FR3054359A1 - METHOD FOR INCRUSTATION OF SOURCE IMAGE IN TARGET IMAGE - Google Patents

Abstract

The invention provides a method of embedding a source image in a target image, comprising: providing a target image of a three-dimensional scene having a target surface delimited by a contour, a front-end object and a light source, providing a multiplicative mask comprising the diffuse reflection information of the light source on the target surface and the transparency of the frontal object, providing an additive mask comprising the specular reflection information of the light source on the target surface, and a piece of information binary for delimiting an area of the target image having the target surface, providing (130) a source image, generating (131) an instantiation of the source image by a three-dimensional transformation so that the source image corresponds to the shape of the image target surface, multiply (132) the instantiation of the source image with the multiplicative mask, add (133) the additive mask to the result, in order to obtain the increment station the source image in the target image.

Description

Technical area

The invention relates to the field of digital image processing, and in particular techniques for embedding a source image in a target image or in a film of target images. The target image may be a photograph or a digital image, and the film may be a film obtained by capture via a camera or by digital generation of an animated film.

Technological background

Inlay techniques, called in English "compositing", such as for example techniques using green or blue backgrounds called in English "green or blue screen compositing" make it possible to digitally embed a fictitious decor in a film or a photograph. These techniques do not make it possible to directly obtain a perfect rendering of the integration, in particular with regard to the transparency of frontal objects placed in the front of the fictitious decoration. Other techniques make it possible to obtain a high-quality photorealistic rendering of an inlay, but require very long and costly calculations and simulations in post production. In addition, the calculations and these simulations must be carried out for and in relation to each unique fictional decoration to be embedded.

summary

An idea on which the invention is based is to provide a method making it possible to superimpose in any limited time and in high quality any source image in a defined target image, for example to allow the integration in real time of the source image in a prerecorded three-dimensional animated film.

One aspect of the invention is to preprogram an additive mask and a multiplicative mask characterizing the target image.

One aspect of the invention is that the additive and multiplicative masks are independent of the source image to be integrated.

One aspect of the invention is to provide a method of embedding a source image into a target image which is rapid by pre-recording the information of diffuse light, brightness and transparency. In particular, its information is pre-recorded in two masks, and not in many layers or alpha channels which are costly to overlay.

According to one embodiment, the invention provides a method of overlaying a source image in a target image, comprising:

provide a target image, the target image being a two-dimensional view of a three-dimensional scene, the view being delimited by a frame according to a shooting angle and an observation point, the scene being illuminated by one or more light sources, the scene comprising: o an environment, o a target object having a visible target surface delimited by a contour in the frame, o a frontal object, at least part of the frontal object being in front of the target surface, part of the front object having a predetermined transparency, the visible target surface reflecting the light coming directly and indirectly from the light source or sources, the method further comprising:

delimit the frame in two zones, a first zone of the frame being delimited by the contour and corresponding to the inside of the contour and a second zone being a zone complementary to the first zone inside the frame, providing a multiplicative mask with dimensions of the framework, by providing:

o a first mask comprising information for combining the diffuse reflection coming directly from the light source or sources on the target surface at each point of the first zone, o a second mask comprising binary information for delimiting the first zone of the second zone and transparency information of its part of the frontal object on the target surface at each point of the first zone, o multiply the second and the first mask to obtain the multiplicative mask.

provide an additive mask with the dimensions of the frame, o additive mask comprising information for combining the specular reflection of the light source (s) on the target surface at each point of the first zone, and o the additive mask corresponding to the target image at each point in its second zone and at the part of the front object at each point in the first zone, provide a first source image, generate an instantiation of the source image to the dimensions of the first zone, instantiation of the source image being generated by a three-dimensional transformation of the source image so that the source image corresponds to the shape of the target surface of the target object, multiply the instantiation of the source image with the multiplicative mask so to obtain an intermediate mask, add the intermediate mask and the additive mask in order to obtain the integration of the source image into the target image.

Thanks to these characteristics, it is possible to obtain a multiplicative mask and an additive mask without performing color or brightness calculations on the source image.

Thanks to these characteristics, the integration of the source image from the multiplicative and additive masks is achievable with any graphics card.

According to embodiments, such a method can include one or more of the following characteristics.

In one embodiment, each point of the multiplicative mask in the first zone has a gray level, the presence of diffuse reflection at a point in the first zone resulting in a first shade of gray whose clarity increases when the intensity of the diffuse reflection at this point increases, the presence of the frontal object at a point in the first zone resulting in a second shade of gray whose clarity decreases when the transparency of the frontal object at this point decreases, the level of gray of the point being defined by the superposition of the first and second shades of gray at this point, each point of the additive mask in the first zone has a color, the color of the point being defined by a combination of a first shade of color and d a second shade of color at this point, the first shade of color corresponding to the specular reflection of the light source or sources on the target surface at this point and a second shade of color ur associated with an alpha channel defining the transparency corresponding to the part of the frontal object superimposed on the target surface at this point.

In one embodiment,

The first mask of the multiplicative mask is formed by a first instantiation of the target image in which the target surface according to the first instantiation is made of a mat material with defined diffuse reflection property and zero specular reflection property, the additive mask is consisting of a second instantiation of the target image in which its target surface according to the second instantiation is made of a material having a defined specular reflection property and a zero diffuse reflection property.

In one embodiment, the color in the second mask of the multiplicative mask is uniformly black at each point of its second zone and the transparency of the front object is simulated by a gray filter at any point of the first zone where the object frontal is present.

In one embodiment, the albedo in the second mask of the multiplicative mask is zero.

In one embodiment, the albedo in the first mask of the multiplicative mask is equal to 100%.

In one embodiment, the material of the target surface according to the first instantiation is white, and the material of the target surface according to the second instantiation is black.

In one embodiment, the material of the target surface according to the first and second instantiation has a defined transparency, for example simulated by an alpha channel.

In one embodiment, providing a target image comprises:

numerically define the three-dimensional scene comprising the target object, generate the two-dimensional view of the three-dimensional scene by digital synthesis of the three-dimensional scene.

In one embodiment, defining the three-dimensional scene numerically is achieved by polygonal synthesis. For example, software selected from the following software list: MAYA ®, MENTAL RAY ®, ARNOLD RENDER ®, RENDELMAN ®, V-RAY ®, RADIANCE ® etc can be used to numerically define the three-dimensional scene.

In one embodiment, the target image is a photograph taken from the three-dimensional scene.

In one embodiment, the multiplicative mask and the additive mask are obtained using a sodium lamp and a prism.

In one embodiment, the material of the target surface according to the first instantiation is mat and green in color.

In one embodiment, the target image according to the first instantiation is a first photograph taken from the three-dimensional scene comprising the target surface according to the first instantiation and the target image according to the second instantiation is a second photograph taken from the three-dimensional scene comprising the target surface according to the second instantiation.

For example, the first and second photographs are obtained using the same camera in movement control, called in English "motion control", the camera being motorized and able to make the same movements successively at different times. For example, the first photograph is acquired while the target surface of the decor is mat and green in order to regain transparency by post-production identification thanks to the green color. For example, the front object is only placed in the three-dimensional scene when the first photo is acquired.

Alternatively, a single photograph is acquired and the target surface is simulated in post-production from the identification of the contour by tracking called in English "tracking" and the lighting of the scene by the light source or sources is also simulated in post-production. This technique is particularly advantageous for embedding a source image in a film capture.

In one embodiment, the target image according to the first instantiation is a first photograph taken of the three-dimensional scene comprising the front object according to a first instantiation in which the front object is black and of albedo zero.

In one embodiment, delimiting the frame into two zones comprises:

numerically identify the coordinates of certain points of the contour.

In one embodiment, the digital identification of the coordinates is carried out by clipping called in English "roïoscopie".

In one embodiment, the method further comprises:

save the multiplicative mask and Additive mask in a digital memory.

In one embodiment, the method further comprises generating a film being a succession of target images, and providing the additive mask and the multiplicative mask for each target image of the film.

In one embodiment, the multiplicative mask and the additive mask are recorded target image by target image in order to embed a single source image in a film.

Brief description of the figures

The invention will be better understood, and other objects, details, characteristics and advantages thereof will appear more clearly during the following description of several particular embodiments of the invention, given solely by way of illustration and without limitation. , with reference to the accompanying drawings.

• Figure 1 is a diagram of a three-dimensional scene comprising a target surface and a transparent frontal object, • Figure 2 is a diagram of the three-dimensional scene of Figure 1 and a virtual camera acquiring a target image of the scene three-dimensional, in which the target surface has an outline, • Figure 3 is a mask constituted by a first instantiation of the target image of Figure 2 in which the target surface is constituted by a white and matt material, • Figure 4 is a mask delimiting the outline of its target surface, • Figure 5 is a mask comprising the transparency of the front object, • Figure 6 is a mask obtained by combining the masks in Figures 4 and 5, • Figure 7 is a multiplicative mask obtained by combining its masks in Figures 3 and 6, • Figure 8 is an additive mask consisting of a second instantiation of the target image of Figure 2 in which the target surface is made of a black material and reflective, • Figure 9 is a source image, • Figure 10 is a diagram of the source image of Figure 9 transformed three-dimensionally so that the source image corresponds to the shape of the target surface according to the point of view of the virtual camera of Figure 2, ® Figure 11 is a mask of the source image adapted to the contour of the target surface of Sa Figure 1, • Figure 12 represents the overlay of the source image in the target image thanks to an inlay process, • Figure 13 is a step diagram which represents an inlay process which can be used to obtain Figure 12 by using the multiplicative mask of Figure 7 and the additive mask of Figure 8.

Detailed description of embodiments

OBTAINING A TARGET IMAGE

FIG. 1 represents a three-dimensional scene 1 comprising a table 5 on which are arranged a frame 3 surrounding a target surface 2, and a transparent veil 4 disposed partly in front of the frame 3. The target surface 2 is delimited by an outline 105.

The three-dimensional scene is generated by digital simulation. For example, each of the objects in the three-dimensional scene is defined by polygonal synthesis. For each object, the shape of the object is defined in three dimensions by the three-dimensional coordinates of points on the surface of the object and an elementary surface between at least three surface points of an object is given texture characteristics. , transparency and color.

For example, the transparent veil 4 is transparent, white in color and has an opaque border 14.

The three-dimensional scene also includes two sources of direct light: the flame 8 of a light 7 on and the sun 6.

The simulation of the lighting of the three-dimensional scene 1 consists in defining points of the three-dimensional scene 1 which emit light, as well as the direction of emission of the light, the intensity of this light and its color. In the example shown, the sun is not covered by clouds and acts as a Lambertian light source of the solar spectrum from its position in the sky, at the zenith.

Photorealistic rendering generation software such as MENTAL RAY ® edited by NVIDIA ARC GmbH allows photorealistic generation of the three-dimensional scene 1.

For example, the software makes it possible to simulate a shadow 9 on the target surface 2 and the table 5. In fact, the software takes into account the transparency of the transparent veil 4 and the light coming from the sun.

Likewise, the simulated luminaire 7 has a translucent property. The software therefore provides that it acts as an indirect light source because the light emitted by the flame 8 also illuminates inside the luminaire 7.

The software also simulates other indirect light sources present in the three-dimensional scene 1, in particular by simulating reflections of light from flame 8 and from the sun 6 on frame 3 and table 5.

FIG. 2 represents the generation of a target image 10 corresponding to a two-dimensional view of the three-dimensional scene 1. This view is delimited by a frame 11 according to a shooting angle and an observation point

12. The generation of the target image 10 is represented symbolically in FIG. 2 by the use of a virtual camera 13.

The target image 10 corresponds to the simulation of the two-dimensional view of the three-dimensional scene 1 by the software.

For example, this simulation is also performed using the MENTAL RAY ® software.

OBTAINING A MULTIPLICATIVE MASK AND A MASK

ADDITIVE

What is explained below explains one way of obtaining masks in order to embed any source image in the target image 10.

FIG. 3 represents a first instantiation 101 of the target image 10 of the three-dimensional scene 1 acquired as explained with reference to FIGS. 1 and 2.

In this first instantiation 101, the frame 3 comprises a target surface 21 according to a first instantiation.

The target surface 21 is defined during the generation of the three-dimensional scene by the software with very precise properties:

the target surface 21 is perfectly mat, that is to say that it has defined diffuse reflection properties and zero specular reflection properties,

- The target surface 21 is of defined color.

In the example shown, the target surface 21 is white in color and is planar.

The target image 10 according to its first instantiation 101 is used to generate a multiplicative mask by a succession of several steps, as described with reference to FIGS. 4 to 7.

In the following description, the term "image" designates a set of pixels each having coordinates defined in the image plane.

Thus, the target image 10 according to its first instantiation 101 is a set of pixels each having coordinates defined in the plane of the first instantiation 101.

Each pixel is coded by a vector of type "Red, Green, Blue" (RGB), in French "Rouge, Vert, Bleu". Such a vector has three components, one for each of the three colors. Optionally; the vector of the pixel also comprises a fourth component called “alpha channel” comprising a value representing the transparency of the pixel.

In the following example, a pixel is represented by a vector of three components, each between 0 and 1. Such a vector is called trichromatic and belongs to the double class. However, other representations exist and can also be used for the invention. For example, the value of a color component in a trichromatic vector of class uint8 takes a value in the range from 0 to 255 under the MATLAB ® calculation software.

The first component of the trichromatic vector is symbolized by the letter "r" and represents the proportion of red in the pixel, between 0 and 1. The second component is symbolized by the letter "g" and represents the proportion of green in the pixel , between 0 and 1. The second component is symbolized by the letter "b" and represents the proportion of blue in the pixel, between 0 and 1.

So a pixel can be represented as follows:

Pixel = <r, g, b>

For example, a red pixel has the following vector: <1, 0, 0>.

FIG. 4 is a mask 15 with the dimensions of the target image 10, which comprises two zones 103 and 104 delimited by the contour 105.

The first zone 104 is empty or white and corresponds to the interior of the contour 105.

For example, the first zone 104 comprises white pixels, that is to say in the previous convention, represented by the following vectors: <1, 1, 1>.

The second zone 103 is complementary to the first zone inside the frame 11. The second zone 103 is uniformly black and of zero albedo.

For example, the second area 103 has black pixels, that is to say in the previous convention, represented by the following vectors: <0, 0, 0>.

The mask 15 is obtained by trimming the two zones along the contour 105 and by fixing the properties of each of the zones.

This mask 15 is used in its generation of the multiplicative mask in combination with other masks, in particular a mask 16.

The mask 15 is also coded by a set of pixels, each pixel being a vector as explained above.

FIG. 5 represents H mask 16, which includes its local transparency information 41 of the transparent veil 4.

The mask 16 is obtained by simulation, by setting a gray level and a zero aibedo at each point of material of the transparent veil 4. The gray level simulates the level of transparency.

A black spot means that the corresponding spot of the transparent veil 4 is opaque. For example, a black place is represented by a black pixel: <0, 0, 0>.

A white spot at the level of the transparent veil 4 on the contrary means that the corresponding spot is entirely transparent.

For example, a white place is represented by a white pixel: <1, 1, 1>.

In particular, the opaque border 14 of the transparent veil 4 is simulated as being a black border 140 on the mask 16.

In the example shown, the transparent veil 4 is not a flat surface. On the contrary, it has been bent by the wind, and this results in a curvature of its surface. Thus, the angle at which each place of the transparent veil is viewed from the point of observation is not always perpendicular. The more this angle moves away from the perpendicularity, the greater the density of transparent veil material present between the observation point and a point on the target surface 2.

This results in the simulation of the mask 16 by a reduction in the transparency at this location, therefore by a darker shade of gray at this location.

For example, here is a shade of light gray: <0.8, 0.8, 0.8> and a shade of dark gray: <0.2, 0.2, 0.2>.

Around place 410, the incidence of gaze on the veil is almost zero. The strongest shade of gray is represented in the figures by a density of black cross greater than the other places in the zone 104.

Around location 411, the incidence of gaze on the veil is almost perpendicular. The weakest shade of gray is represented in His figures by a density of black cross lower than in the other places of the zone 104.

The absence of a transparent veil 4 is simulated on the mask 16 by an empty or white area.

The complementary area of area 104 is empty or white, by simulation.

The mask 16 thus characterized is used with the mask 15 in order to obtain a third mask 17 shown in FIG. 6.

Mask 17 is obtained by superimposing mask 15 and the mask

16. The equivalence of the following terms should already be noted with regard to operations performed on masks: overlay, addition also called screen in English.

The superposition of two masks is carried out simply by an operation relating to the pixels of the masks.

For example, let A be a first mask, also called a "layer", and B a second mask, also called a "layer". These two masks A and B are coded by pixels A _xy and B _xy represented by trichromatic vectors.

The symbols “x” and “y” signify the coordinates of the pixels in the mask.

By convention, we will note for the following:

A _xy = <r _A , g _A , b _A >

®xy ⁼ < ^r B'dB 'bg>

With "r" red component, "g" green component and "b" blue component of the pixel in question.

Then, the operation of superposition, or addition of A _xy , the pixel of coordinates x, y of the mask A, with the pixel B _xy , pixel of coordinates x, y of the mask B is the following:

addition (A _xy , B _xy ) =

with: A _xy + B _xy = (r _A + r _B g _A + g _B b _A + b _B )> and: A _xy B _xy = (r _A xr _B g _A * g _B b _A xb _B )>

For example, the overlay operation thus described can be carried out by using the "overlay" modes, also called "screen" in English, in Adobe Photoshop ® software.

Thus, the mask 17 is obtained by the operation described, for each pixel:

mask 17 = addition (mask 15, mask 16)

In this example, the masks 15 and 16 are superimposed without any operation on a possible alpha channel. The vectors of the pixels of the masks 15 and 16 are only three-dimensional, that is to say three-color. The vectors of the pixels do not exhibit ^4th component, that is to say do not exhibit the "alpha channel". Alternatively, the vectors of the pixels of the masks 15 and 16 have a fourth residual component which is simply ignored in the calculation of the mask 17. The mask 17 resulting from the superposition described above has a black area 103 and a black border 140 and d 'null albedo. Local transparency information 41 is gray and has zero albedo. Field 104 is white or empty.

In other words, the pixels at the level of the veil 4 are coded by trichromatic vectors without the need for a fourth component, called "alpha channel", coding for transparency. Indeed, transparency is represented by the gray color. The multiplicative mask 18 is obtained by multiplying the first instantiation of target image 101 with the mask 17.

The multiplication operation is carried out on the pixels of the masks 18 and

For example, let A be a first mask, and B a second mask, of pixels

17.

A _xy = <r _A , g _A , b _A >

Βχγ ^{= <} - ^γ Β'3β '^ β>

Then, the multiplication operation of A _xy , the pixel of coordinates x, y of the mask A, with the pixel B _xy , pixel of coordinates x, y of the mask B is the following:

multiplication (A _xy , B _xy ) = A _xy B _xy with. A _xy B _xy = (r _A xr _B g _A xg _B b _A xb _B )>

Thus, multiplicative mask 18 = multiplication (mask 17, first instantiation of target image 101)

For example, in Adobe Photoshop ® software, the layer mode corresponding to the operation described above is the Product mode, also called Multiply in English.

This multiplicative mask 18 is shown in Figure 7.

In the multiplicative mask 18, the lighting information in diffuse reflection on the target surface 21 comes from the first instantiation of the target image 101. Thus, the shadow 9 is present in the multiplicative mask. The other elements identical to the other figures are represented by the same reference numbers and have the same properties.

The multiplicative mask 18 therefore includes all of the transparency information of the frontal object, here the transparent veil 4, and of diffuse reflection on the target surface.

Figure 8 shows an additive mask 19 comprising all the specular reflection information of the target surface.

The additive mask 19 is obtained by the generation of a second instantiation 102 of the target image 10 of the three-dimensional scene 1 acquired as explained with reference to Figures 1 and 2.

In this second instantiation 102, the frame 3 comprises a target surface 22 according to a second instantiation.

The target surface 22 is defined during the generation of the three-dimensional scene with very precise properties:

- the target surface 22 is perfectly reflective, that is to say that it has defined specular reflection properties. On the other hand, its diffuse reflection properties are zero,

- The target surface 22 is black in color.

In the example shown, its target surface 22 is a black plane mirror.

The simulation of the second instantiation 102 generates a reflection 80 of its light from its flame 8 on the target surface 22 and a reflection 70 of the luminaire 7 on the target surface 22.

Thus, a multiplicative mask 18 and an additive mask 19 have been obtained by simulation methods, starting from the same three-dimensional scene 1 according to two instantiations.

These two masks 18 and 19 are recorded in association with each other in a memory.

The process for obtaining masks described with reference to FIGS. 2 to 8 can be repeated as much as necessary to construct a film, target image 10 by target image 10.

The target image 10 is not entirely characterized by the additive mask 19 alone or by the multiplicative mask 18 alone.

The target image 10 is entirely characterized by the additive mask 19 and the multiplicative mask 18 taken together and applied as described below. In particular, the transparency information initially coded by an alpha channel in the target image 10 is transformed so as to be coded by the RGB color, in the additive 19 and multiplicative masks 18. The information coding for the transparency is therefore separated in two masks: additive masks 19 and multiplicative masks 18.

The vectors of the pixels of the additive 19 and multiplicative 18 masks are only three-dimensional, that is to say three-color. The vectors of the pixels do not have a 4 ^th component, that is to say do not have an "alpha channel". Alternatively, its vectors of the additive 19 and multiplicative pixel 18 have a fourth residual component which is simply ignored in the overlay process described below.

INTEGRATING ANY SOURCE IMAGE INTO THE TARGET IMAGE

The complete process for obtaining a target image in which a source image is embedded is described with reference to Figure 13.

The method of overlaying a source image 23 in the target image 10 characterized by the multiplicative 18 and additive 19 masks is illustrated with reference to FIGS. 9 to 12.

A first step in the overlay 130 is to provide a source image 23.

In the example shown, the source image 23 is a photograph of a married couple.

A second step 131 of the overlay consists in generating the instantiation 230 of the source image by a three-dimensional transformation of the source image 23 to obtain the mask 20.

Figure 10 shows the three-dimensional deformation of the source image 23 so that the edges 123 of the source image 23 correspond to the contour 105 of the target image 10 in the frame 11.

This three-dimensional deformation is generated digitally, a simple deformation taking account of the contour 105.

The three-dimensional deformation does not modify the color or brightness of the source image 23. In other words, this deformation is a displacement of the pixels making up the source image 23.

This three-dimensional deformation makes it possible to obtain a mask 20 comprising an instantiation 230 of the source image 23, as shown in FIG. 11.

A third step 132 of the inlay consists in multiplying the mask 20 by the multiplicative mask 18, in order to obtain an intermediate mask.

This multiplication is the operation described above.

A last step 133 of the overlay consists in adding the additive mask 19 to the intermediate mask in order to obtain the overlay of the source image in the target image.

This addition is the operation described above.

FIG. 12 represents the final image 200 resulting from the integration of the source image 23 into the target image 10. The source image 23 is present in the environment defined in the three-dimensional scene 1, as if the image source 23 was already present during the generation of the three-dimensional scene 1, arranged in the frame 3, and that a two-dimensional view had been generated according to the observation point 12. In fact, S'ombres 9, les reflections 70 and 80 and its transparent superposition of the transparent veil 4 on the instantiation 230 of the source image 23 are present on the final image 200.

Thus, the invention provides a method of digital overlay inexpensive in computation, of any source image 23 in a target image 10, the multiplicative and additive masks of which are precalculated.

Advantageously, the multiplicative 18 and additive 19 masks being prerecorded, ie the inlaying process is very rapid and allows an inlaying of a source image 23 in real time in a prerecorded animation film.

An apparatus programmed to implement the method described above was produced using a smartphone, the memory of which included the multiplicative 18 and additive 19 masks prerecorded for each target image 10 of an animated film of 3 minutes. It took 20 seconds for the smartphone to compute a picture from the smartphone's image gallery in the animated film.

Although the methods of obtaining multiplicative and additive masks and inlays described above are envisaged for animation films or generations of digital images, they are of course not limited thereto.

Indeed, the three-dimensional scene 1 can be a real scene in which a real frame 3 is provided.

In this case, the target image 10 is a digital photograph of this real scene.

Obtaining the first instantiation of the target image 101 is then possible by taking a first digital photograph in which the target surface 21 is a matt and green, blue or white surface arranged in the frame 3 before taking the digital photograph. of target image 10.

The diffuse reflection information can then be obtained from the first digital photograph by virtue of the matt property of the target surface 21.

The transparency information of the transparent veil 4 or of any front object can be obtained in post-production, for example by a selection of zones based on the color, for example by trimming the meshes of the transparent veil 4.

Thus, a first digital photograph makes it possible to obtain the multiplicative mask 18.

Obtaining the second instantiation of the target image 102 is also possible by taking a second digital photograph of the same three-dimensional scene 1 according to the same observation point, by replacing the target surface 21 in the frame 3 by a surface target 22 bright and black, for example by a black mirror with the dimensions of the frame 3. This second digital photograph then constitutes the additive mask 19.

The integration of any source image into the target image 10 of the real scene is then possible by performing exactly Its same steps 130 to 133 described with reference to Figure 13.

Although the invention has been described in connection with several particular embodiments, it is obvious that it is in no way limited thereto and that it includes all the technical equivalents of the means described as well as their combinations if these fall within the scope of the invention.

In particular, although the target surface described in the illustrative example of the figures is planar, it can be of any shape. For example, the target surface can be curved.

In particular, although a particular way has been described for generating the additive mask 18, other ways can be envisaged. In particular, the use of an alpha channel for calculating the mask 17 is conceivable.

The use of the verb "behave", "understand" or "include" and its conjugate forms do not exclude its presence from other elements or steps than those set out in a claim. The use of the indefinite article "a" or "an" for an element or a stage does not exclude, unless otherwise stated, the presence of a plurality of such elements or stages.

In the claims, any reference sign in parentheses cannot be interpreted as a limitation of the claim.

Claims (10)

1. Method for overlaying a source image (23) in a target image (10), comprising: providing a target image (10), the target image being a two-dimensional view of a three-dimensional scene (1), the view being delimited by a frame (11) at a viewing angle and an observation point (12), the scene being illuminated by one or more light sources (8, 6), the scene comprising: o an environment (5), o a target object having a visible target surface (2) delimited by a contour (105 ) in the frame (11), o a front object (4), at least part of the front object being in front of the target surface (2), the part of the front object having a predetermined transparency, o the target surface visible (2) reflecting the light coming directly and indirectly from the light source (s) (6, 7, 8), the method further comprising: - delimiting the frame (11) into two zones (103, 104), a first zone (104) of the frame being delimited by the contour (105) and corresponding to the interior of the contour and a second zone (104) being a zone complementary to the first zone inside the frame (11), provide a multiplicative mask (18) to the dimensions of the frame (11), by providing: a first mask comprising information (9) of combination of the diffuse reflection coming directly from the light source or sources (6, 8) on the target surface (2) at each point of the first zone (104), a second mask (17) comprising binary information for delimiting its first area (104) of the second area (103) and transparency information (41) of the part of the front object (4) on the target surface (2) at each point in the first area (104), o multiply the second and Se first mask to obtain the multiplicative mask (18). - provide an additive mask (19) to the dimensions of the frame (11), o Additive mask comprising information (70, 80) for combining its specular reflection from the light source or sources (6, 7, 8) on the target surface at each point of the first zone (104), and o the additive mask corresponding to the target image (10) at each point of the second zone (103) and to the part of the front object (4) in each point of the first zone (103), - provide a source image (23), - generating an instantiation (230) of the source image (23) to the dimensions of the first zone (104), the instantiation of the source image being generated by a three-dimensional transformation of the source image so that the source image (23) corresponds to the shape of the target surface (2) of the target object, multiply the instantiation of the source image (230) with the multiplicative mask (18) in order to obtain an intermediate mask, - add the intermediate mask and the additive mask (19) in order to obtain the integration of the source image (23) in the target image (10). 2. Method according to claim 1, in which: each point of the multiplicative mask (18) in the first zone (104) has a gray level, the presence of diffuse reflection at a point in the first zone (104) resulting in a first shade of gray (9) whose clarity increases when the intensity of the diffuse reflection increases at this point, the presence of the frontal object (4) at a point in the first zone (104) resulting in a second shade of gray (41) whose clarity decreases when the transparency of the front object (4) at this point decreases, the gray level of the point being defined by the superposition of the first and second shades of gray at this point, each point of the additive mask (19) in the first zone (104) has a color, the color of the point being defined by a combination of a first shade of color and a second shade of color at this point, Its first shade of color corresponding to the specular reflection (70, 80) of the light source (s) (6, 7, 8) on the target surface (2) at this point and a second shade of color associated with an alpha channel defining the transparency corresponding to the part of the front object (4) superimposed on the target surface (2) at this point. 3. Method according to claim 1 or 2, in which: the first mask of the multiplicative mask (18) consists of a first instantiation of the target image in which the target surface according to its first instantiation (21) consists of a mat material of defined diffuse reflection property and of specular reflection property zero, - the additive mask (19) consists of a second instantiation of the target image in which the target surface according to the second instantiation (22) is made of a material having a defined specular reflection property and a zero diffuse reflection property . 4. Method according to any one of claims 1 to 3, in which the color in the second mask of the multiplicative mask (18) is uniformly black at each point of the second zone (103) and the transparency of the front object ( 4) is simulated by a gray filter (41) at any point in the first zone (104) where the front object is present. 5. The method of claim 3, wherein the material of the target surface according to the first instantiation (21) is white, and which the material of the target surface according to the second instantiation (22) is black. 6. Method according to any one of claims 1 to 5, in which providing a target image comprises: numerically define the three-dimensional scene (1) comprising the target object, generate the two-dimensional view of the three-dimensional scene (1) by digital synthesis of the three-dimensional scene (1). 7. Method according to any one of claims 1 to 5, in which the target image (10) is a photograph taken of the three-dimensional scene. 8. The method of claim 3, wherein the target image according to the first instantiation is a first photograph taken from the three-dimensional scene (1) comprising the target surface according to the first instantiation (21) and the target image according to the second instantiation is a second photograph taken from the three-dimensional scene (1) comprising the target surface according to the second instantiation (22). 9. Method according to claim 3 or 8, in which the target image according to the first instantiation is a first photograph taken from the three-dimensional scene (1) comprising the front object (4) according to a first instantiation in which the front object (4) is black and of zero albedo. 5 10. Method according to any one of claims 1 to 9, further comprising: - save the multiplicative mask (18) and Additive mask (19) in a digital memory. 1/6