+

WO2001078007A2 - Procede de traitement d'une image - Google Patents

Procede de traitement d'une image Download PDF

Info

Publication number
WO2001078007A2
WO2001078007A2 PCT/IL2001/000321 IL0100321W WO0178007A2 WO 2001078007 A2 WO2001078007 A2 WO 2001078007A2 IL 0100321 W IL0100321 W IL 0100321W WO 0178007 A2 WO0178007 A2 WO 0178007A2
Authority
WO
WIPO (PCT)
Prior art keywords
image
points
pixel set
continuous pixel
calculating
Prior art date
Application number
PCT/IL2001/000321
Other languages
English (en)
Other versions
WO2001078007A3 (fr
Inventor
Ornit Maimon
Shmuel Kiro
Original Assignee
Ioimage Ltd.
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Ioimage Ltd. filed Critical Ioimage Ltd.
Priority to AU2001250600A priority Critical patent/AU2001250600A1/en
Publication of WO2001078007A2 publication Critical patent/WO2001078007A2/fr
Publication of WO2001078007A3 publication Critical patent/WO2001078007A3/fr

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/20—Analysis of motion
    • G06T7/246—Analysis of motion using feature-based methods, e.g. the tracking of corners or segments

Definitions

  • This invention relates to methods of processing a digitized image.
  • ⁇ n image may be digitized by dividing it into a large number of small regions referred to as pixels.
  • the location of each pixel in the image is described by a two-dimensional vector (k. 1) having integral coordinates.
  • the pixel w hose location is (k, 1) is simply referred to as the pixel (k. 1).
  • the set of ⁇ > pixels (k. I ) forms a discrete set in the set of ordered pairs of real numbers.
  • R " . ⁇ digital image is described by one or more color functions / ' defined on the set of pixels.
  • ⁇ , ⁇ ) is the intensity of a particular color at the pixel (k.l) in the image.
  • Movement of the camera is described by means of a tracking transformation T that maps entities in one image into the same entities in the second image.
  • T is then defined on the reference points by manually assigning to each reference point a corresponding pixel in the second image based upon the visual inspection of the two images.
  • the value of T on other pixels in the first image is then calculated by interpolation of the manually defined values of T on the set of reference pixels, for example, as described in Amar Mitchie, 1994, pages 45-80. and in Klette et al. 1998, pages 129-175.
  • the need for visual inspection of the two images and manual selection of pixels makes this method tedious and slow.
  • camera stabilization refers to rectification of a sequence of images obtained over time in order to compensate for unwanted camera movement. In principal, this may be accomplished by first applying the process of camera tracking as just described to sequential pairs of images in the sequence. This would generate a sequence of tracking transfo ⁇ ations T Desi(,x,y), where T Tin(x,y) is a tracking transformation from the n 1 image in the sequence obtained at time t n to the (n+ 1 )" image obtained at time t, 1+
  • the transformations T fur could then be replaced with new tracking transformations that vary smoothly in time and that could thus be used to rectify the sequence of images. In reality, however, the need for visual inspection of the images and manual selection of pixels for each sequential pair of images makes camera tracking in a long sequence impractical.
  • I he present invention provides a method for processing a digital image not requiring visual inspection of the image.
  • the invention provides a method for performing camera tracking.
  • Camera tracking in accordance with this aspect of the invention, may be used for camera stabilization in a time sequence of i s images.
  • the invention provides a method of camera stabilization that does not require visual inspection of the images.
  • the domain of /u is a set Ku of vectors (x.y) in R ⁇ containing (k.l ) and at least one other pixel different from (k,l).
  • Ku may be for example, a 0 square with (k.l) at its center and having an edge length of n pixels.
  • K contains (n+l ) ⁇ pixel points including (k,l).
  • the functions /u are preferably continuous functions, and are most preferably differentiable functions in each variable.
  • an object referred to herein as an "edge".
  • an edge is an elongated region in the image along which there is an abrupt change in the intensity of the color.
  • intersections are used as objects.
  • node an object, referred to herein as a "node " is used.
  • ⁇ node is defined as two intersecting edges together with the point of their intersection.
  • a corner consists of three points, and is constructed as follows. Two edge points are found that are near to each other (e.g., the distance between them not exceeding 1.5 pixels) for which the gradients of the functions i defined at each of the two points are nearly perpendicular. The third point is the one found at the intersection of the two gradient vectors.
  • a ridge is an elongated region along which the color intensity has a local maximum or minimum.
  • Camera tracking in accordance with the invention involves identifying objects of the same type in two images. A tracking transformation is then calculated that associates objects in one image with objects in the second image.
  • the mapping preferably takes into account one or more properties of the objects: I . The location of the objects in the continuous pixel space.
  • Camera stabilization of a sequence I Anlagen of a time sequence of images involves calculating a tracking transformation for each pair of consecutive images in the sequence. This generates a sequence of tracking transformations T Titan(,x,y).
  • T effet(x,y) is a tracking transformation from the n 1 1 image l n in the sequence obtained at time t n to the (n+1) 1 image I n+ ⁇ obtained at time t perennial i .
  • the transformations T n are then replaced in a process of regression with new tracking transformations that vary smoothly in time.
  • the new tracking transformations are then used to produce a new sequence of images. Since the process may be performed automatically, it may be applied to long sequences of images. 5
  • the invention thus provides a method for processing a digital image, the image having a color function /defined on a finite set A of points in R , the method comprising the steps of:
  • the invention provides a method for camera tracking between first and second digital images, the first and second images having color f unctions and /. respectively, defined on a finite set A of points in R .
  • the 0 method comprising the steps of:
  • the invention provides a computer program product comprising a computer useable medium having computer readable program code embodied therein for processing a digital image, the image having a color function /defined on a finite set A of points in R .
  • the computer program product comprising:
  • the invention provides a program storage device readable by machine, tangibly embodying a program of instructions executable by the machine to perform method steps for processing a digital image, the image having a color function / defined on a finite set A of points in R " , the method comprising the steps of:
  • the invention provides a program storage device readable by machine, tangibly embodying a program of instmctions executable by the machine to perform method steps for camera tracking between first and second digital images, the first and second images having color functions / and /,
  • the invention provides a computer program product comprising a computer useable medium having computer readable program code embodied therein for camera tracking between first and second digital images, the first and second images having color functions/ and /, respectively, defined on a finite set A of points in R ⁇ the computer program product comprising computer readable program code for causing the computer to:
  • (aaad) computer readable program code for causing the computer to calculate one or more objects based upon the one or more functions fu, the objects being regions in the continuous pixel set of the image;
  • FIG. 1 is a flow-chart diagram for calculating an edge in a digital image
  • Fig. 2 is a flow-chart diagram for calculating a ridge in a digital image
  • Fig. 3 is a tabular representation of a digital image
  • Fig. 4 shows edge points, provisional edges, and edges in the continuous pixel space of the image of Fig. 3;
  • ⁇ o Fig. 5 is a flow-chart diagram for calculating an intersection in a digital image
  • Fig. 6 is a flow-chart diagram for calculating a node in a digital image
  • Fig. 7 is a flow-chart diagram for calculating a corner in a digital image
  • Fig. 8a is a flow-chart diagram for camera tracking
  • i s Fig. 8b is another flow chart diagram for camera tracking
  • Fig. 9 is a flow-chart diagram for camera stabilization.
  • the present invention relates to a process of processing a digital image.
  • the image is represented by a color function /defined over a discrete set of points in
  • K k. i may be for example, a square with (k,l) at its center and having an edge length of n pixels. In this case. K contains (n+l ) ⁇ pixel points including (k,l).
  • 25 functions /u are preferably continuous functions, and are most preferably differentiable functions in each variable.
  • the functions fu. may be, for example, second order polynomial approximations detemiined by constraints fu —f at all pixel points in Ku including (k.l).
  • the union of the sets Ku fom is the "continuous pixel sei " .
  • an object referred to herein as an "edge " , is used.
  • An edge is an elongated region in an image along which there is an abrupt change in the intensity of a color described by an abrupt change in the values of the corresponding color function/
  • Fig. I shows a flow chart diagram of a method for finding an edge in an image
  • Vfu dx dv
  • the predetemiined constant may be. for example. 200.
  • This set is referred to as the first provisional set of edge points (step 110).
  • Points are deleted from the first provisional set of edge points to produce the second set of provisional edge points by the following process. For each point (k.l) in the first provisional set. a line Lk.i is formed in the continuous pixel space passing through the point (k.l) and having the direction of the gradient V/. ⁇ (k,l). A neighborhood Nu of (k.l) is calculated containing at least one pixel point in addition to (k,l).
  • a function Pu(x,y) is calculated, defined for all real vectors (x,y) lying on the line Lu in Nu, that is an approximation to the color function/along this line (step 115).
  • (k.l) is not located near the midpoint along Lu between the maximum and the minimum of P . ⁇ (x.y)-P . ⁇ (k.l) in Nk.i (for example, if the difference in the distances exceeds 0.5 pixels), then (k,l) is also not included in the second provisional set of edge points.
  • the second provisional set of edge points consists of all the points in the first set that have not been excluded by this process (step 120).
  • the set of edge points is fo ⁇ ried by replacing each point in the second provisional set with the point (x, y) lying on the line Lu at the midpoint between the positions of the maximum and minimum of Pu(x,y)-Pu(k,l) in Nu (step 125).
  • An edge point may or may not have integral l o coordinates.
  • the distance along the line Lu between the positions of the maximum and the minimum is referred to herein as the "width " of the edge at (x,y).
  • the width of each edge point is calculated.
  • a provisional edge is fo ⁇ rted as follows.
  • a first edge point is selected (step 135).
  • a second edge point is sought near the first edge point (for i s example, within a distance of 1.5 pixels) such that the change in the direction of the gradient between the points is small, and the change in color (the function Pu) is also small (step 140). If such an edge point is found, the process returns to step 135 with the new edge point being selected. If such an edge point does not exist, the number of edge points o selected is determined (step 145). If the number of edge points selected is less than a predetermined number, e.g. 2, then a new edge point is sought that has not yet been selected (step 155). If such a new edge point exists, all previously selected edge points are deselected (step 160) and the process returns to step 135 with the new edge point being selected. If
  • the process te ⁇ rtinates at step 165. If at step 145 it is found that the number of selected edge points is at least the predetermined number, then the subset of the set of edge points calculated by this process is referred to as a "provisional edge". 5.
  • the provisional edge is used to define an edge that is a continuous, sectionally smooth, curve in R . For example, three adjacent points in the provisional edge may be joined by a parabola, and the edge defined as the spline calculated by the parabolas. The process then 5 terminates in step 165.
  • intersections an object is used referred to herein as an "intersection".
  • a tangent is drawn at an endpoint of each of two edges and the two tangents intersect substantially perpendicularly to each other, the two edges are referred to as ⁇ o "intersecting edges " and the point of intersection of the two tangents is referred to as the intersection of the two edges.
  • intersections are used as objects.
  • substantially perpendicular is meant, for example, that the angle between the tangents is 90° ⁇ 30°.
  • Fig. 5 shows a flow chart diagram of a process for calculating an 15 intersection.
  • step 510 two edges in the image are selected. A tangent is then drawn at an endpoint of each of the two edges (step 515). It is then determined whether the two tangents are substantially perpendicular (step 520). If no, the process terminates in step 525. If yes, the intersection of the two tangents is calculated (step 530) which is the intersection. The process then terminates at step 20 535.
  • node an object is used referred to herein as a "node ".
  • a node is defined as two intersecting edges together with the associated intersection.
  • Fig. 6 shows a flow chart diagram of a process for calculating a node.
  • step 61 two edges in the image are selected. A tangent is then drawn at an endpoint of each of the two edges (step 615). It is then detemiined whether the two tangents are substantially perpendicular (step 620). If no. the process te ⁇ riinates in step 625. I f yes. the point of intersection of the two tangents is calculated (step 630). In step 635. the union of the point of intersection with the two edges is o formed, to produce the node. The process then temiinates at step 640. Fourth Embodiment: Corners
  • a corner is constructed as follows. Two edge points are found that are near to each other (e.g.. the distance between them not exceeding 1.5 pixels) for which the s gradients of the functions defined at each of the two points are nearly perpendicular. The third point is the one found at the intersection of the two gradient vectors.
  • Fig. 7 describes a process for calculating a comer.
  • two edge points located near each other are selected. It is then dete ⁇ riined whether the color ⁇ o gradients at the two points are substantially perpendicular to each other (step 715). If no. the process temiinates at step 720. If yes, the intersection point of the gradients is calculated (step 725). At step 730, the union is formed of the two edge points and the intersection points. The process then temiinates at step 735.
  • Fifth Embodiment: Ridges 15 In yet another preferred embodiment of the invention, an object, referred to herein as a "ridge ". is used.
  • a ridge is an elongated region in the continuous pixel set along which the color intensity has a relative maximum or minimum.
  • twice differentiable functions/ . ⁇ (x,y) are calculated.
  • Fig. 2 shows a flow chart diagram of a method for finding a ridge in an image.
  • the first and second predete ⁇ riined constants may be. for example. 200, and 1 . respectively. This set is referred to as the set of ridge points.
  • a provisional ridge is fo ⁇ ed as follows.
  • a first provisional ridge point 5 is selected (step 220).
  • a second ridge point is then sought near the first ridge point (for example, within a distance of 1 .5 pixels) such that the change in the direction of an eigenvector of the Hessian between the points is small, and the change in color (the function fu) is also small (step 225). If such a ridge point is found, the process returns to step 220
  • step 230 the number of ridge points that has been selected is detemiined (step 230). If the number does not exceed a predetemiined number, e.g. 2, then at step 235 a new ridge point is sought that has not yet been selected. If such a new ridge point exists, all selected ridge points are deselected in step 250 and the process i returns to step 220 with the new ridge point being selected. If at step 235 it is determined that a new ridge point that has not been selected, does not exist, then the process terminates at step 245.
  • a predetemiined number e.g. 2
  • provisional ridge is used to define a ridge that is a continuous.
  • objects are first identified in each of two images.
  • the objects are then used as reference points for defining a tracking transfomiation T between the two images.
  • T is first defined on objects in the first image.
  • Objects in the first image are mapped into objects of the same type in the second image, for example, as described in Amar Mitchie, 1994, pages 45-80. and in Klette et al, 1998, pages 129-175.
  • T preferable takes into account one or more of the following features of the objects:
  • the tracking transformation is then extended by interpolation to points in the continuous pixel space not belonging to objects, for example as disclosed in
  • the tracking transformation is preferably an affine transformation of the form Ax+b.
  • A is a 2X2 matrix and b is a two dimensional vector.
  • I f the number of objects identified in the images by the method of the invention is below a predetermined number, for example 4, reference points may be found in the images by prior art methods.
  • optical flow or warping techniques may be applied to the two images as is known in the art to produce a mapping ⁇ from the pixel set of the first image into the pixel set of the second image.
  • Points p, in the first pixel set at which a quality parameter of the mapping ⁇ is above a predetermined level are selection.
  • the objects and reference points p, together are used to define a tracking transfomiation T between the two images. T is first defined on the objects obtained by the method of the invention and on the reference points found by prior art methods.
  • Objects in the first image are mapped by T into objects of the same type in the second image as described above.
  • Point p, in the first image are mapped under T to ⁇ (pi).
  • the tracking transfomiation T is then extended by interpolation to points in the continuous pixel space.
  • Fig. 8b shows a flow chart diagram of a process for carrying out camera tracking between two images in accordance with this embodiment of the invention, s when the number of objects is below a predetemiined level.
  • steps 850 objects are identified in the two images.
  • step 855 reference points in the two images are obtained by prior art methods.
  • a tracking travel number T then defined in the step 860 mapping objects in the first image into objects of the second image and mapping one or more reference points in the first image into reference points in the ⁇ o second image.
  • the transformation T is extended to a transformation mapping point in the first image into the second image.
  • Fig. 9 shows a flow chart diagram of a process for carrying out camera tracking between two images in accordance with this embodiment of the invention.
  • objects are identified in the two images.
  • a tracking transfomiation is i s then defined in step 815 mapping objects in the first image into objects of the second image.
  • the transfomiation T is then extended in step 820 to a transformation mapping points in the first image into the second image.
  • the invention may be used for camera stabilization to rectify images for unwanted camera motion that occurred while calculating a sequence of digital images over a time interval.
  • This aspect of the invention uses tracking transformations T effet(x.y), where T n (x,y) is a tracking transformation from the n 11 image in the sequence obtained at time t n to the (n+1) 1 image obtained at time t n+ ⁇
  • each value of t. S(t. x. y) is calculated for all points (x.y) in the continuous pixel space of the sequence.
  • S may be calculated by regression of the sequence of the transformations S n (x,y) as described in Amar Mitchie, 1994, pages 45-80. and in Klette et al. 1998, pages 129-175.
  • the sequence of images is then rectified by producing the sequence of images S(t ⁇ , x, y).
  • Fig. 9 shows a flow-chart diagram for a process of camera stabilization in accordance with this embodiment of the invention.
  • a sequence of ⁇ o images I Republic is calculated for n from 1 to N.
  • a tracking transfomiation T n (x,y) is then calculated from the image Ikie to the image I Vietnamese+ ⁇ , for n from 1 to N-l(step 915).
  • , and S n+ ⁇ T n+ rS n for n from 1 to N- l (step 920).
  • the transfomiation S is then calculated based upon the S n . (step 925).
  • the sequence of transfo ⁇ ations U n is then calculated by
  • step 935 the sequence of images I', is calculated where I' n is the image whose color function / crown is calculated as /,.
  • the process terminates at step 940.
  • Example 1 f ig. 3 shows a representation of a digital image in the form of a table.
  • the table defines a color function / on the set of pixel points (k.l) where k and 1 are integers from 1 to 20.
  • k is the set of all vectors having real coordinates lying in a square having (k, 1) at its center and having an edge of length 2.
  • Each of 5 the sets Ku thus contains the 9 points (k+a. 1+b) where a and b independently take on the ⁇ alues -1. 0. and 1.
  • fu is the least squares polynomial in two variables determined by the values of/on the 9 points of Ku-
  • Fig. 4a shows the continuous pixel set for the image depicted in Fig. 3. It consists of the set of points (x.y) having real coordinates between 0 and 20. Fig. 4a also shows the 29 edge points of the image described in Fig. 3 that were calculated in accordance of the invention. The 29 edge points are indicated in Fig. 4a by dots, for example the edge points 4a. 4b, and 4c. Fig. 4b shows the four provisional edges in the image described in Fig. 3, calculated in accordance with the invention. The edge points in each provisional edge are enclosed in a circle 4d, 4e 4f and 4g. A parabola was drawn through each set of three consecutive points in each provisional edge.
  • the invention contemplates a suitably programmed computer capable of executing the invention.
  • the invention contemplates a computer program being readable by a computer for executing the method of the invention.
  • the invention further contemplates a machine-readable memory, tangibly embodying a program of instructions executable by the machine for executing the method of the invention.

Landscapes

  • Engineering & Computer Science (AREA)
  • Multimedia (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Processing Of Color Television Signals (AREA)
  • Image Analysis (AREA)

Abstract

L'invention concerne un procédé et un système de traitement d'une image numérique comprenant une fonction de couleur f définie sur un ensemble fini A de points dans R2. Pour chacun de points (k,l) dans A, on obtient une zone K¿k,l? dans R?2¿ contenant (k,l) et contenant un ou plusieurs points supplémentaires (k¿i?,li) de A. Un ensemble de pixels continus de l'image est alors formé, cet ensemble de pixels continus constituant une union des ensembles Kk,l. Une fonction fk,l est définie sur Kk,l sur la base des valeurs de f sur (k,l) et sur les points (ki,li). Un ou plusieurs objets sont définis sur la base des fonctions fk,l, les objets constituant des ensembles de points dans l'ensemble de pixels continus de l'image. Ledit procédé peut être utilisé dans la poursuite de caméra ou la stabilisation de caméra.
PCT/IL2001/000321 2000-04-09 2001-04-05 Procede de traitement d'une image WO2001078007A2 (fr)

Priority Applications (1)

Application Number Priority Date Filing Date Title
AU2001250600A AU2001250600A1 (en) 2000-04-09 2001-04-05 Method for processing an image

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
IL135560 2000-04-09
IL13556000A IL135560A0 (en) 2000-04-09 2000-04-09 Method for processing an image

Publications (2)

Publication Number Publication Date
WO2001078007A2 true WO2001078007A2 (fr) 2001-10-18
WO2001078007A3 WO2001078007A3 (fr) 2002-12-19

Family

ID=11074038

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/IL2001/000321 WO2001078007A2 (fr) 2000-04-09 2001-04-05 Procede de traitement d'une image

Country Status (3)

Country Link
AU (1) AU2001250600A1 (fr)
IL (1) IL135560A0 (fr)
WO (1) WO2001078007A2 (fr)

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2003102870A3 (fr) * 2002-05-31 2004-07-01 Qinetiq Ltd Mise en correspondance d'entites entre ensembles de donnees
EP2579569A4 (fr) * 2010-06-04 2014-10-08 Panasonic Ip Corp America Dispositif de traitement d'images, procédé de traitement d'images, circuit intégré et programme
US9654751B2 (en) 2006-12-21 2017-05-16 Thomson Licensing Method, apparatus and system for providing color grading for displays

Non-Patent Citations (3)

* Cited by examiner, † Cited by third party
Title
GUESTRIN C ET AL: "Fast software image stabilization with color registration" INTELLIGENT ROBOTS AND SYSTEMS, 1998. PROCEEDINGS., 1998 IEEE/RSJ INTERNATIONAL CONFERENCE ON VICTORIA, BC, CANADA 13-17 OCT. 1998, NEW YORK, NY, USA,IEEE, US, 13 October 1998 (1998-10-13), pages 19-24, XP010311349 ISBN: 0-7803-4465-0 *
MORIMOTO C ET AL: "Fast electronic digital image stabilization" PROCEEDINGS OF THE 13TH INTERNATIONAL CONFERENCE ON PATTERN RECOGNITION, PROCEEDINGS OF 13TH INTERNATIONAL CONFERENCE ON PATTERN RECOGNITION, VIENNA, AUSTRIA, 25-29 AUG. 1996, pages 284-288 vol.3, XP002214895 1996, Los Alamitos, CA, USA, IEEE Comput. Soc. Press, USA ISBN: 0-8186-7282-X *
ZIOU D AND TABBONE S: "Edge Detection Techniques- An Overview" TECHNICAL REPORT, NO. 195, DEPT MATH & INFORMATIQUE, UNIVERSITE DE SHERBROOKE, CANADA, [Online] 1997, pages 1-41, XP002214896 Retrieved from the Internet: <URL:http://citeseer.nj.nec.com/ziou97edge .html> [retrieved on 2002-09-26] *

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2003102870A3 (fr) * 2002-05-31 2004-07-01 Qinetiq Ltd Mise en correspondance d'entites entre ensembles de donnees
US7124022B2 (en) 2002-05-31 2006-10-17 Qinetiq Limited Feature mapping between data sets
US9654751B2 (en) 2006-12-21 2017-05-16 Thomson Licensing Method, apparatus and system for providing color grading for displays
EP2579569A4 (fr) * 2010-06-04 2014-10-08 Panasonic Ip Corp America Dispositif de traitement d'images, procédé de traitement d'images, circuit intégré et programme

Also Published As

Publication number Publication date
WO2001078007A3 (fr) 2002-12-19
AU2001250600A1 (en) 2001-10-23
IL135560A0 (en) 2001-05-20

Similar Documents

Publication Publication Date Title
US6393142B1 (en) Method and apparatus for adaptive stripe based patch matching for depth estimation
US5734743A (en) Image processing method and apparatus for block-based corresponding point extraction
Chen et al. Range data acquisition using color structured lighting and stereo vision
Avidan et al. Novel view synthesis in tensor space
Laveau et al. 3-D scene representation as a collection of images
Tomasi et al. Stereo matching as a nearest-neighbor problem
US6608622B1 (en) Multi-viewpoint image processing method and apparatus
Quan Invariants of 6 points from 3 uncalibrated images
KR100891549B1 (ko) 깊이 카메라를 이용하여 보완한 깊이 정보 생성 방법 및장치, 그리고 그 방법을 수행하는 프로그램이 기록된 기록매체
JP2000182038A (ja) 歪まされた情報からの遠近法的矯正デ―タの生成方法及びこれに用いる装置並びにコンピュ―タプログラム製品
JP3067207B2 (ja) 画像シーケンスにおける運動補正及び弾性変形法
US6011863A (en) Cylindrical rectification to minimize epipolar distortion
Robert et al. Weakly-calibrated stereo perception for rover navigation
Cherian et al. Accurate 3D ground plane estimation from a single image
WO2016129612A1 (fr) Procédé de reconstruction d&#39;une scène tridimensionnelle (3d)
WO2001078007A2 (fr) Procede de traitement d&#39;une image
Bober et al. On accurate and robust estimation of fundamental matrix
Han et al. Identification of cornerpoints of two-dimensional images using a line search method
CN110148206B (zh) 多空间的融合方法
Connor et al. Novel View Specification and Synthesis.
Chan et al. Dense stereo correspondence based on recursive adaptive size multi-windowing
Agrawal et al. Trinocular stereo using shortest paths and the ordering constraint
Stein Geometric and photometric constraints: Motion and structure from three views
Wu et al. Projective rectification based on relative modification and size extension for stereo image pairs
Rother et al. Linear multi view reconstruction with missing data

Legal Events

Date Code Title Description
AK Designated states

Kind code of ref document: A2

Designated state(s): AE AG AL AM AT AU AZ BA BB BG BR BY BZ CA CH CN CO CR CU CZ DE DK DM DZ EE ES FI GB GD GE GH GM HR HU ID IL IN IS JP KE KG KP KR KZ LC LK LR LS LT LU LV MA MD MG MK MN MW MX MZ NO NZ PL PT RO RU SD SE SG SI SK SL TJ TM TR TT TZ UA UG US UZ VN YU ZA ZW

AL Designated countries for regional patents

Kind code of ref document: A2

Designated state(s): GH GM KE LS MW MZ SD SL SZ TZ UG ZW AM AZ BY KG KZ MD RU TJ TM AT BE CH CY DE DK ES FI FR GB GR IE IT LU MC NL PT SE TR BF BJ CF CG CI CM GA GN GW ML MR NE SN TD TG

121 Ep: the epo has been informed by wipo that ep was designated in this application
DFPE Request for preliminary examination filed prior to expiration of 19th month from priority date (pct application filed before 20040101)
AK Designated states

Kind code of ref document: A3

Designated state(s): AE AG AL AM AT AU AZ BA BB BG BR BY BZ CA CH CN CO CR CU CZ DE DK DM DZ EE ES FI GB GD GE GH GM HR HU ID IL IN IS JP KE KG KP KR KZ LC LK LR LS LT LU LV MA MD MG MK MN MW MX MZ NO NZ PL PT RO RU SD SE SG SI SK SL TJ TM TR TT TZ UA UG US UZ VN YU ZA ZW

AL Designated countries for regional patents

Kind code of ref document: A3

Designated state(s): GH GM KE LS MW MZ SD SL SZ TZ UG ZW AM AZ BY KG KZ MD RU TJ TM AT BE CH CY DE DK ES FI FR GB GR IE IT LU MC NL PT SE TR BF BJ CF CG CI CM GA GN GW ML MR NE SN TD TG

122 Ep: pct application non-entry in european phase
NENP Non-entry into the national phase

Ref country code: JP

点击 这是indexloc提供的php浏览器服务，不要输入任何密码和下载