Threshold Clustering for Improving Image Retrieval Fuzzy -Means Intalzed by Fxed Threshold lusterng for Improvng Image Retreval NAWARA HANSIRI, SIRIPORN SUPRATID,HOM KIMPAN 3 Faculty of Informaton Technology Rangst Unversty Muang-Ake, Paholyotn Road, Patumtan,

More information