EP2135221A1 - Procédé de codage de données répresentatives d'une texture multidimensionnelle, dispositif de codage, procédé et dispositif de décodage, signal et programme correspondants - Google Patents
Procédé de codage de données répresentatives d'une texture multidimensionnelle, dispositif de codage, procédé et dispositif de décodage, signal et programme correspondantsInfo
- Publication number
- EP2135221A1 EP2135221A1 EP08762021A EP08762021A EP2135221A1 EP 2135221 A1 EP2135221 A1 EP 2135221A1 EP 08762021 A EP08762021 A EP 08762021A EP 08762021 A EP08762021 A EP 08762021A EP 2135221 A1 EP2135221 A1 EP 2135221A1
- Authority
- EP
- European Patent Office
- Prior art keywords
- data
- dimensions
- texture
- wavelet
- decomposition
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Ceased
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T9/00—Image coding
- G06T9/001—Model-based coding, e.g. wire frame
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/60—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using transform coding
- H04N19/62—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using transform coding by frequency transforming in three dimensions [3D]
Definitions
- the present invention is in the field of computer graphics, and more specifically in the field of coding and data compression for viewing virtual scenes.
- the invention relates to a method for encoding multidimensional textures, for transmitting and generating photo-realistic synthetic images.
- the shape of the surface is placed at the coarsest level, the macroscopic level, and corresponds to the geometry of the object whose surface is represented.
- this geometry is most often expressed by a mesh, that is to say a set of polygons.
- BRDF Bilateral Reflectance Distribution Functions
- An intermediate layer encapsulates geometric details, such as small bumps on a granular surface for example. From an application point of view, this layer is often represented by a method called normal disturbances or "bump mapping", or by a method of geometric variation of the points of the mesh along their normal, called “Displacement mapping” method. This level disturbs the upper layer because the local variations of the surface cause a local variation in light intensity, due to the creation of an occlusion or a shadow for example.
- BTF Bilateral Texture Functions
- BTF BTF (x, y, ⁇ v , ⁇ v , ⁇ !, ⁇ l )
- - x and y represent spatial coordinates expressed in parametric coordinates
- - ⁇ v and ⁇ v define the direction of a point of view in polar coordinates
- a BTF thus corresponds to a set of images where each image is associated with a point of view and a direction of light.
- This capture method makes it possible to model both the mesostructure, the micro-structure and all the interactions that exist between these two layers, for the same material.
- fine sampling is necessary, ie a BTF corresponds to a large number of images.
- the first approach is to approximate a multidimensional texture by continuous function models.
- a texture expressed as a BTF is approximated by BRDF function models, or by bi-quadratic polynomials.
- this BTF being expressed by a set of images, each being the photograph of the surface of a material corresponding to the texture according to a given point of view and direction of light, classifying differently these data is obtained for each pixel its variation according to the direction of the light considered and according to the point of view considered.
- this BTF is approximated by a set of BRDF function models:
- z is the projection of the spatial coordinates x and y initially defining a pixel in the BTF, on a single dimension
- v is the projection of the initial polar coordinates ⁇ v and ⁇ v initially defining the direction of a point of view in the BTF, on a single dimension
- - / is the projection of the initial polar coordinates ⁇ , and ⁇ , initially defining a direction of light in the BTF, on a single dimension
- BRDF z (v, I) is a BRDF function model for the z pixel, in which the direction of light and the direction of a point of view are each expressed in a single dimension.
- v is the projection of the polar coordinates ⁇ v and ⁇ v initially defining the direction of a point of view in the BTF, on a single dimension
- the approximation-based methods according to this first approach make it possible to obtain a compact and continuous representation of the textures, which adapts well to the current rendering algorithms, generally carried out directly on programmable graphic cards. However, they are limited by the complexity of the calculations required. Indeed, if a singular value decomposition of a BTF is sufficient to obtain an approximation by bi-quadratic polynomials, the approximation of the same function by BRDF function models is often very complex. Moreover, these approximations do not make it possible to make the variety of the effects captured during the acquisition of the BTFs. Effects related to light-wise or light-directional displacements would require, to be taken into account, the use of even more complex BRDF function models, or different polynomials, as described in the articles:
- This method consists in finding the directions in space that best represent the correlations of the dataset constituted by the BTF.
- a new base is then defined, resulting from the orthogonalization of the analysis domain by a search of the eigenvectors and eigenvalues of the covariance matrix associated with the BTF.
- the axes of the new associated marker are such that the variance of the initial samples of the BTF is maximum.
- the eigenvectors are then sorted in order of importance, according to their corresponding eigenvalues, in descending order. We keep only the most influential axes.
- the initial data of the BTF are then projected in this new space of reduced dimensions. We thus obtain, by linear combination, an approximation of the original signal in this new base:
- PCA Component Analysis
- CGI Computer Graphics International
- a spatial tensor represents the projection of the spatial coordinates JC and y on a single dimension
- a tensor linked to the point of view represents the projection of the polar coordinates ⁇ v and ⁇ v on a single dimension
- a tensor related to the direction of light represents the projection of the polar coordinates ⁇ , and ⁇ , on a single dimension.
- the methods based decomposition according to this second approach improve the qualitative results compared to the methods based approximation, the choice of the number of principal components determining the quality of the textures compressed by such methods. They further define a compact and discrete representation of multidimensional textures. This compactness is due to the fact that the vectors resulting from the decomposition are ranked in order of importance and only the most representative vectors are taken into account. However this decimation is random and does not allow to define which details will be hidden during the transformation. It also avoids certain singularities captured during the acquisition process, the substituted axes corresponding to zones of weak correlation. This lack of frequency information limits the flexibility of these methods: they do not make it possible to directly provide a representation according to different levels of detail, by a progressive decoding of the information.
- a multi-resolution representation of the textures compressed by these methods requires, as in the article "Level-of-detail representation of bidirectional texture functions for real-time rendering", additional transformations.
- the decomposition-based multi-resolution method mentioned in this article actually produces a texture representation by resolution and therefore does not correspond to a single multi-resolution representation allowing progressive decoding of the information based on a choice of resolution.
- the existing methods of texture compression therefore do not allow to preserve all the captured details of a texture, nor to provide a progressive decoding of the texture information in order to adapt to the material constraints during the transmission of this texture.
- information for example related to the power of the graphics cards receivers or network throughput used.
- this ability is necessary to generalize the use of textures to decentralized systems or to a wider range of devices, such as mobile devices.
- the present invention aims to solve the disadvantages of the prior art by providing in particular a method and a data encoding device representative of a multidimensional texture, using the properties of the wavelet analysis on all dimensions of this texture.
- the invention proposes a method for encoding data representative of a multidimensional texture, said data being initially organized on a number of dimensions at least equal to three, at least one of said dimensions being linked to a rendering parameter of said texture, characterized in that it comprises the steps of:
- the compression ratio obtained is better than in the prior art, while keeping a good level of detail and requiring only a small calculation complexity, This allows to rebuild in real time the initial data during rendering.
- multilinear algebra is sometimes expensive during synthesis, ie say during decoding, the tensors used in the invention are hollow, that is to say that most of their coefficients are zero, which, in practice, involves few calculations.
- the high compression ratio obtained by the coding method according to the invention makes it possible to maintain a texture representation of very good quality, which in particular retains all the effects related to the displacement of the point of view or the direction of the light.
- the wavelet analysis corresponds to a frequency analysis of an initial signal, which is broken down into frequency bands. As a result, it is possible to target the decimation frequency and this for each dimension of the signal. This makes it possible to preserve all the singularities of the compressed texture according to the invention.
- This analysis is also a multi-resolution analysis, it produces a representation of the initial signal by scale, in which one qualifies of coarser level the coefficients low-frequency and of the finest level the reconstruction starting from the coefficients high-frequency . This feature allows you to define different levels of detail from a single texture representation.
- Another advantage related to wavelet analysis is to make the coding according to the invention extremely configurable: according to whether a representation of the best possible quality or a most compact representation is preferred, a lossless or lossy compression.
- the coding method according to the invention is compatible with the "peer-to-peer” virtual navigation systems, which require fault-tolerance and great flexibility of information transmission. In particular, it makes it possible to transmit texture data from different transmitters in parallel to receivers of very different capacities.
- the coding method according to the invention allows dynamic and fast access to texture data coded according to the invention. Indeed, it is not necessary to reconstruct the set of initial data to each new image of the same texture when rendered, because the coding of the data according to the invention allows a local reconstruction of areas of interest.
- the coding method according to the invention comprises a preliminary step of reorganization of said data representative of a multidimensional texture according to a predetermined number of dimensions.
- This reorganization step of the texture data makes it possible to simplify the wavelet decomposition, by reducing the number of dimensions in particular.
- said data are arranged so as to maximize the correlation between two successive samples of said data in at least one dimension.
- said data representative of a multidimensional texture represent colors according to the YUV color coding system.
- YUV color coding system instead of a conventional red / green / blue RGB coding system, to represent the colors of the coded texture according to the invention, makes it possible to obtain, for an equivalent visual quality, better compression rates. Indeed, since the human eye is less sensitive to chromatic variations and more to luminance, this choice makes it possible, during the quantization step associated with the coding, to code less precisely the parameters U and V representative of the chromaticity, than the parameter Y representative of the luminance.
- the compression step of the coding method according to the invention uses a "zerotree" type coding.
- This type of coding makes it possible to optimize the compression of the texture data that has been analyzed in wavelets.
- the invention also relates to a method for decoding data representative of a multidimensional texture, said data being initially organized on a number of dimensions at least equal to three, at least one of said dimensions being linked to a rendering parameter of said texture, characterized in that it comprises the steps of: - decompression of said data, and wavelet synthesis on said data dimensions obtained from said decompression.
- the invention also relates to a signal representative of a multidimensional texture initially represented by data organized on a number of dimensions at least equal to three, at least one of said dimensions being linked to a rendering parameter of said texture, characterized in that said data has been encoded by wavelet analysis on said dimensions, and then by compression of data obtained from the result of said analysis.
- the invention further relates to a device for encoding data representative of a multidimensional texture, said data being initially organized over a number of dimensions of at least three, at least one of said dimensions being linked to a rendering parameter of said texture, characterized in that it comprises: - wavelet analysis means of said data on said dimensions,
- the invention also relates to a device for decoding data representative of a multidimensional texture, said data being initially organized on a number of dimensions of at least three, at least one of said dimensions being linked to a rendering parameter of said texture, characterized in that it comprises:
- the decoding method, the signal, the coding device and the decoding device have advantages similar to those of the coding method according to the invention. 1 0
- the invention finally relates to a computer program comprising instructions for implementing the coding method according to the invention or the decoding method according to the invention, when it is executed on a computer.
- FIG. 1 represents the sampling of the acquisition of a BTF modeling a texture
- FIG. 2 is a table illustrating the values of this sampling
- FIG. 3 represents an interpretation of this BTF
- FIG. 4 represents an embodiment of the coding method and the decoding method according to the invention
- FIG. 5 represents steps of the coding method according to the invention, as described in this embodiment,
- FIG. 6 represents a level of wavelet decomposition of texture data
- FIG. 7 represents a wavelet decomposition of texture data
- FIG. 8 represents a decomposition tree in wavelet packets of texture data
- FIG. 9 represents the allocation of costs to the elements of this decomposition in wavelet packets
- FIG. 10 represents another decomposition tree in texture data wavelet packets
- FIG. 11 represents a coded texture data representation format according to the invention
- FIG. 12 represents steps of the decoding method according to the invention as described in this embodiment.
- the coding method according to the invention is used to compress a multidimensional texture expressed in the form of a six-dimensional BTF.
- wavelet decomposition can be generalized to any number of dimensions
- the method according to the invention is also applicable to multidimensional textures represented differently, for example in the form of a four-dimensional "Polynomial Texture Map" or "Time-varying BTF". "in seven dimensions.
- the BTF expressing this texture is the result of an acquisition process shown in Figure 1.
- the BTF is sampled according to a hemisphere of shots.
- each point of the texture is thus photographed in a polar coordinate point of view direction ⁇ v and ⁇ v , and in a direction of the light of polar coordinates ⁇ and ⁇ .
- the table TAB of FIG. 2 indicates the number of photographs taken for a given latitude ⁇ v or ⁇ .
- the angle ⁇ v varies from 72 degrees to 72 degrees, which corresponds to 5 photographs taken for this latitude.
- FIG. 3 illustrates this interpretation of the BTF: each of the images I of these 6400 images represents the texture in parametric coordinates x and y, and corresponds to a point of view direction ( ⁇ Vl ⁇ v ) and a direction of light ( ⁇ , q> ⁇ ).
- the coding method according to the invention uses wavelets which are obviously wavelets of the "second generation" type, because the wavelets of the first generation are not suitable. to this type of sampling.
- the images I forming the data of the BTF are stored in a database BDD shown in FIG. 4, the coding method according to the invention being implemented in this embodiment in a software manner in a computer ORD having access to this database. of data.
- the calculations necessary for the coding method according to the invention are carried out in the CPU of the computer ORD. As a variant, given the importance of the size of the data on which the calculations necessary for their coding are performed, these calculations are carried out on several computers operating in parallel.
- the coding method according to the invention is represented in the form of an algorithm comprising steps C1 to C3.
- the first step C1 is a reorganization of the data of the BTF whose data are initially organized in six dimensions, in a reduced number of dimensions, so as to limit the complexity of the calculations.
- this number of dimensions depends on the applications using the compressed data, depending on whether less complex calculations or a higher compression ratio are preferred. In fact, the more dimensions we keep, the more we exploit the inter-dimensional coherence and the more 1 es
- BTF (x, y, ⁇ v , ⁇ v , ⁇ ,, ⁇ ,) BTF (x, y, v, /),
- v is the projection of the initial polar coordinates ⁇ v and ⁇ v initially defining the direction of a point of view in the BTF, on a single dimension
- / is the projection of the initial polar coordinates ⁇ , and ⁇ t initially defining a direction of light in the BTF, in one dimension.
- the BTF data is not rearranged, wavelet analysis is then done in six dimensions.
- two types of wavelets are used, for example: non-separable two-dimensional wavelets in order to adapt the analysis filters to the sampling of the BTF according to the point of view and the light directions , and separable one-dimensional wavelets for spatial analysis.
- the initial data of the BTF are rearranged in three dimensions according to a decomposition called "reflectance field", that is to say that it is expressed by point of view v and that the direction of the light is projected on one dimension:
- BTF (x, y, ⁇ v , ⁇ v , ⁇ ⁇ , ⁇ ,) ⁇ BTF v (x, y) ⁇ , Vv
- the data of the BTF are rearranged in five dimensions, the direction light being projected on one dimension:
- BTF (x, y, ⁇ v , ⁇ v , ⁇ ⁇ , ⁇ ⁇ ) BTF (x, y, ⁇ v , ⁇ v , l)
- the projection v corresponding to the doublet ( ⁇ v , ⁇ v ) is ranked in ascending order according to ⁇ v and ⁇ v , that is to say that according to the dimension of the point of view v the images I are classified in the following order of sampling: (0,0), (15,0), (15,72) I ..., (75,345).
- the projection / corresponding to the doublet ( ⁇ ,, ⁇ t ) is ranked in increasing order according to ⁇ , and ⁇ ,.
- the images I are classified so as to maximize the correlation between two successive images in the direction of light.
- This classification is performed for example in the following manner on all the images having in common the same point of view:
- a root image is chosen from this set, for example the projection image corresponding to the doublet (0,0) in polar coordinates,
- the next image is then iteratively searched from the previous image, this image corresponding to the image minimizing the difference between the previous image and the set of images that have not already been classified.
- This classification of the images I in order of likelihood along the axis corresponding to the projection /, allows in the following to improve the compression during the step C3.
- BTF data is coded for RGB acquisition.
- this color coding format does not make it possible to exploit the perception factor to its maximum.
- a change of RGB color space towards the color space YUV is preferably carried out, where Y represents the luminance, and U and V the chrominance. .
- This change has an impact on the continuation of the treatment because the utility of such a modification is to provide a larger quantity of information for the luminance with respect to the chrominance during the quantification in step C3. 2
- the second step C2 of the coding method is the wavelet analysis of the data thus reorganized and reformatted, on the four dimensions chosen in step C1.
- wavelet analysis means that a wavelet or wavelet packet signal is broken down into successive wavelet transforms. A wavelet analysis therefore covers at least one wavelet decomposition.
- This wavelet decomposition is done according to the organization chosen in step C1.
- the use of second-generation wavelets is necessary because of irregular sampling intervals and border areas. It should also be noted that even assuming that new samples are synthesized in step C1 in order to regularize these intervals, the analysis domain still being limited, the second generation wavelets are still necessary.
- step C1 if the BTF data has been rearranged in step C1 by a number of dimensions greater than four, the calculations are weighted during the wavelet decomposition to transform an irregular interval into a regular interval and back to the case. regular figure. This weighting makes the wavelet decomposition more faithful to reality, that is to say that during decompression the decompressed data make it possible to synthesize new, very realistic texture views.
- wavelet base more complex, higher order. This allows, however, at the cost of more complex calculations and longer, to maintain despite compression a texture representation of very good resolution.
- bi-orthogonal wavelets which are practical for their ease of construction via the "lifting scheme” method, or geometric wavelets are used, considering the spatial configuration of the texture as a geometry in a space of the same dimension.
- mesh-based wavelets using the quadrilateral as primitive, and applying conventional subdivision techniques such as the "Butterfly" technique.
- the basic use of wavelets having two zero moments is a good compromise to reconcile quality of restitution and speed of calculation.
- the wavelet decomposition in step C2 is performed at each decomposition level according to the diagram of FIG. 6.
- S J be the signal formed by the set of data ⁇ s ⁇ J , mn ⁇ of the jth wavelet decomposition level in the Haar base, reorganized data obtained at the end of step C1, and wherein:
- k is an index representing the kth value of the signal along the axis of the spatial coordinate x
- p is an index representing the p-th value of the signal along the axis of the spatial coordinate y
- - m is an index representing the mth value of the signal along the axis of the projection /
- n is an index representing the nth value of the signal along the axis of the projection v.
- the decomposition is done dimension by dimension, by block of data each corresponding to the set of values of the BTF according to one dimension, when each of the three other dimensions is fixed to a given value.
- the order of processing of these dimensions is chosen according to the correlation of the data of the BTF.
- the spatial correlation of the texture being stronger than its correlation during a change of direction of the light, itself stronger than during a change of direction of point of view, the wavelet decomposition is carried out first following the index k, then following the index p, then following the index m and finally following the index n.
- the signal S J is first subjected along the axis of the spatial coordinate x to the following functions:
- the low-pass filter h performs the averages of the data ⁇ s k J pmn ⁇ according to the direction in question, which produces the data ⁇ s ⁇ J pmn ⁇ defined by:
- the decomposition takes place according to the method of the "lifting scheme", which means for the decomposition in the Haar base that the difference between two values of the signal S J is first calculated and then the sum is calculated. of these two values from this difference:
- the local character of the wavelet decomposition allows the decomposition to be performed "out of core", that is to say that all the data to be processed is not entirely contained in the main memory. This character also allows data to be processed in parallel by multiple processors.
- the field of analysis does not need to be segmented to be treated in its entirety.
- the data ) are then subjected to the same functions but according to the index p: 5 - the operator s separates the data ⁇ s k J ! p ' mn ⁇ and the data [d k J ? pnm ⁇ according to their even or odd indices, the filter h produced from the data ⁇ s k ' pmn ⁇ the data ⁇ s k J ', p . mn ⁇ and from the data ⁇ d k J f pm ⁇ ⁇ the data ⁇ d k J f p ', mn ⁇ ,
- the filter h produced from the data ⁇ s k Jh p , mn ⁇ , the data ⁇ s k J ! p "! m ',” ⁇ defined by: - and the filter g produced from the data defined by: f j jhltg _ ⁇ h _ ⁇ rjhh
- the filter h performs the averages of data ⁇ S ym k J ', n ⁇ according to the index n, which produces the data ⁇ s / ⁇ y ** ⁇ defined by: ojhhh, Qjhhh ⁇ j hhhh _ ° k. 'p : mX2n') " * " J * ', pW (2fl' + 1) ⁇ k'p'm'n ' ⁇ j
- the filter g calculates the differences between the data ⁇ s ⁇ ,. ⁇ according to the index n, which produces the data defined by:
- the set of data produced during this last decomposition according to the index n forms the result of the wavelet decomposition of the signal S J , ie the data of the at least one wavelet decomposition level of the data obtained at the end of the wavelet decomposition. step C1.
- We distinguish in this result the data low-frequency low-resolution, and other data that form a signal J ⁇ ⁇ of higher frequency.
- the decomposition detailed above is a conventional wavelet decomposition of the initial signal S J formed from the data obtained in step C1.
- this decomposition shown in FIG. 7 only the low frequency signal S J ⁇ ⁇ is re-decomposed at the next decomposition level, the detail signal d J ⁇ ⁇ being retained.
- the signal S J "x is decomposed into wavelets and produces a signal S J ⁇ 2 of lower frequency that the signal S J ', and a signal J ⁇ 2 of lower frequency than the signal J ⁇ ⁇ and at the decomposition level following the signal S J ⁇ 2 is itself decomposed into wavelets and so on.
- the last wavelet decomposition on the signal S 1 produces the signals d 0 and S 0 .
- the data obtained at the end of the step C1 being the images I coded according to the YUV format and classified according to 80 directions of viewpoint and 80 directions of the light, supposing that these images are of resolution 256 * 256, one s for example, stops after three levels of wavelet decomposition.
- the result of the wavelet analysis is then formed of the signals S 0 , and d ° to d J ⁇ ⁇ where J is three.
- the 5 ° low frequency signal contains 10 * 10 * 32 * 32 colors coded in the YUV color space.
- step C2 in order to code the texture optimally, in this step C2 the data obtained at the end of step C1 are broken down into wavelet packets.
- This decomposition is represented on the tree of FIG. 8.
- the root of the tree corresponds to the initial signal S J containing the data obtained at the end of step C1.
- the next level is the result of a wavelet transformation iteration, i.e., the low frequency signal S i and the high frequency signal I x .
- the recursive decomposition of these signals completes the lower levels of the tree.
- the decomposition of the signal S J ⁇ X gives two signals S J ⁇ 2 and d J ⁇ 2
- the decomposition of the signal d Ix also gives two signals S ⁇ J ⁇ and d ⁇ .
- This decomposition into wavelet packets makes it possible to choose an optimal decomposition tree according to a given criterion, such as entropy of the coding, a predetermined threshold, or the distortion generated by the coding. For this we assign a cost function to each wavelet decomposition, that is, say at each node of the tree shown in Figure 8. This cost function corresponds to the criterion chosen to optimize the tree.
- the value of the cost function at the decomposition node corresponding to the signal S J is:
- the decomposition into wavelet packets will stop on the left branch of the tree at the signals S J ⁇ 2 and d J ⁇ 2 , which will not be decomposed, because the sum of their costs is greater than or equal to the cost of their parent signal S J ⁇ ⁇ .
- the value of the cost function at the decomposition node corresponding to the signal S J is:
- This cost function defines Shannon's entropy, which measures the amount of information, that is, different coefficients in a decomposition. This criterion is useful for the coding method according to the invention because it keeps the decomposition of lower entropic cost.
- the wavelet analysis performed at this step C2 is an integer decomposition into wavelets or an entire decomposition into wavelet packets.
- Such a method is described in the article "Reversible image compression via multi-resolution representation and predictive coding,” by A. Said and W. Pearlman, published in 1993 at an international conference “Visual Communications and Image Processing ". This further optimizes the wavelet analysis by limiting the size of the result of this analysis. Indeed, the data from the analysis are thus represented with fewer resources, the size in bytes of an integer being smaller than that of a decimal number.
- any wavelet basis can be modified to perform such an entire decomposition. For example, the transformation by Haar rondelette of the signal S J according to the index k becomes:
- step C3 of the coding method according to the invention is the compression of the data obtained at the end of step C2, the result of the wavelet analysis of the BTF data reorganized and reformatted in step C1.
- this compression uses a so-called coding
- a coefficient of the signal d J ⁇ 2 should be related to a coefficient in each of the 16 signals from the decomposition of the signal J ⁇ ⁇ .
- the signal S ⁇ 1 issuing from this decomposition is itself re-decomposed, in particular into a signal Si / ⁇ - ⁇ of level coarser than that of the signal d J ⁇ 2 .
- the coefficients of the signals S ⁇ 1 and d ⁇ are related, not to the coefficients of the signal J ⁇ 2 , but to the coefficients of the signal J ⁇ % .
- a bit stream is obtained which encodes the wavelet packet decomposition performed in step C2.
- the data thus obtained are much smaller in size than at the end of step C2. Indeed, the fact of exploiting the decrease of the coefficients across the scales, added to the fact that the wavelet analysis produces numerous coefficients close to zero, allows the "zerotree” coding method to obtain a high rate of compression.
- the data obtained are organized in order of importance, as shown in Figure 11.
- the ZTO data encode coefficients corresponding to the coarsest level of detail, while the data ZT 1 encodes detail coefficients at a somewhat less coarse level, the ZT2 data of the detail coefficients at an even finer level, and so on.
- ZTO data is information that best describes the original data.
- a progressive representation of the texture is available insofar as the data after the ZTO data makes it possible to progressively refine the coarse representation of the texture contained in these ZTO data.
- a coarse representation of the texture is obtained.
- this step C3 instead of using a "zerotree” coding, a “dynamic Huffman 11 " type coding combined with a non-uniform scalar quantization is used, in particular the Y component of the data resulting from the step C2 is less quantified than the components U and V.
- the pitch of this quantization is for example adequately fixed for a fixed compressed data size.
- This process called “Rate Allocation”, is used in the standard "Joint Photography Experts Group (JPEG
- JPEG Job Photographic Experts Group
- the quantization step is calculated iteratively until the desired compression ratio is reached.
- This aftertreatment is useful for example when transmitting the compressed data over a network at a fixed rate.
- such a quantification is adaptable in order to quantify less certain areas of interest of the texture, if such zones are identified, compared to other zones of lesser interest.
- the terminal T Upon reception of the data flow F, the terminal T stores in memory all the data received, or only the first data, for example the data ZTO, ZT 1 and ZT2 if the terminal T has a limited memory capacity, or if a failure communication stops the sending of data just after sending the ZT2 data.
- the terminal T after channel decoding of the signal carrying the data stream F, decodes the data received according to the steps D1 to D3 shown in FIG. 12.
- the step D1 is the decompression of the received data.
- the terminal T uses a "zerotree" decoding algorithm inverse to that used in step C3 for coding, that is to say that it uses the same rules for associating the coefficients of a level of resolution to another to retrieve data from wavelet packet decomposition. If only the data ZTO, ZT1 and ZT2 are decompressed, only the coefficients of the first resolution levels of the wavelet packet decomposition of the step C2 are obtained.
- the terminal T is able to decode only a portion of the data received, for example to make the texture only in certain directions of viewpoints or certain directions of light.
- Step D2 is the wavelet synthesis of the decompressed data in step D1. Since the "zerotree" coding is a function of the structure of the wavelet packet decomposition tree retained in step C2, the terminal T easily reconstitutes all or part of the reordered texture data and reformatted at the end of step C1. All that is required is to perform inverse Haar transforms in the order of wavelet packet decomposition, dimension by dimension.
- step D3 is the six-dimensional reorganization of the data obtained at the end of step D2.
- the terminal T thus obtains a texture expressed in the form of a BTF 1 that can be used directly to render it on a screen. If only the data ZTO, ZT1 and ZT2 were synthesized in step D2, this BTF is in fact a rough representation of the compressed texture according to the invention.
Landscapes
- Engineering & Computer Science (AREA)
- Multimedia (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Signal Processing (AREA)
- Compression Or Coding Systems Of Tv Signals (AREA)
- Compression Of Band Width Or Redundancy In Fax (AREA)
- Image Generation (AREA)
Abstract
Description
Claims
Applications Claiming Priority (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| FR0752998 | 2007-02-01 | ||
| PCT/FR2008/050161 WO2008110719A1 (fr) | 2007-02-01 | 2008-01-31 | Procede de codage de donnees representatives d'une texture multidimensionnelle, dispositif de codage, procede et dispositif de decodage, signal et programme correspondants |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| EP2135221A1 true EP2135221A1 (fr) | 2009-12-23 |
Family
ID=38654952
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| EP08762021A Ceased EP2135221A1 (fr) | 2007-02-01 | 2008-01-31 | Procédé de codage de données répresentatives d'une texture multidimensionnelle, dispositif de codage, procédé et dispositif de décodage, signal et programme correspondants |
Country Status (4)
| Country | Link |
|---|---|
| US (1) | US20100053153A1 (fr) |
| EP (1) | EP2135221A1 (fr) |
| JP (1) | JP5102310B2 (fr) |
| WO (1) | WO2008110719A1 (fr) |
Families Citing this family (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| FR2941548A1 (fr) * | 2009-01-28 | 2010-07-30 | France Telecom | Procede de representation d'un materiau |
| US8378859B2 (en) | 2010-07-16 | 2013-02-19 | Apple Inc. | Memory compression technique with low latency per pixel |
| US8989509B2 (en) | 2012-12-18 | 2015-03-24 | Apple Inc. | Streaming wavelet transform |
| JP6410451B2 (ja) * | 2014-03-31 | 2018-10-24 | キヤノン株式会社 | 情報処理装置、計測システム、情報処理方法およびプログラム。 |
| US9392212B1 (en) * | 2014-04-17 | 2016-07-12 | Visionary Vr, Inc. | System and method for presenting virtual reality content to a user |
| US9665170B1 (en) | 2015-06-10 | 2017-05-30 | Visionary Vr, Inc. | System and method for presenting virtual reality content to a user based on body posture |
| CN105898305B (zh) * | 2016-04-12 | 2019-02-15 | 上海兆芯集成电路有限公司 | 基于无损联合图像专家小组格式的图像压缩与解压缩方法 |
| CN110192106B (zh) * | 2017-01-16 | 2021-09-28 | 株式会社岛津制作所 | 数据解析装置以及数据解析用程序 |
Family Cites Families (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6057884A (en) * | 1997-06-05 | 2000-05-02 | General Instrument Corporation | Temporal and spatial scaleable coding for video object planes |
| US6661927B1 (en) * | 2000-07-27 | 2003-12-09 | Motorola, Inc. | System and method for efficiently encoding an image by prioritizing groups of spatially correlated coefficients based on an activity measure |
| FR2827409B1 (fr) * | 2001-07-10 | 2004-10-15 | France Telecom | Procede de codage d'une image par ondelettes permettant une transmission adaptative de coefficients d'ondelettes, signal systeme et dispositifs correspondants |
| EP1296288B1 (fr) * | 2001-09-21 | 2006-04-19 | Interuniversitair Microelektronica Centrum Vzw | Un dispositif et méthode pour un codage à blocs avec une mémoire FIFO bidimensionnelle |
| JP3901644B2 (ja) * | 2003-01-30 | 2007-04-04 | 株式会社東芝 | テクスチャ画像圧縮装置及び方法、テクスチャ画像抽出装置及び方法、データ構造、記憶媒体 |
| FR2856548A1 (fr) * | 2003-06-18 | 2004-12-24 | France Telecom | Procede de representation d'une sequence d'images par modeles 3d, signal et dispositifs correspondants |
| FR2876821A1 (fr) * | 2004-10-14 | 2006-04-21 | France Telecom | Procede de decodage local d'un train binaire de coefficients d'ondelettes |
| US20070019740A1 (en) * | 2005-07-25 | 2007-01-25 | Texas Instruments Incorporated | Video coding for 3d rendering |
| US8340193B2 (en) * | 2006-08-04 | 2012-12-25 | Microsoft Corporation | Wyner-Ziv and wavelet video coding |
| US20080095235A1 (en) * | 2006-10-20 | 2008-04-24 | Motorola, Inc. | Method and apparatus for intra-frame spatial scalable video coding |
-
2008
- 2008-01-31 WO PCT/FR2008/050161 patent/WO2008110719A1/fr not_active Ceased
- 2008-01-31 JP JP2009547740A patent/JP5102310B2/ja not_active Expired - Fee Related
- 2008-01-31 EP EP08762021A patent/EP2135221A1/fr not_active Ceased
- 2008-01-31 US US12/524,744 patent/US20100053153A1/en not_active Abandoned
Non-Patent Citations (1)
| Title |
|---|
| See references of WO2008110719A1 * |
Also Published As
| Publication number | Publication date |
|---|---|
| WO2008110719A1 (fr) | 2008-09-18 |
| JP5102310B2 (ja) | 2012-12-19 |
| US20100053153A1 (en) | 2010-03-04 |
| JP2010518667A (ja) | 2010-05-27 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| EP2135221A1 (fr) | Procédé de codage de données répresentatives d'une texture multidimensionnelle, dispositif de codage, procédé et dispositif de décodage, signal et programme correspondants | |
| TW202147849A (zh) | 點雲壓縮方法、編碼器、解碼器及儲存媒介 | |
| Balsa Rodríguez et al. | A survey of compressed GPU-based direct volume rendering | |
| Guarda et al. | Deep learning-based point cloud coding: A behavior and performance study | |
| Hu et al. | An adaptive two-layer light field compression scheme using GNN-based reconstruction | |
| CA2346342C (fr) | Compression et codage de reseau maille tridimensionnel | |
| Lin et al. | 3d model streaming based on jpeg 2000 | |
| Liu et al. | Mobile volumetric video streaming system through implicit neural representation | |
| US11893684B1 (en) | Systems and methods for signal-based point cloud representation | |
| Feng et al. | Neural subspaces for light fields | |
| EP1574068B1 (fr) | Procede de codage d'une image par ondelettes et procede de decodage correspondant | |
| Rudolph et al. | Learned compression of point cloud geometry and attributes in a single model through multimodal rate-control | |
| FR2876821A1 (fr) | Procede de decodage local d'un train binaire de coefficients d'ondelettes | |
| US12469179B2 (en) | System and methods for upsampling of decompressed data after lossy compression using a neural network | |
| EP4078959B1 (fr) | Procede d'encodage d'une image numerique en vue de sa compression | |
| Cui et al. | Palette-based Color Attribute Compression for Point Cloud Data. | |
| WO2006053842A1 (fr) | Procede de codage d'images codees par ondelettes a controle de debit, dispositif de codage et programme d'ordinateur correspondants | |
| Hakami | Wavelet based multimedia data compression techniques | |
| Freitas | Sensing and Reconstruction of Plenoptic Point Clouds | |
| de Carvalho Freitas | Sensing and reconstruction of plenoptic point clouds | |
| WO2004112398A1 (fr) | Procede et dispositif de codage et de decodage d’une sequence d’images video | |
| Pushpalatha et al. | Interpolative Model on Hueristic Projection Transform for Image Compression in Cloud Services. | |
| FR2929431A1 (fr) | Procede et dispositif de classification des echantillons representatifs d'un signal numerique d'image | |
| Siddeq | Novel methods of image compression for 3D reconstruction | |
| Pratapa et al. | RANDM: Random Access Depth Map Compression Using Range-Partitioning and Global Dictionary |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PUAI | Public reference made under article 153(3) epc to a published international application that has entered the european phase |
Free format text: ORIGINAL CODE: 0009012 |
|
| 17P | Request for examination filed |
Effective date: 20090813 |
|
| AK | Designated contracting states |
Kind code of ref document: A1 Designated state(s): AT BE BG CH CY CZ DE DK EE ES FI FR GB GR HR HU IE IS IT LI LT LU LV MC MT NL NO PL PT RO SE SI SK TR |
|
| RIN1 | Information on inventor provided before grant (corrected) |
Inventor name: BARIL, JEROME Inventor name: GIOIA, PATRICK |
|
| 17Q | First examination report despatched |
Effective date: 20091216 |
|
| DAX | Request for extension of the european patent (deleted) | ||
| REG | Reference to a national code |
Ref country code: DE Ref legal event code: R003 |
|
| STAA | Information on the status of an ep patent application or granted ep patent |
Free format text: STATUS: THE APPLICATION HAS BEEN REFUSED |
|
| 18R | Application refused |
Effective date: 20121018 |