Search

EP-4740177-A1 - METHOD FOR 3D RECONSTRUCTION

EP4740177A1EP 4740177 A1EP4740177 A1EP 4740177A1EP-4740177-A1

Abstract

The present invention relates to a method for 3D reconstruction of a scene (2). The method comprises the step of providing a first camera (3) having a known position and orientation. The method further comprises the step of determining a first set of projection lines corresponding to pixels (11) of said first camera (3), said projection lines being defined by the position and orientation of said camera (3) and the corresponding pixel (11). The method further comprises the step of projecting a light beam (13) on the scene (2) by means of a light source (12), thereby defining a dot (1) on said scene (2). The method further comprises the step of determining the pixel corresponding to an image of the dot (1) on said first camera (3), thereby defining a corresponding first projection line (9) from the first set of projection lines. The method further comprises the step of determining a second projection line (10) from a second set of projection lines either corresponding to instantaneous positions and orientations of the light beam (13) or to an image of the dot (1) on a second camera (4). The method further comprises the step of calculating a 3D location (17) of said dot (1) on the scene (2) based on said first and second projection lines (9, 10), wherein said 3D location (17) is defined in a region (18) in which the first projection line (9) and second projection line (10) are closest to each other.

Inventors

  • Pataridis, Konstantinos
  • Rogge, Ségolène

Assignees

  • VoxelSensors SRL

Dates

Publication Date
20260513
Application Date
20240701

Claims (15)

  1. 1. A method for 3D reconstruction of a scene (2), comprising the steps of: providing a first camera (3); projecting a light beam (13) on the scene (2) by means of a light source (12) thereby defining a dot (1) on said scene (2), wherein said first camera (3) having a known relative position and orientation to said light source (12) or to a second camera (4); determining a first set of projection lines corresponding to pixels (11) of said first camera (3), said projection lines being defined by the known relative position and orientation of said first camera (3), and the corresponding pixel (11); determining the pixel corresponding to an image of the dot (1) on said first camera (3), thereby defining a corresponding first projection line (9) from the first set of projection lines; determining a second projection line (10) from a second set of projection lines either corresponding to instantaneous positions and orientations of the light beam (13) or to an image of the dot (1) on said second camera (4); calculating a 3D location (17) of said dot (1) on the scene (2) based on said first and second projection lines (9, 10), wherein said 3D location (17) is defined in a region (18) in which the first projection line (9) and the second projection line (10) are closest to each other.
  2. 2. A method according to claim 1, wherein said region (18) is defined on a perpendicular line (35) on said first projection line (9) and second projection line (10), preferably wherein said 3D location (17) is midway between said projection lines (9, 10).
  3. 3. A method according to any of claims 1 or 2, wherein in case of absence of pre-defined projection lines in said first and/or second sets of projection lines, the method further comprises the steps of: defining said first projection line (9) on said first set of projection lines, and/or defining said second projection line (10) on said second set of projection lines.
  4. 4. A method according to any of claims 1 to 3, wherein the method further comprises the step of updating the projection lines in said first and/or second set of projection lines.
  5. 5. A method according to any of claims 1 to 4, wherein the method comprises the step of scanning said light beam (13) on said scene (2) along a trajectory (16), wherein said trajectory (16) comprises a plurality of points (14, 15) on said scene (2).
  6. 6. A method according to any of claims 1 to 5, wherein said first and/or second sets of projection lines are progressively filled with projection lines.
  7. 7. A method according to any of claims 1 to 6, wherein the method comprises the step of arranging the plurality of sensing units (11) in M rows and N columns, wherein each projection line (9, 10) or each predetermined number of projection lines (9, 10) corresponds to at least one pixel (11).
  8. 8. A method according to any of claims 1 to 7 , wherein the method comprises the step of providing a singlephoton detector within each pixel (11), preferably SPADs.
  9. 9. An optical sensing system (100) for 3D reconstruction of a scene (2), comprising: a first camera (3) wherein said first camera (3) comprises a plurality of pixels (11), each of said pixels (11) comprising a photodetector a light source (12), wherein said light source (12) is adapted to project a light beam on the scene (2) thereby defining a dot (1) on said scene (2), wherein said first camera (3) having a known relative position and orientation to said light source (12) or to a second camera (4), optics able to produce an image of said scene (2) on said camera (3) or cameras (3, 4), and a memory element comprising a first and a second set of projection lines, wherein said first set of projection lines correspond to pixels (11) of said first camera (3), said projection lines being defined by the known relative position and orientation of said first camera (3), and the corresponding pixel (11), wherein the system (100) is adapted to determine the pixel corresponding to an image of the dot (1) on said first camera (3), thereby defining a corresponding first projection line (9) from the first set of projection lines, the system (100) being further adapted to determine a second projection line (10) from the second set of projection lines, either corresponding to instantaneous positions and orientations of the light beam (13) or to an image of the dot (1) on said second camera (4), wherein the system (100) is adapted to calculate a 3D location (17) of said dot (1) on the scene (2) based on said first and second projection lines (9, 10), wherein said 3D location (17) is defined in a region (18) in which the first projection line (9) and second projection line (10) are closest to each other.
  10. 10. An optical sensing system (100) according to claim 9, wherein said region (18) is defined on a perpendicular line (35) on said first projection line (9) and second projection line (10), preferably wherein said 3D location (17) is midway between said projection lines (9, 10).
  11. 11. An optical sensing system (100) according to any of claims 9 to 10, wherein the system (100) is adapted, in case of absence of pre-defined projection lines in said first and/or second set of projection lines , to define the first projection line (9) on said first set of projection lines, and to define the second projection line (10) on said second set of projection lines.
  12. 12. An optical sensing system (100) according to any of claims 9 to 11, wherein the system (100) further comprises scanning means, wherein said scanning means is adapted to scan said light beam (13) on said scene (2) along a trajectory (16), wherein said trajectory (16) comprises a plurality of points (14, 15) on said scene (2).
  13. 13. An optical sensing system (100) according to any of claims 9 to 12, wherein the system (100) is adapted to progressively fill said memory element with projection lines.
  14. 14. An optical sensing system (100) according to any of claims 9 to 13, wherein the pixels (11) are arranged in M rows and N columns, wherein each projection line (9, 10) or each predetermined number of projection lines (9, 10) correspond to at least one pixel.
  15. 15. An optical sensing system (100) according to any of claims 9 to 14, wherein the photodetectors are singlephoton detectors, preferably SPADs.

Description

METHOD FOR 3D RECONSTRUCTION TECHNICAL FIELD The present invention relates to a method for 3D reconstruction. BACKGROUND There are several ways to reconstructing a 3D scene, one of which is triangulation, wherein two cameras are used to capture information of a scene, and by triangulating the data (2D) of the two cameras. In a first step, 2D data of the scene is obtained by each camera, after which the 2D data of both cameras is triangulated to obtain the 3D information of the scene. This can be done by performing said triangulation using mathematical operations at each point in the scene. However, this process is time- and energy-consuming. There is a need, therefore, for a method that allows low-power and low-latency 3D reconstruction. The present invention aims at resolving, at least in part, the problems mentioned above. SUMMARY OF THE INVENTION It is an object of embodiments of the present invention to provide a fast and low-power method for 3D reconstruction. The above objective is accomplished by a system and method according to the present invention. In a first aspect, the present invention relates to a method for 3D reconstruction of a scene, comprising the steps of: providing a first camera having a known position and orientation; determining a first set of projection lines corresponding to pixels of said first camera, said projection lines being defined by the position and orientation of said camera and the corresponding pixel; projecting a light beam on the scene by means of a light source thereby defining a dot on said scene; determining the pixel corresponding to an image of the dot on said first camera, thereby defining a corresponding first projection line from the first set of projection lines; determining a second projection line from a second set of projection lines either corresponding to instantaneous positions and orientations of the light beam or to an image of the dot on a second camera; calculating a 3D location of said dot on the scene based on said first and second projection lines, wherein said 3D location is defined in a region in which the first projection line and second projection line are closest to each other. It is an advantage of embodiments of the present invention that a fast 3D reconstruction is obtained, since the 3D location is obtained based on projection lines that are previously defined e.g. pre-computed or predefined. It is an advantage of embodiments of the present invention that the power consumption is reduced since less processing needs to be done. It is an advantage of embodiments of the present invention that the amount of memory needed is low since the method only needs to store projection lines. It is an advantage of embodiments of the present invention that the needed processing time and memory do not scale when having more than two optical sensors since the projection lines are pre-computed. Preferred embodiments of the first aspect of the invention comprise one or a suitable combination of more than one of the following features. Said region is preferably defined on a perpendicular line on said first projection line and said second projection line, preferably wherein said 3D location is midway between said projection lines. It is an advantage of embodiments of the present invention that this assumption makes it easy and fast to find the 3D location while at the same time providing a good estimate of the 3D location. In the absence of pre-defined projection lines in said first and/or set of second projection lines, the method preferably further comprises the steps of defining said first projection line on said first set of projection lines and/or defining said second projection line on said second set of projection lines. It is an advantage of embodiments of the present invention that, in the event of changes in the system, such as calibration, the sets (e.g. look-up tables) can be refilled. The method further comprises the step of updating the projection lines in said first and/or second projection lines. It is an advantage of embodiments of the present invention that, in case of recalibration, the projection lines are updated. The method preferably comprises the step of scanning said light beam on said scene along a trajectory, wherein said trajectory comprises a plurality of points on said scene. It is an advantage of embodiments of the present invention that the need for stereo-matching is eliminated since the method relies on the position of the light spot. The first and/or second sets of projection lines are preferably progressively filled with projection lines. It is an advantage of embodiments of the present invention that the look-up tables can be used if the needed information is available or refilled if the needed information is unavailable so that it can be used in the future. The method preferably comprises the step of arranging the plurality of sensing units in M rows and N columns, wherein each projection line or each predetermined nu