WO2007071677A1

WO2007071677A1 - Enhanced image processing method for preventing image piracy

Info

Publication number: WO2007071677A1
Application number: PCT/EP2006/069919
Authority: WO
Inventors: Laurent Blonde; Pascal Bourdon; Julien Thollot
Original assignee: Thomson Licensing
Priority date: 2005-12-20
Filing date: 2006-12-19
Publication date: 2007-06-28

Abstract

Processing of a source image IS0 includes a step of breaking the image down into a series of n component images (IC1, IC2, ..., ICk,, ICn) in a colour space that includes the absolute brightness quantity and is, according to the invention, independent of any display device. When the images in the series are displayed one after the other at a frequency n times greater than the display frequency of the source image and higher than the colour fusion frequency of the human eye, the resulting fused image IF is identical or nearly identical to the source image. The invention enables differences between component images to be accentuated and the interference level to be raised without adversely affecting the image display.

Description

PIRACY OF IMAGES.

The invention relates to a method for processing a source image (I _s ) comprising a step of decomposing this image (I _s ) into a series of n "component" images (I _C1 , I _C2 , ..., Ick - - -. 'in) ⁰ M ^are different, said decomposition being adapted for the successive display of the images of this series to a frequency higher than the color fusion frequency for the human eye to produce a merged image ( I _F ) identical to said source image. This frequency is n times greater than the display frequency of the source image. Such an image processing method is described in WO05 / 027529 - THOMSON.

The identity or quasi-identity between the source image and the resulting image of the temporal fusion of the component images means that any differences between these two images are not perceptible to the eye. In general, for displaying an image using an image display device comprising a display screen having a two-dimensional matrix of polychromatic elementary displays, and means for controlling each elementary display according to of a triplet of video data, each pixel of this image is associated on the one hand with a display of the screen, on the other hand with a triplet of video data (D _R , D _G , D _B ), and control each display of the screen which is associated with a pixel of the image using the video data triple (D _R , D _G , D _B ) associated with this pixel. By convention, each triplet of video data associated with a pixel of the image to be displayed forms, in a color space linked to the display device, the coordinates of what is called the color vector of this pixel. By definition, the set of possible values of the triplets of video data or color vectors describes, in this color space associated with the device, a range or "gamut" three-dimensional color.

In the case of displaying television sequences, the video data is generally normalized, for example in the PAL system or in the NTSC system, both of which are so-called luminance-chrominance systems. For the display of an image, a PAL image display device therefore receives, in the form of electrical signals, triplets of video data, generally noted (Y, U, V), which correspond to the set color vectors and thus pixels of this image, in the color PAL space linked to this device. In the same way, for the display of an image, an NTSC image display device therefore receives, in the form of electrical signals, the video data triplets, generally noted (Y, I, Q), which correspond to each other. to the set of color vectors and thus pixels of this image, in the NTSC color space linked to this device. By extension, we thus speak of the spaces YUV and YIQ, where Y designates the luminance, and where U and V, where I and Q designate the chrominance.

When applied to the display of an image sequence using an image display device, the image processing method described in the aforementioned WO05 / 027529 results in a method image display. As indicated in this document, such a display method then makes it possible:

on the one hand, to display any sequence of processed source images so that the observer perceives the source images of this sequence as if none of them were processed;

- On the other hand, scrambling the shooting of the display of this sequence of processed images, that would try to perform a malicious person, including using a camcorder not synchronized to the sequence of source images. In this document, the decomposition of the source images is performed in a color space noted YUV, that is to say a color PAL space; this color space therefore comprises the luminance Y; as indicated above, this color space depends on the image display device used to display the images of the sequence. By repeating the notations given in this document, to decompose a pixel of the source image I ₁ , to which corresponds a coordinate color vector (Y ₁ , U ₁ , V ₁ ), in this color space, this vector is decomposed. "Source" in two "component" vectors respectively of coordinates (Y ₃ , U ₃ , V ₃ ), (Y ₄ , U ₄ , V ₄ ), where Y ₁ = Y ₃ = Y ₄ ; still according to this document, the values of U ₃ , V ₃ , U ₄ , V ₄ are determined so that the ends C ₃ and C ₄ of these two "component" vectors:

have a centroid which corresponds to the end C ₁ of the color source vector, that is to say are symmetrical with respect to this end C ₁ in the same color PAL space linked to the display device;

- be located in the color gamut of this device.

Thus, the essential criteria for the decomposition of a color vector from one pixel of the source image into two color vectors each associated with a pixel of a component image are that in the YUV color space related to the display device, the points C ₁ , C ₃ , and C ₄ are aligned, belong to the color gamut of the device, and that the Euclidean distance C ₁ C ₃ is equal to the Euclidean distance C ₁ C ₄ . The prior art cites the transposition of color vectors in other spaces for image processing with other purposes, such as for example US6640005 and US2003 / 184779. It has been found that the successive visualization, at a frequency greater than the color melting frequency for the human eye, of component images of a source image, which are constructed by the decomposition of the color vectors of this source image as described herein. above and in the document WO05 / 027529, produced a merged image (I _F ) which could have some defects with respect to the source image. Thus, to avoid illegal shooting of image sequences, there is a risk of inadvertently degrading the display of these sequences. Moreover, when seeking to improve the interference by exacerbating the differences between the component images, the risk of degradation of the display is further increased. An object of the invention is to limit or eliminate this disadvantage. For this purpose, the subject of the invention is a method for processing a source image (I ₈₀ ) comprising a step of decomposition into a series of n "component" images (I _C1 , I _C2 , ..., Ick- - -. 'en) in a color space which, according to the invention, is independent of any display device. The invention particularly relates to a method for processing a source image (I ₈₀ ) comprising a step of decomposing this image (I ₈₀ ) into a series of n "component" images (I _C1 , I _C2 , .. ., Ick - - - '') ⁰ M ^are different, in which:

at each "source" pixel (E _SOj ) of the source image (I ₈₀ ) which is decomposed corresponds to a "component" pixel in each of the n images

"Components" (I _C1 , I _C2 , ..., l _C k> - -> ¹ Cn) - and, each pixel of said images, source or components, being associated with a color vector in a given color space, said decomposition is performed in such a way that the associated color vector, in this space, at each source pixel ( E _SOj ) which is decomposed to be equal to the resultant, divided by n, of the associated color vectors, in this same space, to the component pixels that correspond to said source pixel.

According to the invention, said color space is independent of any display device.

The source image comprises q pixels which are decomposed, the other pixels of this source image then not being decomposed; the pixels that are not decomposed are unchanged in each of the component images; on the contrary, each decomposed pixel corresponds to a series of n component pixels, at least two of the pixels of this series being different, so that the n "component" images (I _C1 , I _C2 , ..., Ick- ..., l _Cn ) actually differentiate; the set of decomposed pixels

(E _S oi, Eso ₂ - - - E ₈₀ J, ..., E _SOq ) of the source image forms a plurality of q decomposed pixels.

More specifically, the subject of the invention is a method for processing a source image (I ₈₀ ) comprising a step of decomposing this image (I ₈₀ ) into a series of n "component" images (I _C1 , I _C2 , ..., Ick- - -. 'En) ⁰ M ^are different,

each pixel of said images, source or components, being associated with a color vector in a color space which comprises the absolute quantity of luminance and which is independent of any display device, the n "component" images (I _C1 , I _C2 , ..., Ick- - -. 'En) ^being differentiated by a plurality of "component" pixels whose resultant color vectors, in said space, is equal to n times the color vectors associated with the decomposed pixels (E _{S (n} , E ₈₀₂ , - -, E _SOj , ..., E _SOq ) of a corresponding plurality of said source image (I ₈₀ ). According to the invention, this decomposition is adapted so that the successive display of the images of this series of n "component" images, at a frequency which is equal to n times the display frequency of said source image and which is greater than the color fusion frequency for the human eye , produce a merged image (I _F ) identical or almost identical to said source image

In a luminance-chrominance color space, the color vector which, multiplied by n, is the resultant of the "component" color vectors, has a luminance equal to the average of the luminance of these color vectors and a chrominance resulting from the chrominance composition of these color vectors. This color vector corresponds to what the eye perceives when a succession of pixels corresponding to these color vectors "components" is displayed at a frequency greater than the melting frequency of the eye.

The invention therefore proposes to perform the decomposition of color vectors in a color space which is independent of any display device. It has been found that, according to the invention, by decomposing the color vectors of a source image in such a visual space independent of any display device, instead of proceeding in a visual space depending on the device of the device. display as the YUV space as in the prior art (see document WO05 / 027529), it is found that the successive visualization, at a frequency greater than the color melting frequency for the human eye, of the component images of this The source image produces a merged image (I _F ) much closer to the source image than in the prior art, even when the differences between the component images are exacerbated to enhance the interference. Thus, while effectively jamming the illegal shooting of image sequences by this image processing, it limits the degradation of the display of these images. According to a first essential variant of the invention, the color space used for the decomposition comprises the absolute quantity of luminance, which is generally expressed in cd / m ² : the absolute space XYZ satisfies this criterion. Preferably, said color space is of XYZ type or linearly derived from this space. The XYZ color space is an absolute visual space in the sense that the X, Y, Z coordinates are each expressed in cd / m ² . The XYZ color space is defined, in particular, as a visual space whose primary colors are adapted so that all the colors perceptible to the human eye are expressed by positive trichromatic components and for one Y of these color components to represent luminance information.

The decomposition of the source images is then performed in a color space denoted XYZ or linearly derived from XYZ, which does not depend on the display device used for displaying the processed image. A color coordinate vector (X ₁ , Y ₁ , Z ₁ ), in this color space, is for example broken down into two "component" vectors respectively of coordinates (X ₃ , Y ₃ , Z ₃ ), (X ₄ , Y ₄ , Z ₄ ), where Y ₁ = Y ₃ = Y ₄ (so we have Y ₃ + Y ₄ = 2 Y ₁ ). The essential criteria for the decomposition of this color vector from one pixel of the source image into two color vectors each associated with a pixel of a component image are then that, in the XYZ color space or linearly derived from this XYZ space (and not in the YUV space linked to the display device), the points C ₁ , C ₃ , and C ₄ corresponding to the ends of the color vectors are aligned, belong to the color gamut of the device. display transposed in this space of XYZ or derivative color, and that the Euclidean distance CiC ₃ is equal to the Euclidean distance C \ C ₄ . We thus have: X ₃ + X ₄ = 2.X ₁ and Z ₃ + Z ₄ = 2.Z ₁ . We note that any linear transformation of this space, using for example a transition matrix of which all the elements are constant, preserves these two last equations; this is why any space drifting linearly from this space can also be used for the implementation of the invention. According to a second preferred variant, said color space is of type Ycd or linearly derived from this space, defined as follows with respect to the space XYZ: c = X / Y and d = Z / Y. The luminance parameter Y of this space Ycd resumes that of the space XYZ. The decomposition of the source images is then carried out in a color space denoted Ycd or linearly derived from Ycd, which does not depend on the display device used for displaying the processed image. A color vector of coordinates (Y ₁ , C ₁ , d {) in this color space, for example, is decomposed into two "component" vectors respectively of coordinates (Y ₃ , C ₃ , d ₃ ), (Y ₄ , C ₄ , d ₄ ), where Y ₁ = Y ₃ = Y ₄ (so we have Y ₃ + Y ₄ = 2 Y ₁ ). The essential criteria for the decomposition of this color vector from one pixel of the source image into two color vectors each associated with a pixel of a component image are whereas, in the color space Ycd or linearly derived from this space Ycd, the points C ₁ , C ₃ , and C ₄ corresponding to the ends of the color vectors are aligned, belong to the color gamut of the transposed display device in this color space Ycd or derivative, and that the Euclidean distance C ₁ C ₃ is equal to the Euclidean distance C \ C " _4. We thus have: C ₃ + C ₄ = 2 C ₁ and d ₃ + d ₄ = 2 d

According to a second essential variant of the invention, said color space is perceptually uniform. Thus, the color space used can be in particular the CIE-LAB space, the CIE-LUV space, the QMH space, or the JCH space; preferably, the CIE-LAB space or a space linearly derived from this space is used. The choice of such a perceptually uniform space advantageously makes it possible to further exacerbate the differences between the component images, this time as they are perceived by the human eye, which further enhances the level of interference; note that the LAB space also includes the luminance magnitude.

Preferably, to implement the invention according to the first or second essential variant, each pixel of said source image being associated with a video data triple (D _R , D _G , D _B ) which conventionally forms the coordinates of the color vector associated with said pixel of this image in a color space associated with said device:

each pixel of said plurality of differentiating pixels being associated with a color source vector, determining the coordinates of said source vector in said device-independent color space by a first processing of said video data associated with said pixel, and determining the coordinates of the color component vectors whose resultant, in said device-independent color space, is equal to n times said color source vector,

and, each color component vector being associated with a pixel of a component image, a video data triplet associated with this pixel is determined by a second processing, inverse of the first, of the coordinates of said color component vector associated with this pixel .

Said first and second processing correspond to color space changes, symmetrical to one another. The invention also relates to a method for displaying a source image sequence at a source display frequency in which at least one source image (I ₈₀ ) is processed by the method according to the invention, according to the first or the second essential variant, to be decomposed into a series of n "component" images (I _C1 , I _C2 ,..., Ick - -... en) ⁰ M ^are different, and in which one displays successively said component images at a component display frequency equal to n times said source display frequency, wherein said component display frequency is greater than the color fusion frequency for the human eye. The invention also relates to a device for displaying an image sequence, where each pixel of an image of this sequence is associated with a video data triple (D _R , D _G , D _B ), comprising:

a display screen provided with a two-dimensional matrix of polychromatic elementary displays, each display being associated with a pixel of the image to be displayed,

processing means capable of processing at least one image to be displayed of said sequence by the method according to the invention, to be decomposed into a series of n "component" images (I _C1 , I _C2 , ..., Ick- - -. 'in) ⁰ M ^are different, and for associating with each pixel of a triplet its component images of video data (D _R, D _G, D _B),

means for controlling each elementary display associated with a pixel of an image to be displayed as a function of a video data triplet associated with this pixel, wherein said control means are adapted to successively display the component images of each series at a frequency higher than the color melting frequency.

As an example of a basic display of a display screen, there may be mentioned a group of three liquid crystal or micro-mirror valves modulating in three different primary colors, or a group of three light-emitting diodes emitting in three different primary colors. . In the case of a projection display device where the light of a source is modulated by a micro-imager of liquid crystal or micro-mirror valves, each elementary display is formed by a valve. The invention will be better understood on reading the description which follows, given by way of nonlimiting example, and with reference to the appended figures in which:

FIG. 1 shows the intersection of different planes Y = constant with the three-dimensional gamut of colors of the display device to which the invention applies, gamut which is represented in an independent color space (Y, c, d) of this device which is used in one embodiment of the invention;

FIG. 2 represents a step of the decomposition of a color vector of a source image into two color vectors of component images, in the same color space (Y, c, d) independent of this device, according to the same embodiment of the invention as that of Figure 1. We will now describe an embodiment of the processing method according to the first essential variant of the invention, applied to the display of a sequence of images using an image display device having a screen comprising a matrix of display elements and having control means for these display elements, in which, to obtain the display of a given image , each pixel of this image is associated with a video data triple (D _R , D _G , D _B ) which, when it is addressed to the display element corresponding to this pixel, via the control means of this device generates the display of this pixel. Each image of the sequence is partitioned into a matrix of pixels, so that each display element corresponds to a pixel of this matrix. Such a display device can be indifferently a digital video projector, a retro-projector, a plasma screen, an LCD screen or another image display screen that is addressable by video data. In the sequence of images to be displayed, a source image to be decomposed I ₈₀ is selected, here in a series of two component images I _C1 , I _C2 . Some pixels of the two component images I _C1 , I _C2 are identical to those of the source image I ₈₀ , others differ from the source image and form a plurality of q differentiating pixels: E _C11 , E _C12 , .. ., E _clj , ...., E _clq for the component image I _C1 , and E _C21 , E _C22 , ..., E _C2j , ...., E _C2q for the component image I _C2 . In each component image I _C1 , I _C2 there are therefore q pixels differentiating from the source image I ₈₀ , the other pixels being identical. The number of differentiating pixels preferably represents at least 10% of the total number of pixels of an image, so that the difference between the component images can be perceptible to the eye. The decomposition of the source image I ₈₀ which will be described below aims to obtain that the fusion of the pixels E _{cl I} and E _C21 , E _cl2 and E _C22 , ..., E _clj and E _C2j , ... , E _clq and E _C2q , identical positions on all the component images I _C1 , I _C2 , generates for the human eye a pixel identical to that (E _{S (n} , E ₈₀₂ , ..., E _SOj , .. ., E _SOq ) of the same position on the source image I _80. By extension, we therefore say that the source pixels (E _{S (n} , E ₈₀₂ , ..., E _SOj , ..., E _SOq ) are decomposed into component pixels (E _C11 , E _C12 ,..., E _clj , ...., E _clq ) for the component I _C1 image, and component pixels (E _C21 , E _C22 , ..., E _C2j , ...., E _C2q ) for the component image I _{C2, the} pixels of the source image I ₈₀ that are to be decomposed therefore form the following plurality: E _{S (n} , E ₈₀₂ , ..., E _SOj , ..., E _SOq We will now explain in detail how to decompose one of these pixels, E _SOj , in pixels of the same position E _clj and E _C2j respectively in the component images I _C1 , I _C2 , the decomposition of the other pixels of this plurality being done in an identical manner. At this pixel E _SOJ of the source image I ₈₀ is associated, as has been seen above, a video data triplet (D _R _SOJ, _SOJ D _G, D _B _SOJ...); the triplets of video data are searched (D _R , _cl , D _G , _clj , D _B, _clj ), (D _R , _C2j , D _G , _C2j , D _B, _C2j ) which are respectively associated with the pixels E _clj and E _C2j component images I _C1 , I _C2 , and which, when displayed by the display device at a frequency higher than the color melting frequency of the human eye, generate a pixel identical to E _SOj ; the invention is described here in the case where the triplets of video data (D _R , C ₁ , D _G , C ₁ , D _B , C ₁ ), (D _R , C ₂ , D _G , C ₂ , D _B , C ₂ ) give rise to at the pixel display E _C1 ; and E _C2 ; as different as possible from each other, so as to improve the scrambling of the images. In accordance with an aim pursued by the invention, it will be shown that, despite a high scrambling factor, the quality of the image display is not deteriorated. It is now considered that the triplet (D _R _S0 , D _G _S0 , D _B _S0 ) associated with the pixel E _SOj represents the coordinates of a vector OP _S0 , called "color vector", in an associated color space. to the display device. It is considered here that the video data as it is to be addressed to the display device is all encoded in 10 bits; each of the video data can therefore take an integer value between 0 and 1023. The three columns of Table 1 below give the coordinates of color reference vectors OO, OR, OG, OB, OC, OM, OY and OW, corresponding, in the order of the lines, to the black, then to each of the primaries of the device (respectively red, green and blue), then to each of the secondary of the device (respectively cyan, magenta, and yellow), then to the white of reference of the device. The ends of these color reference vectors therefore delimit a cube in this color space, also called a three-dimensional gamut, within which, including limits, are all the color vectors that can be displayed by the display elements of the color. the screen.

Table 1

DR D _G D _B XYZYC d

OO O O O 0.00 0.00 0.00 0.00 0.00 0.00

OR 1023 O O 19.19 8.99 0.97 8.99 2.13 0.11

OG O 1023 O 10.27 28.28 2.92 28.28 0.36 0.10

OB O O 1023 10.02 3.95 50.64 3.95 2.54 12.82

OC O 1023 1023 20.29 32.23 53.56 32.23 0.63 1.66

OM 1023 O 1023 29.22 12.94 51.61 12.94 2.26 3.99

OY 1023 1023 O 29.46 37.27 3.89 37.27 0.79 0.10

OW 1023 1023 1023 39.49 41.22 54.53 41.22 0.96 1.32

We will now transpose this video data in a known color space XYZ, which is independent of the device, then in a new color space Ycd, also derived from XYZ and independent of the device. In this new color space advantageously used for the implementation of the invention, the coordinates Y, C, D of each color vector OP are expressed as follows: Y = Y, c = XfY, d = ZfY. In this new color space, one of the trichromatic components Y represents the luminance of the pixel, and the other two trichromatic components c, d are independent of the luminance and represent the chrominance. Table 1 gives the correspondence between the coordinate values of the eight color reference vectors OO, OR, OG, OB, OC, OM, OY and OW. when expressed in the video data space or device-specific color space, when expressed in XYZ space, and when expressed in the new color space used here for decomposition. The correspondence D _R , D _G , D _B -> XYZ is established in a manner known per se, for example using known methods of colorimetric characterization of a display device, such as those described in the referenced IEC standard. 61966. To establish this correspondence, it is also possible to use the known spectral visual functions x (λ), y (λ), z (λ) which are characteristic of XYZ color systems. Other known correspondence methods can also be used. The correspondence XYZ -> Ycd is established as previously defined. Since the O, R, G, B, C, M, Y, and W ends of these eight different color reference vectors are the vertices of the three-dimensional gamut of the device that forms a cube in the video data space, this table gives the coordinates of the vertices of this same three-dimensional gamut in the new color space according to the invention. In this last space, the three-dimensional gamut forms a polyhedron whose vertices are formed by the ends of the eight color reference vectors 00, OR, OG, OB, OC, OM, OY and OW.

FIG. 1 represents, in the two-dimensional coordinate system of the chrominance components c and d, different intersections of planes Y (luminance) = constant with this three-dimensional polyhedral gamut, these planes and this gamut thus being represented in the new independent color space of the device previously defined: intersections of the Y = O plane in phantom-dotted dashed line, Y = 10 dotted line, Y = 15 dashed line, and Y = 35 solid line. These intersections are limited by two-dimensional polygons. In this two-dimensional reference of the chrominance components c and d, the projection R ', G', B ', C, M', Y 'and W on these planes Y = constant of the end points R, G , B, C, M, Y and W of the OR, OG, OB, OC, OM, OY and OW color reference vectors of Table 1. The coordinates of these points R ', G', B ', C, M ', Y' and W are given in columns c and d of Table 1. _Returning now to the pixel E _SOj of the source image I ₈₀ to be decomposed, which is associated with the triplet (D _R , _S0 , D _G , _S0 , D _B, _S0 ) which represents, in the color space associated with the device display, the coordinates of the color vector OP _SOj associated with this pixel. Therefore sought the triplet (YSOJ ^c n _j -. _SOJ d) the coordinates of this same vector source color OP _SOJ, expressed, this time, in the new color space according to the invention. For this purpose, linear interpolation is performed from color reference vectors, on the one hand, which frame the color vector OP _SOj , on the other hand for which the correspondence D _R , D _G , D _B -> XYZ -> Ycd has been established, as previously mentioned; such a linear interpolation method is known per se and will not be described here in detail. This gives the triplet (YSOJ ^c n _j -. Of _SOJ) coordinates of the color source vector OP _SOJ expressed in the new space for independent color of the device. With reference to FIG. 2, we search for the intersection of the three-dimensional gamut of this device in this new color space with the plane Y = Y _SOj ; this intersection forms a two-dimensional luminance gamut 1 Y = Yso _j> ^e * represents the set of colors accessible to the display device for this luminance Y = Y _SOj . In this two-dimensional gamut, the end P _{SO j} of the color vector OP _{SO j is positioned} ; in the two-dimensional reference of the chrominance components c and d situated in this two-dimensional gamut, the coordinates of this point P _SOj are therefore c _SOj , d _SOj .

We will now describe an optional step of the method which makes it possible to obtain pixels E _clj and E _{C2j that are} as different as possible from one another, so as to improve the scrambling of the images. We now define a zone, called symmetric two-dimensional gamut 2, which is symmetrical with the two-dimensional gamut with respect to the point P _SOj , always in the same plane Y = Y _SOj ; then defining an area, called a reduced two-dimensional gamut 3, which corresponds to the intersection of the two-dimensional gamut and the symmetric two-dimensional gamut; concentric circles 4, 5, 6, centered on P _SOj , which are included in the reduced two-dimensional gamut or which have a line of intersection with this reduced two-dimensional gamut, are drawn; specifically, these circles are defined as follows: the limit circle 4 is centered on P _SOj and passes through the limit points P ₁ JL and P _2jL , symmetrical of P _SOj , which are the farthest from each other in the reduced gamut 3; we call L _SOj the distance Pso _j ^ i _j L ⁼ Pso _j Pi _j L -

the minimum circle 5 is centered on P _SOj and has a radius equal to 0.5 x L _SOj ; - The average circle 6 is centered on P _SOj and has a radius equal to 0.8 x L _SOj ;

The radius L _SOj of the limit circle is deduced from the equation representing algebraically the reduced two-dimensional gamut 3 and from the equation expressing that the point P _SOj is the centroid of the limit points P ₁ JL and P _2jL . The factors 0.5 and 0.8 correspond to possible values of a factor of scrambling K _SOj specific to the pixel E _SOj ; this factor can be common to all the pixels of the source image S ₀ to be broken down; conversely, this factor can be variable according to the pixels of the source image S ₀ to be decomposed, preferably inversely proportional to the motion vector of this pixel, so as to advantageously reduce the interference rate in the parts of the image subject to strong movement; it can be common to all the source images to decompose, or on the contrary be variable according to the source images. In general, to decompose the color vector OP _SOj associated with the pixel E _SOj into two color vectors OP _clj , OP _C2j associated with the pixels E _clj and E _C2j respectively of the component image I _C1 and the component image I _C2 we choose two symmetrical points P _clj , P _C2j on a circle centered on P _SOj , and evaluate the respective coordinates (c _clj , d _clj ), (c _C2j , d _C2j ) of these points in the two-dimensional coordinate system c, d previously defined. In general, the ends of the color component vectors are chosen so that their centroid, in the color space according to the invention, coincides approximately with the P _{SO j} end of the color source vector. To obtain pixels E _C1 ; and E _C2 ; as different as possible from one another, for example, the circle 6 whose radius is equal to 0.8 L _{SOj is chosen} , which corresponds to a mean level of image interference (K _SOj = 0.8 ). By continuing the decomposition of the pixel E _SOj of the source image S ₀ into two pixels E _clj and E _C2j respectively of the component image I _C1 and the component image I _C2 , we define Y _clj = Y _C2j = Y _SOj , is obtained triples _(j Yci. c _clj, DCij) (Yc2j> ^c C2p ⁰ Oj) ^ ^{1 eχ} Pri ^ment ^'are coordinates of the two color vectors _clj OP, OP _C2j in the new color space. By an inverse transformation of the previously defined linear interpolation, which makes it possible to pass from the expression of the coordinates of a color vector in the color space linked to the device to the expression of the coordinates of the same vector in the new space of color of the invention, the triplets (D _R , _cl , D _G , _cl , D _B, _clj ), (D _R , _C2j , D _G , _C2j , D _B, _C2j ) are _calculated which express the coordinates of the same two color vectors OP _clj , OP _C2j , this time in the color space linked to the device.

Thus, when the image display device generates the succession of the component images I _C1 , I _C2 at a frequency greater than the melting frequency of the colors of the human eye, the pixels E ₀ e. Eq ₂ of the component images I _C1 , I _C2 , will be displayed successively from the following triplets of video data (D _R , _cl , D _G , _cl , D _B, _clj ), (D _R , _C2j , D _G , _C2j , D _B _C2j ), which will generate, because of the merging of the colors, a pixel identical to the pixel E _SOj of the source image E ₈₀ . Conversely, in the image obtained from an illegal recording using a non-synchronized camera, the observer will see two images I _C1 , I _C2 , which will be all the more distinct as the pixels that compose them are associated with different color vectors. Thanks to the decomposition of the color vectors in a color space independent of any display device, such as the Ycd space just used, it is found that the successive visualization at a frequency greater than the melting frequency colors for the human eye, component images of this source image produce a merged image (I _F ) much closer to the source image than in the prior art, even when the differences between the component images are exacerbated to improve scrambling, according to the optional variant described above. Thus, even if we strongly blur the illegal shooting of image sequences by this image processing, we obtain a good image display quality. In the description of the present invention, the transformations of the expression of the coordinates of the color vectors between different color spaces have been made by linear interpolation; other types of known transformations can be used without departing from the invention. The present invention has been described with reference to a source image decomposition into two component images; decompositions into one higher number of component images can be envisaged without departing from the invention; by generalizing, a source image can thus be decomposed into a series of n "component" images: I _C1 , I _C2 , ..., Ick- _Cn ; alternatively, the number n, greater than or equal to 2, may vary depending on the source image to be decomposed; indeed, as the color melting frequency depends on the brightness of the images, it is possible to consider a higher number n for low melting frequencies, and vice versa. Other visual color spaces independent of the display device can be envisaged without departing from the invention, in particular the linearly derived spaces of the XYZ space.

Preferably, according to a second essential variant of the invention, as a device-independent color space, a perceptually uniform space is chosen, such as the CIE-LAB space (also called Lab), the CIE-LUV space (also called Luv), or the QMH space, or the JCH space; for the QCH space, Q is the brightness (brightness), C is the colorfulness, and H is the huequadrature or hueangle; for the space JCH, J designates the luminance ("lightness"), C designates the "chroma", and H designates as previously the hue. The choice of such a perceptually uniform space advantageously makes it possible to further exacerbate the differences between the component images, this time as they are perceived by the human eye, which further enhances the level of interference; indeed, the choice of such a color space makes it possible to specifically maximize the differences in perception between the component images of the same source image. The example that has just been described can be implemented using, this time, space transformation formulas adapted to the perceptually uniform space, formulas that are within the reach of those skilled in the art. As a variant of this example, instead of maximizing the Euclidean distances between ends of the pixel color vectors of the component images as previously described, the color differences Δh between these pixels are maximized. It is found that by performing the decomposition in a LAB space in the same way as in the Ycd space as previously described, one obtains, for the same source image, substantially different component images. It is constant that the use a perceptually uniform space for decomposition can improve the scrambling of images. Note that the LAB and LUV spaces include the luminance magnitude.

The invention applies to other embodiments of the method of displaying a sequence of images, without departing from the scope of the claims below.

Claims

A method for processing a source image (I ₈₀ ) comprising a step of decomposing this image (I ₈₀ ) into a series of n images

"Components" (I _C1 , I _C2 , ..., Ick- - -. 'En) ⁰ M ^are different, in which:

"Components" (I _C1 , I _C2 , ..., l _C k> - -> ¹ Cn) -

and, each pixel of said images, source or components, being associated with a color vector in a given color space, said decomposition is performed in such a way that the associated color vector, in this space, at each source pixel ( E _SOj ) which is decomposed to be equal to the resultant, divided by n, of the associated color vectors, in this same space, to the component pixels that correspond to said source pixel, characterized in that said color space is independent of any other device. display.

2. Processing method according to claim 1, characterized in that said color space comprises the absolute magnitude of luminance and is of type XYZ or linearly derived from this space.

3. Processing method according to claim 1, characterized in that said color space comprises the absolute magnitude of luminance and is of type Ycd or linearly derived from this space, defined as follows with respect to the space XYZ: c = X / Y and d = Z / Y.

4. Method according to claim 1 characterized in that said color space is perceptually uniform.

5. Method according to claim 4 characterized in that said color space is the CIE-LAB space or a space linearly derived from this space.

A method for processing a source image (I ₈₀ ) according to any one of the preceding claims wherein, each pixel of said source image is associated with a video data triple (D _R , D _G , D _B ) which form by convention the coordinates of the color vector associated with said pixel of this image in a color space associated with said device:

each pixel of said plurality of differentiating pixels being associated with a color source vector, determining the coordinates of said source vector in said device-independent color space by a first processing of said video data associated with said pixel,

and determining the coordinates of the color component vectors whose resultant, in said device-independent color space, is equal to n times said color source vector, and each color component vector being associated with a color pixel. a component image, a triplet of video data associated with this pixel is determined by a second processing, inverse of the first, of the coordinates of said color component vector associated with this pixel.

A method of displaying a source image sequence at a source display frequency in which at least one source image (I ₈₀ ) is processed by the method of any one of claims 1 to 6 to be decomposed in a series of n "component" images (I _C1 , I _C2 , ..., Ick- ..., l _Cn ) which are different, and in which said component images are successively displayed at an equal component display frequency once said source display frequency, characterized in that said component display frequency is greater than the color fusion frequency for the human eye.

Apparatus for displaying a sequence of images, wherein each pixel of an image of this sequence is associated with a video data triple (D _R , D _G , D _B ), comprising:

processing means capable of processing at least one image to be displayed of said sequence by the method according to any one of claims 1 to 6, to be decomposed into a series of n "component" images (I _C1 , I _C2 , ..., Ick- - -. 'En) ⁰ M ^are different, and to associate with each pixel of its component images a triplet of video data (D _R , D _G , D _B ), means for controlling each elementary display associated with a pixel of an image to be displayed as a function of a video data triplet associated with this pixel, characterized in that said control means are adapted to display successively the component images of each series at a frequency greater than the color melting frequency.