Parametric ego-motion estimation for vehicle surround analysis using an omnidirectional camera

Size: px
Start display at page:

Download "Parametric ego-motion estimation for vehicle surround analysis using an omnidirectional camera"

Transcription

1 Mchine Vision nd Applictions (25) 16: Digitl Oject Identifier (DOI) 1.17/s z Mchine Vision nd Applictions Prmetric ego-motion estimtion for vehicle surround nlysis using n omnidirectionl cmer Trk Gndhi, Mohn Trivedi Computer Vision nd Rootics Reserch Lortory, University of Cliforni t Sn Diego, L Joll, CA, USA (e-mil: {tgndhi,trivedi}@ucsd.edu) Received: 1 August 23 / Accepted: 8 July 24 Pulished online: 3 Ferury 25 c Springer-Verlg 25 Astrct. Omnidirectionl cmers tht give 36 pnormic view of the surroundings hve recently een used in mny pplictions such s rootics, nvigtion, nd surveillnce. This pper descries the ppliction of prmetric egomotion estimtion for vehicle detection to perform surround nlysis using n utomoile-mounted cmer. For this purpose, the prmetric plnr motion model is integrted with the trnsformtions to compenste distortion in omnidirectionl imges. The frmework is used to detect ojects with independent motion or height ove the rod. Cmer clirtion s well s the pproximte vehicle speed otined from CAN us re integrted with the motion informtion from sptil nd temporl grdients using Byesin pproch. The pproch is tested for vrious configurtions of n utomoilemounted omni cmer s well s rectiliner cmer. Successful detection nd trcking of moving vehicles nd genertion of surround mp re demonstrted for ppliction to intelligent driver support. Keywords: Motion estimtion Pnormic vision Intelligent vehicles Driver support systems Collision voidnce 1 Introduction nd motivtion Omnidirectionl cmers tht give pnormic view of surroundings hve ecome very populr in mchine vision. Benosmn nd Kng [5] give comprehensive description of pnormic imging systems nd their pplictions. There is considerle interest in motion nlysis from moving pltforms using omni cmers, since pnormic views help in deling with miguities ssocited with ego-motion of the pltforms [15]. In prticulr, vehicle surround nlysis system tht monitors the presence of other vehicles in ll directions is importnt for online s well s offline pplictions. Online systems re useful for intelligent driver support. On the other hnd, offline processing of video sequences is useful for study of ehviorl ptterns of the driver in order to develop etter tools for driver ssistnce. For such systems, complete surround nlysis system tht monitors the lnes nd vehicles round the driver Fig. 1,. Imges from omni cmer mounted on n utomoile. This cmer hs verticl FOV of 5 ove the horizon nd covers only nery surroundings ut gives lrger vehicle imges. This cmer hs verticl resolution of 15 ove the horizon nd covers frther surroundings, ut with smller vehicle imges is very importnt. An omni cmer mounted on the utomoile could provide complete pnormic view of the surroundings nd would e very pproprite for performing such tsk. The min contriution of this pper is to perform moving oject detection from omni imge sequences using direct prmetric motion estimtion method nd pply it to video sequences otined from n utomoile-mounted cmer to detect nd trck neighoring vehicles. Figure 1 shows the imges from omni cmers in different configurtions used for this work. It is seen tht the cmer covers 36 field of view (FOV) round its center. However, the imge it produces is distorted, with stright lines trnsformed into curves. Directly unwrping the imge to perspective imge would introduce severe lur in perspective imge, cusing prolems for susequent steps in motion nlysis. Insted, the omni cmer trnsformtions re comined with the motion trnsformtions to compenste the ego-motion in the omni domin itself. 1.1 Relted work in motion nlysis Motion estimtion from moving omni cmers hs recently een topic of gret interest. Rectiliner cmers usully hve smller FOV, due to which the focus of expnsion often lies outside the imge, cusing motion estimtion to e sensitive to the cmer orienttion. Also, the motion field produced y

2 86 T. Gndhi, M. Trivedi: Prmetric ego-motion compenstion trnsltion in the horizontl direction is similr to tht due to rottion out the verticl xis. As noted y Gluckmn nd Nyr [15], omni cmers void oth these prolems due to their wide FOV. They project the imge motion on sphericl surfce using Jcoins of trnsformtions to determine egomotion of moving pltform in terms of trnsltion nd rottion of the cmer. Vsslo et l. [3] propose generl Jcoin function tht cn descrie wide vriety of omni cmers. Shkerni et l. [26] use the concept of ck-projection flow, where the imge motion is projected to virtul curved surfce in plce of sphericl surfce to simplify the Jcoins. Using this concept, they hve dpted ego-motion lgorithms for rectiliner cmers for use with omni sensors. Svood et l. [28] use feture correspondences to estimte the essentil mtrix etween two frmes using the 8-point lgorithm. They lso note tht the motion estimtion is more stle with omni cmers compred to rectiliner cmers. Most of these methods first compute motion of imge pixels nd then use the motion vectors to estimte the motion prmeters. However, due to the perture prolem [17], the full motion informtion is relile only ner cornerlike points. The edge points only hve motion informtion norml to the edge. Direct methods cn optimlly use the motion informtion from edges s well s corners to get prmeters of motion. Direct methods hve often een used with rectiliner cmers for plnr motion estimtion, ostcle detection, nd motion segmenttion [7,21,2]. To distinguish ojects of interest from extrneous fetures, the ground is usully pproximted y plnr surfce whose ego-motion is modeled using projective trnsform [25,23] or its linerized version [3]. Using this model, the ego-motion of the ground is compensted in order to seprte the ojects with independent motion or height. 1.2 Relted work on intelligent vehicles In recent yers, considerle reserch hs een conducted on developing intelligent vehicles hving driver support systems tht enhnce sfety. Computer vision techniques hve een pplied to detecting lnes, other vehicles, nd pedestrins to wrn the driver of dngers such s lne deprture nd possile collision with other ojects. Stereo cmers re especilly useful for detecting ostcles in front tht re fr from the driver. Bertozzi nd Broggi [6] use stereo cmers for lne nd ostcle detection. They model the rod s plnr surfce nd use inverse perspective trnsform to register the rod plne etween two imges. The ostcles ove the rod would hve residul disprity nd re esily detected. In the cse of curved rods, [24] crete V-disprity imge sed on clustering similr disprities on ech imge row. A line or curve in this imge corresponds to stright or curved rod, respectively, nd the vehicles on the rod form other distinctive ptterns. Omni cmers with their pnormic FOV show gret potentil in intelligent vehicle pplictions. In [18], n omni cmer mounted inside the cr otined view of the driver s well s of the surroundings. The driver s pose ws estimted using hidden Mrkov models nd ws used to generte the driver s view of the surroundings using the sme cmer. In [2], feture-sed methods detecting specific chrcteristics of vehicles, such s wheels, were used to detect nd trck vehicles. Single-cmer motion nlysis hs een used for seprting ego-motion of the ckground to detect vehicles nd other ostcles on the rod. Roust rel-time motion compenstion for the rod plne for this purpose is descried in [23]. In [1], system for video-sed driver ssistnce involving lne nd ostcle detection using rectiliner cmer is descried. Direct prmetric motion estimtion discussed in the previous section is especilly useful for vehicle pplictions, since most of the fetures on the rod re line-sed nd very few corner fetures re ville. The direct estimtion pproch ws generlized for motion compenstion using omni cmers in [13,18], where prmeters of plnr homogrphy were estimted. A modifiction of tht pproch is used here s in [14] to estimte the vehicle ego-motion in terms of liner nd ngulr velocities. These re used to compenste the ego-motion for the rod plne nd detect vehicles hving residul motion to generte complete surround view showing the position nd trcks of the vehicles. 2 Ego-motion estimtion nd compenstion system The system lock digrm is shown in Fig. 2. The inputs to the system re sequence of imges from n omni cmer mounted on utomoile, the vehicle speed from the CAN us which gives informtion out the vehicle stte, nd the nominl clirtion of the cmer with respect to the rod plne. The stte of the vehicle contining vehicle velocity nd clirtion re used to compute the wrping prmeters to compenste the imge motion etween two frmes for points on the rod plne. The wrping trnsform is composition of the omni cmer trnsform nd the plnr motion model. It trnsforms the omni imge coordintes to perspective coordintes, pplies the plnr motion prmeters to compenste the rod motion, nd converts them ck to the omni view. Two consecutive frmes from the imge sequence re tken, nd the wrping prmeters re used to trnsform one imge to nother, to compenste the motion of the rod s much s possile. The ojects with independent motion nd height would hve lrge residul motion, mking it possile to seprte them from rod fetures. Fig. 2. System for ego-motion compenstion from moving pltform. The inputs to the system re the video sequence from n omni cmer nd vehicle speed informtion extrcted from the CAN us of the cr tht provides numer of vriles of the cr s dynmics. The output is surround mp with detected vehicles nd their trcks

3 T. Gndhi, M. Trivedi: Prmetric ego-motion compenstion 87 However, the fetures on the rod my lso hve some residul motion due to errors in the vehicle speed nd clirtion prmeters. To correct for these errors, sptil nd temporl grdients of the motion-compensted imges re otined. Byesin estimtion similr to [23] is pplied with grdients s oservtions to updte the prior knowledge of the stte of the vehicle using Klmn filter mesurement updte equtions. To minimize the effect of outliers, only the grdients stisfying constrint on the residul re used in the estimtion process. The updted vehicle stte is used to recompute the wrping prmeters, nd the residul grdients re recomputed. The process is repeted in corse-to-fine itertive mnner. The grdients computed using the finlly updted stte of the vehicle re used to seprte the vehicle fetures from the rod fetures. The vehicle fetures re comined using constrints on vehicle length nd seprtion to otin los corresponding to vehicles tht re trcked over the numer of frmes. The surround mp is generted y unwrping the omni imge to give pln view nd superimposing the vehicle los nd trcks over the resulting imge. The following sections descrie the processing steps in detil. 3 Motion trnsformtions for omni cmer Let c denote nominl cmer coordinte system, sed on the known cmer clirtion, with the Z-xis long the cmer xis nd the X Y plne eing the imging plne. Due to cmer virtions nd drift, the ctul cmer system t ny given time is ssumed to hve smll rottion with respect to this system due to virtions nd drift. Use of the nominl system llows us to tret smll rottions s ngulr displcement vectors. The ego-motion of the cmer is then descried using stte vector x contining the cmer liner velocity V, ngulr velocity W, nd ngulr displcement A etween nominl cmer system c nd ctul system, ll expressed in nominl cmer system c. 3.1 Plnr motion model To detect ostcles in the pth of moving cmer, the rod is modeled s plnr surfce. Let P nd P denote the perspective projections of point on the rod plne in coordinte systems corresponding to two positions nd of the moving cmer. These re relted y: λ P = λ RP + D = λ [RP + D/λ ], (1) where R nd D denote the rottion nd trnsltion etween the cmer positions, nd λ,λ depend on the distnce of the ctul 3D point. Let the eqution of the rod plne t the cmer position e: K T (λ P )=1, (2) where K is vector norml to the rod plne in the coordinte system of cmer position. Sustituting the vlue of λ from Eq. 2 in Eq. (1), it is seen tht P nd P re relted y projective trnsform [11]: λ P = λ [ R + DK T ] P = λ HP, (3) where H = R+DK T is known s the projective trnsform or homogrphy. This reltion hs een widely used to estimte plnr motion for rectiliner cmers. If the ngulr displcements with respect to the nominl cmer clirtion re smll, the mtrices cn e expressed s: R I W t, D [I W t A ] V t, K [I A ] K, (4) where W nd A represent the skew symmetric mtrices constructed from vectors W nd A, nd K represents the plne norml in the nominl cmer coordinte system. 3.2 Omni cmer trnsform To pply the ego-motion estimtion method to omni cmers, one needs the mpping from the cmer coordinte system to the pixel domin nd vice vers. Given this trnsformtion nd the plnr motion model, one cn generte trnsformtion tht compenstes the motion of the plnr surfce in the omni pixel domin. In prticulr, the omni cmer used in this work consists of hyperolic mirror nd cmer plced on its xis, with the center of projection of the cmer on one of the focl points of the hyperol. It elongs to clss of cmers known s centrl pnormic ctdioptric cmers [5]. These cmers hve single viewpoint tht permits the imge to e suitly trnsformed to otin perspective views. The geometry of hyperolic omni cmer is shown in Fig. 3. According to the mirror geometry, light ry from the oject towrd the viewpoint t the first focus O is reflected so tht it psses through the second focus, where conventionl rectiliner cmer is plced. The eqution of the hyperoloid is given y: (Z c) 2 2 X2 + Y 2 2 =1, (5) where c = Let P =(X, Y, Z) T denote the homogenous coordintes of the perspective trnsform of ny 3D point λp on ry OP, where λ is the scle fctor depending on the distnce of the 3D point from the origin. It cn e shown [1,19,26] tht the reflection in the mirror gives the point p =( x, y) T on the imge plne of the cmer using: ( ) x q 1 p = = y q 2 Z + q 3 P ( ) X, (6) Y where q 1 = c 2 2,q 2 = c 2 + 2,q 3 =2c, (7) nd P = X 2 + Y 2 + Z 2. Note tht the expression for imge coordintes p is independent of the scle fctor λ. The pixel coordintes w = (u, v) T re then otined y using the clirtion mtrix K of the conventionl cmer composed of the focl lengths f u,f v, opticl center coordintes (u,v ) T, nd cmer skew s, or: u x f u s u x v = K y = f v v y. (8)

4 88 T. Gndhi, M. Trivedi: Prmetric ego-motion compenstion deteriortes if the entire imge is trnsformed t one time. Hence, s noted y Dniilidis [8], it is desirle to perform motion estimtion directly in the omni domin, ut use the ove trnsformtions to mp the loctions to the perspective domin s required. Since the internl prmeters of the omni cmer re to e mesured only once, specilized setup ws used to otin the clirtion. The omni cmer ws set on tripod nd leveled to hve verticl cmer xis. A numer of fetures with known coordintes were tken on the ground nd verticl pole to cover the FOV of the omni cmer. The FOV covered y the omni cmer mps into the ellipse s seen in Fig. 3. The cmer center nd spect rtio were computed from the ellipse prmeters. Using these prmeters, the imge coordintes (u, v) cn e normlized to give (u,v ) corresponding to origin s center nd unit spect rtio. Assuming rdil symmetry round the imge center, we hve: d = u 2 + v 2 = X2 + Y 2 c 1 Z + c 2 P, (11) d [y pixels] c z/r Fig.3.Geometry of hyperolic omni cmer. The rys towrd the first focus of the mirror re reflected towrd the second focus nd imged y norml cmer. FOV of omni cmer with numer of points with known coordintes. c Curve fitting for internl prmeter estimtion where c 1 = q 2 /(q 1 f v ) nd c 2 = q 3 /(q 1 f v ). Using the known world nd imge coordintes of these points, the liner equtions in c 1 nd c 2 re formed nd solved using lest squres: dzc 1 + d P c 2 = X 2 + Y 2. (12) Figure 3 shows the plot of d ginst Z/ P of the smple points nd the curve fitted using estimted prmeters. It is seen tht the curve models the omni mpping quite fithfully. Nonliner lest squres cn then e used for improving the ccurcy. Though the method is designed for centrl pnormic cmers, if the distnce to the oserved scene is lrge compred to the mirror size, the method cn lso e pplied to noncentrl pnormic cmers provided the mpping from oject ry directions to pixel coordintes is known. In fct, it ws oserved tht for hyperolic mirror, the FOV is concentrted within close distnce round the cmer, which mde it somewht difficult to detect ojects frther from the cmer where resolution ws poor. Noncentrl cmers my e prticulrly useful since they give more flexiility in djusting the cmer resolution in different prts of the imge, s descried in [16]. This trnsform cn e used to wrp n omni imge to pln perspective view. To convert perspective view ck to omni view within scle fctor, the inverse trnsformtion cn e used: x u y 1 = K 1 v 1, (9) X q 1 x Y q 1 y. (1) Z q 2 q 3 x2 + y 2 +1 It should e noted tht the trnsformtion of omni to perspective view involves very different mgnifictions in different prts of the imge. For this reson, the qulity of the imge 4 Ego-motion estimtion To estimte the ego-motion prmeters, the prmetric imge motion is sustituted into the opticl flow constrint [17]: g u u + g v v + g t =, (13) where g u,g v re sptil grdients nd g t is the temporl grdient. Since the imge motion ( u, v) t ech point i cn e represented s function of the incrementl stte vector x, the opticl flow constrint Eq. 13 for imge points 1... N cn e expressed s: z = c( x)+v C x + v, (14)

5 T. Gndhi, M. Trivedi: Prmetric ego-motion compenstion 89 Tle 1. Chin of functions nd Jcoins leding from stte vector x to opticl flow constrint c. Rows 4 nd 5 correspond to the omni cmer trnsform tht converts the cmer coordintes to pixel coordintes V x = W H = R + DK T H = R + D.K T + D( K) T A R = W t h 1 h 2 h 3 R I W t D =(I W t A ) t V ( W t + A )V t H = h 4 h 5 h 6 D [I A ] V t K = A K h 7 h 8 h 9 K [I A ] K V/ V i = e i, W / W i = A / A i =(e i) ( ) T X h 1 h 2 h 3 X X Y Z P h = h 1... h 9 Y h 4 h 5 h 6 Y = h X Y Z Z h 7 h 8 h 9 Z X Y Z ( ) ( ) ( ) ( ) ( ) T x x q P = XY Z = = 1 X p q y y 2 Z+q 3 P P Y = 1 q 3xX q 1 P q 3xY q 3xZ (q 2 Z+q 3 P ) P q 3yX q 3yY q 1 P q 3yZ ( ) T u f u s u x ( ) w p = xy v = f v v y = f u s p f v ( T ( ) w = uv) ( ) u u ( ) c c = g u g v = g t + η w v v = g u g v where (g u u + g v v) 1 c( x) =. (g u u + g v v) N, (g t ) 1 z =. (g t ) N, (15) v is the vector of mesurement noise in the time grdients, nd C = c/ x is the Jcoin mtrix computed using the chin rule s in [13]. The function c(x) is nonliner. Row i of the N 9 Jcoin mtrix is given y the chin rule: C i = ( ) ( ) ci c w p P h =, (16) x w p P h x i where P = (X,Y,Z ) T, p = (x,y ) T nd w = (u,v ) T re the coordintes of the point in the cmer, imge, nd pixel coordinte systems for cmer position, nd h is the vector of elements of H. The individul Jcoins re computed similrly to [13]. The reltionship etween these vriles, nd their Jcoins, re shown in Tle 1. Since the points hving very low texture do not contriute much to the estimtion of motion prmeters, only those imge points hving grdient mgnitude ove threshold vlue re selected for performing estimtion. Alterntively, nonmximl suppression is performed on the imge grdients nd the imge points with locl mxim re used. This wy, insted of computing Jcoins using multiple imge trnsforms over the entire imge, the Jcoins re computed only t the selected points tht hve significnt informtion for estimting prmeters. The estimtes of the stte x nd its covrince P re itertively updted using the mesurement updte equtions of the iterted extended Klmn filter [4]: P [ C T R 1 C + P 1 ] 1 (17) ˆx ˆx + ˆx = ˆx + P [ C T R 1 z P 1 (ˆx x ) ]. (18) However, the opticl flow constrint eqution is stisfied only for smll imge displcements up to 1 or 2 pixels. To estimte lrger motions, corse-to-fine pyrmidl frmework [22,27] is used. In this frmework, multiresolution Gussin pyrmid is constructed for djcent imges in the sequence. The motion prmeters re first computed t the corsest level, nd the imge points t the next finer level re wrped using the computed motion prmeters. The residul motion is computed t the finer level, nd the process is repeted until the finest level. Note tht since the resolution of the mirror is not constnt, formtion of Gussin pyrmid could hve errors in the neighorhood. However, since the pyrmid is used itertively in corse-to-fine mnner, the errors t lower resolution re expected to e corrected t higher resolution. The prmeters cn lso e updted from frme to frme using time updte equtions of the Klmn filter: ˆx Bˆx, P BPB T + Q, (19) where B nd Q re determined from system dynmics. 4.1 Outlier removl The ove estimte is optiml only when ll points relly elong to the plnr surfce nd when the underlying noise distriutions re Gussin. However, the estimtion is highly sensitive to the presence of outliers, i.e., points not stisfying the rod motion model. These fetures should e seprted using roust method. For this purpose, first the region of

6 9 T. Gndhi, M. Trivedi: Prmetric ego-motion compenstion interest of rod is determined using clirtion informtion, nd the processing is done only in tht region to void extrneous fetures. To detect outliers, n pproch similr to the dt snooping pproch discussed in [9] hs een dpted for Byesin estimtion. In this pproch, the error residul of ech feture is compred with the expected residul covrince t every itertion, nd the fetures re reclssified s inliers or outliers. If point z i is not included in the estimtion of ˆx i.e., is currently clssified s n outlier then the covrince of its residul is: V [ z i C i ˆx] =V [ z i ]+CV [ˆx] C T = R + C i PC T i. (2) However, if z i is included in the estimtion of ˆx i.e., is currently clssified s n inlier then it cn e shown tht the covrince of its residul is given y: V [ z i C i ˆx] =R C i PC T i < R. (21) Hence, to clssify in the next itertion, the residul is compred with its covrince ccording to whether it is currently n outlier or n inlier. If the Mhlnois norm is greter thn given threshold, the point is clssified s outlier, otherwise s n inlier. Alterntively, roust-m estimtion [12] could e used to reduce the effect of outliers y itertively reweighting the contriution of smples ccording to their error residuls. 4.2 Algorithm for motion prmeter estimtion The lgorithm for itertive estimtion of motion prmeters is descried elow: Form Gussin pyrmid from imges A nd B from consecutive frmes. Set the initil prmeters nd the covrince mtrix to their priors s: ˆx = x nd P = P. Proceeding from the corsest to the finest level, perform multiple itertions of the following steps: 1. Wrp imge B using current estimte ˆx of motion prmeters to form imge W (B; ˆx). 2. Otin sptil nd temporl grdients etween imge A nd the wrped imge W (B; ˆx). 3. Use opticl flow constrint with prmetric motion model on inlier points to pply incrementl correction in motion prmeters nd their covrinces ccording to Eqs. 17 nd Compre the residuls of ll points with their expected covrinces in Eqs. 2 nd 21 to reclssify them s inliers nd outliers. 5 Vehicle detection nd trcking After motion compenstion, the fetures on the rod plne would e ligned etween the two frmes, wheres those due to ostcles would e misligned. Imge difference etween the frmes would therefore enhnce the ostcles nd suppress the rod fetures. To reduce the dependence on locl texture, the normlized frme difference [29] is used. This is given t ech pixel y: g t g 2 u + gv 2 k + gu 2 + gv 2, (22) where g u,g v re sptil grdients, g t is the temporl grdient fter motion compenstion, nd denotes Gussin weighted verging performed over K K neighorhood of ech pixel. In fct, the normlized difference is smoothed version of the norml opticl flow nd hence depends on the mount of motion ner the point. Due to the untextured interior of vehicle, los re usully detected t the sides of the vehicle. To get the full vehicle, it is ssumed tht if two los re within threshold distnce (5. m) in the direction of the cr s motion, they constitute vehicle. To detect this sitution, the originl imge is unwrped using the flt plne trnsform, nd morphologicl closing is performed on the trnsformed imge using 1 N verticl msk. After the los corresponding to moving ojects re identified, nery los re clustered nd trcked over frmes using Klmn filtering [4]. The points on the lo tht re nerest to the cmer center usully correspond to the rod plne nd re mrked s n ostcle mp. The vehicle position on the rod is computed y projecting the trck loction on the ostcle mp. Since the ostcle mp is ssumed to e on the rod plne, the loction of the vehicle cn e otined y inverse perspective trnsform. 6 Experimentl studies The ego-motion compenstion pproch ws pplied for detecting vehicles from n omni cmer mounted on n utomoile tested used for intelligent vehicle reserch. The tested is instrumented with numer of cmers nd computers to cpture synchronized video of the surroundings. In ddition, the CAN us of the vehicle gives informtion on vehicle speed, pedl nd rke positions, rdr, etc. The vehicle ws driven on freewy s well s on city rods. The mximum vehicle speed for the test ws 65 miles per hour (29 m/s). The ctul vehicle speed, otined from the CAN us, ws used for initil motion estimte. The first test run ws conducted with n omni cmer hving verticl FOV of only 5 ove the horizon. For this reson, only the vehicles ner the cr were oserved, ut the resolution ws s lrge s possile. To get s little of the cr s possile, the cmer ws rised 18 in. (45 cm) ove the cr using specilly designed fixture. Figure 4 shows n imge from the omni cmer on the cr eing driven on the freewy. The estimted prmetric motion is shown using red rrows. Note tht the motion is estimted only in the designted region of interest, which excludes the cr ody. Figure 4 shows the clssifiction of points into inliers (gry), outliers (white), nd unused (lck) points. The estimtion is done using only the inlier points. An imge with the normlized frme difference etween the motion-compensted frmes is shown in Fig. 4c, which enhnces the regions corresponding to independently moving vehicles. Figure 4d shows the detection nd trcking of vehicles mrked with trck ID nd the coordintes in the rod plne. The omni imge ws trnsformed to otin

7 T. Gndhi, M. Trivedi: Prmetric ego-motion compenstion 91 c d e Fig.4. Imge from sequence using n omni cmer mounted on moving cr with estimted prmetric motion of rod plne. Clssifiction of points into inliers (gry), outliers (white), nd unused (lck). c Normlized difference etween motion-compensted imges. d Detection nd trcking of moving vehicles mrked with trck ID nd the coordintes in rod plne. e Surround view generted y trnsforming the omni imge longitudinl position: Z [m] longitudinl position: Z [m] time [s] time [s] Fig. 5. Plot of the longitudinl position of vehicle trcks on two sides of the cr ginst time. The trcks re color coded s red, yellow, nd green in order of incresing lterl distnce from the cmer Fig. 6. Surround nlysis in different situtions with top-mounted cmer. City rod. Freewy

8 92 T. Gndhi, M. Trivedi: Prmetric ego-motion compenstion c d e Fig.7.Imge from sequence using n omni cmer with wider FOV mounted on moving cr. The rnge of the cmer is incresed ut the resolution is decresed. Clssifiction of points into inliers (gry), outliers (white), nd unused (lck). c Normlized difference etween motion-compensted imges. d Detection nd trcking of moving vehicles mrked with trck ID nd the coordintes in rod plne. e Surround view generted y dewrping omni imge Fig. 8. Smples showing surround vehicle detection with wider FOV omni cmer 1 1 longitudinl position: Z [m] longitudinl position: Z [m] time [s] time [s] Fig. 9. Plot of longitudinl position of vehicle trcks ginst time on two sides of cr ginst time. The trcks re color coded s red, yellow, nd green in order of incresing lterl distnce from the cmer

9 T. Gndhi, M. Trivedi: Prmetric ego-motion compenstion 93 c d e longitudinl position: Z [m] time [s] Fig. 1. Imge from sequence using side cmer mounted on moving cr with estimted prmetric motion of the rod plne. Clssifiction of points into inliers (gry), outliers (white), nd unused (lck). c Normlized difference etween motion-compensted imges. d Detection nd trcking of moving vehicles mrked with trck ID nd the coordintes in the rod plne. e Surround view generted y pplying inverse perspective trnsform. f Plot of longitudinl position of vehicle trcks ginst time. The trcks re color coded s red, yellow, nd green in order of incresing lterl distnce from the cmer f the pln view of the cr surround, s shown in Fig. 4e. The longitudinl position of the cr with reference to cmer ws recorded for ech trck. Figure 5 shows the plots of trck positions ginst time seprtely for vehicles on two sides of the cmer. The test run lso contined sections driven on city rods tht hd lne mrks nd other fetures tht were more prominent compred to the freewy. Figure 6 shows exmples of moving vehicle detection on city rods s well s in freewy conditions. The second test run ws conducted using n omni cmer with FOV 15 ove the horizon. It ws noted tht the cmer cn see vehicles t greter distnce thn the previous cmer. The trdeoff ws lower resolution due to which the vehicles hd smller imge size, mking them slightly more difficult to detect. Figure 7 shows the result of surround vehicle detection t lrger longitudinl distnce from the cmer. Figure 8 shows more exmples of vehicle detection. Figure 9 shows the plots of trck positions ginst time seprtely for vehicles on two sides of the cmer. It should e noted tht the simplified version of the surround nlysis lgorithm developed in this pper cn lso e used with commonly ville rectiliner cmers. We conducted severl experiments where video strems were cquired using rectiliner cmer mounted on cr window to get rer side view on the driver s side. Figure 1 shows the results of the detection lgorithm. Figure 1e shows the top view generted y pplying the inverse perspective trnsformtion using the known clirtion. Insted of the full surround view, which cn e cquired using n omni cmer, only prtil view on one side of the vehicle ws otined. 7 Summry This pper descried n pproch to oject detection using ego-motion compenstion from utomoile-mounted omni cmers using direct prmetric motion estimtion. The rod ws modeled s plnr surfce, nd the equtions for plnr motion trnsform were comined with the omni cmer trnsform. An opticl flow constrint ws used to optimlly comine the prior knowledge of ego-motion prmeters with the informtion in the imge grdients. Corse-to-fine motion estimtion ws used, nd the motion etween the frmes ws compensted t ech itertion. Experimentl results demonstrted vehicle detection in two different configurtions of omni cmers tht otin ner nd fr views of the surround.

10 94 T. Gndhi, M. Trivedi: Prmetric ego-motion compenstion Acknowledgements. This reserch ws supported y UC Discovery Progrm digitl medi grnt in collortion with the Nissn Reserch Center. We lso thnk our collegues from the CVRR Lortory for their contriutions nd support. We lso thnk Dr. Erwin Boer for his suggestions on visulizing the results. Finlly we thnk the reviewers for their insightful comments which helped us to improve the qulity of the pper. Author, plese note. The list of References in your dt is not exctly the sme s tht in your printout. Plese check crefully. Thnk you. References 1. Achler O, Trivedi MM (22) Rel-time trffic flow nlysis using omnidirectionl video network nd fltplne trnsformtion. In: Workshop on intelligent trnsporttion systems. Chicgo 2. Achler O, Trivedi MM (24) Vehicle wheel detector using 2d filter nks. In: Proc. IEEE symposium on intelligent vehicles, pp Adiv G (1985) Determining three-dimensionl motion nd structure from opticl flow generted y severl moving ojects. IEEE Trns Pttern Anl Mch Intell 7(4): Br-Shlom Y, Li XR, Kirurjn T (21) Estimtion with pplictions to trcking nd nvigtion. Wiley, New York 5. Benosmn R, Kng SB (21) Pnormic vision: sensors, theory, nd pplictions. Springer, Berlin Heidelerg New York 6. Bertozzi M, Broggi A (1998) Gold: prllel rel-time stereo vision system for generic ostcle nd lne detection. IEEE Trns Imge Proc 7(1): Blck MJ, Anndn P (1996) The roust estimtion of multiple motions: Prmetric nd piecewise-smooth flow fields. Comput Vis Imge Understnd 63(1): Dniilidis K, Mkdi A, Bulow T (22) Imge processing in ctdioptric plnes: sptiotemporl derivtives nd opticl flow computtion. In: IEEE workshop on omnidirectionl vision, pp Dnuser G, Stricker M (1998) Prmetric model fitting: from inlier chrcteriztion to outlier detection. IEEE Trns Pttern Anl Mch Intell 2(2): Enkelmnn W (21) Video-sed driver ssistnce: from sic functions to pplictions. Int J Comput Vis 45(3): Fugers O (1993) Three-dimensionl computer vision: geometric viewpoint. MIT Press, Cmridge, MA 12. Forsyth D, Ponce J (23) Computer vision: modern pproch. Prentice-Hll, Upper Sddle River, NJ 13. Gndhi T, Trivedi MM (23) Motion nlysis of omnidirectionl video strems for moile sentry. In: 1st ACM interntionl workshop on video surveillnce, pp 49 58, Berkeley, CA 14. Gndhi T, Trivedi MM (24) Motion sed vehicle surround nlysis using omni-directionl cmer. In: Proc. IEEE symposium intelligent vehicles, pp Gluckmn J, Nyr S (1998) Ego-motion nd omnidirectionl cmers. In: Proc. interntionl conference on computer vision, pp Hicks RA, Bjcsy R (1999) Reflective surfces s computtionl sensors. In: Proc. 2nd workshop on perception for moile gents, pp Horn B, Schunck B (1981) Determining opticl flow. In: DARPA81, pp Hung, K, Trivedi MM, Gndhi T (23) Driver s view nd vehicle surround estimtion using omnidirectionl video strem. In: IEEE symposium intelligent vehicles, Columus, OH, pp Hung KC, Trivedi MM (23) Video rrys for rel-time trcking of persons, hed nd fce in n intelligent room. Mch Vis Appl 14(2): Irni M, Anndn P (1998) A unified pproch to moving oject detection in 2D nd 3D scenes. IEEE Trns Pttern Anl Mch Intell 2(6): Irni M, Rousso B, Peleg S (1994) Computing occluding nd trnsprent motions. Int J Comput Vis 12: Jähne B, Hußecker H, Geißler P (1999) Hndook of Computer Vision nd Applictions, vol 2, chp 14, pp Acdemic Press, Sn Diego, CA 23. Kruger W (1999) Roust rel time ground plne motion compenstion from moving vehicle. Mch Vis Appl 11: Lyrde R, Auert D, Trel J-P (22) Rel time ostcle detection in stereovision on non flt rod geometry through v- disprity representtion. In: IEEE symposium intelligent vehicles, 2: Lourkis MIA, Orphnoudkis SC (1998) Visul detection of ostcles ssuming loclly plnr ground. In: Asin conference on computer vision, 2: Shkerni O, Vidl R, Sstry S (23) Omnidirectionl egomotion estimtion from ck-projection flow. In: IEEE workshop on omnidirectionl vision 27. Simoncelli EP (1993) Corse-to-fine estimtion of visul motion. In: Proc. 8th workshop on imge nd multidimensionl signl processing, Cnnes, Frnce, pp Svood T, Pjdl T, Hlváč V (1998) Motion estimtion using centrl pnormic cmers. In: IEEE interntionl conference on intelligent vehicles, pp Trucco E, Verri A (1998) Computer vision nd pplictions: guide for students nd prctitioners. Prentice Hll, Upper Sddle River, NJ 3. Vssllo RF, Sntos-Victor J, Schneeeli HJ (22) A generl pproch for egomotion estimtion with omnidirectionl imges. In: IEEE workshop on omnidirectionl vision, pp Trk Gndhi received his chelor of technology degree in computer science nd engineering t the Indin Institute of Technology, Bomy. He erned his M.S. nd Ph.D. from the Pennsylvni Stte University in computer science nd engineering, specilizing in computer vision. He worked t Adept Technology, Inc. on designing lgorithms for rootic systems. Currently, he is postdoctorl scholr t the Computer Vision nd Rootics Reserch lortory t the University of Cliforni t Sn Diego. His interests include computer vision, motion nlysis, imge processing, rootics, trget detection, nd pttern recognition. He is working on projects involving intelligent driver ssistnce, motion-sed event detection, trffic flow nlysis, nd structurl helth monitoring of ridges.

11 T. Gndhi, M. Trivedi: Prmetric ego-motion compenstion 95 Mohn Mnuhi Trivedi is professor of electricl nd computer engineering t the University of Cliforni t Sn Diego. Trivedi hs rod rnge of reserch interests in the intelligent systems, computer vision, intelligent ( smrt ) environments, intelligent vehicles nd trnsporttion systems, nd humn-mchine interfce res. He estlished the Computer Vision nd Rootics Reserch Lortory t UCSD. Currently, Trivedi nd his tem re pursuing systems-oriented reserch in distriuted video rrys nd ctive vision, omnidirectionl vision, humn ody modeling nd movement nlysis, fce nd ffect nlysis, nd intelligent vehicles nd interctive pulic spces. He serves on the executive committee of the Cliforni Institute for Telecommuniction nd Informtion Technologies [Cl-(IT) 2 ] s the leder of the Intelligent Trnsporttion nd Telemtics Lyer t UCSD. He lso serves s chrter memer of the executive committee of the University of Cliforni systemwide Digitl Medi Innovtion Progrm (DiMI). He serves regulrly s consultnt to industry nd government gencies in the USA nd rod. Trivedi ws editor-in-chief of the Mchine Vision nd Applictions Journl during He hs served on the editoril ords of journls nd progrm committees of severl mjor conferences. He served s chirmn of the Rootics Technicl Committee of the IEEE Computer Society. He ws elected to serve on the dministrtive committee (BoG) of the IEEE Systems, Mn nd Cyernetics Society. Trivedi hs received the Distinguished Alumnus wrd from the Uth Stte University nd Pioneer (Technicl Activities) nd Meritorious Service wrds from the IEEE Computer Society. He is fellow of the Interntionl Society for Opticl Engineering (SPIE).

Motion analysis for event detection and tracking with a mobile omnidirectional camera

Motion analysis for event detection and tracking with a mobile omnidirectional camera Multimedi Systems 10: 131 143 (2004) Digitl Oject Identifier (DOI) 10.1007/s00530-004-0146-3 Multimedi Systems Motion nlysis for event detection nd trcking with moile omnidirectionl cmer Trk Gndhi, Mohn

More information

Before We Begin. Introduction to Spatial Domain Filtering. Introduction to Digital Image Processing. Overview (1): Administrative Details (1):

Before We Begin. Introduction to Spatial Domain Filtering. Introduction to Digital Image Processing. Overview (1): Administrative Details (1): Overview (): Before We Begin Administrtive detils Review some questions to consider Winter 2006 Imge Enhncement in the Sptil Domin: Bsics of Sptil Filtering, Smoothing Sptil Filters, Order Sttistics Filters

More information

Fig.1. Let a source of monochromatic light be incident on a slit of finite width a, as shown in Fig. 1.

Fig.1. Let a source of monochromatic light be incident on a slit of finite width a, as shown in Fig. 1. Answer on Question #5692, Physics, Optics Stte slient fetures of single slit Frunhofer diffrction pttern. The slit is verticl nd illuminted by point source. Also, obtin n expression for intensity distribution

More information

Spectral Analysis of MCDF Operations in Image Processing

Spectral Analysis of MCDF Operations in Image Processing Spectrl Anlysis of MCDF Opertions in Imge Processing ZHIQIANG MA 1,2 WANWU GUO 3 1 School of Computer Science, Northest Norml University Chngchun, Jilin, Chin 2 Deprtment of Computer Science, JilinUniversity

More information

CHAPTER III IMAGE DEWARPING (CALIBRATION) PROCEDURE

CHAPTER III IMAGE DEWARPING (CALIBRATION) PROCEDURE CHAPTER III IMAGE DEWARPING (CALIBRATION) PROCEDURE 3.1 Scheimpflug Configurtion nd Perspective Distortion Scheimpflug criterion were found out to be the best lyout configurtion for Stereoscopic PIV, becuse

More information

P(r)dr = probability of generating a random number in the interval dr near r. For this probability idea to make sense we must have

P(r)dr = probability of generating a random number in the interval dr near r. For this probability idea to make sense we must have Rndom Numers nd Monte Crlo Methods Rndom Numer Methods The integrtion methods discussed so fr ll re sed upon mking polynomil pproximtions to the integrnd. Another clss of numericl methods relies upon using

More information

II. THE ALGORITHM. A. Depth Map Processing

II. THE ALGORITHM. A. Depth Map Processing Lerning Plnr Geometric Scene Context Using Stereo Vision Pul G. Bumstrck, Bryn D. Brudevold, nd Pul D. Reynolds {pbumstrck,brynb,pulr2}@stnford.edu CS229 Finl Project Report December 15, 2006 Abstrct A

More information

On the Detection of Step Edges in Algorithms Based on Gradient Vector Analysis

On the Detection of Step Edges in Algorithms Based on Gradient Vector Analysis On the Detection of Step Edges in Algorithms Bsed on Grdient Vector Anlysis A. Lrr6, E. Montseny Computer Engineering Dept. Universitt Rovir i Virgili Crreter de Slou sin 43006 Trrgon, Spin Emil: lrre@etse.urv.es

More information

GENERATING ORTHOIMAGES FOR CLOSE-RANGE OBJECTS BY AUTOMATICALLY DETECTING BREAKLINES

GENERATING ORTHOIMAGES FOR CLOSE-RANGE OBJECTS BY AUTOMATICALLY DETECTING BREAKLINES GENEATING OTHOIMAGES FO CLOSE-ANGE OBJECTS BY AUTOMATICALLY DETECTING BEAKLINES Efstrtios Stylinidis 1, Lzros Sechidis 1, Petros Ptis 1, Spiros Sptls 2 Aristotle University of Thessloniki 1 Deprtment of

More information

Complete Coverage Path Planning of Mobile Robot Based on Dynamic Programming Algorithm Peng Zhou, Zhong-min Wang, Zhen-nan Li, Yang Li

Complete Coverage Path Planning of Mobile Robot Based on Dynamic Programming Algorithm Peng Zhou, Zhong-min Wang, Zhen-nan Li, Yang Li 2nd Interntionl Conference on Electronic & Mechnicl Engineering nd Informtion Technology (EMEIT-212) Complete Coverge Pth Plnning of Mobile Robot Bsed on Dynmic Progrmming Algorithm Peng Zhou, Zhong-min

More information

A multiview 3D modeling system based on stereo vision techniques

A multiview 3D modeling system based on stereo vision techniques Mchine Vision nd Applictions (2005) 16: 148 156 Digitl Oject Identifier (DOI) 10.1007/s00138-004-0165-2 Mchine Vision nd Applictions A multiview 3D modeling system sed on stereo vision techniques Soon-Yong

More information

Lecture 10 Evolutionary Computation: Evolution strategies and genetic programming

Lecture 10 Evolutionary Computation: Evolution strategies and genetic programming Lecture 10 Evolutionry Computtion: Evolution strtegies nd genetic progrmming Evolution strtegies Genetic progrmming Summry Negnevitsky, Person Eduction, 2011 1 Evolution Strtegies Another pproch to simulting

More information

USING HOUGH TRANSFORM IN LINE EXTRACTION

USING HOUGH TRANSFORM IN LINE EXTRACTION Stylinidis, Efstrtios USING HOUGH TRANSFORM IN LINE EXTRACTION Efstrtios STYLIANIDIS, Petros PATIAS The Aristotle University of Thessloniki, Deprtment of Cdstre Photogrmmetry nd Crtogrphy Univ. Box 473,

More information

A dual of the rectangle-segmentation problem for binary matrices

A dual of the rectangle-segmentation problem for binary matrices A dul of the rectngle-segmenttion prolem for inry mtrices Thoms Klinowski Astrct We consider the prolem to decompose inry mtrix into smll numer of inry mtrices whose -entries form rectngle. We show tht

More information

TOWARDS GRADIENT BASED AERODYNAMIC OPTIMIZATION OF WIND TURBINE BLADES USING OVERSET GRIDS

TOWARDS GRADIENT BASED AERODYNAMIC OPTIMIZATION OF WIND TURBINE BLADES USING OVERSET GRIDS TOWARDS GRADIENT BASED AERODYNAMIC OPTIMIZATION OF WIND TURBINE BLADES USING OVERSET GRIDS S. H. Jongsm E. T. A. vn de Weide H. W. M. Hoeijmkers Overset symposium 10-18-2012 Deprtment of mechnicl engineering

More information

2 Computing all Intersections of a Set of Segments Line Segment Intersection

2 Computing all Intersections of a Set of Segments Line Segment Intersection 15-451/651: Design & Anlysis of Algorithms Novemer 14, 2016 Lecture #21 Sweep-Line nd Segment Intersection lst chnged: Novemer 8, 2017 1 Preliminries The sweep-line prdigm is very powerful lgorithmic design

More information

ZZ - Advanced Math Review 2017

ZZ - Advanced Math Review 2017 ZZ - Advnced Mth Review Mtrix Multipliction Given! nd! find the sum of the elements of the product BA First, rewrite the mtrices in the correct order to multiply The product is BA hs order x since B is

More information

6.3 Volumes. Just as area is always positive, so is volume and our attitudes towards finding it.

6.3 Volumes. Just as area is always positive, so is volume and our attitudes towards finding it. 6.3 Volumes Just s re is lwys positive, so is volume nd our ttitudes towrds finding it. Let s review how to find the volume of regulr geometric prism, tht is, 3-dimensionl oject with two regulr fces seprted

More information

1 Drawing 3D Objects in Adobe Illustrator

1 Drawing 3D Objects in Adobe Illustrator Drwing 3D Objects in Adobe Illustrtor 1 1 Drwing 3D Objects in Adobe Illustrtor This Tutoril will show you how to drw simple objects with three-dimensionl ppernce. At first we will drw rrows indicting

More information

Machine vision system for surface inspection on brushed industrial parts.

Machine vision system for surface inspection on brushed industrial parts. Mchine vision system for surfce inspection on rushed industril prts. Nicols Bonnot, Rlph Seulin, Frederic Merienne Lortoire Le2i, CNRS UMR 5158, University of Burgundy, Le Creusot, Frnce. ABSTRACT This

More information

SUPPLEMENTARY INFORMATION

SUPPLEMENTARY INFORMATION Supplementry Figure y (m) x (m) prllel perpendiculr Distnce (m) Bird Stndrd devition for distnce (m) c 6 prllel perpendiculr 4 doi:.8/nture99 SUPPLEMENTARY FIGURE Confirmtion tht movement within the flock

More information

Geometric transformations

Geometric transformations Geometric trnsformtions Computer Grphics Some slides re bsed on Shy Shlom slides from TAU mn n n m m T A,,,,,, 2 1 2 22 12 1 21 11 Rows become columns nd columns become rows nm n n m m A,,,,,, 1 1 2 22

More information

L. Yaroslavsky. Fundamentals of Digital Image Processing. Course

L. Yaroslavsky. Fundamentals of Digital Image Processing. Course L. Yroslvsky. Fundmentls of Digitl Imge Processing. Course 0555.330 Lecture. Imge enhncement.. Imge enhncement s n imge processing tsk. Clssifiction of imge enhncement methods Imge enhncement is processing

More information

Section 10.4 Hyperbolas

Section 10.4 Hyperbolas 66 Section 10.4 Hyperbols Objective : Definition of hyperbol & hyperbols centered t (0, 0). The third type of conic we will study is the hyperbol. It is defined in the sme mnner tht we defined the prbol

More information

LETKF compared to 4DVAR for assimilation of surface pressure observations in IFS

LETKF compared to 4DVAR for assimilation of surface pressure observations in IFS LETKF compred to 4DVAR for ssimiltion of surfce pressure oservtions in IFS Pu Escrià, Mssimo Bonvit, Mts Hmrud, Lrs Isksen nd Pul Poli Interntionl Conference on Ensemle Methods in Geophysicl Sciences Toulouse,

More information

Math 35 Review Sheet, Spring 2014

Math 35 Review Sheet, Spring 2014 Mth 35 Review heet, pring 2014 For the finl exm, do ny 12 of the 15 questions in 3 hours. They re worth 8 points ech, mking 96, with 4 more points for netness! Put ll your work nd nswers in the provided

More information

International Conference on Mechanics, Materials and Structural Engineering (ICMMSE 2016)

International Conference on Mechanics, Materials and Structural Engineering (ICMMSE 2016) \ Interntionl Conference on Mechnics, Mterils nd tructurl Engineering (ICMME 2016) Reserch on the Method to Clibrte tructure Prmeters of Line tructured Light Vision ensor Mingng Niu1,, Kngnin Zho1, b,

More information

HW Stereotactic Targeting

HW Stereotactic Targeting HW Stereotctic Trgeting We re bout to perform stereotctic rdiosurgery with the Gmm Knife under CT guidnce. We instrument the ptient with bse ring nd for CT scnning we ttch fiducil cge (FC). Above: bse

More information

Digital approximation to extended depth of field in no telecentric imaging systems

Digital approximation to extended depth of field in no telecentric imaging systems Journl of Physics: Conference Series Digitl pproximtion to extended depth of field in no telecentric imging systems To cite this rticle: J E Meneses nd C R Contrers 0 J Phys: Conf Ser 74 0040 View the

More information

Tilt-Sensing with Kionix MEMS Accelerometers

Tilt-Sensing with Kionix MEMS Accelerometers Tilt-Sensing with Kionix MEMS Accelerometers Introduction Tilt/Inclintion sensing is common ppliction for low-g ccelerometers. This ppliction note describes how to use Kionix MEMS low-g ccelerometers to

More information

Stained Glass Design. Teaching Goals:

Stained Glass Design. Teaching Goals: Stined Glss Design Time required 45-90 minutes Teching Gols: 1. Students pply grphic methods to design vrious shpes on the plne.. Students pply geometric trnsformtions of grphs of functions in order to

More information

Topic 3: 2D Transformations 9/10/2016. Today s Topics. Transformations. Lets start out simple. Points as Homogeneous 2D Point Coords

Topic 3: 2D Transformations 9/10/2016. Today s Topics. Transformations. Lets start out simple. Points as Homogeneous 2D Point Coords Tody s Topics 3. Trnsformtions in 2D 4. Coordinte-free geometry 5. (curves & surfces) Topic 3: 2D Trnsformtions 6. Trnsformtions in 3D Simple Trnsformtions Homogeneous coordintes Homogeneous 2D trnsformtions

More information

Modeling and Simulation of Short Range 3D Triangulation-Based Laser Scanning System

Modeling and Simulation of Short Range 3D Triangulation-Based Laser Scanning System Modeling nd Simultion of Short Rnge 3D Tringultion-Bsed Lser Scnning System Theodor Borngiu Anmri Dogr Alexndru Dumitrche April 14, 2008 Abstrct In this pper, simultion environment for short rnge 3D lser

More information

Analysis of Computed Diffraction Pattern Diagram for Measuring Yarn Twist Angle

Analysis of Computed Diffraction Pattern Diagram for Measuring Yarn Twist Angle Textiles nd Light ndustril Science nd Technology (TLST) Volume 3, 2014 DO: 10.14355/tlist.2014.0301.01 http://www.tlist-journl.org Anlysis of Computed Diffrction Pttern Digrm for Mesuring Yrn Twist Angle

More information

A Tautology Checker loosely related to Stålmarck s Algorithm by Martin Richards

A Tautology Checker loosely related to Stålmarck s Algorithm by Martin Richards A Tutology Checker loosely relted to Stålmrck s Algorithm y Mrtin Richrds mr@cl.cm.c.uk http://www.cl.cm.c.uk/users/mr/ University Computer Lortory New Museum Site Pemroke Street Cmridge, CB2 3QG Mrtin

More information

Slides for Data Mining by I. H. Witten and E. Frank

Slides for Data Mining by I. H. Witten and E. Frank Slides for Dt Mining y I. H. Witten nd E. Frnk Simplicity first Simple lgorithms often work very well! There re mny kinds of simple structure, eg: One ttriute does ll the work All ttriutes contriute eqully

More information

Statistical classification of spatial relationships among mathematical symbols

Statistical classification of spatial relationships among mathematical symbols 2009 10th Interntionl Conference on Document Anlysis nd Recognition Sttisticl clssifiction of sptil reltionships mong mthemticl symbols Wl Aly, Seiichi Uchid Deprtment of Intelligent Systems, Kyushu University

More information

Tree Structured Symmetrical Systems of Linear Equations and their Graphical Solution

Tree Structured Symmetrical Systems of Linear Equations and their Graphical Solution Proceedings of the World Congress on Engineering nd Computer Science 4 Vol I WCECS 4, -4 October, 4, Sn Frncisco, USA Tree Structured Symmetricl Systems of Liner Equtions nd their Grphicl Solution Jime

More information

MTH 146 Conics Supplement

MTH 146 Conics Supplement 105- Review of Conics MTH 146 Conics Supplement In this section we review conics If ou ne more detils thn re present in the notes, r through section 105 of the ook Definition: A prol is the set of points

More information

OUTPUT DELIVERY SYSTEM

OUTPUT DELIVERY SYSTEM Differences in ODS formtting for HTML with Proc Print nd Proc Report Lur L. M. Thornton, USDA-ARS, Animl Improvement Progrms Lortory, Beltsville, MD ABSTRACT While Proc Print is terrific tool for dt checking

More information

Computer Vision and Image Understanding

Computer Vision and Image Understanding Computer Vision nd Imge Understnding 116 (2012) 25 37 Contents lists ville t SciVerse ScienceDirect Computer Vision nd Imge Understnding journl homepge: www.elsevier.com/locte/cviu A systemtic pproch for

More information

Control Camera and Light Source Positions using Image Gradient Information

Control Camera and Light Source Positions using Image Gradient Information IEEE Int. Conf. on Rootics nd Automtion, ICRA'7 Rom, Itli, April 27 Control Cmer nd Light Source Positions using Imge Grdient Informtion Eric Mrchnd Astrct In this pper, we propose n originl pproch to

More information

COLOUR IMAGE MATCHING FOR DTM GENERATION AND HOUSE EXTRACTION

COLOUR IMAGE MATCHING FOR DTM GENERATION AND HOUSE EXTRACTION Hee Ju Prk OLOUR IMAGE MATHING FOR DTM GENERATION AND HOUSE EXTRATION Hee Ju PARK, Petr ZINMMERMANN * Swiss Federl Institute of Technology, Zuric Switzerlnd Institute for Geodesy nd Photogrmmetry heeju@ns.shingu-c.c.kr

More information

Class-XI Mathematics Conic Sections Chapter-11 Chapter Notes Key Concepts

Class-XI Mathematics Conic Sections Chapter-11 Chapter Notes Key Concepts Clss-XI Mthemtics Conic Sections Chpter-11 Chpter Notes Key Concepts 1. Let be fixed verticl line nd m be nother line intersecting it t fixed point V nd inclined to it t nd ngle On rotting the line m round

More information

10.5 Graphing Quadratic Functions

10.5 Graphing Quadratic Functions 0.5 Grphing Qudrtic Functions Now tht we cn solve qudrtic equtions, we wnt to lern how to grph the function ssocited with the qudrtic eqution. We cll this the qudrtic function. Grphs of Qudrtic Functions

More information

Video-rate Image Segmentation by means of Region Splitting and Merging

Video-rate Image Segmentation by means of Region Splitting and Merging Video-rte Imge Segmenttion y mens of Region Splitting nd Merging Knur Anej, Florence Lguzet, Lionel Lcssgne, Alin Merigot Institute for Fundmentl Electronics, University of Pris South Orsy, Frnce knur.nej@gmil.com,

More information

About the Finite Element Analysis for Beam-Hinged Frame. Duan Jin1,a, Li Yun-gui1

About the Finite Element Analysis for Beam-Hinged Frame. Duan Jin1,a, Li Yun-gui1 Advnces in Engineering Reserch (AER), volume 143 6th Interntionl Conference on Energy nd Environmentl Protection (ICEEP 2017) About the Finite Element Anlysis for Bem-Hinged Frme Dun Jin1,, Li Yun-gui1

More information

File Manager Quick Reference Guide. June Prepared for the Mayo Clinic Enterprise Kahua Deployment

File Manager Quick Reference Guide. June Prepared for the Mayo Clinic Enterprise Kahua Deployment File Mnger Quick Reference Guide June 2018 Prepred for the Myo Clinic Enterprise Khu Deployment NVIGTION IN FILE MNGER To nvigte in File Mnger, users will mke use of the left pne to nvigte nd further pnes

More information

AVolumePreservingMapfromCubetoOctahedron

AVolumePreservingMapfromCubetoOctahedron Globl Journl of Science Frontier Reserch: F Mthemtics nd Decision Sciences Volume 18 Issue 1 Version 1.0 er 018 Type: Double Blind Peer Reviewed Interntionl Reserch Journl Publisher: Globl Journls Online

More information

Image Segmentation Using Wavelet and watershed transform

Image Segmentation Using Wavelet and watershed transform Imge Segmenttion Using Wvelet nd wtershed trnsform Atollh Hdddi, Mhmod R. Shei, Mohmmd J. Vldn Zoej, Ali mohmmdzdeh Fculty of Geodesy nd Geomtics Engineering, K. N. Toosi University of Technology, Vli_Asr

More information

Date: 9.1. Conics: Parabolas

Date: 9.1. Conics: Parabolas Dte: 9. Conics: Prols Preclculus H. Notes: Unit 9 Conics Conic Sections: curves tht re formed y the intersection of plne nd doulenpped cone Syllus Ojectives:. The student will grph reltions or functions,

More information

CALIBRATION OF A STRUCTURED LIGHT-BASED STEREO CATADIOPTRIC SENSOR *

CALIBRATION OF A STRUCTURED LIGHT-BASED STEREO CATADIOPTRIC SENSOR * AIBRATION OF A STRUTURED IGHT-BASED STEREO ATADIOPTRI SENSOR * Rdu Orghidn, Joquim Slvi, El Mustph Mouddi omputer Vision nd Rootics Group, Institute of Informtics nd Applictions, Universitt de Giron, Avd.

More information

Subband coding of image sequences using multiple vector quantizers. Emanuel Martins, Vitor Silva and Luís de Sá

Subband coding of image sequences using multiple vector quantizers. Emanuel Martins, Vitor Silva and Luís de Sá Sund coding of imge sequences using multiple vector quntizers Emnuel Mrtins, Vitor Silv nd Luís de Sá Instituto de Telecomunicções, Deprtmento de Engenhri Electrotécnic Pólo II d Universidde de Coimr,

More information

In the last lecture, we discussed how valid tokens may be specified by regular expressions.

In the last lecture, we discussed how valid tokens may be specified by regular expressions. LECTURE 5 Scnning SYNTAX ANALYSIS We know from our previous lectures tht the process of verifying the syntx of the progrm is performed in two stges: Scnning: Identifying nd verifying tokens in progrm.

More information

MA1008. Calculus and Linear Algebra for Engineers. Course Notes for Section B. Stephen Wills. Department of Mathematics. University College Cork

MA1008. Calculus and Linear Algebra for Engineers. Course Notes for Section B. Stephen Wills. Department of Mathematics. University College Cork MA1008 Clculus nd Liner Algebr for Engineers Course Notes for Section B Stephen Wills Deprtment of Mthemtics University College Cork s.wills@ucc.ie http://euclid.ucc.ie/pges/stff/wills/teching/m1008/ma1008.html

More information

RGB D TERRAIN PERCEPTION AND DENSE MAPPING FOR LEGGED ROBOTS

RGB D TERRAIN PERCEPTION AND DENSE MAPPING FOR LEGGED ROBOTS Int. J. Appl. Mth. Comput. Sci., 206, Vol. 26, No., 8 97 DOI: 0.55/mcs-206-0006 RGB D TERRAIN PERCEPTION AND DENSE MAPPING FOR LEGGED ROBOTS DOMINIK BELTER,, PRZEMYSŁAW ŁABECKI, PÉTER FANKHAUSER, ROLAND

More information

Computing offsets of freeform curves using quadratic trigonometric splines

Computing offsets of freeform curves using quadratic trigonometric splines Computing offsets of freeform curves using qudrtic trigonometric splines JIULONG GU, JAE-DEUK YUN, YOONG-HO JUNG*, TAE-GYEONG KIM,JEONG-WOON LEE, BONG-JUN KIM School of Mechnicl Engineering Pusn Ntionl

More information

From Dependencies to Evaluation Strategies

From Dependencies to Evaluation Strategies From Dependencies to Evlution Strtegies Possile strtegies: 1 let the user define the evlution order 2 utomtic strtegy sed on the dependencies: use locl dependencies to determine which ttriutes to compute

More information

Today. CS 188: Artificial Intelligence Fall Recap: Search. Example: Pancake Problem. Example: Pancake Problem. General Tree Search.

Today. CS 188: Artificial Intelligence Fall Recap: Search. Example: Pancake Problem. Example: Pancake Problem. General Tree Search. CS 88: Artificil Intelligence Fll 00 Lecture : A* Serch 9//00 A* Serch rph Serch Tody Heuristic Design Dn Klein UC Berkeley Multiple slides from Sturt Russell or Andrew Moore Recp: Serch Exmple: Pncke

More information

Algorithm Design (5) Text Search

Algorithm Design (5) Text Search Algorithm Design (5) Text Serch Tkshi Chikym School of Engineering The University of Tokyo Text Serch Find sustring tht mtches the given key string in text dt of lrge mount Key string: chr x[m] Text Dt:

More information

Rigid Body Transformations

Rigid Body Transformations igid od Kinemtics igid od Trnsformtions Vij Kumr igid od Kinemtics emrk out Nottion Vectors,,, u, v, p, q, Potentil for Confusion! Mtrices,, C, g, h, igid od Kinemtics The vector nd its skew smmetric mtri

More information

Presentation Martin Randers

Presentation Martin Randers Presenttion Mrtin Rnders Outline Introduction Algorithms Implementtion nd experiments Memory consumption Summry Introduction Introduction Evolution of species cn e modelled in trees Trees consist of nodes

More information

1. SEQUENCES INVOLVING EXPONENTIAL GROWTH (GEOMETRIC SEQUENCES)

1. SEQUENCES INVOLVING EXPONENTIAL GROWTH (GEOMETRIC SEQUENCES) Numbers nd Opertions, Algebr, nd Functions 45. SEQUENCES INVOLVING EXPONENTIAL GROWTH (GEOMETRIC SEQUENCES) In sequence of terms involving eponentil growth, which the testing service lso clls geometric

More information

Grade 7/8 Math Circles Geometric Arithmetic October 31, 2012

Grade 7/8 Math Circles Geometric Arithmetic October 31, 2012 Fculty of Mthemtics Wterloo, Ontrio N2L 3G1 Grde 7/8 Mth Circles Geometric Arithmetic Octoer 31, 2012 Centre for Eduction in Mthemtics nd Computing Ancient Greece hs given irth to some of the most importnt

More information

CS380: Computer Graphics Modeling Transformations. Sung-Eui Yoon ( 윤성의 ) Course URL:

CS380: Computer Graphics Modeling Transformations. Sung-Eui Yoon ( 윤성의 ) Course URL: CS38: Computer Grphics Modeling Trnsformtions Sung-Eui Yoon ( 윤성의 ) Course URL: http://sgl.kist.c.kr/~sungeui/cg/ Clss Ojectives (Ch. 3.5) Know the clssic dt processing steps, rendering pipeline, for rendering

More information

Slicer Method Comparison Using Open-source 3D Printer

Slicer Method Comparison Using Open-source 3D Printer IOP Conference Series: Erth nd Environmentl Science PAPER OPEN ACCESS Slicer Method Comprison Using Open-source 3D Printer To cite this rticle: M K A Mohd Ariffin et l 2018 IOP Conf. Ser.: Erth Environ.

More information

Multiresolution Interactive Modeling with Efficient Visualization

Multiresolution Interactive Modeling with Efficient Visualization Multiresolution Interctive Modeling with Efficient Visuliztion Jen-Dniel Deschênes Ptrick Héert Philippe Lmert Jen-Nicols Ouellet Drgn Tuic Computer Vision nd Systems Lortory heert@gel.ulvl.c, Lvl University,

More information

Department of Electrical and Computer Systems Engineering

Department of Electrical and Computer Systems Engineering Deprtment of Electricl nd Computer Systems Engineering Technicl Report MECSE--24 Aircrft Pose Estimtion from Homogrphy E Beets, S Boukir, nd D Suter MECSE--24: "Aircrft Pose Estimtion from Homogrphy",

More information

Solving Problems by Searching. CS 486/686: Introduction to Artificial Intelligence Winter 2016

Solving Problems by Searching. CS 486/686: Introduction to Artificial Intelligence Winter 2016 Solving Prolems y Serching CS 486/686: Introduction to Artificil Intelligence Winter 2016 1 Introduction Serch ws one of the first topics studied in AI - Newell nd Simon (1961) Generl Prolem Solver Centrl

More information

4452 Mathematical Modeling Lecture 4: Lagrange Multipliers

4452 Mathematical Modeling Lecture 4: Lagrange Multipliers Mth Modeling Lecture 4: Lgrnge Multipliers Pge 4452 Mthemticl Modeling Lecture 4: Lgrnge Multipliers Lgrnge multipliers re high powered mthemticl technique to find the mximum nd minimum of multidimensionl

More information

F. R. K. Chung y. University ofpennsylvania. Philadelphia, Pennsylvania R. L. Graham. AT&T Labs - Research. March 2,1997.

F. R. K. Chung y. University ofpennsylvania. Philadelphia, Pennsylvania R. L. Graham. AT&T Labs - Research. March 2,1997. Forced convex n-gons in the plne F. R. K. Chung y University ofpennsylvni Phildelphi, Pennsylvni 19104 R. L. Grhm AT&T Ls - Reserch Murry Hill, New Jersey 07974 Mrch 2,1997 Astrct In seminl pper from 1935,

More information

Planning with Reachable Distances: Fast Enforcement of Closure Constraints

Planning with Reachable Distances: Fast Enforcement of Closure Constraints Plnning with Rechle Distnces: Fst Enforcement of Closure Constrints Xinyu Tng, Shwn Thoms, nd Nncy M. Amto Astrct Motion plnning for closed-chin systems is prticulrly difficult due to dditionl closure

More information

INVESTIGATION OF RESAMPLING EFFECTS ON IRS-1D PAN DATA

INVESTIGATION OF RESAMPLING EFFECTS ON IRS-1D PAN DATA INVESTIGATION OF RESAMPLING EFFECTS ON IRS-D PAN DATA Smpth Kumr P.,*, Onkr Dikshit nd YVS Murthy Geo-Informtics Division, Ntionl Remote Sensing Agency, Hyderd, Indi.-(smpth_k, murthy_yvs)@nrs.gov.in Deprtment

More information

Unit 5 Vocabulary. A function is a special relationship where each input has a single output.

Unit 5 Vocabulary. A function is a special relationship where each input has a single output. MODULE 3 Terms Definition Picture/Exmple/Nottion 1 Function Nottion Function nottion is n efficient nd effective wy to write functions of ll types. This nottion llows you to identify the input vlue with

More information

A Fast Imaging Algorithm for Near Field SAR

A Fast Imaging Algorithm for Near Field SAR Journl of Computing nd Electronic Informtion Mngement ISSN: 2413-1660 A Fst Imging Algorithm for Ner Field SAR Guoping Chen, Lin Zhng, * College of Optoelectronic Engineering, Chongqing University of Posts

More information

A New Learning Algorithm for the MAXQ Hierarchical Reinforcement Learning Method

A New Learning Algorithm for the MAXQ Hierarchical Reinforcement Learning Method A New Lerning Algorithm for the MAXQ Hierrchicl Reinforcement Lerning Method Frzneh Mirzzdeh 1, Bbk Behsz 2, nd Hmid Beigy 1 1 Deprtment of Computer Engineering, Shrif University of Technology, Tehrn,

More information

Announcements. CS 188: Artificial Intelligence Fall Recap: Search. Today. General Tree Search. Uniform Cost. Lecture 3: A* Search 9/4/2007

Announcements. CS 188: Artificial Intelligence Fall Recap: Search. Today. General Tree Search. Uniform Cost. Lecture 3: A* Search 9/4/2007 CS 88: Artificil Intelligence Fll 2007 Lecture : A* Serch 9/4/2007 Dn Klein UC Berkeley Mny slides over the course dpted from either Sturt Russell or Andrew Moore Announcements Sections: New section 06:

More information

Topics in Analytic Geometry

Topics in Analytic Geometry Nme Chpter 10 Topics in Anltic Geometr Section 10.1 Lines Objective: In this lesson ou lerned how to find the inclintion of line, the ngle between two lines, nd the distnce between point nd line. Importnt

More information

15. 3D-Reconstruction from Vanishing Points

15. 3D-Reconstruction from Vanishing Points 15. 3D-Reconstruction from Vnishing Points Christin B.U. Perwss 1 nd Jon Lseny 2 1 Cvendish Lortory, Cmridge 2 C. U. Engineering Deprtment, Cmridge 15.1 Introduction 3D-reconstruction is currently n ctive

More information

Agilent Mass Hunter Software

Agilent Mass Hunter Software Agilent Mss Hunter Softwre Quick Strt Guide Use this guide to get strted with the Mss Hunter softwre. Wht is Mss Hunter Softwre? Mss Hunter is n integrl prt of Agilent TOF softwre (version A.02.00). Mss

More information

analysisof natural imagery [1], [2]. The analysis of textural properties

analysisof natural imagery [1], [2]. The analysis of textural properties 272 of integrting multiple texturl fetures into single oundry determintion. The process is designed to simulte ctul perception of texturl discontinuities. Success of the system is demonstrted on pictures

More information

ON THE DEHN COMPLEX OF VIRTUAL LINKS

ON THE DEHN COMPLEX OF VIRTUAL LINKS ON THE DEHN COMPLEX OF VIRTUAL LINKS RACHEL BYRD, JENS HARLANDER Astrct. A virtul link comes with vriety of link complements. This rticle is concerned with the Dehn spce, pseudo mnifold with oundry, nd

More information

A Study on Eye Gaze Estimation Method Based on Cornea Model of Human Eye

A Study on Eye Gaze Estimation Method Based on Cornea Model of Human Eye A Study on Eye Gze Estimtion Method Bsed on Corne Model of Humn Eye Eui Chul Lee 1 nd Kng Ryoung Prk 2 1 Dept. of Computer Science, Sngmyung University, 7 Hongji-dong, Jongro-Ku, Seoul, Republic of Kore

More information

Lily Yen and Mogens Hansen

Lily Yen and Mogens Hansen SKOLID / SKOLID No. 8 Lily Yen nd Mogens Hnsen Skolid hs joined Mthemticl Myhem which is eing reformtted s stnd-lone mthemtics journl for high school students. Solutions to prolems tht ppered in the lst

More information

CS321 Languages and Compiler Design I. Winter 2012 Lecture 5

CS321 Languages and Compiler Design I. Winter 2012 Lecture 5 CS321 Lnguges nd Compiler Design I Winter 2012 Lecture 5 1 FINITE AUTOMATA A non-deterministic finite utomton (NFA) consists of: An input lphet Σ, e.g. Σ =,. A set of sttes S, e.g. S = {1, 3, 5, 7, 11,

More information

A Transportation Problem Analysed by a New Ranking Method

A Transportation Problem Analysed by a New Ranking Method (IJIRSE) Interntionl Journl of Innovtive Reserch in Science & Engineering ISSN (Online) 7-07 A Trnsporttion Problem Anlysed by New Rnking Method Dr. A. Shy Sudh P. Chinthiy Associte Professor PG Scholr

More information

Qubit allocation for quantum circuit compilers

Qubit allocation for quantum circuit compilers Quit lloction for quntum circuit compilers Nov. 10, 2017 JIQ 2017 Mrcos Yukio Sirichi Sylvin Collnge Vinícius Fernndes dos Sntos Fernndo Mgno Quintão Pereir Compilers for quntum computing The first genertion

More information

Ray surface intersections

Ray surface intersections Ry surfce intersections Some primitives Finite primitives: polygons spheres, cylinders, cones prts of generl qudrics Infinite primitives: plnes infinite cylinders nd cones generl qudrics A finite primitive

More information

Paracatadioptric Camera Calibration Using Lines

Paracatadioptric Camera Calibration Using Lines Prctdioptric Cmer Clibrtion Using Lines Joo P Brreto, Helder Arujo Institute of Systems nd Robotics - Dept of Electricl nd Computer Engineering University of Coimbr, Coimbr, Portugl Abstrct Prctdioptric

More information

The Nature of Light. Light is a propagating electromagnetic waves

The Nature of Light. Light is a propagating electromagnetic waves The Nture of Light Light is propgting electromgnetic wves Index of Refrction n: In mterils, light intercts with toms/molecules nd trvels slower thn it cn in vcuum, e.g., vwter The opticl property of trnsprent

More information

Efficient K-NN Search in Polyphonic Music Databases Using a Lower Bounding Mechanism

Efficient K-NN Search in Polyphonic Music Databases Using a Lower Bounding Mechanism Efficient K-NN Serch in Polyphonic Music Dtses Using Lower Bounding Mechnism Ning-Hn Liu Deprtment of Computer Science Ntionl Tsing Hu University Hsinchu,Tiwn 300, R.O.C 886-3-575679 nhliou@yhoo.com.tw

More information

Fig.25: the Role of LEX

Fig.25: the Role of LEX The Lnguge for Specifying Lexicl Anlyzer We shll now study how to uild lexicl nlyzer from specifiction of tokens in the form of list of regulr expressions The discussion centers round the design of n existing

More information

LU Decomposition. Mechanical Engineering Majors. Authors: Autar Kaw

LU Decomposition. Mechanical Engineering Majors. Authors: Autar Kaw LU Decomposition Mechnicl Engineering Mjors Authors: Autr Kw Trnsforming Numericl Methods Eduction for STEM Undergrdutes // LU Decomposition LU Decomposition LU Decomposition is nother method to solve

More information

Section 9.2 Hyperbolas

Section 9.2 Hyperbolas Section 9. Hperols 597 Section 9. Hperols In the lst section, we lerned tht plnets hve pproimtel ellipticl orits round the sun. When n oject like comet is moving quickl, it is le to escpe the grvittionl

More information

A Sparse Grid Representation for Dynamic Three-Dimensional Worlds

A Sparse Grid Representation for Dynamic Three-Dimensional Worlds A Sprse Grid Representtion for Dynmic Three-Dimensionl Worlds Nthn R. Sturtevnt Deprtment of Computer Science University of Denver Denver, CO, 80208 sturtevnt@cs.du.edu Astrct Grid representtions offer

More information

An Expressive Hybrid Model for the Composition of Cardinal Directions

An Expressive Hybrid Model for the Composition of Cardinal Directions An Expressive Hyrid Model for the Composition of Crdinl Directions Ah Lin Kor nd Brndon Bennett School of Computing, University of Leeds, Leeds LS2 9JT, UK e-mil:{lin,brndon}@comp.leeds.c.uk Astrct In

More information

Paracatadioptric Camera Calibration Using Lines

Paracatadioptric Camera Calibration Using Lines Prctdioptric Cmer Clibrtion Using Lines Joo P Brreto, Helder Arujo Institute of Systems nd Robotics - Dept of Electricl nd Computer Engineering University of Coimbr, 33 Coimbr, Portugl Abstrct Prctdioptric

More information

The gamuts of input and output colour imaging media

The gamuts of input and output colour imaging media In Proceedings of IS&T/SPIE Electronic Imging 1 The gmuts of input nd output colour imging medi án Morovic,* Pei Li Sun* nd Peter Morovic * Colour & Imging Institute, University of Dery, UK School of Informtion

More information

Announcements. CS 188: Artificial Intelligence Fall Recap: Search. Today. Example: Pancake Problem. Example: Pancake Problem

Announcements. CS 188: Artificial Intelligence Fall Recap: Search. Today. Example: Pancake Problem. Example: Pancake Problem Announcements Project : erch It s live! Due 9/. trt erly nd sk questions. It s longer thn most! Need prtner? Come up fter clss or try Pizz ections: cn go to ny, ut hve priority in your own C 88: Artificil

More information

If you are at the university, either physically or via the VPN, you can download the chapters of this book as PDFs.

If you are at the university, either physically or via the VPN, you can download the chapters of this book as PDFs. Lecture 5 Wlks, Trils, Pths nd Connectedness Reding: Some of the mteril in this lecture comes from Section 1.2 of Dieter Jungnickel (2008), Grphs, Networks nd Algorithms, 3rd edition, which is ville online

More information