(19)
(11)EP 3 080 781 B1

(12)EUROPEAN PATENT SPECIFICATION

(45)Mention of the grant of the patent:
13.02.2019 Bulletin 2019/07

(21)Application number: 14809691.0

(22)Date of filing:  05.12.2014
(51)International Patent Classification (IPC): 
G06T 15/40(2011.01)
G06T 17/20(2006.01)
G06T 19/00(2011.01)
(86)International application number:
PCT/GB2014/053616
(87)International publication number:
WO 2015/087055 (18.06.2015 Gazette  2015/24)

(54)

IMAGE RENDERING OF LASER SCAN DATA

BILDWIEDERGABE VON LASERABTASTDATEN

RENDU D'IMAGE DE DONNÉES DE BALAYAGE LASER


(84)Designated Contracting States:
AL AT BE BG CH CY CZ DE DK EE ES FI FR GB GR HR HU IE IS IT LI LT LU LV MC MK MT NL NO PL PT RO RS SE SI SK SM TR

(30)Priority: 13.12.2013 GB 201322113

(43)Date of publication of application:
19.10.2016 Bulletin 2016/42

(73)Proprietor: Aveva Solutions Limited
Cambridge, Cambridgeshire CB3 0HB (GB)

(72)Inventors:
  • FREEDMAN, Aaron
    Hobe Sound, FL 33455-3949 (US)
  • ELTON, Paul
    Cambridge Cambridgeshire CB23 3LG (GB)

(74)Representative: Piotrowicz, Pawel Jan Andrzej et al
Venner Shipley LLP Byron House Cambridge Business Park Cowley Road
Cambridge CB4 0WZ
Cambridge CB4 0WZ (GB)


(56)References cited: : 
US-A1- 2010 079 454
  
  • GOPI M ET AL: "A fast and efficient projection-based approach for surface reconstruction", COMPUTER GRAPHICS AND IMAGE PROCESSING, 2002. PROCEEDINGS. XV BRAZILIA N SYMPOSIUM ON FORTALEZA-CE, BRAZIL 7-10 OCT. 2002, LOS ALAMITOS, CA, USA,IEEE COMPUT. SOC, US, 7 October 2002 (2002-10-07), pages 179-186, XP010624507, DOI: 10.1109/SIBGRA.2002.1167141 ISBN: 978-0-7695-1846-6
  • V. SEQUEIRA ET AL: "Automated reconstruction of 3D models from real environments", ISPRS JOURNAL OF PHOTOGRAMMETRY AND REMOTE SENSING, vol. 54, no. 1, 1 February 1999 (1999-02-01), pages 1-22, XP055110110, ISSN: 0924-2716, DOI: 10.1016/S0924-2716(98)00026-4
  • RUSINKIEWICZ S ET AL: "QSPLAT: A MULTIRESOLUTION POINT RENDERING SYSTEM FOR LARGE MESHES", COMPUTER GRAPHICS. SIGGRAPH 2000 CONFERENCE PROCEEDINGS. NEW ORLEANS, LA, JULY 23 - 28, 2000; [COMPUTER GRAPHICS PROCEEDINGS. SIGGRAPH], NEW YORK, NY : ACM, US, 23 July 2000 (2000-07-23), pages 343-352, XP001003573, ISBN: 978-1-58113-208-3
  
Note: Within nine months from the publication of the mention of the grant of the European patent, any person may give notice to the European Patent Office of opposition to the European patent granted. Notice of opposition shall be filed in a written reasoned statement. It shall not be deemed to have been filed until the opposition fee has been paid. (Art. 99(1) European Patent Convention).


Description

Field of the Invention



[0001] The present invention relates to a method of, and a system for, rendering an image of laser scan data.

Background



[0002] A three-dimensional laser scanner can be used to survey an environment such as a process plant, vessel or other facility. A typical scanner includes a laser rangefinder which can measure a distance between the scanner and a point on a surface which is in view. By sweeping through a field of view (typically 360 degrees horizontally and nearly 180 vertically), the scanner can capture a set of ranges (herein referred to as "laser scan data") for the surrounding environment. These can be used to generate a set of points in three-dimensional space, often referred to as a "point cloud". An example of a point cloud is described in EP 1176 393 A2.

[0003] Multiple scans can be performed at different positions in an environment and point clouds from different scans can be combined to produce a combined (or "aggregated") point cloud covering a wider area. An example of combining point cloud data can be found in WO 2004/003844 A1.

[0004] In addition to acquiring range data, the scanner can also capture images of the surrounding environment by measuring intensity of reflected laser light or using a camera.

[0005] The point cloud(s) and images can be used to visualize and/or analyze an environment using a point cloud viewer application or a three-dimensional computer-aided design (CAD) application. Typically, these applications fall into two categories, namely those that work with points from individual scans and those that work with points combined from multiple scans.

[0006] One of the simplest applications of laser scanning is to display an image captured by an individual scan. Because the image from a laser scan is spherical, covering the area around the laser scanner, the software application can map the image onto the inside of a sphere. The application can display a portion of the sphere on a computer screen.

[0007] The user can rotate the view in order to view different portions of the entire image. This presentation is called a "bubble view".

[0008] In bubble view, the user can select a spot on the image and retrieve the three-dimensional coordinate of that location using the point cloud data for that laser scan. By selecting two points, the user can measure distances.

[0009] One type of application can overlay a three-dimensional CAD model in a bubble view. Because the application knows the three-dimensional location of the points in bubble view, it can obscure the appropriate portions of the CAD model behind the bubble view. This combined image can be useful when designing new areas of the facility.

[0010] An appealing feature of a bubble view is that it looks realistic. Realism comes from the image captured at the scanner location. A limitation of bubble views is, however, that they can only be produced for the locations at which a laser scanner was positioned. A user can select a bubble view and rotate to the left and right or up and down, but he/she cannot move forward, backward, horizontally or vertically in order to view the environment from a different perspective.

[0011] To allow free roaming, some software applications work with a combined point cloud from multiple scans. Using such an application, a user chooses a location within a facility and a viewing direction. The application then displays each point in the combined point cloud around that location from the point of view of the user. The user can move the viewing location and direction to see the points from different perspectives.

[0012] Some applications can display a CAD model in the same three-dimensional space as the combined point cloud. A user can then measure distances between locations in the CAD model and points in the point cloud. The user can also determine if portions of the point cloud intersect portions of the CAD model.

[0013] Although displaying a combined point cloud allows the user to view points from more than one perspective, this approach can have one or more drawbacks.

[0014] Displaying individual points tends to be computationally expensive. Gaps can appear in a representation of a scanned surface at close distances and so it can become difficult to discern the surfaces.
M. Gopi et al. "A fast and efficient projection-based approach for surface reconstruction", Proceedings of the XV Brazilian symposium on Computer Graphics And Image Processing, pages 179 to 186 (2002) describes an algorithm for reconstructing the surfaces of unorganized point clouds using a 250 MHz, R10000 SGI Onyx2.

[0015] V. Sequeira et al., "Automated reconstruction of 3D models from real environments", ISPRS Journal Of Photogrammetry And Remote Sensing, volume 54, pages 1 to 22 (1999) discloses construction of textured 3D scene models of building interiors.
US 2010/089454 A1 describes performing tessellation in a single pass through a graphics processor.
S. Rusinkiewicz et al., "QSPLAT: A Multiresolution Point Rendering System For Large Meshes", Conference Proceedings SIGGRAPH 2000 describes representing and progressively displaying these meshes that combines a multiresolution hierarchy based on bounding spheres with a rendering system based on points.

[0016] G. Turk et al., "Zippered Polygon Meshes from Range Images", Conference Proceedings SIGGRAPH 1994 discloses a method for combining a collection of range images into a single polygonal mesh, wherein scans are acquired and combined one at a time, which allows to acquire and combine large numbers of scans with minimal storage overhead.

Summary



[0017] According to a first aspect of the present invention there is provided a method of rendering an image of three-dimensional laser scan data. The method comprises providing a plurality of sets of laser scan data for a set of scan as spherical displacement maps wherein each set of laser scan data corresponds to a respective scan; a graphics processing unit (GPU) generating a tessellation pattern which comprises a set of triangles by sampling the spherical displacement maps, wherein said generating of the tessellation pattern is performed by said graphics processing unit, converting each triangle into raster data comprising pixels each pixel including a value of depth, and writing a pixel into a depth buffer if the pixel has a lower depth value than a corresponding pixel stored in the depth buffer, and discarding a pixel if the pixel has a higher depth value than a corresponding pixel stored in the depth buffer, wherein pixel data generated from different scans is written to the same depth buffer.
By preserving laser scan data for a given scan as a set of points (as opposed to aggregating laser scan data for multiple scans) and by taking advantage of the fact that the laser scan data can be provided in the form of a displacement map which can be handled directly by a graphics system, an image of the laser scan data can be rendered efficiently and/or quickly. This can be particularly helpful when combining images from multiple scans since each scan can be processed independently and the images from different scans can be easily combined in a common buffer. This allows efficient/fast rendering not only of static images, but also moving images, for example, as a user "walks through" the environment.
Providing the set of laser scan data for the given scan as the spherical displacement map may comprise receiving a set of laser scan data generated by a laser scanner at a given location and copying a value of range for a given laser scan point at a given azimuth and a given elevation from the laser scan data into a respective texel of a two-dimensional texture (or higher-dimensional texture) at a texel position corresponding to the given azimuth and the given elevation.
The graphics processing unit may be configured using Microsoft (RTM) DirectX (RTM) 11 (or later) or OpenGL 4.4 (or later) application programming interface (API).

[0018] The method may further comprise identifying discontinuities between adjacent points in the spherical displacement map and marking the adjacent points. Marking the

[0019] The method may further comprise generating a normal map in dependence upon said spherical displacement map. The displacement map and normal map may be combined in one texture. The one texture may comprise at least four channels.

[0020] Generating the normal map may comprise calculating a normal for each point in the spherical displacement map and storing the normal in the normal map. The normal may comprise first, second and third vector component values.

[0021] The method may further comprise generating a blending texture in dependence upon the spherical displacement map. The blending texture may comprise an array of blending texels, each blending texel comprising a value which depends on distance from a discontinuity.

[0022] The method may further comprise generating a patch map for a given scan, the patch map comprising a plurality of patches.

[0023] The patch map may comprise polygonal patches, each patch having three or more vertices. The patches may be the same shape. The patches may be the same size. The patches may be rectangles. If the patches are rectangles, then the patch map may comprise positions of opposite vertices.

[0024] Position, shape and/or size of the patches may depend on discontinuities in the spherical displacement map.

[0025] The patch map may include a relative tessellation level for each patch. The method may comprise setting the relative tessellation level of a given patch in dependence upon discontinuities. The method may comprise setting the relative tessellation level of a given patch in dependence upon normal variance across the given patch.

[0026] The method may comprise calculating an absolute tessellation level for a given patch in dependence upon a view position and/or visibility of the given patch.

[0027] The method may further comprise determining whether a vertex forming part of a triangle in the tessellation pattern is marked as being on or adjacent to a discontinuity and, in dependence upon determining that the vertex is invalid, culling (or "discarding") the triangle.

[0028] The method may further comprise determining whether a vertex forming part of a triangle in the tessellation pattern is marked as being on or adjacent to a discontinuity and, in dependence upon determining that the vertex is invalid, culling (or "discarding") the triangle.
The method may further comprise generating a set of pixels for the scan and performing a depth test.
The method may comprise colouring a pixel in dependence upon the normal of the pixel. The method may comprise colouring a pixel in dependence upon intensity and/or colour in a corresponding part of an image.
According to a second aspect of the present invention there is provided a computer program which comprises instructions for performing the method.
According to a third aspect of the present invention there is provided a computer readable medium or non-transitory computer-readable medium which stores the computer program.
According to a fourth aspect of the present invention there is provided computer system comprising memory and at least one processing unit comprising at least one graphics processing unit. The memory stores a plurality of sets of laser scan data for a set of scans as spherical displacement maps, wherein each set of laser scan data corresponds to a respective scan and the at least one graphics processing unit is (are) configured to generate a tessellation pattern comprising a set of triangles (72) by sampling the spherical displacement maps, convert each triangle into raster data comprising pixels each pixel including a value of depth, and writing a pixel into a depth buffer if the pixel has a lower depth value than a corresponding pixel stored in the depth buffer, and discarding a pixel if the pixel has a higher depth value than a corresponding pixel stored in the depth buffer, wherein pixel data generated from different scans is written to the same depth buffer.
The at least one processing unit may comprise at least one central processing unit. The at least one processing unit may comprise one processing unit, for example, one graphical processing unit.

[0029] The at least one processing unit may be configurable using a Microsoft (RTM) DirectX (RTM) 11 (or later) application programming interface. The at least one processing unit may be configurable using OpenGL application programming interface.

Brief Description of the Drawings



[0030] Certain embodiments of the present invention will now be described, by way of example, with reference to the accompanying drawings, in which:

Figure 1 is a schematic block diagram of a system for obtaining and processing laser scan data, and rendering an image;

Figure 2 is a schematic block diagram of a computer system which can be used to process laser scan data and render an image;

Figure 3 is a process flow diagram of a method of processing laser scan data;

Figure 4 illustrates generation of a spherical displacement map;

Figure 5 shows a discontinuity between two surfaces;

Figure 6 shows updating a spherical displacement map to mark a discontinuity;

Figure 7 illustrates generation of a normal map;

Figure 8 shows generation of a blending map;

Figure 9 illustrates an arrangement of regular, rectangular patches;

Figure 10 illustrates an arrangement of irregular, rectangular patches;

Figure 11 illustrates an arrangement of irregular, non-rectangular patches;

Figure 12 shows a patch map;

Figure 13 shows adding relative tessellation data to a patch map;

Figure 14 is a schematic block diagram of a graphics pipeline and graphics memory;

Figure 15 illustrates output of a hull shader;

Figure 16 illustrates tessellation levels for a patch;

Figure 17 illustrates output of a tessellation shader and input of a domain shader;

Figure 18 illustrates operation of a domain shader using a texture which includes a displacement map and a normal map;

Figure 19 illustrates output of a vertex list from a domain shader;

Figure 20 shows operation of a geometry shader which culls invalid triangles;

Figure 21 illustrates operation of a rasterizer;

Figure 22 illustrates operation of a pixel shader;

Figure 23 is process flow diagram of a depth test;

Figure 24 illustrates parallel processing of scans;

Figure 25 illustrates a combined point cloud generated from multiple scans;

Figure 25a is a magnified view of part of the point cloud shown in Figure 25;

Figure 26 shows a combined tessellated image generated from individual scans of the point cloud shown in Figure 25;

Figure 26a is a magnified view of part of the tessellated image shown in Figure 26;

Figure 26b is a magnified view of another part of the tessellated image shown in Figure 26;

Figure 27 shows a combined rendered image generated using the tessellation shown in Figure 26; and

Figure 28 shows a combined rendered image generated from individual scans of the point cloud shown in Figure 25 from a different view point.


Detailed Description of the Certain Embodiments


System overview



[0031] Referring to Figure 1, a system 1 for generating and processing laser scan data and for rendering an image is shown.

[0032] The system 1 includes one or more three-dimensional laser scanners 2 for surveying an environment 3 which includes a number of target surfaces 4. The, or each, laser scanner 2 includes a laser scanning unit 5 which generates laser scan data 6 (herein referred to simply as "scan data"), an optional camera 7 which can be used to generate image data 8, for example in the form a JPEG file, and on-board storage 9 for storing data 6, 8. The, or each, laser scanner 2 includes processor(s) 10 and memory 11 which can be used to process the laser scan data 6, for example, to format the data. A separate computer system (not shown) can be used to process the data.

[0033] The laser scanning unit 5 generates an element of scan data 6 for a point by emitting a pulsed laser beam 12 in a given direction (i.e. at given a horizontal angle and a given vertical angle), sensing the reflected beam 13 that is reflected off a target surface 4, back to the laser scanner 2, and determining a range to the target surface 4 based on time of flight of the laser beam 12, 13. A set of scan data 6 can be acquired by scanning the laser beam 12 horizontally and vertically so as to build up a set of points around the scanner 2. Each point in the scan data 6 is provided in the form of a set of Cartesian coordinates, i.e. each point is expressed in (x, y, z). Points in a set of data 6 are ordered by azimuth and elevation.

[0034] The scan data 6 and, optionally, image data 8 are supplied to a computer system 15 which includes a scan data processing module 16 which generates a set 17 of textures 18, 19, 20, 21, some textures combined into a single texture 22, and a patch map 23 for each scan. The computer system 15 also includes storage 24 in the form of one or more hard drives for storing data. The hard drive(s) may be hard disk drives, solid-state drive, optical drive or other form of suitable storage.

[0035] As will be explained in more detail later, the set of textures 17 include a spherical displacement map 18 (herein also referred to as a "depth map") and a normal map 19 obtained from scan data 6. As will also be explained in more detail later, the displacement map 18 and the normal map 19 are combined in a single, 4-channel texture 22 (herein referred to as a "combined texture").

[0036] The set of textures 17 can include a blending map 20. The set of textures 17 can include a colour map 21 obtained from the image data 8.

[0037] The computer system 15 includes a user input devices 25 (such as a mouse and/or keyboard), a rendering system 26 and a display 27 for displaying an image 28 from a view point 29 which is supplied by the user via a user input device 25. The rendering system 26 produces triangulated three-dimensional surfaces using the set of textures 17 and the patch map 23 obtained from one or more different scans and renders the surfaces in real time, from any view point, combining surfaces obtained from the scan(s) in an image 28.

[0038] The scan data processing module 16 and the rendering system 26 may be implemented in different computer systems. For example, the scan data processing module 16 may be implemented all or in part by a laser scanner 2. Alternatively, the laser scanner 2 may generate the displacement map 18 and a first computer system may generate the other texture(s) 19, 20, 21 and the patch map 23, or just the patch map 23, and supply texture(s) 18, 19, 20, 21 and patch map 23 to a second computer system which implements the rendering system 26.

[0039] Referring also to Figure 2, the computer system 15 is shown in more detail.

[0040] The computer system 15 includes one or central processing units (CPUs) 31 having respective memory caches (not shown), system memory 32, a graphics module 33, for example in the form of a graphics card, which includes a graphics processing unit (GPU) 34 and graphics memory 35 (which may be referred to as "video RAM"), and an input/output (I/O) interface 36 operatively connected by a bus 37. An example of a suitable graphics module 33 is an NVIDIA (RTM) GeForce 460 GPU with 1 GB of video RAM.

[0041] The I/O interface 36 is operatively connected to bus and/or network interface(s) 38 (such as USB interface or WLAN interface) for receiving scan data 6 from the, or each, scanner 2. The I/O interface 36 is also operatively connected to user input devices 25 and the storage 24, for example, in the form of one or more hard disk drives and/or solid-state drives. Some peripheral devices, such as removable storage, and other computer components are not shown. The computer system 15 may have a different configuration from that shown in Figure 2.

[0042] The scan processing module 16 is implemented in software. Computer code 39 for implementing the scan processing module 16 is held in storage 24 and loaded into memory 32 for execution by the CPU(s) 31.

[0043] The rendering system 26 is preferably implemented using a GPU so as to take advantage of the enhanced graphics processing capabilities of a GPU, in particular tessellation. However, the rendering system 26 can be implemented using one or more CPUs.

[0044] As will be explained in more detail later, the rendering system 26 implements a graphics pipeline 60 (Figure 14) which includes a plurality of modules 61, 62, 63, 64, 65, 66, 67, 68 (Figure 14) including programmable and fixed-function shaders 61, 62, 63, 64, 65, 67, 68 and a rasterizer 66.

Scan processing



[0045] Referring to Figures 1, 2 and 3, operation of the scan data processing module 16 will now be described in more detail.

[0046] The module 16 loads a set of scan data 6 for a scan from a scanner 2 via a bus or network (not shown) or portable storage (not shown) and/or from storage 24 (step S1). The scan data 6 is in Cartesian coordinates and so the module 16 converts and stores the scan data 6 as a spherical displacement map 18 (step S2).

[0047] Referring also to Figure 4, a spherical coordinate system 41 is shown. The spherical coordinate system 41 includes an origin 42 (i.e. a point of reference in the scanner 2) instance 855 and 22 respectively. The threshold, ΔdTH, may be, for instance 300. Thus, the difference in depth Δd = 855 - 22 = 833 and Δd > ΔdTH and so the module 16 marks the discontinuity by setting both values equal to zero.

[0048] Adjacent texels 46 in u- and/or v-directions can be compared.

[0049] Referring to Figures 1, 2, 3 and 7, the scan data processing module 16 generates a normal map 19 from the displacement map 18 and stores the normal map 19 (step S4).

[0050] For each texel 46 in the displacement map 18, the module 16 calculates the Cartesian coordinates from the spherical coordinates represented by the texel 46. The module calculates a normal 49 for a given texel 46 (which is shown shaded in Figure 7) by taking the cross product of a first vector between horizontal neighbouring texels 46 and a second vector between vertical neighbouring texels 46 which are adjacent to the texel 46, in Cartesian coordinates. The module 16 does not calculate a normal 49 across a discontinuity.

[0051] The module 16 stores the x-, y- and z-components (nx, ny, nz) of the normal 49 in the texel 46 in the second, third and fourth elements 462, 463, 464.

[0052] Referring to Figures 1, 2, 3 and 8, the scan data processing module 16 can, optionally, generate a blending texture 20 that contains information used to smooth the edge of a surface near a discontinuity (step S5).

[0053] Each texel 50 in the blending texture 20 contains a blending value, s, which lies in a range between 0 and 1.

[0054] The scan data processing module 16 identifies texels 46 in the displacement map 18 which correspond to points along the discontinuity and sets corresponding texels 50 in the blending texture 20 (i.e. having the same values of u and v) to have a blending value of 0. The scan data processing module 16 generates further blending values which gradually propagate through the blending texture 20, i.e. values which gradually increase. For example, as shown in Figure 8, a value of 0.5 is given to a texel 50 which is a first-nearest neighbour to a discontinuity, a value of 0.8 is given to a texel 50 which is a second-nearest neighbour and a value of 1 is given to a texel 50 which is a third-nearest neighbour.

[0055] For example, as shown in Figure 5, two adjacent scan points fall on either side of an edge 47. As shown in Figure 6, the corresponding texels 46 have depth values, for instance 855 and 22 respectively. The threshold, ΔdTH, maybe, for instance 300. Thus, the difference in depth Δd = 855 - 22 = 833 and Δd > ΔdTH and so the module 16 marks the discontinuity by setting both values equal to zero.

[0056] Adjacent texels 46 in u- and/or v-directions can be compared.

[0057] Referring to Figures 1, 2, 3 and 7, the scan data processing module 16 generates a normal map 19 from the displacement map 18 and stores the normal map 19 (step S4).

[0058] For each texel 46 in the displacement map 18, the module 16 calculates the Cartesian coordinates from the spherical coordinates represented by the texel 46. The module calculates a normal 49 for a given texel 46 (which is shown shaded in Figure 7) by taking the cross product of a first vector between horizontal neighbouring texels 46 and a second vector between vertical neighbouring texels 46 which are adjacent to the texel 46, in Cartesian coordinates. The module 16 does not calculate a normal 49 across a discontinuity.

[0059] The module 16 stores the x-, y- and z-components (nx, ny, nz) of the normal 49 in the texel 46 in the second, third and fourth elements 462, 463, 464.

[0060] Referring to Figures 1, 2, 3 and 8, the scan data processing module 16 can, optionally, generate a blending texture 20 that contains information used to smooth the edge of a surface near a discontinuity (step S5).

[0061] Each texel 50 in the blending texture 20 contains a blending value, s, which lies in a range between 0 and 1.

[0062] The scan data processing module 16 identifies texels 46 in the displacement map 18 which correspond to points along the discontinuity and sets corresponding texels 50 in the blending texture 20 (i.e. having the same values of u and v) to have a blending value of 0. The scan data processing module 16 generates further blending values which gradually propagate through the blending texture 20, i.e. values which gradually increase. For example, as shown in Figure 8, a value of 0.5 is given to a texel 50 which is a first-nearest neighbour to a discontinuity, a value of 0.8 is given to a texel 50 which is a second-nearest neighbour and a value of 1 is given to a texel 50 which is a third-nearest neighbour.

[0063] Referring to Figures 1, 2, 3 and 9, the scan data processing module 16 breaks a scan into patches 51 and stores a patch map 23 (step S6).

[0064] Each patch 51 defines a region of the scan textures in (u,v) coordinates. Each patch has a set of vertices 52.

[0065] The patches 51 can be any shape and can be regularly or irregularly distributed in the scan. The shape and position of the patches 51 can be predetermined or based on any aspect of the scan, such as locations of discontinuities or variance of normals.

[0066] For example, Figure 9 shows an arrangement of rectangular patches 51 which are all the same shape and size.

[0067] In another example, Figure 10 shows an arrangement of rectangular patches 51 of different sizes which take into account the location of discontinuities 47. For example, as shown in Figure 10, instead of using fixed-sized patches 51, the scan data processing module 16 may break a larger patch 51 into smaller patches 511, 512, 513, 514, 515 according to whether a discontinuity is found within the patch 51.

[0068] In yet another example, Figure 11 shows an arrangement in which the scan data processing module 16 generates non-rectangular patches 516, 517, 518, 519 having edges which take into account (for example, are aligned with) a forking discontinuity 47. For clarity, the outer edges of the patches 516, 517, 518, 519 are shown truncated so that they appear as straight vertical or horizontal edges. However, it will be appreciated that the patches need not be truncated and can have an irregular outline, for example, as shown in chain for patch 516 in Figure 11.

[0069] Referring also to Figure 12, for each patch 51, a set of elements 54 are stored in the patch map 23 for each patch 51, each element 54 comprising the position of a patch vertex 52. In the case of rectangular patch 51 (such as that shown in Figure 9), the patch 51 can be specified using only two vertices 52, for example, upper left and lower right vertices 52. However, there may be more than two elements 54, particularly for irregularly-shaped patches.

[0070] Referring to Figures 1, 2, 3 and 13, the module 16 calculates a relative tessellation level 55 for each patch 51 (step S7).

[0071] The relative tessellation level 55 represents how much a patch 51 will be tessellated relative to the other patches 51. An absolute tessellation level is calculated later, after view-dependent factors or other factors are applied. The relative tessellation level 55 can be predetermined or based on any aspect of the patch such as the location of discontinuities or the variance of normals.

[0072] The displacement map 18, the normal map 19, the blending texture 20 and the patch map 23 can be produced once for each scan and stored for later use.

Rendering



[0073] Referring to Figures 1, 2 and 14, the scan data processing module 16 loads the textures 18, 19, 20 and the patch map 23 for each scan into the rendering system 26. The rendering system 26 is used to produce tessellated surface(s) of the scan using information in the patch map 23. Any rendering system that supports tessellation can be used. Preferably, a rendering system which can process 100 million triangles per second or at a higher rate is used. The rendering system can run on a CPU or GPU.

[0074] The rendering system 26 employs Microsoft (RTM) DirectX 11 (RTM) running on the GPU 34. However, other graphics systems, such as OpenGL 4.x, can be used.

[0075] The textures 18, 19, 20 and patch map 23 are stored in a form which can be efficiently processing by the rendering system 26.

[0076] The textures 18, 19, 20 are stored as two-dimensional texture resources and the patch map 23 is stored as a vertex buffer with a topology of a one-control-point patch list. As explained earlier, the displacement and normal maps 18, 19 are combined into a combined texture 22 with four channels.

[0077] The textures 18, 19, 20 and the vertex buffer 23, i.e. patch map 23, are sent to the GPU 34.

[0078] Referring to Figure 14, a graphics pipeline 60 for the graphics system is shown.

[0079] Referring also to Figures 18 and 19, each point 73 in each patch 51 is processed by the domain shader 64. The domain shader 64 samples the combined depth/normal texture 22 and projects tessellated points into three-dimensional space.

[0080] The domain shader 64 calculates the position of each point 73 in the tessellation pattern 71 in a patch 51. The domain shader 64 samples the combined depth/normal texture 22 at the (u, v) coordinate of the point 73 to retrieves values of depth and normal. If the sample includes a texel 46 that is next to a discontinuity, the point is marked invalid.

[0081] The domain shader 64 calculates the azimuth and elevation from the (u, v) coordinate using information in the patch 51. The domain shader 64 then converts the spherical coordinates (i.e. azimuth, elevation, depth) into view coordinates 74 either directly or (as shown in Figure 18) in stages, namely converting the spherical coordinates into the Cartesian position 75 in the coordinate system of the scan and then transforms the position and normal into world coordinates and then into view coordinates 74. The coordinates 74 are output in the form of a vertex list 76.

[0082] The domain shader 64 effectively displaces points on a spherical surface by the depth specified in the (spherical) displacement map 19.

[0083] Referring to Figure 20, each triangle 72 in a tessellated patch 71 is then processed by the geometry shader 65. The geometry shader 65 culls triangles 72 at discontinuities.

[0084] The geometry shader 65 determines if a triangle 72 contains any invalid vertices 77. A vertex 74 is invalid if it is next to a discontinuity. If an invalid vertex 77 is present, then the geometry shader 65 discards the triangle 72, i.e. discards the invalid triangle 78. The geometry shader 65 also determines if the triangle 72 is at an angle which is too oblique to the laser scanner position. If so, the triangle 72 is also discarded. The geometry shader 65 outputs an updated vertex list 79.

[0085] Referring to Figure 21, the rasterizer 66 converts each triangle 72 in the updated vertex list 79 into raster data 80 comprising pixels 81. Each pixel 81 includes a value of depth 82.

[0086] Each rasterized triangle 72 is then processed by the pixel shader 67.

[0087] Referring also to Figures 18 and 19, each point 73 in each patch 51 is processed by the domain shader 64. The domain shader 64 samples the combined depth/normal texture 22 and projects tessellated points into three-dimensional space.

[0088] The domain shader 64 calculates the position of each point 73 in the tessellation pattern 71 in a patch 51. The domain shader 64 samples the combined depth/normal texture 22 at the (u, v) coordinate of the point 73 to retrieves values of depth and normal. If the sample includes a texel 46 that is next to a discontinuity, the point is marked invalid.

[0089] The domain shader 64 calculates the azimuth and elevation from the (u, v) coordinate using information in the patch 51. The domain shader 64 then converts the spherical coordinates (i.e. azimuth, elevation, depth) into view coordinates 74 either directly or (as shown in Figure 18) in stages, namely converting the spherical coordinates into the Cartesian position 75 in the coordinate system of the scan and then transforms the position and normal into world coordinates and then into view coordinates 74. The coordinates 74 are output in the form of a vertex list 76.

[0090] The domain shader 64 effectively displaces points on a spherical surface by the depth specified in the (spherical) displacement map 19.

[0091] Referring to Figure 20, each triangle 72 in a tessellated patch 71 is then processed by the geometry shader 65. The geometry shader 65 culls triangles 72 at discontinuities.

[0092] The geometry shader 65 determines if a triangle 72 contains any invalid vertices 77. A vertex 74 is invalid if it is next to a discontinuity. If an invalid vertex 77 is present, then the geometry shader 65 discards the triangle 72, i.e. discards the invalid triangle 78. The geometry shader 65 also determines if the triangle 72 is at an angle which is too oblique to the laser scanner position. If so, the triangle 72 is also discarded. The geometry shader 65 outputs an updated vertex list 79.

[0093] Referring to Figure 21, the rasterizer 66 converts each triangle 72 in the updated vertex list 79 into raster data 80 comprising pixels 81. Each pixel 81 includes a value of depth 82.

[0094] Each rasterized triangle 72 is then processed by the pixel shader 67.

[0095] Referring to Figure 22, the pixel shader 67 samples the blending texture 20 to determine if a pixel 81 is too close to a discontinuity 47 (Figure 5). If the pixel 81 is too close to a discontinuity, then it is discarded. The pixel shader 67 samples the combined depth/normal texture 22 to retrieve the normal 49 (Figure 7) at the pixel 81. The pixel shader 67 transforms the normal 49 (Figure 7) into a vector (not shown) in world space. The pixel shader 67 calculates a colour 83 of a pixel 81 and outputs raster data 84 for including the depth 82 and colour 83 for each pixel 81.

[0096] Referring to Figures 14 and 23, when multiple scans are rendered into one image, scan data can be combined using the depth buffer 69.

[0097] The pixel shader 67 can measure quality of the depth by sampling, at a given location and offset, the depth 82. The pixel shader 67 can favour higher-quality samples at the same screen coordinate. The pixel shader 67 can use information in the blending texture 20 and/or the combined depth/normal texture 22 to determine the quality of the sample.

[0098] The output merger stage 68 performs, among other things, a depth test.

[0099] Referring also to Figure 24, each pixel 81 is compared to a corresponding pixel in the depth buffer 69 (steps S11, S12 & S13). If the incoming pixel 81 has a lower value of depth than the corresponding pixel stored in the depth buffer 69, then the incoming pixel 81 is written into the buffer 69 (step S14). If the incoming pixel 81 has a higher value of depth than the corresponding pixel stored in the depth buffer 69, then the incoming pixel 81 is discarded (step S15).

[0100] Pixel data generated from several scans can be written to the same depth buffer 69 and, thus, an image comprising data from several scans can be formed.

[0101] Referring to Figures 25 to 28, the method of processing laser scan data and rendering an image is illustrated for a set of scans at different stages during the process.

[0102] Figure 25 is an image 91 of point cloud generated from multiple scans. Figure 25a is a magnified view of part 91a of the image 91 shown in Figure 25.

[0103] The image 91 shows the interior of a process plant which includes, among other things, a floor 92, a ceiling 93, storage tanks 94, a row of three reaction vessels 95, various pipes 96, a row of four control panels 97, cable trays 98 suspended from the ceiling 93, lighting units 99 (or "luminaries") and a row of four pallet-type liquid storage containers 100.

[0104] Figure 25a is a more detailed view of a middle reaction vessel 95, pipes 96 and a corresponding control panel 97. The image 91, 91a is formed from point cloud points 45. In this case, each point 45 has a respective colour captured at time of scanning.

[0105] As shown in Figure 25a, relatively distant objects, such as the control panel 97, are visible even though relatively near objects, such as the reaction vessel 95, is in front of it and so should obscure the distant object. This effect becomes more pronounced when the image is magnified.

[0106] Figure 26 shows a combined tessellated image 101 generated from the individual scans of the point cloud shown in Figure 25. Figures 26a and 26b are magnified view of parts 111a, 111b of the image 111 shown in Figure 26.

[0107] Referring to Figures 25a and 26b, points 45 from the point cloud are used to form tessellation patterns 71 made up of triangles 72.

[0108] Figure 26b shows triangles 72 having different sizes, in other words, have different tessellation levels.

[0109] Objects which should be hidden, such as the reaction vessels 95, are still transparent since the triangles 72 have not been rasterized and, thus, filled in.

[0110] Figure 27 shows a combined rendered image 28, 111 generated using tessellation.

[0111] As shown in Figure 27, relatively distant objects, such as control panels 97, are now obscured by relatively near objects, such as reaction vessels 95.

[0112] As explained earlier, the rendering system 26 (Figure 1) can render images in real time using multiple scans. Furthermore, the system 26 (Figure 1) allows position to be changed and generate tessellated images viewed from different positions. Position can be incrementally or continuously changed allowing the user to "walk" or "fly" through the surveyed environment.

[0113] Figure 28 shows another combined rendered image 28, 111' generated from the same point cloud shown in Figure 25, but from a different view point.

[0114] As shown in Figure 28, the image 111' shows the same objects, such as the control panels 97 and reaction vessels 95, but from another view point. The image 111 shown in Figure 27 is taken from a point of view which is in front of doors 112 shown Figure 28.

[0115] Figure 28 shows the positions of the scanner 2 (Figure 1) when carrying out a scan. Each dark circular region 113 corresponds to an area under the scanner 2 where the scanner 2 does not log points.

[0116] A comparison of the point cloud image 91 (Figure 25) and the rendered image 111' (Figure 29) shows a visual artefact associated with displaying an image 91 of point cloud generated from multiple scans. The rendered image 111' shows front faces of the control panels 97. Each front face bears a manufacturer label 113 and warning signs 114. However, these labels 113 and signs 114 are visible in the point cloud image 91 when they would not be visible in the real world.

[0117] By not aggregating scan data for individual scans into one, larger set of scan data, the system can take advantage of the order in which scan data are collected. Each set of scan data can be treated as a spherical displacement map and can be sampled to generate a tessellation pattern using computationally-efficient techniques. Further graphical processes, such as rasterization, can be performed and only then are images from different scans combined. Thus, the system allows an image of the scan data to be rendered efficiently and/or quickly, particularly if the graphics capabilities of a GPU are used.

[0118] It will be appreciated that various modifications may be made to the embodiments hereinbefore described. Such modifications may involve equivalent and other features which are already known in the design, manufacture and use of laser scan systems and/or graphics processing systems, and component parts thereof, and which maybe used instead of or in addition to features already described herein. Features of one embodiment may be replaced or supplemented by features of another embodiment.


Claims

1. A method of rendering an image of three-dimensional laser scan data (6), the method comprising:

providing a plurality of sets of laser scan data (6) for a set of scans as spherical displacement maps (18), wherein each set of laser scan data corresponds to a respective scan;

generating a tessellation pattern (71) which comprises a set of triangles (72) by sampling the spherical displacement maps, wherein said generating of the tessellation pattern is performed by a graphics processing unit (34); and

converting each triangle into raster data (80) comprising pixels (81) each pixel including a value of depth (82);

writing a pixel (81) into a depth buffer (69) if the pixel (81) has a lower depth value (82) than a corresponding pixel stored in the depth buffer (69), and discarding a pixel (81) if the pixel (81) has a higher depth value (82) than a corresponding pixel stored in the depth buffer (69);

wherein pixel data generated from different scans is written to the same depth buffer (69).


 
2. A method according to claim 1, wherein providing each set of laser scan data for the set of scans as a spherical displacement map comprises:

receiving a set of laser scan data (6) generated by a laser scanner (2) at a given location; and

copying a value of range for a given laser scan point at a given azimuth and a given elevation from the laser scan data into a respective texel (46) of an at least two-dimensional texture at a texel position corresponding to the given azimuth and the given elevation.


 
3. A method according to any preceding claim, further comprising:

identifying discontinuities (47) between adjacent points in each spherical displacement map; and

marking said adjacent points.


 
4. A method according to any preceding claim, further comprising:
generating a set of normal maps (19) in dependence upon the set of said spherical displacement maps (18).
 
5. A method according to any preceding claim, wherein generating the set of normal maps (19) comprises calculating a normal (49) for each point in the set of spherical displacement map and storing the normal in a set of normal maps.
 
6. A method according to any preceding claim, further comprising:
generating a blending texture (20) in dependence upon each said spherical displacement map (18).
 
7. A method according to any preceding claim, further comprising:
generating a patch map (23) for a given scan, the patch map comprising a plurality of patches (51).
 
8. A method according to claim 7, wherein the patch map includes a relative tessellation level for each patch and, the method comprising:
setting the relative tessellation level (55) of a given patch in dependence upon discontinuities.
 
9. A method according to any preceding claim, further comprising:

determining whether a vertex (74, 77) forming part of a triangle (72, 78) in the tessellation pattern (71) is marked as being on or adjacent to a discontinuity;

in dependence upon determining that the vertex (77) is invalid, culling the triangle (78).


 
10. A computer program (39) which comprises instructions for performing a method according to any preceding claim.
 
11. A computer readable medium which stores a computer program according to claim 11.
 
12. A computer system comprising:

memory (32);

at least one processing unit (31, 34) comprising at least one graphics processing unit (34);
wherein the memory stores a plurality of sets of laser scan data (6) for a set of scans as spherical displacement maps (18), wherein each set of laser scan data corresponds to a respective scan and the at least one graphics processing unit (34) is configured to generate a tessellation pattern (71) comprising a set of triangles (72) by sampling the spherical displacement maps;

convert each triangle into raster data (80) comprising pixels (81) each pixel including a value of depth (82);

write a pixel (81) into a depth buffer (69) if the pixel (81) has a lower depth value (82) than a corresponding pixel stored in the depth buffer (69), and discard a pixel (81) if the pixel (81) has a higher depth value (82) than a corresponding pixel stored in the depth buffer (69);

wherein pixel data generated from different scans is written to the same depth buffer (69).


 
13. A computer system according to claim 13, wherein the at least one graphics processing unit (34) is configurable using a Microsoft (RTM) DirectX (RTM) 11 application programming interface.
 


Ansprüche

1. Verfahren zur Rasterung eines Bilds aus dreidimensionalen Laserscandaten (6), wobei das Verfahren Folgendes umfasst:

Bereitstellen mehrerer Sätze Laserscandaten (6) für einen Satz Scans als kugelförmige Displacement-Maps (18) wobei jeder Satz Laserscandaten einem jeweiligen Scan entspricht;

Generieren eines Tessellationsmusters (71), das einen Satz Dreiecke (72) umfasst, durch Abtasten der kugelförmigen Displacement-Maps, wobei das genannte Generieren des Tessellationsmusters von einem Graphikprozessor (34) durchgeführt wird; und

Umwandeln jedes Dreiecks in Rasterdaten (80), die Pixel (81) umfassen, wobei jedes Pixel einen Tiefenwert (82) aufweist;

Schreiben eines Pixels (81) in einen Tiefenpuffer (69), falls das Pixel (81) einen tieferen Tiefenwert (82) als ein entsprechendes im Tiefenpuffer (69) gespeichertes Pixel hat, und Verwerfen eines Pixels (81), falls das Pixel (81) einen höheren Tiefenwert (82) als ein entsprechendes im Tiefenpuffer (69) gespeichertes Pixel hat;

wobei aus verschiedenen Scans generierte Pixeldaten in den gleichen Tiefenpuffer (69) geschrieben werden.


 
2. Verfahren nach Anspruch 1, wobei das Bereitstellen von jedem Satz Laserscandaten für den Satz Scans als eine kugelförmige Displacement-Map Folgendes umfasst:

Empfangen eines Satzes Laserscandaten (6), die durch einen Laser-Scanner (2) an einer bestimmten Position generiert wurden; und

Kopieren eines Bereichswerts für einen bestimmten Laserscanpunkt an einem bestimmten Azimut und einer bestimmten Höhe von den Laserscandaten in einen jeweiligen Texel (46) von einer wenigstens zweidimensionalen Textur an einer Texelposition, die dem bestimmten Azimut und der bestimmten Höhe entspricht.


 
3. Verfahren nach einem der vorhergehenden Ansprüche, das ferner Folgendes umfasst:

Identifizieren von Diskontinuitäten (47) zwischen benachbarten Punkten in jeder kugelförmigen Displacement-Map; und

Markieren der genannten benachbarten Punkte.


 
4. Verfahren nach einem der vorhergehenden Ansprüche, das ferner Folgendes umfasst:
Generieren eines Satzes Normalen-Maps (19) in Abhängigkeit von dem Satz der genannten kugelförmigen Displacement-Maps (18).
 
5. Verfahren nach einem der vorhergehenden Ansprüche, wobei das Generieren des Satzes Normalen-Maps (19) das Berechnen einer Normalen (49) für jeden Punkt im Satz der kugelförmigen Displacement-Map und Speichern der Normalen in einem Satz Normalen-Maps umfasst.
 
6. Verfahren nach einem der vorhergehenden Ansprüche, das ferner Folgendes umfasst:
Generieren einer Blending-Textur (20) in Abhängigkeit von jeder genannten kugelförmigen Displacement-Map (18).
 
7. Verfahren nach einem der vorhergehenden Ansprüche, das ferner Folgendes umfasst:
Generieren einer Patch-Map (23) für einen bestimmten Scan, wobei die Patch-Map mehrere Patches (51) umfasst.
 
8. Verfahren nach Anspruch 7, wobei die Patch-Map ein relatives Tesselationsniveau für jeden Patch aufweist und wobei das Verfahren Folgendes umfasst:
Setzen des relativen Tessellationsniveaus (55) eines bestimmten Patchs in Abhängigkeit von Diskontinuitäten.
 
9. Verfahren nach einem der vorhergehenden Ansprüche, das ferner Folgendes umfasst:

Bestimmen, ob ein Vertex (74, 77), der Teil eines Dreiecks (72, 78) im Tessellationsmuster (71) bildet, als auf oder angrenzend an eine(r) Diskontinuität befindlich markiert ist;

in Abhängigkeit vom Bestimmen, dass der Vertex (77) ungültig ist, Entfernen des Dreiecks (78).


 
10. Computerprogramm (39), das Anweisungen zum Durchführen eines Verfahrens nach einem der vorhergehenden Ansprüche umfasst.
 
11. Computerlesbares Medium, das ein Computerprogramm nach Anspruch 11 speichert.
 
12. Computersystem, umfassend:

Speicher (32);

wenigstens einen Prozessor (31, 34), der wenigstens einen Graphikprozessor (34) umfasst;

wobei der Speicher mehrere Sätze Laserscandaten (6) für einen Satz Scans als kugelförmige Displacement-Maps (18) speichert, wobei jeder Satz Laserscandaten einem jeweiligen Scan entspricht und der wenigstens eine Graphikprozessor (34) konfiguriert ist zum Generieren eines Tessellationsmusters (71), das einen Satz Dreiecke (72) umfasst, durch Abtasten der kugelförmigen Displacement-Maps;

Umwandeln jedes Dreiecks in Rasterdaten (80), die Pixel (81) umfassen, wobei jedes Pixel einen Tiefenwert (82) aufweist;

Schreiben eines Pixels (81) in einen Tiefenpuffer (69), falls das Pixel (81) einen tieferen Tiefenwert (82) als ein entsprechendes im Tiefenpuffer (69) gespeichertes Pixel hat, und Verwerfen eines Pixels (81), falls das Pixel (81) einen höheren Tiefenwert (82) als ein entsprechendes im Tiefenpuffer (69) gespeichertes Pixel hat;

wobei aus verschiedenen Scans generierte Pixeldaten in den gleichen Tiefenpuffer (69) geschrieben werden.


 
13. Computersystem nach Anspruch 13, wobei der wenigstens eine Graphikprozessor (34) unter Verwendung einer Microsoft (RTM) DirectX (RTM) 11 Anwendungsprogrammierschnittstelle konfigurierbar ist.
 


Revendications

1. Procédé de restitution d'une image de données de balayage laser tridimensionnelles (6), le procédé consistant à :

fournir une pluralité d'ensembles de données de balayage laser (6) à un ensemble de balayages en tant que cartes de déplacement sphérique (18), chaque ensemble de données de balayage laser correspondant à un balayage respectif ;

générer un patron de tessellation (71) qui comprend un ensemble de triangles (72) en échantillonnant les cartes de déplacement sphérique, ladite génération du patron de tessellation étant réalisée par une unité de traitement graphique (34) ; et

convertir chaque triangle en données matricielles (80) comprenant des pixels (81), chaque pixel incluant une valeur de profondeur (82) ;

écrire un pixel (81) dans une mémoire tampon de profondeur (69) si le pixel (81) a une valeur de profondeur (82) inférieure à un pixel correspondant stocké dans la mémoire tampon de profondeur (69), et rejeter un pixel (81) si le pixel (81) a une valeur de profondeur (82) supérieure à un pixel correspondant stocké dans la mémoire tampon de profondeur (69) ;

les données de pixel générées à partir de différents balayages étant écrites dans la même mémoire tampon de profondeur (69).


 
2. Procédé selon la revendication 1, dans lequel la fourniture de chaque ensemble de données de balayage laser à l'ensemble de balayages en tant que carte de déplacement sphérique consiste à :

recevoir un ensemble de données de balayage laser (6) généré par un dispositif de balayage laser (2) à un emplacement donné ; et

copier une valeur d'un intervalle d'un point de balayage laser donnée à un azimut donné et à une élévation donnée, provenant des données de balayage laser, dans un texel (46) respectif d'une texture au moins bidimensionnelle dans une position de texel correspondant à l'azimut donné et à l'élévation donnée.


 
3. Procédé selon l'une quelconque des revendications précédentes, consistant en outre à :

identifier des discontinuités (47) entre des points adjacents dans chaque carte de déplacement sphérique ; et

marquer lesdits points adjacents.


 
4. Procédé selon l'une quelconque des revendications précédentes, consistant en outre à :
générer un ensemble de cartes normales (19) selon l'ensemble desdites cartes de déplacement sphérique (18).
 
5. Procédé selon l'une quelconque des revendications précédentes, dans lequel la génération de l'ensemble de cartes normales (19) consiste à calculer une normale (49) pour chaque point de l'ensemble de cartes de déplacement sphérique, et stocker la normale dans un ensemble de cartes normales.
 
6. Procédé selon l'une quelconque des revendications précédentes, consistant en outre à :
générer une texture de mélange (20) selon chacune desdites cartes de déplacement sphérique (18).
 
7. Procédé selon l'une quelconque des revendications précédentes, consistant en outre à :
générer une carte de corrections (23) pour un balayage donné, la carte de corrections comprenant une pluralité de corrections (51).
 
8. Procédé selon la revendication 7, dans lequel la carte de corrections comprend un niveau de tessellation relatif pour chaque correction, et le procédé consistant à :
régler le niveau de tessellation relatif (55) d'une correction donnée selon des discontinuités.
 
9. Procédé selon l'une quelconque des revendications précédentes, consistant en outre à :

déterminer si un sommet (74, 77) faisant partie d'un triangle (72, 78) dans le patron de tessellation (71) est marqué comme étant sur une discontinuité ou adjacent à une discontinuité ;

selon la détermination que le sommet (77) n'est pas valide, mettre au rebut le triangle (78).


 
10. Programme informatique (39) comprenant des instructions pour réaliser un procédé selon l'une quelconque des revendications précédentes.
 
11. Support lisible par ordinateur qui stocke un programme informatique selon la revendication 11.
 
12. Système informatique, comprenant :

une mémoire (32) ;

au moins une unité de traitement (31, 34) comprenant au moins une unité de traitement graphique (34) ;

la mémoire stockant une pluralité d'ensembles de données de balayage laser (6) à un ensemble de balayages en tant que cartes de déplacement sphérique (18), chaque ensemble de données de balayage laser correspondant à un balayage respectif, et l'au moins une unité de traitement graphique (34) étant configurée pour générer un patron de tessellation (71) comprenant un ensemble de triangles (72) en échantillonnant les cartes de déplacement sphérique ;

convertir chaque triangle en données matricielles (80) comprenant des pixels (81), chaque pixel incluant une valeur de profondeur (82) ;

écrire un pixel (81) dans une mémoire tampon de profondeur (69) si le pixel (81) a une valeur de profondeur (82) inférieure à un pixel correspondant stocké dans la mémoire tampon de profondeur (69), et rejeter un pixel (81) si le pixel (81) a une valeur de profondeur (82) supérieure à un pixel correspondant stocké dans la mémoire tampon de profondeur (69) ;

les données de pixel générées à partir de différents balayages étant écrites dans la même mémoire tampon de profondeur (69).


 
13. Système informatique selon la revendication 13, dans lequel l'au moins une unité de traitement graphique (34) peut être configurée au moyen d'une interface de programme d'application Microsoft (RTM) DirectX (RTM) 11.
 




Drawing






















































































Cited references

REFERENCES CITED IN THE DESCRIPTION



This list of references cited by the applicant is for the reader's convenience only. It does not form part of the European patent document. Even though great care has been taken in compiling the references, errors or omissions cannot be excluded and the EPO disclaims all liability in this regard.

Patent documents cited in the description




Non-patent literature cited in the description