EP2183633A2

EP2183633A2 - Panoramic projection device and method implemented by said device

Info

Publication number: EP2183633A2
Application number: EP08827211A
Authority: EP
Inventors: Robert André Marcel STEHLE; Michael Bray; Laurent Sarry; Eric Zeghers
Original assignee: MB Optique; Universite Clermont Auvergne
Current assignee: MB Optique; Universite Clermont Auvergne
Priority date: 2007-08-09
Filing date: 2008-08-08
Publication date: 2010-05-12
Also published as: FR2919934B1; EP2385405B1; HK1163824A1; EP2385405B9; US20110211175A1; EP2385405A1; WO2009022087A3; CA2695909A1; FR2919934A1; JP5486113B2; EP2385405B8; KR20100081971A; ES2394588T3; JP2013210663A; KR101319777B1; JP2010536061A; CA2695909C; WO2009022087A2; US8272752B2

Abstract

The invention relates to a projection device (1) that comprises: a projector (2) adapted for projecting a mother image along a projection axis (11); and a convex mirror adapted for at least partially reflecting the mother image and projecting it onto a three-dimensional environment (7, 8a, 8b) such as the ground, a ceiling, one or more side walls and the content of a room. When combining the projector with a convex mirror, the mother image can for example be projected into a solid angle higher than 2p steradians (i.e. a half-sphere), or even substantially equal to or slightly lower than 4p steradians (i.e. a sphere). The invention also relates to a method implemented by said device. The field of the invention is particularly that of video projectors for virtual or mixed reality applications.

Description

"Panoramic projection device and method implemented in this device"

Technical area

The present invention relates to a device for projecting an image, for example on the surfaces of a room. It also relates to a method implemented in this device.

The field of the invention is more particularly that of video projectors for virtual or mixed reality applications.

For a virtual reality application, the device according to the invention allows for example a user to project an image on the surfaces of a room, so that the projected image creates a virtual reality environment, an observer observing the image. projected image having the illusion of viewing a scene from a virtual point of view.

For a mixed reality application, the device according to the invention makes it possible, for example, for a user to project an image onto the surfaces of a room, so that the projected image creates a mixed reality environment by combining with real elements. of the room. State of the art

Examples of projection device of a sequence of images on a volume are known, this device comprising a single projector. The projector is arranged to project an image in a projection direction, the projected image growing as it propagates at a solid projection angle centered on the projection direction. The volume in which the device can project the image is limited by the opening of the projector, that is to say by the value of the solid projection angle. Thus, such a device whose projector is placed in the middle of a rectangular room can not project an image simultaneously on the four side walls of the room.

Projection devices are also known comprising several projectors, each projector illuminating a facet or a particular object of the volume. These devices have the disadvantage of understanding several projectors, which represents an additional cost, and which generally requires coordinating the projections of the projectors.

The object of the present invention is to provide a device for projecting an image from a single projector, without the volume or environment on which the image can be projected is limited by an opening of the projector.

DESCRIPTION OF THE INVENTION This object is achieved with a projection device, comprising:

a projector arranged to project a mother image along a projection axis, and

a convex mirror arranged to reflect at least part of the mother image and project it onto a three-dimensional environment.

The mirror is convex because it includes a convex surface arranged to reflect at least part of the mother image and project it on the environment.

Preferably, the projector is arranged to successively project different mother images, such as for example successive mother images of a video stream. These different mother images can have different sizes, and different positions relative to the projection axis.

The mirror may be a convex mirror having substantially a symmetry of revolution about an axis of revolution. In this case, the axis of revolution of the mirror and the projection axis are preferably substantially merged. In this document, we say that two axes are substantially confused either when they are exactly confused, or when they are confused to a near imperfection.

Preferably, the mirror is arranged to reflect the mother image by deforming the mother image. For example, the projector may be arranged to project the mother image at a first solid angle, and the mirror may be arranged to project the mother image at a second solid angle greater than the first solid angle. Preferably, the first solid angle is less than 2π steradians, that is to say, a half-sphere. In this document, it is said that the mother image is projected in a solid angle if the size of this mother image increases according to this solid angle as it propagates along a projection direction.

The second solid angle can have any value between 0 and 4π steradians, the value of the second solid angle depending in particular on the position of the mother image with respect to the axis of projection, the incidence of the axis projection on the mirror, and the size of the mother image. For a given mother image, the second solid angle may be greater than 2π steradians (ie a half-sphere), or even substantially equal to or slightly less than 4π steradians (ie a sphere ).

Similarly, the mirror can be arranged to project the mother image in an angle around the projection axis, the value of this angle can be between 0 degrees and 360 degrees, and depending in particular on the position of the image relative to the projection axis, the incidence of the projection axis on the mirror, and the size of the mother image. For a given mother image, the mirror may be arranged to project the mother image in an angle greater than 180 degrees about the projection axis, or even substantially equal to 360 degrees around the projection axis. In this document, an angle about an axis is defined in a plane perpendicular to that axis.

The device according to the invention may further comprise means for acquiring at least one image of the three-dimensional environment. An image of the environment can image part or all of the three-dimensional environment. The acquisition means may be arranged to acquire at least one image of the environment from light rays emitted by the environment and not reflected by the mirror. The acquisition means may comprise a camera, and means for rotating the camera about an axis of rotation. Preferably, the acquisition means are arranged to acquire several images of the environment, the sum of these images imaged the environment at 360 degrees around an axis such as the axis of rotation. Preferably, the projection axis and the axis of rotation are substantially merged. In addition, the device according to the invention may furthermore comprise means for modeling the three-dimensional environment from at least one image of the environment.

The device according to the invention may furthermore comprise means for constructing the mother image as a function of a position of an observer and possibly of a modeling of the three-dimensional environment, so that the observer observing from his position a projection of the mother image on the environment observes a scene from a given virtual point of view.

In addition, the device according to the invention may further comprise means for measuring at least one radiometric characteristic of the three-dimensional environment, and means for correcting the mother image as a function of at least one measured radiometric characteristic.

Finally, the device according to the invention may further comprise means for fixing the mirror to a wall, floor or ceiling.

According to another aspect of the invention, there is provided a projection method implemented in a device according to the invention, comprising:

a projection of a mother image along a projection axis,

a reflection of at least part of the mother image by a convex mirror, and

a projection on a three-dimensional environment of the mother image reflected by the convex mirror.

Part of the mother image may not be reflected by the mirror, this part of the mother image being directly projected onto the three-dimensional environment by the designer.

Preferably, the mirror reflects the mother image by deforming the mother image. For example, the projector can project the mother image at a first solid angle, and the mirror can project the mother image into a second solid angle larger than the first solid angle. Preferably, the first solid angle is less than 2π steradians, that is to say a half-sphere.

The second solid angle can have any value between 0 and 4π steradians, the value of the second solid angle depending in particular on the position of the image with respect to the axis of projection, the incidence of the axis projection on the mirror, and the size of the mother image. For a given mother image, the second solid angle may be greater than 2π steradians (ie a half-sphere), or even substantially equal to or slightly less than 4π steradians (ie a sphere ). Similarly, the mirror can project the mother image in an angle around the projection axis, the value of this angle can be between 0 degrees and 360 degrees. For a given mother image, the mirror can project the mother image at an angle greater than 180 degrees around the projection axis, or even substantially equal to 360 degrees around the projection axis.

The method according to the invention may further comprise an acquisition of at least one image of the three-dimensional environment.

Acquiring an image can include:

a positioning of a camera at an angular position about an axis of rotation, and

an acquisition of an image of the environment for this angular position of the camera, from light rays emitted by the three-dimensional environment.

The method according to the invention may further comprise a modeling of the three-dimensional environment, the mother image comprising at least one modeling point, the acquisition of images comprising an acquisition of at least one image of the environment for modeling, modeling comprising for each modeling point and for each image of the environment acquired for the modeling: a calculation of the trajectory of a light ray coming from the modeling point, reflected by the mirror, and projected onto a first point of the environment,

a calculation of the trajectory of a light ray coming from the first point of the environment and imaged by a point of the image of the environment acquired for the modeling, and

- A calculation of an intersection of the path of the light beam from the modeling point and the path of the light beam from the first point of the environment. The greater the number of modeling points, the higher the modeling resolution. The modeling points can be distributed over several mother images projected successively. Preferably, the images of the environment acquired for the modeling comprise several images of the environment, the sum of these images imaging the environment at 360 degrees around an axis such as the axis of rotation. It is thus possible to model the environment at 360 degrees around this axis.

The trajectory calculations may depend on calculation parameters, the method according to the invention may comprise a calibration of the parameters, the mother image comprising at least one calibration point, and the acquisition of images comprising an acquisition of at least one an image of the environment for calibration.

The images of the environment acquired for the calibration may comprise at least a couple of different images but which imitate a common part of the environment, the calibration comprising for each calibration point and for each pair of images:

an association of the calibration point with a point of the first image of the couple and with a point of the second image of the couple, the point of the first image of the couple and the point of the second image of the couple imaging a second point of the environment on which the calibration point is projected,

a calculation of a theoretical position of the point of the second image of the couple, starting from the position of the point of the first image of the couple, and of the calculation parameters, the calibration possibly comprising a modification of at least one calculation parameters, so as to minimize the sum, on all the calibration points, of the distances between the position of the point of the second image of the torque and the theoretical position of this point of the second image of the couple.

Calibration may also include for each calibration point and for each image of the environment acquired for calibration: a calculation of the trajectory of a light beam coming from the calibration point, reflected by the mirror, and projected onto a third point of the environment,

a calculation of the trajectory of a light ray coming from the third point of the environment and imaged by a point of the image of the environment acquired for the calibration, the calibration possibly further comprising a modification of at least one calculation parameters, so as to minimize the sum, on all the calibration points, of the distances between the trajectory of the light beam projected on the third point of the environment and the trajectory of the light beam coming from the third point of the environment .

For each of the calibrations described above, the greater the number of calibration points, and the better the calibration of the parameters. The calibration points can be distributed over several mother images projected successively. Preferably, the images of the environment acquired for the calibration comprise several images of the environment, the sum of these images imaged the environment at 360 degrees around an axis such as the axis of rotation.

Dots of the mother image can be used both as a modeling point and as a calibration point. Thus, first points of the environment, second points of the environment, and third points of the environment can be confused. Similarly, images of the environment can be acquired for both modeling and calibration. In this document, a "dot" may have a surface other than zero. Indeed, a "point" of the mother image (such as a modeling or calibration point) can typically consist of at least one pixel, preferably a single pixel. The size of a point of the environment and the size of a point of an image of the environment depend for example on the optical characteristics of the projector or the camera

(focal lengths, ...), the distance between the projector and the point of the environment, etc.

In addition, "distance between two trajectories" is understood to mean the length of the shortest straight path between the two trajectories. The method according to the invention may further comprise a construction of the mother image as a function of a position of an observer, and possibly a three-dimensional environment modeling, so that the observer observing from his position the projection of the mother image on the environment observes a scene from a given virtual point of view.

Finally, the method according to the invention may further comprise a measurement of at least one radiometric characteristic of the three-dimensional environment, and a correction of the mother image as a function of the at least one measured radiometric characteristic.

DESCRIPTION OF THE FIGURES AND EMBODIMENTS Other advantages and particularities of the invention will appear on reading the detailed description of implementations and non-limiting embodiments, and the following appended drawings: FIG. 1 is a view profile of an embodiment of the device according to the invention,

FIG. 2 is a view from above of this device,

FIG. 3 illustrates a mother image before its projection by this device, as it is displayed on the matrix of the projector of the device, FIG. 4 illustrates six perspectives of a scene, and

FIG. 5 illustrates the construction of the mother image of FIG. 3, from the perspectives of FIG. 4.

We will therefore describe, with reference to Figures 1 to 5, an embodiment of device 1 according to the invention implementing a method according to the invention.

The device 1 comprises a projector 2, a convex mirror 3, a camera 4, and means 5 for rotating the camera 4 about an axis of rotation 6. The device 1 is surrounded by a three-dimensional environment comprising different elements serving as projection screen, in particular lateral walls 7, a floor 8a, a ceiling 8b of a room, as well as objects included in the room. Each of these elements may not be flat, such as a curved object, or a ceiling of a cellar. The operation of the device 1 is as follows. We initially have a scene that we want to project on the three-dimensional environment with a desired eye impression for an observer. The device 1 can cover practically all the space of the environment by using on the one hand the video projector 2 which projects a mother image in an angularly limited beam, and on the other hand the recovery mirror 3 placed in a ceiling and which reflects the angularly limited beam to direct it to the three-dimensional environment in a solid angle preferably as large as possible. In order for the projection to be faithful to the desired eye impression, the device 1 takes into account the specificities of the projection screens: position, color, reflection coefficients, interaction with each other.

For this, it exploits images captured by the acquisition camera 4 which records the environment illuminated by different colored or primitive geometric backgrounds emitted by a matrix of the video projector. If the camera rotates around the projector's optical transmission axis, then it will be able to inform about all objects that serve as projection screen, with parallax effects, and without shadow area.

The information collected by the camera is processed and introduced in terms of the correction of the mother image to modify it and generate a corrected mother image that will, she, the necessary changes to a faithful reproduction of the scene on the screens, so that the observer has an impression of observing this scene with the desired eye impression.

The projector 2 is a matrix video projector, comprising a matrix 9 of pixels on which is displayed the mother image, and further comprising a lens 10. The projector is arranged to project via the lens 10 the mother image along the screen. a projection axis 11 orthogonal to the wavefront of the projected image, the size of the mother image increasing as it is propagated along the projection axis 11 according to a first solid projection angle 18 centered on the projection axis 11 and less than 2π steradians. Typically, the projector projects the mother image into a projection cone having an apex angle of 30 degrees, i.e., in a first solid angle 18 of π / 4 steradians. The optical axis 13 of the objective 10 substantially passes through the center of the matrix 9, and is substantially coincidental with the projection axis 11.

The mirror 2 is fixed to the ceiling 8b, the convex surface of the mirror 3 is oriented towards the projector 2, and the objective 10 is oriented towards the mirror 3, so that the convex surface of the mirror 3 is arranged to reflect at least part the mother image projected in the first solid angle by the projector, and to project on all or part of the three-dimensional environment and in a second solid angle the part of the mother image reflected. Thus, the device 1 is a catadioptric projection device, because it associates the matrix video projector 2 with the convex mirror 3 for reflecting the mother image in the three-dimensional environment 7, 8a, 8b.

In the example illustrated in Figure 1, the projection of the mother image comprises a projection of several light rays 14 to 17. The mirror 3 is arranged to reflect a portion 14 to 16 of these light rays. The mirror 3 is arranged to reflect some of these rays 14 towards the ground 8a, rays projected sufficiently close to the axis of projection 11 by the projector 2 being even reflected by the mirror on the projector 2. The mirror 3 is also arranged to reflect some of these rays 15 to the walls 7, and some of these rays 16 to the ceiling 8b. The mirror 3 is furthermore arranged so as not to reflect a peripheral part of the mother image, the light rays 17 coming from this part of the mother image being directly projected onto the three-dimensional environment 7, 8a, 8b by the projector 2 .

The mirror 3 has a symmetry of revolution about an axis of revolution 12. The mirror 3 can for example comprise a sphere portion whose convex surface is oriented towards the projector 2. To simplify the geometry of the device 1, the axis of revolution 12 of the mirror 3 is substantially coincident with the axis of rotation 6 of the camera 4, the projection axis 11, and the optical axis 13.

Thanks to the convex mirror 3, and unlike a projection device according to the state of the art, the device 1 is arranged to project the mother image in a panoramic manner around the projection axis 11. The mirror may for example projecting a given mother image in an angle greater than 180 degrees around the projection axis 11. For example, a mother image centered on the axis of revolution 12 and thus projection 11 is reflected and projected by the mirror to 360 degrees around the axis of projection 11, that is to say all around the axis of projection. On the other hand, for a mother image which, when it is displayed on the matrix 9, occupies only a part of the matrix without cutting or surrounding the axis of revolution 12 and thus projection 11, this mother image is reflected. and projected by the mirror 3 in an angle that can have any value less than 360 degrees around the projection axis. Depending on the mother image, this mother image can be projected on one or all of the side walls. The mirror 3 is positioned relative to the projector 2 so that the first solid projection angle 18 encompasses the convex surface of the mirror 3. The mirror 3 is arranged to project the mother image into a second solid angle, the size of the mirror. mother image increasing as its propagation from the mirror towards the environment according to the second solid angle. In the particular case where the axis of revolution 12 and the projection axis 11 are substantially merged, the second solid angle is also centered on the projection axis 11. The second solid angle can have any value between 0 and 4π steradians, the value of the second solid angle depending in particular on the position of the mother image with respect to the axis of revolution of the mirror or the projection axis, and the size of the mother image on the matrix 9. For a mother image centered on the axis of revolution of the mirror, and thus on the matrix 9, this second solid angle is typically greater than 2π steradians. As illustrated in FIG. 1, the second solid angle may even be slightly less than 4π steradians, the mirror being arranged to reflect and then project the mother image of the ground 8a to the ceiling 8b. The mother image can be projected onto all the side walls, floor and ceiling of the room. In FIG. 1, the second solid angle is delimited by the portions, located between the mirror 3 and the ceiling 8b, of the light rays 16. The camera 4 and the means for rotating the camera about its axis of rotation 6 allow to acquire images of the three-dimensional environment 7, 8a, 8b. The camera can be a CCD camera connected to an acquisition card, or a webcam with USB connection, or a simple digital camera with USB or wireless link (Bluetooth or WIFI). The rotation of the camera can be provided by a stepping motor, the transmission between the motor and a movement of the camera about its axis 6 is provided by any transmission means, for example a timing belt, and the rotation guidance of the camera is provided by a mechanical assembly, for example a large diameter bearing surrounding the projection axis 11, and allowing the light rays of the mother image projected by the projector to pass. For a position of the camera around its axis, the camera receives directly light rays emitted by the environment and not reflected by the mirror. From these light rays 20, the camera thus acquires an image of a part of the three-dimensional environment, that is to say of a part of the walls 7, of the ground 8a, of the ceiling 8b, and of certain objects contained in the room.

The camera 4 and the rotating means 5 form means of omnidirectional acquisition of the three-dimensional environment, these acquisition means being able to acquire images of parts of the three-dimensional environment distributed over more than 180 degrees around the projection axis 11. For example, a rotation of the camera 4 360 degrees around its axis of rotation 6 to acquire images of the three-dimensional environment all around the axis of rotation 6, and therefore all around the projection axis 11. However, there may be blind spots, that is to say parts of the three-dimensional environment that can not be seen by the camera, for example parts of the ground 8a hidden by the projector 2 , or parts of the ceiling 8b hidden by the mirror 3. To limit these blind spots the camera has a relatively wide field of view 19, typically of the order of 90 degrees and therefore a very short focal length. Similarly, to limit these blind spots, the camera is placed between the projector 2 and the mirror 3, so that the optical axis of the camera is substantially perpendicular to the projection axis 11, or is slightly inclined relative to the perpendicular to the projection axis. Finally, again to limit these blind spots, the rotating means 5 can be arranged to perform a series of successive rotations of the camera 4 360 degrees around the axis of rotation 6, and to change for each of these rotations successive the inclination of the camera relative to the perpendicular to the axis of projection 11 or rotation.

The camera is thus arranged to perceive the three-dimensional environment. The device 1 further comprises calculation means 21 connected to the camera 4. These calculation means typically comprise means for implementing algorithms or software, such as for example a processor, a central unit of a computer, or an electronic card. These calculation means are arranged to take into account a perception of the three-dimensional environment by the camera 4, then to perform on the mother image different types of correction according to the perception of the three-dimensional environment.

First, the computing means are arranged to model the three-dimensional environment from one or more images of the environment acquired by the camera. The greater the number of images used for the modeling, the more these images are distributed around the axis of rotation 6, and the larger the part of the three-dimensional environment that can be modeled. The calculation means are arranged to construct in three dimensions a modeling of the environment by triangulation between the mirror 3 and the acquisition means 4, 5. The calculation means are furthermore arranged to calculate, from the modeling, a distortion to be applied to the mother image according to a position of an observer 22, so that the observer observing from his position the projection of the mother image on the three-dimensional environment observes a scene from a given virtual point of view. The means of calculation thus make it possible to construct the mother image to give the observer the illusion of a virtual point of view.

On the other hand, the camera comprises means for measuring at least one radiometric characteristic of the three-dimensional environment, such as colorimetric characteristics or reflectance functions of the three-dimensional environment. The calculation means are further arranged to correct the mother image as a function of the measured radiometric characteristics, so that the observer 22 observes the projection of the mother image on the environment with colorimetric renditions and desired light intensities. Thus, the means of calculation allow the mother image to be faithfully reproduced from the point of view of colorimetry and light intensity, taking into account, in particular, the color, texture, and reflectance of each element of the three-dimensional environment on which the image mother is projected, and also taking into account the light sources present in the three-dimensional environment.

A method implemented in the device 1 comprises the following steps:

a projection, by the projector 2, of the mother image along the optical projection axis 11,

a reflection of the mother image by the mirror 3, and

a projection on the three-dimensional environment of the mother image reflected by the mirror 3.

As indicated above, the mirror projects the mother image preferably in a panoramic manner, for example in an angle A greater than

180 degrees around the axis of projection, or even 360 degrees around the axis of projection 11 that is to say on all the walls 7 of the room. In practice, this angle A can have any value between zero and 360 degrees, this value depending on the size of the mother image at the level of the matrix 9, and the position of the mother image on the matrix 9.

FIG. 3 illustrates an example of mother image 23 as it is displayed on the matrix 9 of the projector 2. The mother image 23 is centered on the matrix 9, and is thus cut substantially in its middle by the axis of projection 11. A portion 24 located at the periphery of the mother image 23, outside a closed line 25 (shown in dotted lines), is not reflected by the mirror 3 after having been projected by the projector 2, but is Directly projected on the three-dimensional environment by the projector 2. The light rays 17 from the pixels of this part 24 do not meet the mirror 3 and are directly projected on the ceiling 8b. This part 24 of the image can be exploited or extinguished; in practice, there is a tendency to extinguish the portion 24 of the matrix that has a direct impact on the environment and to limit itself to lighting the mirror alone. The other part 26 of the mother image 23, located inside the line 25, is projected by the projector 2 and then reflected by the mirror 3 on the environment three-dimensional. The shape of the line 25 depends on the shape of the mirror 3. In the case where the convex mirror has a shape of revolution, such as a sphere portion, the closed line 25 is a circle. The mother image 23 is projected on the three-dimensional environment by the projector 2 in the second solid angle greater than 2π steradians, or even substantially equal to or slightly less than 4π steradians.

The method further comprises constructing the mother image 23 according to a position of the observer 22, so that the observer 22 observing from its position the projection of the mother image on the environment observes a scene from a virtual point of view P. This construction is carried out as follows. The scene is stored in the projector 2 by storage means, in the form of six perspectives 27 to 32 illustrated in Figure 4. All these perspectives are views of the scene from the point of view P, these perspectives being views of the scene according to the six directions carried by the three axes of a three-dimensional orthogonal reference. These six perspectives are each contained in a square, so that by placing these six perspectives on the six inner faces of a cube 33 shown in Figure 5, the observer placed in the center of this cube observes the scene from point of view P. These perspectives can be photographs of a real landscape, or can be computer-generated images.

The six perspectives are then processed by an algorithm implemented by the calculation means 21, so as to participate in the construction of the mother image 23 projected by the projector 2. For this, the calculation means use a modeling of the three-dimensional environment, comprising a determination of the coordinates, in the frame linked to the projector 2, of points of the environment on which the mother image is projected. This modeling allows, for each point or pixel of the mother image 23, a creation of a correspondence table between the coordinates (u, v) of the pixel in the mother image 23, the coordinates of the point M _r of the three-dimensional environment on which this pixel is projected, and the direction of the line D connecting the point M _r and the position of the observer 22, then to associate with this pixel the color and the brightness of a point M _v , this point M _v being a point from one of six perspectives 27 to 32 and being positioned at the intersection of the cube 33 and the right line D, the cube 33 being positioned in the environment 7, 8a, 8b so that its center, that is to say the point of view P, is substantially confused with the position of the observer 22. The observer's position may be defined by default as the position of the projector 2, or may be any and defined by a user by means of capture or determined by means to locate the observer.

In one embodiment, the coordinates of the points of the environment on which the points of the mother image are projected are defined by a user, for example by means of input means enabling the user to define the dimensions of the image. room in which is the device 1, and for defining the position of the projector 2 and the mirror 3 in this room. In another embodiment, the geometry of the environment is known by an external information source, comprising optical sensors such as laser or ultrasound.

Finally, in a last embodiment of the method according to the invention, the modeling of the environment comprises: a projection, by the projector, of a modeling mother image comprising at least one modeling point;

a reflection of the mother image of modeling by the mirror 3;

a projection on the three-dimensional environment of the modeling mother image reflected by the mirror 3; and an acquisition of at least one image of the environment acquired for modeling, these images of the environment being preferably acquired for different angular positions of the camera 4 around the axis of rotation 6; and

- for each modeling point and for each image of the environment acquired for modeling:

An association of the modeling point with a point of the image of the environment acquired for modeling, this point of the image of the environment imaging a first point of the environment 34 on which is projected the modeling point, a calculation of the trajectory of a light beam 15 from the modeling point, reflected by the mirror, and projected on the first point of the environment, a calculation the trajectory of a light ray 20 coming from the first point of the environment 34 and imaged by the point of the image of the environment acquired for modeling, and a calculation of an intersection of the trajectory of the light beam 15 from the modeling point and the trajectory of the light ray 20 from the first point of the environment 34, the coordinates of the intersection point between these two paths being substantially the coordinates of the first point of the environment.

Thanks to several modeling points, we obtain the coordinates of several first points of the environment.

Calculated trajectories are simple because they consist of straight lines or straight portions in the case of geometrical optics. The trajectory calculations depend on calculation parameters, which will be described in more detail later.

In practice, because of possible imperfections in the knowledge of these calculation parameters, the calculated trajectory of the light beam 15 coming from the modeling point and the calculated trajectory of the light ray 20 coming from the first point of the environment do not intersect each other. not, but are very close to one place. By approximation, the coordinates of the middle of the shortest segment connecting these two trajectories are considered as the coordinates of the point of intersection between these two trajectories.

The modeling thus makes it possible, for each coordinate modeling point (u, v) on the mother image 23, and therefore by extension for each pixel of coordinates (u, v) on the mother image 23, to create the correspondence between the coordinates (u, v) of this modeling point or pixel, the coordinates of the first point of the environment 34 (or M _r ) on which this modeling point or pixel is projected, and the direction of the D right connecting this modeling point 34 and the position of the observer 22. All these coordinates and directions are preferably expressed in the reference frame of the projector 2.

The association can for example be performed by not projecting a mother image, the camera observing an image of the neutral environment, then projecting a mother image comprising a single modeling point, the camera observing another image of the image. environment corresponding to the neutral image to which is added the signal of the modeling point reflected by the mirror, projected on the environment, and re-transmitted to the camera. In one variant, the association may for example be carried out by projecting a mother image comprising several modeling points distributed according to a particular distribution, so that an image of the environment acquired for the modeling comprises several points reproducing substantially and at least in part this distribution, each of these points of the image of the environment can then easily be associated with one of the modeling points.

The higher the density of modeling points on the mother image, the better the resolution of the modeling of the three-dimensional environment. If for the modeling, each mother image comprises only one modeling point, the modeling comprises successive projections of mother images each having a different modeling point, the resolution of the modeling increasing with the number of projected successive mother images. .

Preferably, the images of the environment acquired for modeling are made for different angular positions adjacent to the camera. Indeed, a single image of the environment does not necessarily image all the points of the environment on which all the modeling points are projected. These neighboring angular positions are preferably distributed over 360 degrees around the axis of rotation, so that two images acquired for two neighboring angular positions imitate a common part of the three-dimensional environment. Thus, the modelization makes it possible to model on 360 degrees around the axis of rotation 6, and thus makes it possible to model the whole of the environment three-dimensional, or almost all of the environment if there are blind spots.

Thus, the three-dimensional coordinates of the points of the environment on which the mother image is projected can be obtained in the reference linked to the projector 2 by triangulating between the mirror 3 on one side and the camera 4 on the other. In fact, a knowledge of parameters of the device 1 makes it possible to calculate the equations of the trajectories of the rays coming from the mirror 3 and the equations of the trajectories of the rays picked up by the camera, and the calculation of their intersections. The three-dimensional environment can therefore be completely modeled, either by an unstructured cloud-like approach of modeling points, or by a mesh-type approach. This last case involves projecting modeling points in an orderly manner to know how to connect the points of the environment 34 obtained. It is also possible to analyze a cloud of points in the environment to recognize plan portions or other geometric shapes.

As previously stated, the trajectory calculations are simple geometrical optics calculations that depend on calculation parameters. These calculations are preferably carried out in the reference frame of the projector 2.

For example, the calculation of the trajectory of a light beam 15 coming from a modeling point, reflected by the mirror, and projected onto the first point of the environment 34 depends in particular on the following parameters:

the position of the modeling point on the matrix 9, this point preferably comprising a single pixel, or possibly comprising a group of assembled pixels,

the dimension of the matrix 9 and the resolution in pixels of the matrix 9,

the relative position of the matrix 9 with respect to the objective 10; in particular, the projection axis 11 and the optical axis 13 of the objective 10 pass substantially through the middle of the matrix 9, but it is possible that there is a slight shift or inclination between this center and these axes 11 , 13, this offset and this inclination to be known or estimated,

the focal length of the objective 10,

the optical distortions or aberrations of the objective 10, the geometry and the dimensions of the mirror 3,

the relative position of the projector 2 with respect to the mirror 3, in particular the distance and the alignment between the projector 2 and the mirror 3; indeed, the projection axis 11 and the axis of revolution 12 are substantially merged, but it is possible that there is a slight shift and / or inclination between these two axes 11, 12, this offset and this inclination to be known or estimated.

Likewise, the calculation of the trajectory of a light ray 20 coming from the first point of the environment 34 depends in particular on the following parameters:

the focal length of the camera 4,

- for the camera sensor, the dimensions and resolution of the sensor on which the three-dimensional environment is imaged,

the optical distortions or aberrations of the camera 4, the relative position of the camera with respect to the projector 2 during the acquisition of the first image; the angular position of the camera 4 must be known, the rotation pitch of the camera about its axis 6 can for example be deduced from the pitch of the motor used to rotate the camera; in addition, the projection axis 11 and the axis of rotation 6 are substantially merged, but it is possible that there is a slight shift and / or inclination between these two axes 11, 6, this offset and this inclination to be known or estimated.

To be able to carry out the modeling step described previously, all these parameters must be known. Some of these parameters can be memorized by storage means. For example the focal lengths, dimensions and resolutions of the camera and the matrix 9 may be provided by a manufacturer. The distortions or aberrations of the camera or lens 10 may for example be known by means of patterns. However, it may be necessary to measure the relative positions of the mirror 3, the camera 4, and the rotational movement of the camera relative to the projector 2. In fact, a user can make approximate alignments of the axes of rotation 6, projection 11, of revolution 12. However, the accuracy of these alignments is generally insufficient to know the transformations between the repositories related to the projector, the mirror and the camera, and is therefore insufficient to correctly perform the trajectory calculations. The method according to the invention therefore preferably comprises an optimization or calibration of the calculation parameters.

The general principle of calibration is to project calibration points on the three-dimensional environment, having known positions in the mother image, and to deduce some of the parameters from the coordinates of the projections of the calibration points in an image acquired. by the camera for a known angular position. The calibration points can be distributed on the mother image in the form of a scatter plot or in the form of lines. Calibration does not at any time use the three-dimensional coordinates of the points of the environment on which the calibration points are projected, which makes it independent of the geometry of the three-dimensional environment. It is also not a matter of detecting characteristic points of the environment as do the standard stereovision calibration methods, but of using the projector 2 to create them. This has the advantage of being able to project calibration points sufficiently contrasted, for example clear calibration points on a dark background, to be detected by the camera whatever the nature of the materials of the environment. A subtracted image procedure makes it possible to increase the contrast of the device 1 in a very significant way, by avoiding the differences in materials and ambient light in the environment. The subtracted image procedure typically comprises subtracting, from an image acquired by the camera during the projection of a mother image including calibration points, an image acquired by the camera during projection of a mother image comprising no calibration points. For this purpose, the camera preferably comprises automatic gain control means, to remain in the range of linearity of the camera sensor avoiding saturations that penalize the subtraction. The calibration then comprises a projection of a mother image comprising at least one calibration point, and an acquisition of at least one image of the environment acquired for the calibration.

The images of the environment acquired for the calibration may comprise at least a couple of different images but which imitate a common part of the environment, a first type of calibration comprising for each calibration point and for each pair of images:

a calculation of a theoretical position of the point of the second image of the couple, from the position of the point of the first image of the couple, and of the calculation parameters,

a calculation of a distance between the position of the point of the second image of the couple and the theoretical position of this point of the second image of the couple. This sequence of steps is followed by a modification of at least one of the calculation parameters, and a new calculation of the theoretical positions, so as to minimize the sum, on all the calibration points, of the distances between the position of the point the second image of the couple and the theoretical position of this point of the second image of the couple. Thus, from image acquisitions for different angular positions of the camera, these images imaged points of the environment regularly distributed in the three-dimensional environment, and by exploiting a redundancy of the points of the environment between two consecutive images. the calculation parameters can be optimized by minimizing the quadratic distance between the points of an image acquired by the camera and the theoretical position of the points of a successive image after rotation.

This first type of calibration makes it possible in particular to optimize camera distortion parameters, and parameters of rigid transformation of the reference frame of the camera 4 with respect to the reference of the axis of rotation 6, such as the angle between the optical axis of the camera 4 and the axis of rotation 6.

A second type of calibration may include, for each calibration point and for each image of the environment acquired for calibration:

a calculation of the trajectory of a light beam coming from the calibration point, reflected by the mirror, and projected onto a third point of the environment, a calculation of the trajectory of a light ray coming from the third point of the environment and imaged by a point of the image of the environment acquired for calibration,

- A calculation of a distance between the trajectory of the light beam projected on the third point of the environment and the path of the light beam from the third point of the environment.

This sequence of steps is followed by a modification of at least one of the calculation parameters, and new calculations of the trajectories of the rays coming from the calibration points and the trajectories of the rays imaged by points of the at least one image. of the environment acquired for the calibration, so as to minimize the sum, on all the calibration points, of the distances between the trajectory of the light beam projected on the third point of the environment and the trajectory of the light ray coming from the third point of the environment.

In other words, the number of intersections between the trajectory of the light beam projected on the third point of the environment and the trajectory of the light beam coming from the third point of the environment is maximized on all the calibration points.

This second type of calibration makes it possible to optimize in particular values for: rigid transformation parameters between the mirror reference frame

3 and the reference frame of the projector 2, such as for example the distance between the mirror and the projector, and an offset or inclination between the axis of revolution 12 of the mirror and the projection axis 11; rigid transformation parameters between the reference frame of the axis of rotation 6 of the camera 4, and the reference frame of the projector 2, such as, for example, the distance between the camera 4 and the projection axis 11, the inclination between the optical axis of the camera 4 and the projection axis 11, and an offset or inclination between the axis of rotation 6 and the projection axis 11.

For the first and second type of calibration, the greater the number of calibration points on the mother image, the better the optimization of the parameters. Finally, the method according to the invention may comprise a measurement of at least one radiometric characteristic of the three-dimensional environment, and a correction of the mother image as a function of at least one measured radiometric characteristic. The correction may include color correction or intensity of the mother image. Thus, the combination of the projector and the camera allows a comparison of the color and intensity reflected by the three-dimensional environment with a color and intensity expected from the projection of the mother image on the three-dimensional environment. This makes it possible to apprehend the restitution of the colors by the environment, and to ensure the fidelity of the projection of the mother image on the environment compared to the scene which served as a basis of construction for the mother image.

Of course, the invention is not limited to the examples that have just been described and many adjustments can be made to these examples without departing from the scope of the invention. In particular, the axes of rotation 6, of projection 11, of revolution

12 and the optical axis 13 may not be substantially merged. The calculations of the trajectories of the light rays emitted by the projector and the trajectories of the light rays emitted by the points of the environment are then more complex, since they must take into account significant offsets and / or inclinations between these axes, the modeling the three-dimensional environment and the construction of the mother image are consequently also more complex.

In addition, the convex surface of the mirror 3 does not necessarily have a shape of sphere portion, this shape of sphere portion making the resolution vertical device 1 inhomogeneous. Indeed, in the device 1 just described, the more a pixel is away from the center of the matrix 9, the larger the portion of the mirror reflecting the light rays 14 to 16 emitted by this pixel, and the larger the point size. of the environment on which this ray is projected is great. A mirror profile 3 is therefore preferably used which tends to homogenize the vertical resolution of the device 1 with respect to a spherical mirror.

Claims

1. Device (1) for projection, comprising:

a projector (2) arranged to project a mother image (23) along a projection axis (11), and

a convex mirror (3) arranged to reflect at least part of the mother image and to project it onto a three-dimensional environment (7, 8a, 8b).

2. Device according to claim 1, characterized in that the mirror (3) is a convex mirror having substantially a symmetry of revolution about an axis of revolution (12).

3. Device according to claim 2, characterized in that the axis of revolution (12) of the mirror and the projection axis (11) are substantially merged.

4. Device according to any one of claims 1 to 3, characterized in that the mirror is arranged to reflect the mother image by deforming the mother image.

5. Device according to claim 4, characterized in that the projector is arranged to project the mother image in a first solid angle, and in that the mirror is arranged to project the mother image in a second solid angle higher than the first solid angle.

6. Device according to any one of claims 1 to 5, characterized in that the mirror is arranged to project the mother image in a solid angle greater than 2π steradians, preferably slightly less than 4π steradians.

7. Device according to any one of claims 1 to 6, characterized in that the mirror is arranged to project the mother image in an angle greater than 180 degrees about the axis of projection (11), preferably substantially equal to 360 degrees around the axis of projection.

8. Device according to any one of claims 1 to 7, characterized in that it further comprises means (4, 5) for acquiring at least one image of the three-dimensional environment.

9. Device according to claim 8, characterized in that the acquisition means are arranged to acquire at least one image of the environment from light rays (20) emitted by the environment and not reflected by the mirror.

10. Device according to claim 8 or 9, characterized in that the acquisition means comprises a camera (4), and means for rotating (5) the camera about an axis of rotation (6).

11. Device according to claim 10, characterized in that the projection axis (11) and the axis of rotation (6) are substantially merged.

12. Device according to any one of claims 8 to 11, characterized in that it further comprises means (21) for modeling the three-dimensional environment from at least one image of the environment.

13. Device according to any one of claims 1 to 12, characterized in that it further comprises means (21) for constructing the mother image according to a position of an observer (22), so that the observer observing from his position a projection of the mother image on the environment observes a scene from a virtual point of view (P).

14. Device according to any one of claims 1 to 13, characterized in that it further comprises means for measuring at least one radiometric characteristic of the three-dimensional environment, and in that it comprises means for correcting the mother image as a function of at least one measured radiometric characteristic.

15. Device according to any one of claims 1 to 14, characterized in that it further comprises means for fixing the mirror

(3) to a wall, floor or ceiling.

16. Projection method implemented in a device according to any one of the preceding claims, comprising: - a projection of a mother image (23) along a projection axis

(U), and

a reflection of at least part of the mother image by a convex mirror (3), and

a projection on a three-dimensional environment (7, 8a, 8b) of the mother image reflected by the convex mirror.

17. The method of claim 16, characterized in that the mirror reflects the mother image by deforming the mother image.

18. The method of claim 17, characterized in that the mother image is projected along the projection axis in a first solid angle, and in that the mirror projects the mother image into a second solid angle greater than the first solid angle.

19. Method according to any one of claims 16 to 18, characterized in that the mirror projects the mother image in a solid angle greater than 2π steradians, preferably slightly less than 4π steradians.

20. Method according to any one of claims 16 to 19, characterized in that the mirror projects the mother image in an angle greater than 180 degrees about the projection axis (11), preferably substantially equal to 360 degrees around the projection axis.

21. Method according to any one of claims 16 to 20, characterized in that it further comprises an acquisition of at least one image of the three-dimensional environment.

22. The method as claimed in claim 21, characterized in that the acquisition of an image comprises:

a positioning of a camera (4) at an angular position about an axis of rotation (6), and

an acquisition of an image of the environment for this angular position of the camera, from light rays (20) emitted by the three-dimensional environment.

23. The method of claim 21 or 22, characterized in that it further comprises a modeling of the three-dimensional environment, the mother image comprising at least one modeling point, the image acquisition comprising an acquisition of at least one image of the environment acquired for the modeling, the modeling comprising for each modeling point and for each image of the environment acquired for modeling: a calculation of the trajectory of a light ray (15) coming from the modeling point, reflected by the mirror, and projected on a first point of the environment,

a calculation of the trajectory of a light ray (20) coming from the first point of the environment (34) and imaged by a point of the image of the environment acquired for modeling, and

- A calculation of an intersection of the path of the light beam (15) from the modeling point and the path of the light beam (20) from the first point of the environment (34).

24. The method according to claim 23, characterized in that the trajectory calculations depend on calculation parameters, in that it comprises a calibration of the parameters, in that the mother image comprises at least one calibration point, and what image acquisition includes a acquiring at least one image of the environment acquired for calibration.

25. The method according to claim 24, characterized in that the images of the environment acquired for the calibration comprise at least a couple of different images but which imitate a common part of the environment, the calibration comprising for each calibration point and for each pair of images:

a calculation of a theoretical position of the point of the second image of the pair, starting from the position of the point of the first image of the pair, and of the calculation parameters.

26. The method of claim 25, characterized in that the calibration further comprises a modification of at least one of the calculation parameters, so as to minimize the sum, on all the calibration points, of the distances between the position of the point. of the second image of the couple and the theoretical position of this point of the second image of the couple.

27. Method according to any one of claims 24 to 26, characterized in that the calibration comprises for each calibration point and for each image of the environment acquired for the calibration:

a calculation of the trajectory of a light beam coming from the calibration point, reflected by the mirror, and projected onto a third point of the environment, a calculation of the trajectory of a light ray coming from the third point of the environment and imaged by a point of the image of the environment acquired for calibration.

28. The method according to claim 27, characterized in that the calibration further comprises a modification of at least one of the calculation parameters, so as to minimize the sum, on all the calibration points, of the distances between the trajectory of the beam. light projected on the third point of the environment and the trajectory of the light beam from the third point of the environment.

29. A method according to any one of claims 16 to 28, characterized in that it further comprises a construction of the mother image according to a position of an observer (22), so that the observer observing from its position the projection of the mother image on the environment observes a scene from a virtual point of view (P).

30. Method according to any one of claims 16 to 29, characterized in that it further comprises a measurement of at least one radiometric characteristic of the three-dimensional environment, and in that it comprises a correction of the mother image according to at least one measured radiometric characteristic.