Estimation of large-amplitude motion and disparity fields: Application to intermediate view reconstruction

Size: px
Start display at page:

Download "Estimation of large-amplitude motion and disparity fields: Application to intermediate view reconstruction"

Transcription

1 c 2000 SPIE. Personal use of this material is permitte. However, permission to reprint/republish this material for avertising or promotional purposes or for creating new collective works for resale or reistribution to servers or lists, or to reuse any copyrighte component of this work in other works must be obtaine from the SPIE. IS&T/SPIE Symposium on Electronic Imaging Visual Communications an Image Proc. Jan , 2001, San Jose, CA, USA Estimation of large-amplitue motion an isparity fiels: Application to intermeiate view reconstruction Mustapha Karouchi, Janusz Konra an Carlos Vázquez INRS-Télécommunications, Institut National e la Recherche Scientifique Place Bonaventure, P.O. 644, Montreal QC, H5A 1C6, Canaa karou@inrs-telecom.uquebec.ca Boston University, Dept. of Electrical an Computer Eng., Boston, MA 02215, USA jkonra@bu.eu ABSTRACT This paper escribes a metho for establishing ense corresponence between two images in a vieo sequence (motion) or in a stereo pair (isparity) in case of large isplacements. In orer to eal with large-amplitue motion or isparity fiels, multi-resolution techniques such as blocks matching an optical flow have been use in the past. Although quite successful, these techniques cannot easily cope with motion/isparity iscontinuities as they o not explicitly exploit image structure. Aitionally, their computational complexity is high; block matching requires examination of numerous vector caniates while optical flow-base techniques are iterative. In this paper, we propose a new approach that aresses both issues. The approach combines feature matching with Delaunay triangulation, an thus reliable long-range corresponences result while the computational complexity is not high (sparse representation). In the propose approach, feature points are foun first using a simple intensity corner etector. Then, corresponence pairs between two images are foun by maximizing cross-correlation over a small winow. Finally, the Delaunay triangulation is applie to the resulting points, an a ense vector fiel is compute by planar interpolation over Delaunay triangles. The resulting vector fiel is continuous everywhere, an thus oes not reflect motion or epth iscontinuities at object bounaries. In orer to improve the renition of such iscontinuities, we propose to further ivie Delaunay triangles whenever the isplacement vectors within a triangle o not allow goo intensity match. The approach has been extensively teste on stereoscopic images in the context of intermeiate view reconstruction where the quality of estimate isparity fiels is critical for final image renering. The first results are very encouraging as the reconstructe images are of high quality, especially at object bounaries, an the computational complexity is lower than that of multi-resolution block matching. Keywors: motion estimation, isparity estimation, optical flow, feature corresponence, Delaunay triangulation, stereoscopic image processing, intermeiate view reconstruction 1. INTRODUCTION Estimation of ense corresponences, such as motion fiels in vieo sequences or isparity fiels in stereoscopic image pairs, in the presence of large-amplitue isplacements, is a ifficult problem. Methos base on block matching can hanle large isplacements but they are computationally involve. Moreover, they cannot reliably isambiguate among caniate blocks since simple cumulative error criteria are typically use. Traitional optical flow methos are incremental as they assume local intensity linearity an compute isplacement fiels iteratively in small steps. They are not irectly applicable to large-amplitue isplacements. To eal with this problem, two ifferent approaches have been propose in the literature. One approach uses a multi-resolution representation that enables estimation of large-amplitue vectors at coarse levels of a multi-resolution pyrami an subsequent refinement at finer resolutions. Although successfully use in both block matching an optical flow methos, multi-resolution implementation oes not exploit explicit image structure to help isambiguate among caniate matches. Thus, vector fiel iscontinuities are not well estimate; in block matching, because of vector constancy over a block, an in optical flow since it is very ifficult to ajust smoothness factors at each resolution level an usually oversmoothe fiels result 1,2. In the secon approach, corresponences are explicitly compute base on image structure, e.g., by means of feature point matching, but the resulting vector fiels are sparse since the vectors are efine only at feature points 3,4.

2 In this paper, we propose a new metho that aresses the above issues. Since the intene application is the reconstruction of intermeiate images in a vieo sequence or of intermeiate views in a stereo pair, reliability an computational efficiency of the approach are very important. In orer to aress reliability, we propose to compute feature-base corresponences first. Although the resulting isplacement vectors are mostly reliable, they are also sparse. Thus, to compute ense vector fiels, an o this with efficiency, we propose triangulation-base low-orer spatial interpolation between feature-point vectors. Since some triangles overlap ifferent motion/isparity regions an thus cannot faithfully represent the corresponing iscontinuity, we etect such triangles, further ivie them, an re-interpolate the isplacement vectors for each new triangle. We apply the estimate vector fiel to intermeiate view reconstruction in stereoscopic pairs. Numerous experimental results are shown in Section APPROACH The propose metho consists of three main components: computation of sparse corresponences base on feature points, computation of ense vector fiel base on these sparse corresponences by means of Delaunay triangulation, an recursive refinement of the triangulation. These three components are escribe next Computation of sparse corresponences Extraction of feature points A feature point etector aims at the localization of significant intensity structures in an image. Often, points of abrupt intensity change, as manifeste by large intensity graient, are use as an inication of a significant image structure. These points shoul be etecte consistently an localize precisely because errors in etection or localization are magnifie in the final results. Corner-point etection is one of the more reliable mechanisms allowing to fin feature points. Three ifferent approaches have been propose in the literature to etect corners: contour-base methos 6,7, signal-base methos 8 an template fitting methos 9,10. In this work, we opt for a metho from the secon class, the so-calle Harris etector 5. We apply the Harris etector to each image of the stereo or motion image pair (I 0, I 1 ). The Harris etector s threshol is fixe epening on the average image graient. In our application, this threshol is selecte so that 800-1,200 corner points are recovere from a pixel image. The final results are not particularly sensitive to the number of corner points etecte Matching of feature points Once feature points are extracte in both images, a corresponence between feature points in the two images must be establishe. This may be one by evaluating similarity between small winows aroun two feature points, one from each image. A pair is sai to have been matche if within a search area it results in the highest similarity (lowest issimilarity) measure. There are many possible measures of similarity. Some are taken from the fiel of signal processing 11 while others from statistics 12. In this paper, to measure the similarity we use a statistically-motivate metho base on cross-correlation 13. For each feature point m 0 in image I 0, a local winow centere at m 0 is correlate with a winow centere at each caniate feature point m 1 within a selecte area S of image I 1. The search area S represents an upper boun on the maximum motion/isparity. A constraint on the correlation measure is applie in orer to eliminate unreliable matches; results with correlation uner a certain threshol are rejecte. Then, cross-valiation is applie to eliminate false matches; if m 1 in I 1 is the best match for m 0 in I 0 within the search area S, then also m 0 must be the best match for m 1 within S. Typically, we foun acceptable matches for 800-1,200 feature points etecte in a pixel image. Although the cross-valiation eliminates almost all unreliable/false matches, a small number of them may go unetecte thus affecting the final results. We remove the remaining unreliable matches by postprocessing escribe in Section As a result, we obtain an ensemble of feature point pairs (corresponences), that can be use as control points for ense vector fiel interpolation Computation of ense vector fiel In this section, we present a metho to generate a ense vector fiel from the compute corresponence pairs. First, Delaunay triangulation is performe, then its noes are post-processe to eliminate potential outliers, an finally a ense vector fiel is compute by low-orer spatial interpolation within each Delaunay triangle.

3 p 2 p 1 p 3 p p 4 p 5 Figure 1. Neighborhoo η(p) of corresponence point p Delaunay triangulation of corresponence points The general Delaunay triangulation can be escribe as follows. Let P be the set of all points in a two-imensional plane to be triangulate. The Delaunay triangulation of P, enote DT (P), is the unique triangulation base on the concept of empty overlap with a circles, i.e., { } DT (P) = (p, q, r) P 3, C(p, q, r) P =, where C(p, q, r) is a circle circumscribe on the tree points p, q, an r. In the case consiere here, P is the set of corresponence points in image I 0. Clearly, more feature points exist in active image areas, i.e., at object bounaries or in texture regions. Thus, smaller triangles will result in these active areas, an larger triangles will appear in homogeneous regions Elimination of potential outliers As mentione in Section 2.1.2, unreliable corresponence points may be foun an use for the Delaunay triangulation. Such unreliable points may aversely affect the ense fiel interpolation escribe in the next section. In orer to eliminate them, we use the following proceure: 1. for each corresponence point p in image I 0 ientify its neighborhoo η(p), i.e., all vertices of triangles containing p, except for p itself (Fig. 1), 2. remove unreliable corresponence points; a corresponence point p is consiere unreliable if: p p > ρ, where p is the corresponence vector at p, p is the average corresponence vector over η(p), an ρ is an ajustable threshol. Using this simple proceure, we are able to remove significant number of outlying corresponences (although some correct matches are also remove) at very low computational cost. We teste the propose proceure on several natural stereoscopic images. Results obtaine with the cross-correlation matching an the above post-processing step are presente in Section 4. Typically, at least 16% of the total number of corresponence points are remove by this post-processing step Interpolation of vectors Given the triangulation an the corresponence vectors at its noes, we compute a ense vector fiel uner the assumption that, in spatial coorinates, motion/isparity varies linearly within each triangle. Given three vertices of a triangle, vector = [ 1, 2 ] T at each pixel within the triangle is compute from an affine (linear) moel. This is equivalent to passing one plane in space (x, y, 1 ) an one plane in space (x, y, 2 ), each with respect to triangle vertices. Let p, q, an r be the three noes of a triangle (Fig. 2). If p, q an r are the corresponence vectors associate with p, q an r, respectively, then the vector x for image point x within the triangle forme by the triplet of corners p, q, an r is given by: x = (1 i j) p + i q + j r, (1)

4 θ n ipq p p q jpr q x m r r Figure 2. Interpolation of isplacement vector at x within triangle with noes p, q an r. where i an j relate the coorinates of point x to those of the noe points p, q, an r. Referring to Fig. 2, the equations for the values of i an j are: i = ( px 2 pn 2 ) 1 2 pq cos( π 2 Θ), j = ( px 2 pm 2 ) 1 2 pr cos( π, (2) 2 Θ) where ( pq pr ) Θ = arccos, pm = pq pr pr px, pn = pr pq px. pq with pq enoting a vector joining points p an q, while m an n are orthogonal projections of x onto pr an pq. The ense vector fiel is etermine by first computing the coefficients i an j (2) for each image point within each triangle, an then by applying them to equation (1). Such vector fiel computation presents very interesting properties both from mathematical an practical point of view it efines a vector fiel that is continuous everywhere an oes so with computational efficiency. However, this metho oes not allow to take into account potential vector fiel iscontinuities. We aress this problem in the next section Triangulation refinement It is clear that the performance of the ense fiel computation escribe above epens on the quality of the Delaunay triangulation. One potential problem exists when a triangle overlaps two ifferent objects in the image or an object an backgroun. Then, such a triangle contains an object bounary where the true motion or isparity vector fiel is likely to be iscontinuous. Clearly, a iscontinuity cannot be reprouce by the linear interpolation propose, an a remey must be sought. One option woul be to evelop a more sophisticate interpolation scheme within each triangle, while another to align the triangles better with object bounaries in the image. We opt for the secon metho an we propose to refine (subivie) the triangles that exhibit excessive preiction error between images I 0 an I 1. The mean-square preiction error between the images I 0 an I 1 is efine as follows: E = s Ω(I 1 ( s + s ) I 0 ( s)) 2, where Ω is the (iscrete) image omain, s = (x, y) is a spatial position in Ωm, an s is a isplacement vector at s. Note that Ω is partitione by a triangular mesh. Let Ψ = {ψ i, i = 1, 2,..., N} be the set of N Delaunay triangles; ψ i is the set of image points within triangle number i. Using this triangular mesh, the mean-square preiction error can be written as follows: E ε(ψ i ), ψ i Ψ where ε(ψ i ) is the mean-square preiction error between I 0 an I 1 within the triangle ψ i. This is an approximation only since each triangle sie is counte twice in the computation (it is consiere to belong to two triangles).

5 (a) Figure 3. Refinement of the triangulation: (a) ivision of the triangle of interest into three triangles; (b) four cases of ivision of the neighboring triangles. (b) In orer, to ientify triangles that are likely to overlap motion or isparity iscontinuity, an that nee to be further ivie, we compute the mean-square preiction error for each triangle an we compare this error to threshol σ. If the absolute value of the preiction error for the particular triangle excees σ, an the triangle is sufficiently large, we ivie it into four triangles accoring to Fig. 3.a. The triangle create in the center has noes (vertices) in the mile of the original triangle s sies. In orer to assure triangulation uniformity, neighboring triangles may be split into up to four triangles (Fig. 3.b). For each univie triangle ψ i, we compute the number n(ψ i ) of neighboring ivie triangles. There are four possibilities (Fig. 3.b): if n(ψ i ) = 0, nothing to be one; if n(ψ i ) = 1, the triangle ψ i is ivie into two sub-triangles with the new vertex place on the sie that ajoins the neighboring ivie triangle, if n(ψ i ) = 2, the triangle ψ i is ivie into three sub-triangles; there are two possibilities: we choose the one for which the triangles create are more compact, if n(ψ i ) = 3, the triangle ψ i is ivie into four sub-triangles. Now, that we have new vertices (an triangles), we nee to compute isplacement vectors associate with them. Let i, j an k be the unknown isplacement vectors corresponing to the new noes i, j an k create in a given triangle ψ i with noes p, q an r. Let p, q an r be the corresponence vectors for ol vertices p, q an r (Fig. 4). We assume that the triangle ψ i is sufficiently compact. Uner the assumption that motion/isparity is locally smooth, the isplacements that minimize the mean-square preiction error in the neighborhoo of each vertex i, j, k shoul belong to the set { p, q, r }. Then, the vector l at vertex l is estimate as follows: l = min s { p, q, r} s W l (I 1 (s + s ) I 0 (s)) 2, l {i, j, k}, (3) where W l is a small winow aroun image point l. We chose 3 3 winow for experiments. Once the vectors at new noes have been compute, the ense vector fiel within each new sub-triangle is compute using the metho escribe in Section Having compute this new ense vector fiel, the refinement proceure is repeate until convergence. The convergence is achieve when no triangles unergo ivision. We can summarize the refinement proceure as follows: 1. choose value of the threshols: σ for the mean-square preiction error E, an γ for the triangle size, 2. for each triangle ψ i, compute the mean-square preiction error ε(ψ i ); if ε(ψ i ) > σ an ψ i γ, where ψ i enotes the size of ψ i, ivie the triangle into four sub-triangles (Fig. 3.a), 3. ensure triangulation uniformity; for all univie triangles compute the number of neighboring newly-ivie triangles an split them accoringly (Fig. 3.b),

6 p i p i q k j q k j r r Figure 4. Original triangle with vertices p, q, r, an the newly create vertices i, j, k. 4. compute isplacement vectors for all the new triplets of vertices by performing minimization (3), 5. re-interpolate the ense vector fiel for each new triangle as escribe in Section 2.2.3, 6. go back to 2 unless no more triangles can be ivie. 3. APPLICATION OF THE METHOD TO THE RECONSTRUCTION OF INTERMEDIATE IMAGES IN STEREOSCOPIC PAIRS The application we are intereste in is the reconstruction of intermeiate views in stereoscopic images. Such a reconstruction has a number of applications in the fiel of stereoscopic imaging. Problems relate to parallax ajustment (e.g., too much stereo) 14, for example, are solve through the reconstruction of images coming from virtual cameras with a baseline (istance between optical centers of the cameras) ifferent than that of the true cameras. Continuous look-aroun, i.e., reconstruction of intermeiate images to allow lateral viewer motion in front of stereoscopic isplay, is another possible application 15. In the stereoscopic literature, two approaches have been propose for the reconstruction of intermeiate views 16,17 : 1. methos base on a 3-D moel of the scene: estimation of a 3-D representation of the scene from two or more images (left an right in the stereo case), followe by projection onto a virtual camera place at the esire reconstruction position, 2. methos base on isparity compensation: estimation of a isparity fiel between left an right images an its application to isparity-compensate reconstruction at the esire camera position. For the complex real-worl scenes, the 3-D moeling is too complex an therefore results are usually imprecise. For the intene entertainment applications, that we envisage, the camera convergence angle is very small an the camera baseline is close to that of the interocular istance, limiting the range of isparities an thus favoring the isparity compensation approach. We are intereste in the case of parallel or almost parallel camera geometry, an in virtual cameras having optical centers on the line joining the optical centers of left an right cameras; the optical axes of the virtual cameras are assume parallel to the optical axes of the true cameras. With this geometry, projections of a point onto a virtual camera are on the line passing through the projections of this point on the left an right camera planes. This allows us to reconstruct any intermeiate view by linear isparity-compensate interpolation, assuming that we are able to estimate the isparity between left an right images. Once the isparity is estimate, the reconstruction of an intermeiate image at position γ (0 < γ < 1, γ=0 being the left image, an γ=1 the right one), uner ieal conitions, shoul be as simple as copying either the left or right pixel to the intermeiate image I γ. But ue to noise, illumination effects an aliasing, the matching pixels, even in the case of perfect isparity estimation, are not ientical. In a simple scenario, an image at position γ can be then reconstructe by a linear isparity-compensate interpolation as follows 17 : I γ ( s + (γ α) s ) = (1 γ)i L ( s α s ) + γi R ( s + (1 α) s ).

7 (a) (b) Figure 5. Original pair from the stereoscopic vieo sequence Flowerpot: (a) left image; (b) right image. Note that the isparity fiel { s, s Ω} is efine at position α, i.e., the vectors pass through points of the regular gri Ω at α. Two reconstruction scenarios are possible. In the first scenario, in orer to reconstruct {I γ, s Ω}, isparities are estimate at α = γ; for each γ a separate isparity fiel must be compute thus imposing extreme computational effort. Alternatively, a single isparity fiel is compute at α, but then for γ α samples of I γ are recovere over a set of irregular spatial locations an further reconstruction of regularly-space samples is neee; for a given α, a regularly-space image {I γ ( s), s Ω} must be reconstructe from irregular samples {I γ ( s + (γ α) s ), s Ω}. In particular, for α=0, i.e., when the isparity fiel is estimate at the left image, a regularly-space image {I γ ( s), s Ω} must be reconstructe from irregular samples: I γ ( s + γ s ) = (1 γ)i L ( s) + γi R ( s + s ). (4) Various approaches have been propose to ate to reconstruct an image from irregularly-space samples, varying from simple polynomial interpolation base on triangulation to more complex iterative algorithms. Complex algorithms like those propose in the literature offer better results at the expense of higher complexity an processing time. A low-orer polynomial interpolation is a simpler solution an offers acceptable results for the current stage of our work. To keep the computational complexity of the overall process low, we use a linear interpolation base on Delaunay triangulation of the irregularly space image samples. Let Γ = { s + γ s, s Ω} be the irregular sampling gri. First, we perform the Delaunay triangulation on Γ. Then, we compute pixel values over the regular gri Ω by linear (planar) interpolation over each triangle in the mesh. The overall process can be escribe as follows: 1. isparity estimation using the algorithm propose in Section 2, 2. computation of the irregularly-sample image through isparity compensation (4), 3. reconstruction of the regularly-sample image using linear interpolation base on Delaunay triangulation. 4. EXPERIMENTAL RESULTS We have teste the propose metho on real stereoscopic images. An example stereo pair Flowerpot (see Acknowlegments) is shown in Fig. 5. First, feature points were extracte from the two images using the Harris corner point etector. Then, corresponence pairs were foun by maximizing a cross-correlation measure. Fig. 6.a shows corresponence vectors that join pairs of corresponence points, before post-processing. Fig. 6.b shows the corresponence points after post-processing

8 that removes outliers (Section 2.2.2). It can be seen that most of the false matches have been eliminate, although some correct matches may have been remove as well. The resulting fiel of corresponence vectors is sparse with more vectors locate in texture areas an very few vectors - in uniform areas. The set of resulting corresponence points was then use to perform the Delaunay triangulation an to compute a ense isparity fiel. In the experiments, the preiction error threshol σ was set to 200. Fig. 7.a shows the ense fiel obtaine with linear interpolation before refinement. Overall, the compute isparity fiel closely matches epth in the original scene espite the fact that it is interpolate from a sparse set of corresponence vectors rather than being compute locally for every pixel. Fig. 7.b shows the same fiel after refining the mesh (triangle subivision) as escribe in Section Although the vector fiels before an after refinement are very similar, local ifferences exist an they can be better seen in the enlarge sub-winows in Fig. 9. The triangulation results, superpose on the reference image I 0 (left image from the stereo pair), are epicte in Fig. 8. Fig. 8.a shows the triangulation of corresponence points with no refinement, while Fig. 8.b shows the triangulation after refinement. Note the appearance of many new triangles in texture areas (plants, fruit stan) an aroun object bounaries (aroun umbrellas, flags, men s pants). (a) (b) Figure 6. Corresponence vectors between images I 0 an I 1 : (a) before post-processing; (b) after post-processing to remove outliers (Section 2.2.1). (a) (b) Figure 7. Disparity fiel obtaine from corresponence points by the propose metho: (a) without mesh refinement; an (b) with mesh refinement. For clarity, both vector fiels have been horizontally an vertically subsample by 18 an 6, respectively.

9 (a) (b) Figure 8. Triangulation (a) with no mesh refinement; an (b) after mesh refinement (preiction threshol σ= 200). Finally, we have applie the compute isparity fiels to intermeiate view reconstruction at γ=0.5 as escribe in Section 3. Fig. 9 shows enlarge pixel winows aroun the flower pot that were extracte from the triangulation image (Fig. 8), from ense vector fiels (Fig. 7) an from images reconstructe at γ=0.5, that we are not showing here in full since they look remarkably similar in print. Clearly, the refinement ha a significant impact on the interpolate ense isparity fiel; the iscontinuities in the isparity fiel are more perceptible after refinement (Fig. 9.). This improvement is ue to the fact that several new triangles have been create at the flower pot ege, especially at the winow frame in the backgroun, allowing for a better renition of epth (isparity) iscontinuity. An improvement is also visible in the reconstructe image with refinement (Fig. 9.f), aroun the winow frame above the flower. Note that the isparities between the two images are greater than 8 pixels/frame in magnitue an, in general, not easy to estimate. Clearly, the reconstructe intermeiate images both with an without refinement, are very close in quality to the original (reference) image (Fig. 10), although they are more blurre ue to the inevitable interpolation process involve in combining images I 0 an I 1, an also ue to the applie non-uniform-to-uniform interpolation. A single-view comparison of the reconstruction, however, is not critical enough. A much more emaning scenario is the reconstruction of a series of intermeiate views an there play-out in time, thus simulating a camera translation between the left- an right-camera positions. We have performe such a simulate camera translation by reconstructing 18 intermeiate images between I 0 an I 1. The overall perception of the camera translation was very natural; it truly seeme that the camera was moving horizontally. While in the sequence reconstructe base on the isparity fiel with no refinement errors appeare at some object bounaries, especially aroun the flower pot where the isparity iscontinuity is significant, most of these errors were eliminate when the refine isparity fiel was use. We have also performe some initial comparisons of the propose reconstruction with a reconstruction base on optical flow-type isparity fiel 14. In these early tests the new metho compare favorably at object bounaries, but further etaile comparisons are require. 5. CONCLUSIONS We propose a new metho for the estimation of large-amplitue motion an isparity fiels. This metho is base on feature corresponences an the interpolation of sparse vector ata. The approach has several benefits: it permits to estimate large isplacements an subsequently take into account motion/isparity iscontinuities. Moreover, the metho is very efficient computationally compare to a typical multi-resolution block matching. Initial results are very promising; the metho prouces high-quality intermeiate views for natural (complex) stereoscopic image sequences. However, the results epen on the presence of texture in the images; the metho works well in sequences with strong textures. 6. ACKNOWLEDGMENTS We tank NHK of Japan for proviing us with the stereoscopic test sequence Flowerpot.

10 (a) (b) (c) () (e) (f) Figure 9. Enlarge winow ( ) from the triangulation (Fig. 8), isparity fiel (Fig. 7) an image reconstructe at γ=0.5: (a,c,e) without mesh refinement; (b,,f) with mesh refinement.

11 Figure 10. Enlarge winow ( ) from the original left image. 7. REFERENCES 1. M. Bober an J. Jittler, Robust motion analysis, in Proc. IEEE Conf. on Computer Vision an Pattern Recognition, pp , E. Memin an P. Perez, Dense estimation an object-base segmentation of optical flow with robust technique, IEEE Trans. Image Processing, vol. 7, no. 5, pp , N. Laurent, P. Lechat an H. Sanson, Limitation of triangles overloaing in mesh-base motion estimation using augmente Lagrangian, in Proc. IEEE Int. Conf. on Image Proc., ICIP 98, vol. II, pp , P.R. Giaccone, D. Greenhill an G.A. Jones, Recovering very large visual motion fiels, in Proc. 10-th Scaninavian Conference on Image Analysis, pp , C. Harris an M. Stephens, A combine corner an ege etector, in Proc. Alvey Vision Conference, pp , H. Asaa an M. Bray, The curvative primal sketch, IEEE Transactions on Pattern Analysis an Machine Intelligence, vol. 8, no. 1, pp. 2-14, G. Meioni an Y. Yasumoto, Corner etection an curve representation using cubic B-splines, Computer Vision, Graphics an Image Processing, vol. 39, no. 1, pp , P. Beauet, Rotationally invariant image operators, in Proc. of 4-th Int. Joint Conference on Pattern Recognition, Tokyo, pp , K. Rohr, Recognition of corners by parametric fitting, Int. Journal of Computer Vision, vol. 9, no. 3, pp , R. Deriche an T. Blaszka, Recovering an characterizing image features using an efficient moel base approach, in Proc. IEEE Conf. on Computer Vision an Pattern Recognition, pp , V.S. Nalwa, A guie tour of computer vision, Aison-Wesley, Z. Puzicha, T. Hofmann an J.M. Buhmann, Non-parametric similarity measures for unsupervise texture segmentation an image retrieval,, Proc. IEEE Conf. on Computer Vision an Pattern Recognition, pp , June P. Pritchett an A. Zisserman, Wie baseline stereo matching, in Proc. 6-th International Conference on Computer Vision, pp , J. Konra, Enhacement of viewer comfort in stereoscopic viewing: parallax ajustement, in Proc. SPIE Stereoscopic Displays an Virtual Reality Systems, vol. 3639, pp , 1999.

12 15. N. Chang an A. Zakhor, View generation for three-imensional scenes from vieo sequences, IEEE Trans. on Image Proc., vol. 6, no. 4, pp , E. Izquiero, Stereo matching for enhance telepresence in three-imensional vieocommunications, IEEE Trans. Circuits Syst. Vieo Technol., vol. 7, no. 4, pp , J. Konra, View reconstruction for 3-D vieo entertaiment: issues, algorithms an applications, in Proc. Int. Conf. on Image Process. an its Applications, pp. 8 12, R. Stasinski an J. Konra, POCS-base image reconstruction from irregulary-space samples, in Proc. IEEE Int. Conf. Image Processing, ICIP-2000, vol. II, pp , C. Vazquez, E. Dubois an J. Konra, Wavelet-base reconstruction of irregularly sample images: application to stereo imaging, in Proc. IEEE Int. Conf. Image Processing, ICIP 2000, vol. II, pp , Ch. Bernar, Wavelets an ill pose problems: optic flow an scatere ata interpolation, PhD thesis, CMAP, Ecole Polytechnique, Paris, 1999.

Refinement of scene depth from stereo camera ego-motion parameters

Refinement of scene depth from stereo camera ego-motion parameters Refinement of scene epth from stereo camera ego-motion parameters Piotr Skulimowski, Pawel Strumillo An algorithm for refinement of isparity (epth) map from stereoscopic sequences is propose. The metho

More information

Shift-map Image Registration

Shift-map Image Registration Shift-map Image Registration Svärm, Linus; Stranmark, Petter Unpublishe: 2010-01-01 Link to publication Citation for publishe version (APA): Svärm, L., & Stranmark, P. (2010). Shift-map Image Registration.

More information

Shift-map Image Registration

Shift-map Image Registration Shift-map Image Registration Linus Svärm Petter Stranmark Centre for Mathematical Sciences, Lun University {linus,petter}@maths.lth.se Abstract Shift-map image processing is a new framework base on energy

More information

Computer Graphics Chapter 7 Three-Dimensional Viewing Viewing

Computer Graphics Chapter 7 Three-Dimensional Viewing Viewing Computer Graphics Chapter 7 Three-Dimensional Viewing Outline Overview of Three-Dimensional Viewing Concepts The Three-Dimensional Viewing Pipeline Three-Dimensional Viewing-Coorinate Parameters Transformation

More information

Fast Window Based Stereo Matching for 3D Scene Reconstruction

Fast Window Based Stereo Matching for 3D Scene Reconstruction The International Arab Journal of Information Technology, Vol. 0, No. 3, May 203 209 Fast Winow Base Stereo Matching for 3D Scene Reconstruction Mohamma Mozammel Chowhury an Mohamma AL-Amin Bhuiyan Department

More information

Image Segmentation using K-means clustering and Thresholding

Image Segmentation using K-means clustering and Thresholding Image Segmentation using Kmeans clustering an Thresholing Preeti Panwar 1, Girhar Gopal 2, Rakesh Kumar 3 1M.Tech Stuent, Department of Computer Science & Applications, Kurukshetra University, Kurukshetra,

More information

Online Appendix to: Generalizing Database Forensics

Online Appendix to: Generalizing Database Forensics Online Appenix to: Generalizing Database Forensics KYRIACOS E. PAVLOU an RICHARD T. SNODGRASS, University of Arizona This appenix presents a step-by-step iscussion of the forensic analysis protocol that

More information

Kinematic Analysis of a Family of 3R Manipulators

Kinematic Analysis of a Family of 3R Manipulators Kinematic Analysis of a Family of R Manipulators Maher Baili, Philippe Wenger an Damien Chablat Institut e Recherche en Communications et Cybernétique e Nantes, UMR C.N.R.S. 6597 1, rue e la Noë, BP 92101,

More information

Dense Disparity Estimation in Ego-motion Reduced Search Space

Dense Disparity Estimation in Ego-motion Reduced Search Space Dense Disparity Estimation in Ego-motion Reuce Search Space Luka Fućek, Ivan Marković, Igor Cvišić, Ivan Petrović University of Zagreb, Faculty of Electrical Engineering an Computing, Croatia (e-mail:

More information

Exercises of PIV. incomplete draft, version 0.0. October 2009

Exercises of PIV. incomplete draft, version 0.0. October 2009 Exercises of PIV incomplete raft, version 0.0 October 2009 1 Images Images are signals efine in 2D or 3D omains. They can be vector value (e.g., color images), real (monocromatic images), complex or binary

More information

A Classification of 3R Orthogonal Manipulators by the Topology of their Workspace

A Classification of 3R Orthogonal Manipulators by the Topology of their Workspace A Classification of R Orthogonal Manipulators by the Topology of their Workspace Maher aili, Philippe Wenger an Damien Chablat Institut e Recherche en Communications et Cybernétique e Nantes, UMR C.N.R.S.

More information

New Geometric Interpretation and Analytic Solution for Quadrilateral Reconstruction

New Geometric Interpretation and Analytic Solution for Quadrilateral Reconstruction New Geometric Interpretation an Analytic Solution for uarilateral Reconstruction Joo-Haeng Lee Convergence Technology Research Lab ETRI Daejeon, 305 777, KOREA Abstract A new geometric framework, calle

More information

Using the disparity space to compute occupancy grids from stereo-vision

Using the disparity space to compute occupancy grids from stereo-vision The 2010 IEEE/RSJ International Conference on Intelligent Robots an Systems October 18-22, 2010, Taipei, Taiwan Using the isparity space to compute occupancy gris from stereo-vision Mathias Perrollaz,

More information

Using Ray Tracing for Site-Specific Indoor Radio Signal Strength Analysis 1

Using Ray Tracing for Site-Specific Indoor Radio Signal Strength Analysis 1 Using Ray Tracing for Site-Specific Inoor Raio Signal Strength Analysis 1 Michael Ni, Stephen Mann, an Jay Black Computer Science Department, University of Waterloo, Waterloo, Ontario, NL G1, Canaa Abstract

More information

SMART IMAGE PROCESSING OF FLOW VISUALIZATION

SMART IMAGE PROCESSING OF FLOW VISUALIZATION SMAR IMAGE PROCESSING OF FLOW VISUALIZAION H Li (A Rinoshika) 1, M akei, M Nakano 1, Y Saito 3 an K Horii 4 1 Department of Mechanical Systems Engineering, Yamagata University, Yamagata 99-851, JAPAN Department

More information

NEW METHOD FOR FINDING A REFERENCE POINT IN FINGERPRINT IMAGES WITH THE USE OF THE IPAN99 ALGORITHM 1. INTRODUCTION 2.

NEW METHOD FOR FINDING A REFERENCE POINT IN FINGERPRINT IMAGES WITH THE USE OF THE IPAN99 ALGORITHM 1. INTRODUCTION 2. JOURNAL OF MEDICAL INFORMATICS & TECHNOLOGIES Vol. 13/009, ISSN 164-6037 Krzysztof WRÓBEL, Rafał DOROZ * fingerprint, reference point, IPAN99 NEW METHOD FOR FINDING A REFERENCE POINT IN FINGERPRINT IMAGES

More information

A Duality Based Approach for Realtime TV-L 1 Optical Flow

A Duality Based Approach for Realtime TV-L 1 Optical Flow A Duality Base Approach for Realtime TV-L 1 Optical Flow C. Zach 1, T. Pock 2, an H. Bischof 2 1 VRVis Research Center 2 Institute for Computer Graphics an Vision, TU Graz Abstract. Variational methos

More information

Data Mining: Clustering

Data Mining: Clustering Bi-Clustering COMP 790-90 Seminar Spring 011 Data Mining: Clustering k t 1 K-means clustering minimizes Where ist ( x, c i t i c t ) ist ( x m j 1 ( x ij i, c c t ) tj ) Clustering by Pattern Similarity

More information

Skyline Community Search in Multi-valued Networks

Skyline Community Search in Multi-valued Networks Syline Community Search in Multi-value Networs Rong-Hua Li Beijing Institute of Technology Beijing, China lironghuascut@gmail.com Jeffrey Xu Yu Chinese University of Hong Kong Hong Kong, China yu@se.cuh.eu.h

More information

Video-based Characters Creating New Human Performances from a Multi-view Video Database

Video-based Characters Creating New Human Performances from a Multi-view Video Database Vieo-base Characters Creating New Human Performances from a Multi-view Vieo Database Feng Xu Yebin Liu? Carsten Stoll? James Tompkin Gaurav Bharaj? Qionghai Dai Hans-Peter Seiel? Jan Kautz Christian Theobalt?

More information

Transient analysis of wave propagation in 3D soil by using the scaled boundary finite element method

Transient analysis of wave propagation in 3D soil by using the scaled boundary finite element method Southern Cross University epublications@scu 23r Australasian Conference on the Mechanics of Structures an Materials 214 Transient analysis of wave propagation in 3D soil by using the scale bounary finite

More information

Estimating Velocity Fields on a Freeway from Low Resolution Video

Estimating Velocity Fields on a Freeway from Low Resolution Video Estimating Velocity Fiels on a Freeway from Low Resolution Vieo Young Cho Department of Statistics University of California, Berkeley Berkeley, CA 94720-3860 Email: young@stat.berkeley.eu John Rice Department

More information

Robust Camera Calibration for an Autonomous Underwater Vehicle

Robust Camera Calibration for an Autonomous Underwater Vehicle obust Camera Calibration for an Autonomous Unerwater Vehicle Matthew Bryant, Davi Wettergreen *, Samer Aballah, Alexaner Zelinsky obotic Systems Laboratory Department of Engineering, FEIT Department of

More information

EFFICIENT STEREO MATCHING BASED ON A NEW CONFIDENCE METRIC. Won-Hee Lee, Yumi Kim, and Jong Beom Ra

EFFICIENT STEREO MATCHING BASED ON A NEW CONFIDENCE METRIC. Won-Hee Lee, Yumi Kim, and Jong Beom Ra th European Signal Processing Conference (EUSIPCO ) Bucharest, omania, August 7-3, EFFICIENT STEEO MATCHING BASED ON A NEW CONFIDENCE METIC Won-Hee Lee, Yumi Kim, an Jong Beom a Department of Electrical

More information

Vision-based Multi-Robot Simultaneous Localization and Mapping

Vision-based Multi-Robot Simultaneous Localization and Mapping Vision-base Multi-Robot Simultaneous Localization an Mapping Hassan Hajjiab an Robert Laganière VIVA Research Lab School of Information Technology an Engineering University of Ottawa, Ottawa, Canaa, K1N

More information

Politehnica University of Timisoara Mobile Computing, Sensors Network and Embedded Systems Laboratory. Testing Techniques

Politehnica University of Timisoara Mobile Computing, Sensors Network and Embedded Systems Laboratory. Testing Techniques Politehnica University of Timisoara Mobile Computing, Sensors Network an Embee Systems Laboratory ing Techniques What is testing? ing is the process of emonstrating that errors are not present. The purpose

More information

Stereo Vision. MAN-522 Computer Vision

Stereo Vision. MAN-522 Computer Vision Stereo Vision MAN-522 Computer Vision What is the goal of stereo vision? The recovery of the 3D structure of a scene using two or more images of the 3D scene, each acquired from a different viewpoint in

More information

Multi-camera tracking algorithm study based on information fusion

Multi-camera tracking algorithm study based on information fusion International Conference on Avance Electronic Science an Technolog (AEST 016) Multi-camera tracking algorithm stu base on information fusion a Guoqiang Wang, Shangfu Li an Xue Wen School of Electronic

More information

Approximation with Active B-spline Curves and Surfaces

Approximation with Active B-spline Curves and Surfaces Approximation with Active B-spline Curves an Surfaces Helmut Pottmann, Stefan Leopolseer, Michael Hofer Institute of Geometry Vienna University of Technology Wiener Hauptstr. 8 10, Vienna, Austria pottmann,leopolseer,hofer

More information

Figure 1: Schematic of an SEM [source: ]

Figure 1: Schematic of an SEM [source:   ] EECI Course: -9 May 1 by R. Sanfelice Hybri Control Systems Eelco van Horssen E.P.v.Horssen@tue.nl Project: Scanning Electron Microscopy Introuction In Scanning Electron Microscopy (SEM) a (bunle) beam

More information

Characterizing Decoding Robustness under Parametric Channel Uncertainty

Characterizing Decoding Robustness under Parametric Channel Uncertainty Characterizing Decoing Robustness uner Parametric Channel Uncertainty Jay D. Wierer, Wahee U. Bajwa, Nigel Boston, an Robert D. Nowak Abstract This paper characterizes the robustness of ecoing uner parametric

More information

Multimodal Stereo Image Registration for Pedestrian Detection

Multimodal Stereo Image Registration for Pedestrian Detection Multimoal Stereo Image Registration for Peestrian Detection Stephen Krotosky an Mohan Trivei Abstract This paper presents an approach for the registration of multimoal imagery for peestrian etection when

More information

Feature Extraction and Rule Classification Algorithm of Digital Mammography based on Rough Set Theory

Feature Extraction and Rule Classification Algorithm of Digital Mammography based on Rough Set Theory Feature Extraction an Rule Classification Algorithm of Digital Mammography base on Rough Set Theory Aboul Ella Hassanien Jafar M. H. Ali. Kuwait University, Faculty of Aministrative Science, Quantitative

More information

Research Article Inviscid Uniform Shear Flow past a Smooth Concave Body

Research Article Inviscid Uniform Shear Flow past a Smooth Concave Body International Engineering Mathematics Volume 04, Article ID 46593, 7 pages http://x.oi.org/0.55/04/46593 Research Article Invisci Uniform Shear Flow past a Smooth Concave Boy Abullah Mura Department of

More information

Rough Set Approach for Classification of Breast Cancer Mammogram Images

Rough Set Approach for Classification of Breast Cancer Mammogram Images Rough Set Approach for Classification of Breast Cancer Mammogram Images Aboul Ella Hassanien Jafar M. H. Ali. Kuwait University, Faculty of Aministrative Science, Quantitative Methos an Information Systems

More information

Animated Surface Pasting

Animated Surface Pasting Animate Surface Pasting Clara Tsang an Stephen Mann Computing Science Department University of Waterloo 200 University Ave W. Waterloo, Ontario Canaa N2L 3G1 e-mail: clftsang@cgl.uwaterloo.ca, smann@cgl.uwaterloo.ca

More information

Cluster Center Initialization Method for K-means Algorithm Over Data Sets with Two Clusters

Cluster Center Initialization Method for K-means Algorithm Over Data Sets with Two Clusters Available online at www.scienceirect.com Proceia Engineering 4 (011 ) 34 38 011 International Conference on Avances in Engineering Cluster Center Initialization Metho for K-means Algorithm Over Data Sets

More information

filtering LETTER An Improved Neighbor Selection Algorithm in Collaborative Taek-Hun KIM a), Student Member and Sung-Bong YANG b), Nonmember

filtering LETTER An Improved Neighbor Selection Algorithm in Collaborative Taek-Hun KIM a), Student Member and Sung-Bong YANG b), Nonmember 107 IEICE TRANS INF & SYST, VOLE88 D, NO5 MAY 005 LETTER An Improve Neighbor Selection Algorithm in Collaborative Filtering Taek-Hun KIM a), Stuent Member an Sung-Bong YANG b), Nonmember SUMMARY Nowaays,

More information

Real Time On Board Stereo Camera Pose through Image Registration*

Real Time On Board Stereo Camera Pose through Image Registration* 28 IEEE Intelligent Vehicles Symposium Einhoven University of Technology Einhoven, The Netherlans, June 4-6, 28 Real Time On Boar Stereo Camera Pose through Image Registration* Fai Dornaika French National

More information

A multiple wavelength unwrapping algorithm for digital fringe profilometry based on spatial shift estimation

A multiple wavelength unwrapping algorithm for digital fringe profilometry based on spatial shift estimation University of Wollongong Research Online Faculty of Engineering an Information Sciences - Papers: Part A Faculty of Engineering an Information Sciences 214 A multiple wavelength unwrapping algorithm for

More information

Classical Mechanics Examples (Lagrange Multipliers)

Classical Mechanics Examples (Lagrange Multipliers) Classical Mechanics Examples (Lagrange Multipliers) Dipan Kumar Ghosh Physics Department, Inian Institute of Technology Bombay Powai, Mumbai 400076 September 3, 015 1 Introuction We have seen that the

More information

A Neural Network Model Based on Graph Matching and Annealing :Application to Hand-Written Digits Recognition

A Neural Network Model Based on Graph Matching and Annealing :Application to Hand-Written Digits Recognition ITERATIOAL JOURAL OF MATHEMATICS AD COMPUTERS I SIMULATIO A eural etwork Moel Base on Graph Matching an Annealing :Application to Han-Written Digits Recognition Kyunghee Lee Abstract We present a neural

More information

Object Recognition Using Colour, Shape and Affine Invariant Ratios

Object Recognition Using Colour, Shape and Affine Invariant Ratios Object Recognition Using Colour, Shape an Affine Invariant Ratios Paul A. Walcott Centre for Information Engineering City University, Lonon EC1V 0HB, Englan P.A.Walcott@city.ac.uk Abstract This paper escribes

More information

THE BAYESIAN RECEIVER OPERATING CHARACTERISTIC CURVE AN EFFECTIVE APPROACH TO EVALUATE THE IDS PERFORMANCE

THE BAYESIAN RECEIVER OPERATING CHARACTERISTIC CURVE AN EFFECTIVE APPROACH TO EVALUATE THE IDS PERFORMANCE БСУ Международна конференция - 2 THE BAYESIAN RECEIVER OPERATING CHARACTERISTIC CURVE AN EFFECTIVE APPROACH TO EVALUATE THE IDS PERFORMANCE Evgeniya Nikolova, Veselina Jecheva Burgas Free University Abstract:

More information

Yet Another Parallel Hypothesis Search for Inverse Entailment Hiroyuki Nishiyama and Hayato Ohwada Faculty of Sci. and Tech. Tokyo University of Scien

Yet Another Parallel Hypothesis Search for Inverse Entailment Hiroyuki Nishiyama and Hayato Ohwada Faculty of Sci. and Tech. Tokyo University of Scien Yet Another Parallel Hypothesis Search for Inverse Entailment Hiroyuki Nishiyama an Hayato Ohwaa Faculty of Sci. an Tech. Tokyo University of Science, 2641 Yamazaki, Noa-shi, CHIBA, 278-8510, Japan hiroyuki@rs.noa.tus.ac.jp,

More information

Zero Disparity Filter Based on Wavelet Representation in an Active Vision System

Zero Disparity Filter Based on Wavelet Representation in an Active Vision System IEEE ICSP'96, Beiing, China Huang Yu etal Zero isparity Filter Base on Wavelet Representation in an Active Vision System Huang Yu, Yuan Baozong Institute o Inormation Science, Northern Jiaotong University,

More information

Dense 3D Reconstruction. Christiano Gava

Dense 3D Reconstruction. Christiano Gava Dense 3D Reconstruction Christiano Gava christiano.gava@dfki.de Outline Previous lecture: structure and motion II Structure and motion loop Triangulation Today: dense 3D reconstruction The matching problem

More information

Classifying Facial Expression with Radial Basis Function Networks, using Gradient Descent and K-means

Classifying Facial Expression with Radial Basis Function Networks, using Gradient Descent and K-means Classifying Facial Expression with Raial Basis Function Networks, using Graient Descent an K-means Neil Allrin Department of Computer Science University of California, San Diego La Jolla, CA 9237 nallrin@cs.ucs.eu

More information

PDE-BASED INTERPOLATION METHOD FOR OPTICALLY VISUALIZED SOUND FIELD. Kohei Yatabe and Yasuhiro Oikawa

PDE-BASED INTERPOLATION METHOD FOR OPTICALLY VISUALIZED SOUND FIELD. Kohei Yatabe and Yasuhiro Oikawa 4 IEEE International Conference on Acoustic, Speech an Signal Processing (ICASSP) PDE-BASED INTERPOLATION METHOD FOR OPTICALLY VISUALIZED SOUND FIELD Kohei Yatabe an Yasuhiro Oikawa Department of Intermeia

More information

Exploring Context with Deep Structured models for Semantic Segmentation

Exploring Context with Deep Structured models for Semantic Segmentation 1 Exploring Context with Deep Structure moels for Semantic Segmentation Guosheng Lin, Chunhua Shen, Anton van en Hengel, Ian Rei between an image patch an a large backgroun image region. Explicitly moeling

More information

STEREOSCOPIC ROBOT VISION SYSTEM

STEREOSCOPIC ROBOT VISION SYSTEM Palinko Oskar, ipl. eng. Facult of Technical Sciences, Department of Inustrial Sstems Engineering an Management, 21 000 Novi Sa, Dositej Obraovic Square 6, Serbia & Montenegro STEREOSCOPIC ROBOT VISION

More information

CAMERAS AND GRAVITY: ESTIMATING PLANAR OBJECT ORIENTATION. Zhaoyin Jia, Andrew Gallagher, Tsuhan Chen

CAMERAS AND GRAVITY: ESTIMATING PLANAR OBJECT ORIENTATION. Zhaoyin Jia, Andrew Gallagher, Tsuhan Chen CAMERAS AND GRAVITY: ESTIMATING PLANAR OBJECT ORIENTATION Zhaoyin Jia, Anrew Gallagher, Tsuhan Chen School of Electrical an Computer Engineering, Cornell University ABSTRACT Photography on a mobile camera

More information

A Versatile Model-Based Visibility Measure for Geometric Primitives

A Versatile Model-Based Visibility Measure for Geometric Primitives A Versatile Moel-Base Visibility Measure for Geometric Primitives Marc M. Ellenrieer 1,LarsKrüger 1, Dirk Stößel 2, an Marc Hanheie 2 1 DaimlerChrysler AG, Research & Technology, 89013 Ulm, Germany 2 Faculty

More information

Evolutionary Optimisation Methods for Template Based Image Registration

Evolutionary Optimisation Methods for Template Based Image Registration Evolutionary Optimisation Methos for Template Base Image Registration Lukasz A Machowski, Tshilizi Marwala School of Electrical an Information Engineering University of Witwatersran, Johannesburg, South

More information

Bends, Jogs, And Wiggles for Railroad Tracks and Vehicle Guide Ways

Bends, Jogs, And Wiggles for Railroad Tracks and Vehicle Guide Ways Ben, Jogs, An Wiggles for Railroa Tracks an Vehicle Guie Ways Louis T. Klauer Jr., PhD, PE. Work Soft 833 Galer Dr. Newtown Square, PA 19073 lklauer@wsof.com Preprint, June 4, 00 Copyright 00 by Louis

More information

AN INVESTIGATION OF FOCUSING AND ANGULAR TECHNIQUES FOR VOLUMETRIC IMAGES BY USING THE 2D CIRCULAR ULTRASONIC PHASED ARRAY

AN INVESTIGATION OF FOCUSING AND ANGULAR TECHNIQUES FOR VOLUMETRIC IMAGES BY USING THE 2D CIRCULAR ULTRASONIC PHASED ARRAY AN INVESTIGATION OF FOCUSING AND ANGULAR TECHNIQUES FOR VOLUMETRIC IMAGES BY USING THE D CIRCULAR ULTRASONIC PHASED ARRAY S. Monal Lonon South Bank University; Engineering an Design 103 Borough Roa, Lonon

More information

Text Particles Multi-band Fusion for Robust Text Detection

Text Particles Multi-band Fusion for Robust Text Detection Text Particles Multi-ban Fusion for Robust Text Detection Pengfei Xu, Rongrong Ji, Hongxun Yao, Xiaoshuai Sun, Tianqiang Liu, an Xianming Liu School of Computer Science an Engineering Harbin Institute

More information

A Comparative Evaluation of Iris and Ocular Recognition Methods on Challenging Ocular Images

A Comparative Evaluation of Iris and Ocular Recognition Methods on Challenging Ocular Images A Comparative Evaluation of Iris an Ocular Recognition Methos on Challenging Ocular Images Vishnu Naresh Boeti Carnegie Mellon University Pittsburgh, PA 523 naresh@cmu.eu Jonathon M Smereka Carnegie Mellon

More information

Learning convex bodies is hard

Learning convex bodies is hard Learning convex boies is har Navin Goyal Microsoft Research Inia navingo@microsoftcom Luis Raemacher Georgia Tech lraemac@ccgatecheu Abstract We show that learning a convex boy in R, given ranom samples

More information

Tight Wavelet Frame Decomposition and Its Application in Image Processing

Tight Wavelet Frame Decomposition and Its Application in Image Processing ITB J. Sci. Vol. 40 A, No., 008, 151-165 151 Tight Wavelet Frame Decomposition an Its Application in Image Processing Mahmu Yunus 1, & Henra Gunawan 1 1 Analysis an Geometry Group, FMIPA ITB, Banung Department

More information

Computer Organization

Computer Organization Computer Organization Douglas Comer Computer Science Department Purue University 250 N. University Street West Lafayette, IN 47907-2066 http://www.cs.purue.eu/people/comer Copyright 2006. All rights reserve.

More information

Sampling the Disparity Space Image

Sampling the Disparity Space Image Accepte for publication, June 2003. To appear in IEEE Transactions on Pattern Analysis an Machine Intelligence (PAMI) Sampling the Disparity Space Image Richar Szeliski Daniel Scharstein Microsoft Research

More information

EE795: Computer Vision and Intelligent Systems

EE795: Computer Vision and Intelligent Systems EE795: Computer Vision and Intelligent Systems Spring 2012 TTh 17:30-18:45 FDH 204 Lecture 14 130307 http://www.ee.unlv.edu/~b1morris/ecg795/ 2 Outline Review Stereo Dense Motion Estimation Translational

More information

[2006] IEEE. Reprinted, with permission, from [Damith C. Herath, Sarath Kodagoda and Gamini Dissanayake, Simultaneous Localisation and Mapping: A

[2006] IEEE. Reprinted, with permission, from [Damith C. Herath, Sarath Kodagoda and Gamini Dissanayake, Simultaneous Localisation and Mapping: A [6] IEEE. Reprinte, with permission, from [Damith C. Herath, Sarath Koagoa an Gamini Dissanayake, Simultaneous Localisation an Mapping: A Stereo Vision Base Approach, Intelligent Robots an Systems, 6 IEEE/RSJ

More information

Calculation on diffraction aperture of cube corner retroreflector

Calculation on diffraction aperture of cube corner retroreflector November 10, 008 / Vol., No. 11 / CHINESE OPTICS LETTERS 8 Calculation on iffraction aperture of cube corner retroreflector Song Li (Ó Ø, Bei Tang (», an Hui Zhou ( ï School of Electronic Information,

More information

A Summary of Projective Geometry

A Summary of Projective Geometry A Summary of Projective Geometry Copyright 22 Acuity Technologies Inc. In the last years a unified approach to creating D models from multiple images has been developed by Beardsley[],Hartley[4,5,9],Torr[,6]

More information

The Reconstruction of Graphs. Dhananjay P. Mehendale Sir Parashurambhau College, Tilak Road, Pune , India. Abstract

The Reconstruction of Graphs. Dhananjay P. Mehendale Sir Parashurambhau College, Tilak Road, Pune , India. Abstract The Reconstruction of Graphs Dhananay P. Mehenale Sir Parashurambhau College, Tila Roa, Pune-4030, Inia. Abstract In this paper we iscuss reconstruction problems for graphs. We evelop some new ieas lie

More information

EXACT SIMULATION OF A BOOLEAN MODEL

EXACT SIMULATION OF A BOOLEAN MODEL Original Research Paper oi:10.5566/ias.v32.p101-105 EXACT SIMULATION OF A BOOLEAN MODEL CHRISTIAN LANTUÉJOULB MinesParisTech 35 rue Saint-Honoré 77305 Fontainebleau France e-mail: christian.lantuejoul@mines-paristech.fr

More information

Coupling the User Interfaces of a Multiuser Program

Coupling the User Interfaces of a Multiuser Program Coupling the User Interfaces of a Multiuser Program PRASUN DEWAN University of North Carolina at Chapel Hill RAJIV CHOUDHARY Intel Corporation We have evelope a new moel for coupling the user-interfaces

More information

4.2 Implicit Differentiation

4.2 Implicit Differentiation 6 Chapter 4 More Derivatives 4. Implicit Differentiation What ou will learn about... Implicitl Define Functions Lenses, Tangents, an Normal Lines Derivatives of Higher Orer Rational Powers of Differentiable

More information

Chapter 5 Proposed models for reconstituting/ adapting three stereoscopes

Chapter 5 Proposed models for reconstituting/ adapting three stereoscopes Chapter 5 Propose moels for reconstituting/ aapting three stereoscopes - 89 - 5. Propose moels for reconstituting/aapting three stereoscopes This chapter offers three contributions in the Stereoscopy area,

More information

More Raster Line Issues. Bresenham Circles. Once More: 8-Pt Symmetry. Only 1 Octant Needed. Spring 2013 CS5600

More Raster Line Issues. Bresenham Circles. Once More: 8-Pt Symmetry. Only 1 Octant Needed. Spring 2013 CS5600 Spring 03 Lecture Set 3 Bresenham Circles Intro to Computer Graphics From Rich Riesenfel Spring 03 More Raster Line Issues Fat lines with multiple pixel with Symmetric lines n point geometry how shoul

More information

MOTION IDENTIFICATION OF PLANAR FREE-FORM OBJECTS

MOTION IDENTIFICATION OF PLANAR FREE-FORM OBJECTS MOTION IDENTIFICATION OF PLANAR FREE-FORM OBJECTS Mustafa Unel Faculty of Engineering an Natural Sciences Sabanci University Orhanli-Tuzla 34956, Istanbul, Turkey email: munel@sabanciuniv.eu ABSTRACT It

More information

Using Vector and Raster-Based Techniques in Categorical Map Generalization

Using Vector and Raster-Based Techniques in Categorical Map Generalization Thir ICA Workshop on Progress in Automate Map Generalization, Ottawa, 12-14 August 1999 1 Using Vector an Raster-Base Techniques in Categorical Map Generalization Beat Peter an Robert Weibel Department

More information

On Effectively Determining the Downlink-to-uplink Sub-frame Width Ratio for Mobile WiMAX Networks Using Spline Extrapolation

On Effectively Determining the Downlink-to-uplink Sub-frame Width Ratio for Mobile WiMAX Networks Using Spline Extrapolation On Effectively Determining the Downlink-to-uplink Sub-frame With Ratio for Mobile WiMAX Networks Using Spline Extrapolation Panagiotis Sarigianniis, Member, IEEE, Member Malamati Louta, Member, IEEE, Member

More information

Dense 3D Reconstruction. Christiano Gava

Dense 3D Reconstruction. Christiano Gava Dense 3D Reconstruction Christiano Gava christiano.gava@dfki.de Outline Previous lecture: structure and motion II Structure and motion loop Triangulation Wide baseline matching (SIFT) Today: dense 3D reconstruction

More information

Optimization Beyond the Convolution: Generalizing Spatial Relations with End-to-End Metric Learning

Optimization Beyond the Convolution: Generalizing Spatial Relations with End-to-End Metric Learning Optimization Beyon the Convolution: Generalizing Spatial Relations with En-to-En Metric Learning Philipp Jun, Anreas Eitel, Nichola Abo an Wolfram Burgar 1 Abstract To operate intelligently in omestic

More information

EXTRACTION OF MAIN AND SECONDARY ROADS IN VHR IMAGES USING A HIGHER-ORDER PHASE FIELD MODEL

EXTRACTION OF MAIN AND SECONDARY ROADS IN VHR IMAGES USING A HIGHER-ORDER PHASE FIELD MODEL EXTRACTION OF MAIN AND SECONDARY ROADS IN VHR IMAGES USING A HIGHER-ORDER PHASE FIELD MODEL Ting Peng a,b, Ian H. Jermyn b, Véronique Prinet a, Josiane Zerubia b a LIAMA & PR, Institute of Automation,

More information

6 Gradient Descent. 6.1 Functions

6 Gradient Descent. 6.1 Functions 6 Graient Descent In this topic we will iscuss optimizing over general functions f. Typically the function is efine f : R! R; that is its omain is multi-imensional (in this case -imensional) an output

More information

Fast Fractal Image Compression using PSO Based Optimization Techniques

Fast Fractal Image Compression using PSO Based Optimization Techniques Fast Fractal Compression using PSO Base Optimization Techniques A.Krishnamoorthy Visiting faculty Department Of ECE University College of Engineering panruti rishpci89@gmail.com S.Buvaneswari Visiting

More information

BIJECTIONS FOR PLANAR MAPS WITH BOUNDARIES

BIJECTIONS FOR PLANAR MAPS WITH BOUNDARIES BIJECTIONS FOR PLANAR MAPS WITH BOUNDARIES OLIVIER BERNARDI AND ÉRIC FUSY Abstract. We present bijections for planar maps with bounaries. In particular, we obtain bijections for triangulations an quarangulations

More information

Local Path Planning with Proximity Sensing for Robot Arm Manipulators. 1. Introduction

Local Path Planning with Proximity Sensing for Robot Arm Manipulators. 1. Introduction Local Path Planning with Proximity Sensing for Robot Arm Manipulators Ewar Cheung an Vlaimir Lumelsky Yale University, Center for Systems Science Department of Electrical Engineering New Haven, Connecticut

More information

Exploring Context with Deep Structured models for Semantic Segmentation

Exploring Context with Deep Structured models for Semantic Segmentation APPEARING IN IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE, APRIL 2017. 1 Exploring Context with Deep Structure moels for Semantic Segmentation Guosheng Lin, Chunhua Shen, Anton van en

More information

Digital fringe profilometry based on triangular fringe patterns and spatial shift estimation

Digital fringe profilometry based on triangular fringe patterns and spatial shift estimation University of Wollongong Research Online Faculty of Engineering an Information Sciences - Papers: Part A Faculty of Engineering an Information Sciences 4 Digital fringe profilometry base on triangular

More information

Research Article Research on Law s Mask Texture Analysis System Reliability

Research Article Research on Law s Mask Texture Analysis System Reliability Research Journal of Applie Sciences, Engineering an Technology 7(19): 4002-4007, 2014 DOI:10.19026/rjaset.7.761 ISSN: 2040-7459; e-issn: 2040-7467 2014 Maxwell Scientific Publication Corp. Submitte: November

More information

Implicit and Explicit Functions

Implicit and Explicit Functions 60_005.q //0 :5 PM Page SECTION.5 Implicit Differentiation Section.5 EXPLORATION Graphing an Implicit Equation How coul ou use a graphing utilit to sketch the graph of the equation? Here are two possible

More information

1 Surprises in high dimensions

1 Surprises in high dimensions 1 Surprises in high imensions Our intuition about space is base on two an three imensions an can often be misleaing in high imensions. It is instructive to analyze the shape an properties of some basic

More information

Generalized Edge Coloring for Channel Assignment in Wireless Networks

Generalized Edge Coloring for Channel Assignment in Wireless Networks Generalize Ege Coloring for Channel Assignment in Wireless Networks Chun-Chen Hsu Institute of Information Science Acaemia Sinica Taipei, Taiwan Da-wei Wang Jan-Jan Wu Institute of Information Science

More information

FINDING OPTICAL DISPERSION OF A PRISM WITH APPLICATION OF MINIMUM DEVIATION ANGLE MEASUREMENT METHOD

FINDING OPTICAL DISPERSION OF A PRISM WITH APPLICATION OF MINIMUM DEVIATION ANGLE MEASUREMENT METHOD Warsaw University of Technology Faculty of Physics Physics Laboratory I P Joanna Konwerska-Hrabowska 6 FINDING OPTICAL DISPERSION OF A PRISM WITH APPLICATION OF MINIMUM DEVIATION ANGLE MEASUREMENT METHOD.

More information

Improving Spatial Reuse of IEEE Based Ad Hoc Networks

Improving Spatial Reuse of IEEE Based Ad Hoc Networks mproving Spatial Reuse of EEE 82.11 Base A Hoc Networks Fengji Ye, Su Yi an Biplab Sikar ECSE Department, Rensselaer Polytechnic nstitute Troy, NY 1218 Abstract n this paper, we evaluate an suggest methos

More information

Iterative Computation of Moment Forms for Subdivision Surfaces

Iterative Computation of Moment Forms for Subdivision Surfaces Iterative Computation of Moment Forms for Subivision Surfaces Jan P. Hakenberg ETH Zürich Ulrich Reif TU Darmstat Figure : Solis boune by Catmull-Clark an Loop subivision surfaces with centrois inicate

More information

An Algorithm for Building an Enterprise Network Topology Using Widespread Data Sources

An Algorithm for Building an Enterprise Network Topology Using Widespread Data Sources An Algorithm for Builing an Enterprise Network Topology Using Wiesprea Data Sources Anton Anreev, Iurii Bogoiavlenskii Petrozavosk State University Petrozavosk, Russia {anreev, ybgv}@cs.petrsu.ru Abstract

More information

A Plane Tracker for AEC-automation Applications

A Plane Tracker for AEC-automation Applications A Plane Tracker for AEC-automation Applications Chen Feng *, an Vineet R. Kamat Department of Civil an Environmental Engineering, University of Michigan, Ann Arbor, USA * Corresponing author (cforrest@umich.eu)

More information

Synthesis Distortion Estimation in 3D Video Using Frequency and Spatial Analysis

Synthesis Distortion Estimation in 3D Video Using Frequency and Spatial Analysis MITSUBISHI EECTRIC RESEARCH ABORATORIES http://www.merl.com Synthesis Distortion Estimation in 3D Vieo Using Frequency an Spatial Analysis Fang,.; Cheung, N-M; Tian, D.; Vetro, A.; Sun, H.; Yu,. TR2013-087

More information

2.7 Implicit Differentiation

2.7 Implicit Differentiation 2.7 Implicit Differentiation [ y] = = = [ x] 1 Sometimes we may be intereste in fining the erivative of an equation that is not solve or able to be solve for a particular epenent variable explicitly. In

More information

Using Geometric Blur for Point Correspondence

Using Geometric Blur for Point Correspondence 1 Using Geometric Blur for Point Correspondence Nisarg Vyas Electrical and Computer Engineering Department, Carnegie Mellon University, Pittsburgh, PA Abstract In computer vision applications, point correspondence

More information

Here are a couple of warnings to my students who may be here to get a copy of what happened on a day that you missed.

Here are a couple of warnings to my students who may be here to get a copy of what happened on a day that you missed. Preface Here are my online notes for my Calculus I course that I teach here at Lamar University. Despite the fact that these are my class notes, they shoul be accessible to anyone wanting to learn Calculus

More information

Message Transport With The User Datagram Protocol

Message Transport With The User Datagram Protocol Message Transport With The User Datagram Protocol User Datagram Protocol (UDP) Use During startup For VoIP an some vieo applications Accounts for less than 10% of Internet traffic Blocke by some ISPs Computer

More information

Regularized Laplacian Zero Crossings as Optimal Edge Integrators

Regularized Laplacian Zero Crossings as Optimal Edge Integrators Regularize aplacian Zero rossings as Optimal Ege Integrators R. KIMME A.M. BRUKSTEIN Department of omputer Science Technion Israel Institute of Technology Technion ity, Haifa 32, Israel Abstract We view

More information

Segmentation and Tracking of Partial Planar Templates

Segmentation and Tracking of Partial Planar Templates Segmentation and Tracking of Partial Planar Templates Abdelsalam Masoud William Hoff Colorado School of Mines Colorado School of Mines Golden, CO 800 Golden, CO 800 amasoud@mines.edu whoff@mines.edu Abstract

More information