Detection of Human Actions from a Single Example

Size: px
Start display at page:

Download "Detection of Human Actions from a Single Example"

Transcription

1 Detecton of Human Actons from a Sngle Example Hae Jong Seo and Peyman Mlanfar Electrcal Engneerng Department Unversty of Calforna at Santa Cruz 1156 Hgh Street, Santa Cruz, CA, {rokaf,mlanfar}@soe.ucsc.edu Abstract We present an algorthm for detectng human actons based upon a sngle gven vdeo example of such actons. The proposed method s unsupervsed, does not requre learnng, segmentaton, or moton estmaton. The novel features employed n our method are based on space-tme locally adaptve regresson kernels. Our method s based on the dense computaton of so-called space-tme local regresson kernels (.e. local descrptors) from a query vdeo, whch measure the lkeness of a voxel to ts spatotemporal surroundngs. Salent features are then extracted from these descrptors usng prncpal components analyss (PCA). These are effcently compared aganst analogous features from the target vdeo usng a matrx generalzaton of the cosne smlarty measure. The algorthm yelds a scalar resemblance volume; each voxel ndcatng the lkelhood of smlarty between the query vdeo and all cubes n the target vdeo. By employng non-parametrc sgnfcance tests and non-maxma suppresson, we accurately detect the presence and locaton of actons smlar to the gven query vdeo. Hgh performance s demonstrated on a challengng set of acton data [8] ndcatng successful detecton of multple complex actons even n the presence of fast motons. 1. Introducton A huge and growng number of vdeos are avalable onlne today. Human actons are one of the most mportant parts n moves, TV shows, and personal vdeos. Analyss of human actons n vdeos s consdered a very mportant component n computer vson systems because of such applcatons as content-based vdeo retreval, vsual survellance, analyss of sports events and more. The generc problem of nterest addressed n ths paper can be brefly descrbed as follows: We are gven a sngle query vdeo of an acton of nterest (for nstance a short Fgure 1. (a) A hand-wavng acton and possbly smlar actons (b) Gven a query vdeo, we want to detect/localze actons of nterest n a target vdeo T. T can be dvded nto a set of overlappng cubes ballet turn), and we are nterested n detectng smlar actons wthn other target vdeos. Detectng human actons from vdeo s a very challengng problem due to the fact that physcal body moton can look very dfferent dependng on the context: 1) smlar actons wth dfferent clothes, or n dfferent llumnaton and background can result n a large appearance varaton; 2) the same actons performed by two dfferent people may look dssmlar n terms of acton speed or frame rate of the vdeo (See Fg. 1 (a)). There have been many efforts to model and recognze human actons broadly by means of parametrc tme-seres approaches, frame-by-frame nonparametrc approaches, and volumetrc approaches. We refer the nterested reader to [13] and references theren for a good summary. Volumetrc approaches tend to outperform the other two approaches. These volumetrc methods do not requre background subtracton, moton estmaton, and complex models of body confguraton and knematcs. They tolerate varatons n appearance, scale, rotaton, and movement to some extent. Methods such as those n [5, 8] whch am at recognzng actons based solely on one query (what we shall call tranngfree) are very useful for vdeo retreval from the web. In these methods, a sngle query vdeo s provded by users and every gallery vdeo n the database s compared wth the sngle query, posng a vdeo-to-vdeo matchng problem. Inspred by ths trend toward tranng-free acton analyss, ths paper presents a novel tranng-free human acton IEEE 12th Internatonal Conference on Computer Vson (ICCV) /09/$ IEEE

2 Fgure 2. Acton detecton system overvew (There are broadly three stages.) detecton framework. Our proposed method s based on the calculaton and use of what we call space-tme local regresson kernels whch are local weghts computed drectly from the gven pxels n both the query and target vdeos. The orgnal motvaton to use these local regresson kernels s the earler successful work on adaptve kernel regresson for mage denosng, nterpolaton [9], deblurrng [10], and (2-D) generc object detecton [6]. Takeda et al. [11] extended the kernel regresson framework to super-resoluton by ntroducng space-tme local steerng kernels whch capture the essental local behavor of a spato-temporal neghborhood. The space-tme local steerng kernel (3-D LSK) s fundamentally based on the comparson of neghborng pxels n both space and tme, thus t mplctly contans nformaton about the local moton of the pxels across tme, thus requrng no explct moton estmaton. The space-tme local steerng kernel s defned as follows: { } det(cs) (xs x) T C s(x s x) K(x s x) = exp, (1) h 2 2h 2 where x s =[x 1,x 2,t] T s s the space-tme coordnates, s [1,,P ], hs a global smoothng parameter, P s the total number of samples n a space-tme local analyss wndow around a sample poston at x, and the matrx C s R (3 3) s a covarance matrx estmated from a collecton of frst dervatves along spatal (x 1,x 2 ) and temporal (t) axes. The covarance matrx C s modfes the shape and sze of the local kernel n a way whch robustly encodes the space-tme local geometrc structures present n vdeos. Normalzaton of ths kernel functon yelds robustness to llumnaton, contrast, and color dfferences. For a more n depth analyss on local steerng kernels, we refer the nterested reader to [6, 9, 10, 11]. Very recently, Shechtman and Iran [7] ntroduced a space-tme local self-smlarty descrptor for acton detecton and showed performance mprovement over ther prevous approach [8]. Ths (ndependently derved) local space-tme self-smlarty descrptor s a specal case of our space-tme local steerng kernel and s also related to a number of other local data adaptve metrcs such as Optmal Space-Tme Adaptaton (OSTA) [2] and Non-Local Means (NLM) [3] whch have been used very successfully for vdeo restoraton n the mage processng communty. Whle the method proposed by Shechtman and Iran [7] s related to our method, ther approach fundamentally dffers from ours n the followng respects: 1) Snce the calculaton of space-tme local steerng kernels s stable n the presence of uncertanty n the data [9], our approach s robust even n the presence of nose; 2) As opposed to [7] flterng out non-nformatve descrptors n order to reduce the tme complexty, we automatcally obtan the most salent feature volumes by applyng Prncpal Components Analyss (PCA) to a collecton of 3-D LSKs. From a practcal standpont, t s mportant to note that the proposed framework operates usng a sngle example of an acton of nterest to fnd smlar matches; does not requre any pror knowledge (learnng) about actons beng sought; and does not requre any pre-processng step or segmentaton of the target vdeo. Fg. 2 shows an overvew of our proposed framework for acton detecton. To summarze the operaton of the overall algorthm, we frst compute the normalzed space-tme local steerng kernels (3-D LSKs) W, W T from both and T. In the second stage, we obtan the salent feature volumes F, F T by projectng the descrptors W, W T to a projecton space A derved from W. In the thrd stage, we compare the feature volumes F T (=a chunk of F T at th poston) and F usng the Matrx Cosne Smlarty measure. The fnal output s gven after a sequence of sgnfcance tests, followed by non-maxma suppresson [4]. Ths paper s organzed as follows. In the next secton, we provde further techncal detals about the varous steps outlned above. In Secton 3, we demonstrate the performance of the system wth expermental results, and we conclude ths paper n Secton Techncal Detals As outlned n the prevous secton, our approach to detect actons conssts broadly of three stages. Assume that we are gven a target vdeo T and that we have a query vdeo, where s generally smaller than T. The task at hand s to detect and locate cubes of T that are smlar to. The frst step s to calculate space-tme local steerng kernels (3-D LSKs). To be more specfc, 3-D LSK functon K(x s x) s densely calculated and normalzed as follows: W I (x s x)= K I(x s x) P s=1 K I(x s x), { s =1,,P, I {, T }. Fg. 3 llustrates what the normalzed versons of 2-D LSKs and 3-D LSKs n varous regons look lke. (2) 1966

3 Fgure 3. (a) Examples of 2-D LSK n varous regons. (b) Examples of space-tme local steerng kernel (3-D LSK) n varous regons. Note that key frame means the frame where the center of 3-D LSK s located. In order to organze W (x s x) s and W T (x s x) s, whch are densely computed from and T, let W,W T be matrces whose columns are vectors w,w T, whch are column-stacked (rasterzed) versons of W (x s x),w T (x s x) respectvely: W = [w, 1, w] n R P n, W T = [wt 1,, w nt T ] RP nt. (3) where n and n T are the number of 3-D LSKs n the query vdeo and the target vdeo T respectvely. As descrbed n Fg. 2, the next step s to apply PCA to W for dmensonalty reducton and to retan only ts salent characterstcs. Applyng PCA to W we can retan the frst (largest) d prncpal components 1 whch form the columns of a matrx A R P d. Next, the lower dmensonal features are computed by projectng W and W T onto A : F = [f, 1, f]=a n T W R d n, F T = [ft 1,, f nt T ]=AT T W T R d nt. (4) Fg. 4 llustrates the prncpal components n A and shows what the features F, F T look lke for the ballet vdeo case. Very recently, Al and Shah [1] proposed a set of knematc features that extract dfferent aspects of moton dynamcs present n the optcal flow. They obtaned bags of knematc modes for acton recognton by applyng PCA to a set of knematc features. We dfferentate our proposed method from [1] n the sense that 1) moton nformaton s mplctly contaned n 3-D LSK whle [1] explctly computes optcal flow; 2) Background subtracton was used as a pre-processng step whle our method s fully automatc; 3) [1] employed multple nstance learnng to a set of all knematc modes n the dataset whle our proposed method does not nvolve any tranng phase. 1 Typcally, d s selected to be a small nteger such as 3 or 4 so that 80 to 90% of the nformaton n the LSKs would be retaned. (.e., d=1 λ P=1 0.8 (to 0.9) where λ λ are the egenvalues.) Fgure 4. Ballet acton : A s learned from a collecton of 3-D LSKs W, and Feature row vectors of F and F T are computed from query and target vdeo T respectvely. Egenvectors and feature vectors were transformed to volume and up-scaled for llustraton purposes. The next step n the proposed framework s the measurement of a dstance between the computed features F, F T. For ths purpose, we employ the nonparametrc detecton framework [6] based on Matrx Cosne Smlarty. The Matrx Cosne Smlarty (MCS) between two feature matrces F, F T whch consst of a set of vectors can be defned as the Frobenus nner product between two normalzed matrces as follows: F T ρ =< F, F T > F = trace( F T ) [ 1, 1], (5) F F F T F f 1 f n f T 1,, ] and F F F F T =[,, F F T F f n T F T F ]. where, F =[ Equaton 5 can be rewrtten as a weghted average of the cosne smlartes ρ(f, f T ) between each par of correspondng feature vectors (.e., columns) n F, F T as follows: n f l T f l T n ρ = = ρ(f F F F T, l f l f l T ) f T l. (6) F F F F T F l=1 l=1 The weghts are represented as the product of f l T f l F F and F T F whch ndcate the relatve mportance of each feature n the feature sets F, F T. Ths measure 2 not only generalzes the cosne smlarty, but also overcomes the dsadvantages of the conventonal Eucldean dstance whch 2 We compute ρ over M target cubes and ths can be effcently mplemented by column-stackng the matrces F, F T and smply computng the cosne smlarty between two long column vectors as follows: ρ n d l=1 j=1 n l=1 f (l,j) f(l,j) T dj=1 f (l,j) 2 nl=1 dj=1 f (l,j) T 2 = ρ(colstack(f ), colstack(f T )) [ 1, 1], where f (l,j),f(l,j) T are elements n l th vector f l and f T l respectvely, and colstack( ) means an operator whch column-stacks a matrx., 1967

4 f l f l T f l f l T Fgure 5. Left: Examples of A) ρ(f, l ft l ) : cosne smlarty, B) F F F T F : weghts, and A B) ρ(f, l ft l ) F F F T F : weghted cosne smlarty. Note that query and target are same as those n Fg. 2(Left). Rght: two sgnfcance tests and non-maxma suppresson [4]are descrbed. s senstve to outlers. Fg. 5(Left) shows examples of the computaton of the MCS, whch ndcate that t provdes a relable measure of smlarty. It s worth notng that Shechtman and Iran [8] proposed 3-D volume correlaton score (global consstency measure between query and target cube) by computng a weghted average of local consstency measures. The dffculty wth that method s that local consstency values should be explctly computed from each correspondng subvolume of the query and target vdeo. Furthermore, the weghts to calculate a global consstency measure are based on a sgmod functon, whch s somewhat ad-hoc. Here, we clam that our measure, MCS s better motvated, more approprate, and more general than ther global consstency measure for acton detecton. The next step s to generate a so-called resemblance volume (RV), whch wll be a volume of voxels ndcatng the lkelhood of smlarty between and T at each spatotemporal poston. As for the fnal test statstc comprsng the values n the resemblance volume, we use the proporton of shared varance (ρ 2 ) to that of the resdual varance (1 ρ 2 ). More specfcally, RV s computed usng the functon f( ) as follows: ρ2 RV : f(ρ )= 1 ρ 2, =0,,M 1. (7) From a quanttatve pont of vew, we note that f(ρ ) s essentally the Lawley-Hotellng Trace statstc [12], whch s used as an effcent test statstc for detectng correlaton between two data sets. Next, we employ a two-step sgnfcance test as shown n Fg 5 (Rght). The frst s an overall threshold (τ 0 )on the RV to decde whether there s any suffcently smlar acton present n the target vdeo at all. If the answer s yes at suffcently hgh confdence, we would then want to know how many actons of nterest are present n the target vdeo and where they are. Therefore, we need two thresholds: an overall threshold 3 τ o as mentoned above, and a threshold 4 τ to detect the (possbly) multple occurrences of the same acton n the target vdeo. After the two sgnfcance tests wth τ o,τ are performed, we employ the dea of non-maxma suppresson [4] for the fnal detecton. We take the volume regon wth the hghest f(ρ ) value and elmnate the possblty that any other acton s detected wthn some radus 5 of the center of that volume regon agan. Ths enables us to avod multple false detectons of nearby actons already detected. Then we terate ths process untl the local maxmum value falls below the threshold τ. Fg. 5 (Rght) shows a graphcal llustraton of sgnfcance tests and non-maxma suppresson [4]. For the sake of completeness, the overall pseudo-code for the algorthm s gven n Algorthm Expermental Results Our method detects the presence and locaton of actons smlar to the gven query and provdes a seres of bound- 3 In a typcal scenaro, we set the overall threshold τ o to be, for nstance, 0.96 whch s about 50% of varance n common (.e., ρ 2 =0.49). In other words, f the maxmal f(ρ ) s just above 0.96, we decde that there exsts at least one acton of nterest. 4 We employ the dea of nonparametrc testng. We compute an emprcal probablty densty functon (PDF) from M samples f(ρ ) and we set τ so as to acheve, for nstance, a 99 % (α =0.99) sgnfcance level n decdng whether the gven values are n the extreme (rght) tals of the dstrbuton. Ths approach s based on the assumpton that n the target vdeo, most cubes do not contan the acton of nterest (n other words, acton of nterest s a relatvely rare event), and therefore, the few matches wll result n values whch are n the tals of the dstrbuton of f(ρ ). 5 The sze of ths excluson regon wll depend on the applcaton at hand and the characterstcs of the query vdeo. 1968

5 Algorthm 1 Tranng-free generc acton detecton : uery vdeo, T : Target vdeo, τ o : Overall threshold, α : Confdence level, P : Sze of space-tme local steerng kernel (3-D LSK) cube. Stage1 : Compute Descrptors Construct W, W T whch are a collecton of normalzed 3-D LSK assocated wth, T. Stage2 : Feature Representaton 1) Apply PCA to W and obtan projecton space A from ts top d egenvectors. 2) Project W and W T onto A to construct F and F T. Stage3 1) Compute Matrx Cosne Smlarty for every target cube T, where [0,,M 1] do ρ =< F F T, > F F F T F and (RV) : f(ρ F )= ρ2 1 ρ 2. end for Then, fnd max f(ρ ). 2) Sgnfcance tests ) If max f(ρ ) >τ o, go on to the next test. Otherwse, there s no acton of nterest n T. ) Threshold RV by τ whch s set to acheve 99 % confdence level (α = 0.99) from the emprcal PDF of f(ρ ). 3) Non-maxma suppresson Apply non-maxma suppresson [4] to RV untl the local maxmum value s below τ. ng cubes wth resemblance volume embedded around detected actons. Note that no background/foreground segmentaton s requred n the proposed method. Ths method can also handle modest amount of varatons n rotaton (up to ±15 degree), and spatal and temporal scale change (up to ±20%). In practce, once gven and T, we downsample both and T by some factor of (3, here) n order to reduce the tme-complexty. We then compute 3-D LSK of sze 3 3 (space) 7 (tme) as descrptors so that every space-tme locaton n and T yelds a 63-dmensonal local descrptor W and W T respectvely. The smoothng parameter h for computng 3-D LSKs was set to 2.1. We end up wth F, F T by reducng dmensonalty from 63 to d =4and then, we obtan RV by computng the MCS measure between F, F T. The threshold τ for each test example was determned by the confdence level α =0.99. We appled our method to 3 dfferent examples :.e. detectng 1) walkng people, 2) ballet turn actons, and 3) multple actons n one vdeo. Shechtman and Iran [8] have tested ther method on these vdeos usng the same query and [5, 7] also tested ther methods on some of these vdeos. We acheved smlar (or even better) performance as compared to the methods n [5, 7, 8]. It s worth notng here that the other acton detecton methods [5, 7, 8] dd not provde ether threshold values or descrbe how they selected threshold values n reportng detecton performance. On the other hand, the threshold values are automatcally chosen n our algorthm wth respect to the confdence level as explaned earler. Fg. 6(A) shows the results of searchng for nstances of walkng people n a target beach vdeo (460 frames of pxels). The query vdeo contans a very short walkng acton movng to the rght (14 frames of pxels) and has a background context whch s not the beach scene. In order to detect walkng actons n ether drecton, we used two queres ( and ts mrror-reflected verson) and generated two RVs. By votng the hgher score among values from two RVs at every space-tme locaton, we ended up wth one RV whch ncludes correct locatons of walkng people n the correct drecton. Fg. 6(A) (a) shows a few sampled frames from. In order to provde better llustraton of T, we dvded T nto 3 non-overlappng sectons. Fg. 6(A) (b) and (c) represent each part of T and ts correspondng RV respectvely. Red color represents hgher resemblance whle blue color denotes lower resemblance values. Fg. 6(A) (d) and (e) show a few frames from T, wth RV and boundng boxes supermposed on them respectvely. Fg. 6 (B) shows the results of detectng ballet turnng acton n a target ballet vdeo (284 frames of pxels). The query vdeo contans a sngle turn of a male dancer (13 frames of pxels). Fg. 6(B) (a) shows a few sampled frames from. Next, Fg. 6(B) (b) and (c) represent each part of T and ts correspondng RV respectvely. Fg. 6(B) (d) and (e) show a few frames from T wth resemblance volumes supermposed on t respectvely. Most of the turns of the two dancers (a male and a female) were detected even though ths vdeo contans very fast movng parts and relatvely large varablty n spatal scale and appearance (the female dancer wearng a skrt) as compared to the gven query. We observed that one of the female dancer turnng actons was mssed because of large spatal scale varaton as compared to the gven. However, we can easly deal wth ths problem by ether adjustng the sgnfcance level or usng mult-scale approach as done n [6]. The detecton result of the proposed method on ths vdeo outperforms that n [5, 8] and compares favorably to that n [7]. Fg. 6(C) shows the results of detectng 4 dfferent actons ( walk, wave, clap, and jump ) whch occur smultaneously n a target vdeo (120 frames of pxels). Four query vdeos were matched aganst the target vdeo ndependently. Fg. 6(C) (a) and (b) show a few sampled frames from and T respectvely. Whte boxes n Fg. 6(C) (a) represent actual regons used for the query. The resultng RVs are shown n Fg. 6(C) (c). In all the above examples, we used the same parameters. It s evdent, based on all the results above, that the proposed tranng-free acton detecton based on 3-D LSK works well and s robust to modest varatons n spato-temporal scale. Our system s desgned wth detecton accuracy as a hgh prorty. A typcal run of the object detecton system takes a lttle over 1 mnute on a target vdeo T (50 frames of pxels, Intel Pentum CPU 2.66 Ghz machne) usng a query (13 frames of ). Most of the runtme s taken up by the computaton of MCS (about 9 seconds, and 16.5 seconds for the computaton of 3-D LSKs from and T respectvely, whch needs to be computed 1969

6 Fgure 6. Results searchng for (A) walkng person on the beach, (B) ballet turn on the ballet vdeo, and (C) multple actons. (A,B): (a) query vdeo (a short walk clp) (b) target vdeo (c) resemblance volumes (RV) (d) a few frames from T (e) frames wth resemblance volume on top of t. (C): (a) four dfferent short vdeo queres. Note that whte boxes represent actual query regons (b) target vdeo T (c) resemblance volumes (RV)s wth respect to each query. only once.) There are many factors that affect the precse tmng of the calculatons, such as query sze, complexty of the vdeo, and LSK sze. Our system runs n Matlab but could be easly mplemented usng mult-threads or parallel programmng as well as General Purpose GPU for whch we expect a sgnfcant gan n speed. 4. Concluson and Dscusson In ths paper, we have proposed a novel acton detecton algorthm by employng space-tme local steerng kernels (3-D LSKs); and by usng a tranng-free nonparametrc detecton scheme based on Matrx Cosne Smlarty (MCS). The proposed method can automatcally detect n the target vdeo the presence, the number, as well as locaton of actons smlar to the gven query vdeo. The proposed method s practcally appealng because t s nonparametrc. The proposed framework s general enough as to be extendable to acton categorzaton usng a nearest neghbor classfer along wth an automatc acton croppng method as smlarly done n [5]. Improvement of the computatonal complexty of the proposed method s also a drecton of future research worth explorng. 5. Acknowledgment Ths work was supported by AFOSR Grant FA References [1] S. Al and M. Shah. Human acton recognton n vdeos usng knematc features and multple nstance learnng. Accepted for publcaton n IEEE Transactons on Pattern Analyss and Machne Intellgence (PAMI), [2] J. Boulanger, C. Kervrann, and P. Bouthemy. Space-tme adaptaton for patch-based mage sequence restoraton. IEEE Transactons on Pattern Analyss and Machne Intellgence, 29: , June [3] A. Buades, B. Coll, and J. M. Morel. Nonlocal mage and move denosng. Internatonal Journal of Computer Vson, 76(2): , [4] F. Devernay. A non-maxma suppresson method for edge detecton wth sub-pxel accuracy. Techncal report, INRIA, (RR-2724), [5] H. Nng, T. Han, D. Walther, M. Lu, and T. Huang. Herarchcal space-tme model enablng effcent search for human actons. IEEE Transactons on Crcuts and Systems for Vdeo Technology, n press, [6] H. J. Seo and P. Mlanfar. Tranng-free, generc object detecton usng locally adaptve regresson kernels. Accepted for publcaton n IEEE Transactons on Pattern Analyss and Machne Intellgence, June [7] E. Shechtman and M. Iran. Matchng local self-smlartes across mages and vdeos. In Proc. of IEEE Conference on Computer Vson and Pattern Recognton (CVPR), pages 1 8, June [8] E. Shechtman and M. Iran. Space-tme behavor-based correlaton -or- how to tell f two underlyng moton felds are smlar wthout computng them? IEEE Transactons on Pattern Analyss and Machne Intellgence, 29: , November [9] H. Takeda, S. Farsu, and P. Mlanfar. Kernel regresson for mage processng and reconstructon. IEEE Transactons on Image Processng, 16(2): , February [10] H. Takeda, S. Farsu, and P. Mlanfar. Deblurrng usng regularzed locally-adaptve kernel regresson. IEEE Transactons on Image Processng, 17: , Aprl [11] H. Takeda, P. Mlanfar, M. Protter, and M. Elad. Superresoluton wthout explct subpxel moton estmaton. Accepted for publcaton n IEEE Transactons on Image Processng, [12] M. Tatsuoka. Multvarate Analyss. Macmllan, [13] P. Turaga, R. Chellappa, V. Subrahmanan, and O. Udrea. Machne recognton of human actvtes: A survey. IEEE Transactons on Crcuts and Systems for Vdeo Technology, 18: , November

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

3D vector computer graphics

3D vector computer graphics 3D vector computer graphcs Paolo Varagnolo: freelance engneer Padova Aprl 2016 Prvate Practce ----------------------------------- 1. Introducton Vector 3D model representaton n computer graphcs requres

More information

Reducing Frame Rate for Object Tracking

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

More information

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

Learning Ensemble of Local PDM-based Regressions. Yen Le Computational Biomedicine Lab Advisor: Prof. Ioannis A. Kakadiaris

Learning Ensemble of Local PDM-based Regressions. Yen Le Computational Biomedicine Lab Advisor: Prof. Ioannis A. Kakadiaris Learnng Ensemble of Local PDM-based Regressons Yen Le Computatonal Bomedcne Lab Advsor: Prof. Ioanns A. Kakadars 1 Problem statement Fttng a statstcal shape model (PDM) for mage segmentaton Callosum segmentaton

More information

Efficient Segmentation and Classification of Remote Sensing Image Using Local Self Similarity

Efficient Segmentation and Classification of Remote Sensing Image Using Local Self Similarity ISSN(Onlne): 2320-9801 ISSN (Prnt): 2320-9798 Internatonal Journal of Innovatve Research n Computer and Communcaton Engneerng (An ISO 3297: 2007 Certfed Organzaton) Vol.2, Specal Issue 1, March 2014 Proceedngs

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

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

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

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

Classifier Swarms for Human Detection in Infrared Imagery

Classifier Swarms for Human Detection in Infrared Imagery Classfer Swarms for Human Detecton n Infrared Imagery Yur Owechko, Swarup Medasan, and Narayan Srnvasa HRL Laboratores, LLC 3011 Malbu Canyon Road, Malbu, CA 90265 {owechko, smedasan, nsrnvasa}@hrl.com

More information

Human Action Recognition Using Dynamic Time Warping Algorithm and Reproducing Kernel Hilbert Space for Matrix Manifold

Human Action Recognition Using Dynamic Time Warping Algorithm and Reproducing Kernel Hilbert Space for Matrix Manifold IJCTA, 10(07), 2017, pp 79-85 Internatonal Scence Press Closed Loop Control of Soft Swtched Forward Converter Usng Intellgent Controller 79 Human Acton Recognton Usng Dynamc Tme Warpng Algorthm and Reproducng

More information

A B-Snake Model Using Statistical and Geometric Information - Applications to Medical Images

A B-Snake Model Using Statistical and Geometric Information - Applications to Medical Images A B-Snake Model Usng Statstcal and Geometrc Informaton - Applcatons to Medcal Images Yue Wang, Eam Khwang Teoh and Dnggang Shen 2 School of Electrcal and Electronc Engneerng, Nanyang Technologcal Unversty

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

An efficient method to build panoramic image mosaics

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

More information

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

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

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

An Entropy-Based Approach to Integrated Information Needs Assessment

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

More information

Detection of hand grasping an object from complex background based on machine learning co-occurrence of local image feature

Detection of hand grasping an object from complex background based on machine learning co-occurrence of local image feature Detecton of hand graspng an object from complex background based on machne learnng co-occurrence of local mage feature Shnya Moroka, Yasuhro Hramoto, Nobutaka Shmada, Tadash Matsuo, Yoshak Shra Rtsumekan

More information

y and the total sum of

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

More information

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

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

Lecture #15 Lecture Notes

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

More information

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

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

More information

Optimizing Document Scoring for Query Retrieval

Optimizing Document Scoring for Query Retrieval Optmzng Document Scorng for Query Retreval Brent Ellwen baellwe@cs.stanford.edu Abstract The goal of ths project was to automate the process of tunng a document query engne. Specfcally, I used machne learnng

More information

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

Appearance-based Statistical Methods for Face Recognition

Appearance-based Statistical Methods for Face Recognition 47th Internatonal Symposum ELMAR-2005, 08-10 June 2005, Zadar, Croata Appearance-based Statstcal Methods for Face Recognton Kresmr Delac 1, Mslav Grgc 2, Panos Latss 3 1 Croatan elecom, Savsa 32, Zagreb,

More information

Online Detection and Classification of Moving Objects Using Progressively Improving Detectors

Online Detection and Classification of Moving Objects Using Progressively Improving Detectors Onlne Detecton and Classfcaton of Movng Objects Usng Progressvely Improvng Detectors Omar Javed Saad Al Mubarak Shah Computer Vson Lab School of Computer Scence Unversty of Central Florda Orlando, FL 32816

More information

Synthesizer 1.0. User s Guide. A Varying Coefficient Meta. nalytic Tool. Z. Krizan Employing Microsoft Excel 2007

Synthesizer 1.0. User s Guide. A Varying Coefficient Meta. nalytic Tool. Z. Krizan Employing Microsoft Excel 2007 Syntheszer 1.0 A Varyng Coeffcent Meta Meta-Analytc nalytc Tool Employng Mcrosoft Excel 007.38.17.5 User s Gude Z. Krzan 009 Table of Contents 1. Introducton and Acknowledgments 3. Operatonal Functons

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

Detecting Irregularities in Images and in Video

Detecting Irregularities in Images and in Video Detectng Irregulartes n Images and n Vdeo Oren Boman Mchal Iran Dept. of Computer Scence and Appled Math The Wezmann Insttute of Scence 76100 Rehovot, Israel Abstract We address the problem of detectng

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

Object-Based Techniques for Image Retrieval

Object-Based Techniques for Image Retrieval 54 Zhang, Gao, & Luo Chapter VII Object-Based Technques for Image Retreval Y. J. Zhang, Tsnghua Unversty, Chna Y. Y. Gao, Tsnghua Unversty, Chna Y. Luo, Tsnghua Unversty, Chna ABSTRACT To overcome the

More information

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

Adaptive Silhouette Extraction and Human Tracking in Dynamic. Environments 1

Adaptive Silhouette Extraction and Human Tracking in Dynamic. Environments 1 Adaptve Slhouette Extracton and Human Trackng n Dynamc Envronments 1 X Chen, Zhha He, Derek Anderson, James Keller, and Marjore Skubc Department of Electrcal and Computer Engneerng Unversty of Mssour,

More information

An Efficient Background Updating Scheme for Real-time Traffic Monitoring

An Efficient Background Updating Scheme for Real-time Traffic Monitoring 2004 IEEE Intellgent Transportaton Systems Conference Washngton, D.C., USA, October 3-6, 2004 WeA1.3 An Effcent Background Updatng Scheme for Real-tme Traffc Montorng Suchendra M. Bhandarkar and Xngzh

More information

Related-Mode Attacks on CTR Encryption Mode

Related-Mode Attacks on CTR Encryption Mode Internatonal Journal of Network Securty, Vol.4, No.3, PP.282 287, May 2007 282 Related-Mode Attacks on CTR Encrypton Mode Dayn Wang, Dongda Ln, and Wenlng Wu (Correspondng author: Dayn Wang) Key Laboratory

More information

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

A Deflected Grid-based Algorithm for Clustering Analysis

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

More information

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 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

1. Introduction. Abstract

1. Introduction. Abstract Image Retreval Usng a Herarchy of Clusters Danela Stan & Ishwar K. Seth Intellgent Informaton Engneerng Laboratory, Department of Computer Scence & Engneerng, Oaland Unversty, Rochester, Mchgan 48309-4478

More information

Classification of Face Images Based on Gender using Dimensionality Reduction Techniques and SVM

Classification of Face Images Based on Gender using Dimensionality Reduction Techniques and SVM Classfcaton of Face Images Based on Gender usng Dmensonalty Reducton Technques and SVM Fahm Mannan 260 266 294 School of Computer Scence McGll Unversty Abstract Ths report presents gender classfcaton based

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

Machine Learning 9. week

Machine Learning 9. week Machne Learnng 9. week Mappng Concept Radal Bass Functons (RBF) RBF Networks 1 Mappng It s probably the best scenaro for the classfcaton of two dataset s to separate them lnearly. As you see n the below

More information

Multi-view 3D Position Estimation of Sports Players

Multi-view 3D Position Estimation of Sports Players Mult-vew 3D Poston Estmaton of Sports Players Robbe Vos and Wlle Brnk Appled Mathematcs Department of Mathematcal Scences Unversty of Stellenbosch, South Afrca Emal: vosrobbe@gmal.com Abstract The problem

More information

Adaptive Transfer Learning

Adaptive Transfer Learning Adaptve Transfer Learnng Bn Cao, Snno Jaln Pan, Yu Zhang, Dt-Yan Yeung, Qang Yang Hong Kong Unversty of Scence and Technology Clear Water Bay, Kowloon, Hong Kong {caobn,snnopan,zhangyu,dyyeung,qyang}@cse.ust.hk

More information

Dynamic Camera Assignment and Handoff

Dynamic Camera Assignment and Handoff 12 Dynamc Camera Assgnment and Handoff Br Bhanu and Ymng L 12.1 Introducton...338 12.2 Techncal Approach...339 12.2.1 Motvaton and Problem Formulaton...339 12.2.2 Game Theoretc Framework...339 12.2.2.1

More information

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

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

More information

Unsupervised Learning and Clustering

Unsupervised Learning and Clustering Unsupervsed Learnng and Clusterng Supervsed vs. Unsupervsed Learnng Up to now we consdered supervsed learnng scenaro, where we are gven 1. samples 1,, n 2. class labels for all samples 1,, n Ths s also

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

Real-time Motion Capture System Using One Video Camera Based on Color and Edge Distribution

Real-time Motion Capture System Using One Video Camera Based on Color and Edge Distribution Real-tme Moton Capture System Usng One Vdeo Camera Based on Color and Edge Dstrbuton YOSHIAKI AKAZAWA, YOSHIHIRO OKADA, AND KOICHI NIIJIMA Graduate School of Informaton Scence and Electrcal Engneerng,

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

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

An Attention Based Method For Motion Detection And Estimation

An Attention Based Method For Motion Detection And Estimation An Attenton Based Method For Moton Detecton And Estmaton Shje Zhang and Fred Stentford Department of Electronc and Electrcal Engneerng Unversty College London, Adastral Par Campus, Ross Buldng Martlesham

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

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

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

Feature Selection for Target Detection in SAR Images

Feature Selection for Target Detection in SAR Images Feature Selecton for Detecton n SAR Images Br Bhanu, Yngqang Ln and Shqn Wang Center for Research n Intellgent Systems Unversty of Calforna, Rversde, CA 95, USA Abstract A genetc algorthm (GA) approach

More information

Fuzzy Logic Based RS Image Classification Using Maximum Likelihood and Mahalanobis Distance Classifiers

Fuzzy Logic Based RS Image Classification Using Maximum Likelihood and Mahalanobis Distance Classifiers Research Artcle Internatonal Journal of Current Engneerng and Technology ISSN 77-46 3 INPRESSCO. All Rghts Reserved. Avalable at http://npressco.com/category/jcet Fuzzy Logc Based RS Image Usng Maxmum

More information

A Probabilistic Approach to Detect Urban Regions from Remotely Sensed Images Based on Combination of Local Features

A Probabilistic Approach to Detect Urban Regions from Remotely Sensed Images Based on Combination of Local Features A Probablstc Approach to Detect Urban Regons from Remotely Sensed Images Based on Combnaton of Local Features Berl Sırmaçek German Aerospace Center (DLR) Remote Sensng Technology Insttute Weßlng, 82234,

More information

Human Face Recognition Using Generalized. Kernel Fisher Discriminant

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

More information

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

Compiler Design. Spring Register Allocation. Sample Exercises and Solutions. Prof. Pedro C. Diniz Compler Desgn Sprng 2014 Regster Allocaton Sample Exercses and Solutons Prof. Pedro C. Dnz USC / Informaton Scences Insttute 4676 Admralty Way, Sute 1001 Marna del Rey, Calforna 90292 pedro@s.edu Regster

More information