EP4085613A1 - Synthese iterative de vues a partir de donnees d'une video multi-vues - Google Patents
Synthese iterative de vues a partir de donnees d'une video multi-vuesInfo
- Publication number
- EP4085613A1 EP4085613A1 EP20845681.4A EP20845681A EP4085613A1 EP 4085613 A1 EP4085613 A1 EP 4085613A1 EP 20845681 A EP20845681 A EP 20845681A EP 4085613 A1 EP4085613 A1 EP 4085613A1
- Authority
- EP
- European Patent Office
- Prior art keywords
- image
- view
- data
- synthesis
- synthesized
- 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.)
- Pending
Links
Classifications
-
- 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
- H04N19/50—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using predictive coding
- H04N19/597—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using predictive coding specially adapted for multi-view video sequence encoding
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T15/00—Three-dimensional [3D] image rendering
- G06T15/04—Texture mapping
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T15/00—Three-dimensional [3D] image rendering
- G06T15/10—Geometric effects
- G06T15/20—Perspective computation
- G06T15/205—Image-based rendering
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/50—Image enhancement or restoration using two or more images, e.g. averaging or subtraction
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/50—Depth or shape recovery
- G06T7/536—Depth or shape recovery from perspective effects, e.g. by using vanishing points
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N13/00—Stereoscopic video systems; Multi-view video systems; Details thereof
- H04N13/10—Processing, recording or transmission of stereoscopic or multi-view image signals
- H04N13/106—Processing image signals
- H04N13/111—Transformation of image signals corresponding to virtual viewpoints, e.g. spatial image interpolation
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N13/00—Stereoscopic video systems; Multi-view video systems; Details thereof
- H04N13/20—Image signal generators
- H04N13/282—Image signal generators for generating image signals corresponding to three or more geometrical viewpoints, e.g. multi-view systems
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N13/00—Stereoscopic video systems; Multi-view video systems; Details thereof
- H04N13/30—Image reproducers
- H04N13/349—Multi-view displays for displaying three or more geometrical viewpoints without viewer tracking
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20212—Image combination
- G06T2207/20221—Image fusion; Image merging
Definitions
- the present invention relates generally to the field of image synthesis.
- the present invention applies more particularly to the synthesis of uncaptured intermediate points of view, from images of several 2D (two-dimensional), 360 °, 180 °, etc. views which are captured to generate an immersive video. , such as in particular a 360 ° video, 180 °, etc.
- the invention can in particular, but not exclusively, be applied to video decoding implemented in current video decoders HEVC (English abbreviation for "High Efficiency Video Coding") and its extensions MV-HEVC (English abbreviation for "Multiview HEVC” ), 3D-HEVC, etc.
- the scene is conventionally captured by a set of cameras.
- These cameras can be:
- the images of such captured views are traditionally encoded and then decoded by the viewer's terminal. However, in order to provide a sufficient quality of experience, and therefore visual quality and good immersion, displaying the captured views is insufficient.
- the images of a multitude of views, called intermediate views, must be calculated from the images of the decoded views.
- a synthesis algorithm is capable, from images of N views, with N> 1, of synthesizing an image from an intermediate point of view located anywhere in space.
- the image of a view considered among N includes a texture component and a depth map which indicates the distance separating the different elements of the scene of the camera that captured the image of this view.
- the image of an intermediate view obtained by synthesis also comprises a texture component synthesized from the N texture components of the images of the N views.
- the depth map is either captured or calculated from the N texture components.
- such a depth map may contain numerous errors which are either linked to the capture, or linked to the calculation, or linked to the compression of the images of the N views.
- an object of the present invention relates to a method for synthesizing an image of a view from data from a multi-view video, implemented in an image synthesis device.
- Such a method comprises an image processing phase as follows:
- the synthesis method according to the invention makes it possible to optimize a compromise - visual quality / computational complexity - since the step of generating image synthesis data does not necessarily use all the texture data of one. image of a video view.
- Another advantage of the iterative synthesis method according to the invention opens up the prospects of a scalability in complexity, in the case where a terminal which requires the display of the image of a synthesized view does not allow the display. of this image with high resolution or high quality.
- an image of a view used to generate the summary data is selected as a function of the result of the analysis of the image of the synthesized view modified during a phase processing preceding the current iteration.
- Such a selection being conditioned on the result of a previous analysis of the image of the synthesized view, it is thus possible to use at each new step of generation of synthesis data at least one image of a view of the multi-video. -views which are as relevant as possible to complete / refine the image synthesis carried out during one or more previous processing phase (s).
- the generation of the summary data uses the summary data generated during a processing phase preceding the current iteration.
- the generation of synthesis data comprises a modification of the synthesis data generated during a processing phase preceding the current iteration, from texture data used in the process. current iteration.
- Such an embodiment advantageously makes it possible to refine the summary data generated during one or more processing phases preceding the current iteration, by taking into account one or more new view (s) considered.
- the summary data generated during the current iteration are combined with summary data generated during a processing phase preceding the current iteration.
- Such an embodiment advantageously makes it possible to obtain, at the current iteration, the image of a synthesized view which contains fewer and fewer occlusion zones.
- the visual quality of the synthesized image is thus improved over the course of the iterations.
- the summary data generated during a processing phase belong to the group comprising:
- the synthesis data generated during a processing phase are associated with an uncertainty value on the correlation of said synthesis data with the corresponding texture data of said at least one image of said. réelle which were used to generate said summary data.
- the generation of synthesis data uses at least one depth image of a view of the multi-view video and / or at least one item of information associated with the. multi-view video.
- the generation of synthesis data advantageously uses data other than texture data of at least one image of a view of the multi-view video. It may be depth data which complements the texture data to enrich the step of generating summary data. In addition to these depth data or alternatively to them, it may be at least one item of information associated with the multi-view video. Such information is representative of a difference or a correlation between the views of the multi-view video (camera parameters, occlusion map, difference in decoding modes, etc.).
- the modification of the image of the summary view calculated in a previous iteration uses summary data generated previously in the current iteration.
- a synthesized image calculated during a processing phase preceding the current iteration is used in addition synthesis data generated in the current iteration and of said at least one image of a view of the multi-view video, to modify the synthesized image calculated in the previous iteration.
- the modification of the image of the synthesized view calculated at a previous iteration comprises the calculation of an image of a synthesized view from said synthesis data generated at l 'current iteration and of the image of said synthesized view calculated at said previous iteration.
- the image of the view synthesized in the previous iteration is used as an input parameter of the step of modifying the image of the synthesized view previously calculated, thus coming in addition to the synthesis data generated at the current iteration and of the image of at least one view of the multi-view video which are used as input to this step.
- the image of the synthesized view calculated at the current iteration thus presents an optimized visual quality.
- the modification of the image of the synthesized view calculated in a previous iteration comprises the following:
- the image of the view synthesized in the previous iteration is used once the image of the synthesized view has been obtained in the current iteration, by being combined / merged with it, in a way to completely or partially complete the occlusion zones still present in the image synthesized at the current iteration.
- the current iteration is implemented in response to a message from the synthesis device, such a message containing: either location information of one or more zones / of one or more pixels the image of the synthesized view calculated previously at the current iteration, which does not meet the synthesis performance criterion,
- Such an embodiment makes it possible to make the synthesis method interactive by iterating an image processing phase in response to a message containing explicit information serving to guide the synthesis device in the choice of the parameters to be considered in order to implement the process. current processing phase.
- the image of a synthesized view can be gradually improved from one iteration to the next, on the basis of parameters chosen according to the result of the analysis of the previously synthesized images.
- the current iteration is also implemented in response to the reception of uncertainty values associated with the summary data generated in the previous iteration.
- Such an embodiment makes it possible to accompany the message sent to restart a current iteration by uncertainty values associated with the summary data generated in the previous iteration, to even better guide the choice of the parameters to be considered for implementing the processing phase at the current iteration.
- the various aforementioned embodiments or characteristics can be added independently or in combination with one another, to the synthesis process defined above.
- the invention also relates to a device for synthesizing an image of a view from multi-view video data, said device comprising a processor which is configured to implement the following:
- the invention also relates to a method for decoding a coded data signal of a multi-view video, implemented in an image decoding device, comprising the following: - decoding images of several coded views, producing a set of images of several decoded views,
- the synthesis data is generated from texture data of at least one of the decoded images of said set.
- the summary data does not need to be transmitted, which allows a significant reduction in the signaling cost of the encoded data.
- the summary data being coded beforehand, they are generated as follows:
- the invention also relates to a computer program comprising instructions for implementing the synthesis method according to the invention or the decoding method integrating the synthesis method according to the invention, according to any one particular embodiments described above, when said program is executed by a processor.
- Such instructions can be stored durably in a non-transient memory medium of the synthesis device implementing the aforementioned synthesis method or of the decoder implementing the aforementioned decoding method.
- This program can use any programming language, and be in the form of source code, object code, or intermediate code between source code and object code, such as in a partially compiled form, or in any other. desirable shape.
- the invention also relates to a recording medium or information medium readable by a computer, and comprising instructions of a computer program as mentioned above.
- the recording medium can be any entity or device capable of storing the program.
- the medium may comprise a storage means, such as a ROM, for example a CD ROM or a microelectronic circuit ROM, or else a magnetic recording means, for example a USB key or a hard disk.
- the recording medium can be a transmissible medium such as an electrical or optical signal, which can be conveyed via an electrical or optical cable, by radio or by other means.
- the program according to the invention can in particular be downloaded from an Internet type network.
- the recording medium can be an integrated circuit in which the program is incorporated, the circuit being adapted to execute or to be used in the execution of the aforementioned synthesis method or of the aforementioned decoding method.
- FIG. 1 represents the progress of an iterative image synthesis process, in a particular embodiment of the invention
- FIG. 2 represents a synthesis device implementing the synthesis method of FIG. 1
- FIG. 3A represents an arrangement of the synthesis device of FIG. 2, in a particular embodiment of the invention
- FIG. 3B represents an arrangement of the synthesis device of FIG. 2, in another particular embodiment of the invention.
- FIG. 4A represents an example of synthesis and image data of a synthesized view obtained during the implementation of a first phase of image processing of the synthesis method of FIG. 1,
- FIG. 4B represents an example of synthesis and image data of a synthesized view obtained during the implementation of an iteration of the image processing phase of FIG. 4A,
- FIG. 5A represents a step of generating summary data, in a particular embodiment of the invention.
- FIG. 5B represents a step of generating summary data, in another particular embodiment of the invention.
- FIG. 6A represents a way of generating synthesis data during the implementation of an iteration of an image processing phase of the synthesis method of FIG. 1, in a particular embodiment of the FIG. 'invention,
- FIG. 6B represents a way of generating synthesis data during the implementation of an iteration of an image processing phase of the synthesis method of FIG. 1, in another particular embodiment of invention,
- FIG. 6C represents a way of generating synthesis data during the implementation of an iteration of an image processing phase of the synthesis method of FIG. 1, in yet another particular embodiment of the invention,
- FIG. 7A represents a way of calculating a synthesized image during the implementation of an iteration of an image processing phase of the synthesis method of FIG. 1, in a particular embodiment of the invention
- FIG. 7B represents a way of calculating a synthesized image during the implementation of an iteration of an image processing phase of the synthesis method of FIG. 1, in another particular embodiment of the 'invention. Detailed description of an embodiment of the invention
- the following describes the implementation of an iterative image synthesis method according to one embodiment of the invention, which uses images from a plurality of views Vi to VN d 'a multi-view video (N> 1), representing a 3D scene respectively according to a plurality of viewing angles or a plurality of positions / orientations.
- Such images are images which have been reconstructed by a decoder, prior to the implementation of the iterative synthesis method.
- the actions executed by the synthesis method are implemented by computer program instructions.
- the synthesis device SYNT has the conventional architecture of a computer and comprises in particular a MEM_S memory, a processing unit UT_S, equipped for example with a PROC_S processor, and controlled by a stored computer program PG_S in MEM_S memory.
- the computer program PG_S comprises instructions for implementing the actions of the synthesis method as described above, when the program is executed by the processor PROC_S.
- the code instructions of the computer program PG_S are for example loaded into a RAM memory (not shown) before being executed by the processor PROC_S.
- the processor PROC_S of the processing unit UT_S notably implements the actions of the synthesis method described below, according to the instructions of the computer program PG_S.
- Such a synthesis device SYNT can be arranged, as illustrated in FIG. 3A, at the output of a decoder DEC having previously reconstructed the images IV1 to IVN, or else be part of such a DEC decoder, as illustrated in FIG. 3B.
- the actions executed by the decoder DEC are implemented by computer program instructions.
- the decoder DEC has the conventional architecture of a computer and notably comprises a MEM_D memory, a UT_D processing unit, equipped for example with a PROC_D processor, and controlled by a computer program PG_D stored in memory MEM_D.
- the computer program PG_D comprises instructions for implementing the actions of the decoding method as described below, when the program is executed by the processor PROC_D.
- the code instructions of the computer program PG_D are for example loaded into a RAM memory (not shown) before being executed by the processor PROC_D.
- the processor PROC_D of the processing unit UT_D notably implements the actions of the decoding method described below, according to the instructions of the computer program PG_D.
- the SYNT synthesis device comprises the following three modules mounted in cascade and controlled by the PROC_S processor:
- an M3 synthesized image analysis module configured to communicate with the M1 module.
- the synthesis device SYNT according to the invention is thus advantageously configured to carry out an image synthesis, in which the estimation of synthesis data is carried out jointly and interactively with the calculation of an image of a synthesized view, the synthesis being iterated as needed to modify / complete / refine the synthesized image obtained, thanks to the communication existing between the module M3 and the module M1.
- such a synthesis method comprises an image processing phase PO which is applied to an image of a view which is part of the plurality of views Vi, ..., VN of a multi-view video, the plurality of views representing a 3D scene respectively according to a plurality of viewing angles or a plurality of positions / orientations.
- Such images are images which have been reconstructed by the decoder DEC of FIGS. 3A and 3B, prior to the implementation of the iterative synthesis method.
- the PO image processing phase includes the following.
- DSo image synthesis data is generated by the module M1 of Figure 2, from texture data of at least one of the images IV1 to IVN.
- This is for example DSo synthesis data generated from texture data of the IVi image.
- this texture data Ti corresponds to only certain pixels of the texture component CTi of the image IVi previously reconstructed.
- a DSo synthesis data map generated for these certain pixels is obtained. Such a card is shown in Figure 4A.
- the generation of the DSo image synthesis data amounts to:
- DSCi image synthesis data which have been previously generated at the encoder from original texture data of the component of CTi texture of the IVi image, then encoded, and
- the generation of the image synthesis data DSo amounts to calculating these image synthesis data DSo from texture data of the image IVi reconstructed beforehand.
- the synthesis data DSo are generated from texture data Ti which corresponds to only certain pixels of the texture component CTi of the image IVi previously reconstructed.
- the calculation resources for the implementation of the step of generating summary data are therefore particularly lightened compared to the depth estimates implemented. in the prior art which consider all the pixels of an image.
- the generated synthesis data DSo is depth data associated with the texture data Ti.
- other types of DSo summary data can of course be considered.
- Examples of DSo summary data can be, in a non-exhaustive way:
- the structural information contained in the contour maps, the angles and the active contours can be used by the iterative image synthesis method according to the invention, for example to avoid ghosting artifacts. This can be achieved by improving the synthesis algorithm or by improving the depth maps.
- Contour detection approaches can include the use of operators such as Sobel, Canny, Prewitt or Roberts.
- the angles can be estimated using a Harris type angle detection operator;
- the SURF algorithm is an extension of the SIFT algorithm, replacing the Gaussian filter in SIFT by an average filter;
- Such statistical characteristics are for example representative of the percentage of texture, or of contour, in given areas of the image of a view. They are used by the SYNT synthesis device in order to synthesize an image with a percentage of texture or contour which approaches it. The same is true for histogram data relating to different areas of the image of a view, the synthesis device SYNT trying to keep this data. histograms for each of these zones for the image of the view synthesized by the latter.
- Machine learning methods such as Convolutional Neural Networks (CNN) can also be used to extract features from reconstructed texture data. useful for image synthesis.
- CNN Convolutional Neural Networks
- DSo data generated at the end of step S1o and more generally the data generated at each iteration.
- Such an expression covers not only partial or partial depth maps, but also the aforementioned data. Other types of data not mentioned here are also possible.
- the synthesis data DSo can also be generated from at least one item of information MD associated with the sequence of multi-view images IVi to IVN and / or d one or more CPi to CPN depth maps available, associated respectively with the texture components CTi to CTN of the images IVi to IVN.
- MD information is, for example, metadata representative of a difference or of a correlation between the views Vi to VN, such as one or more camera parameters, one or more occlusion maps, differences in decoding modes, etc. .
- Such MD information is for example transmitted in a stream received by the iterative synthesis device SYNT illustrated in FIG. 2.
- the MD information may also be already available at the level of the synthesis device.
- the depth estimation is implemented for example using a monocular neural network, d 'a contour based depth estimation algorithm or T-junctions, etc .; ;
- the depth estimation is implemented for example using a stereoscopic matching algorithm
- the DSo synthesis data are associated with an uncertainty value on the correlation of said synthesis data with the corresponding texture data of the IVi image.
- these uncertainty values are grouped together in a map such as for example:
- an OM occupancy map which is a binary map indicating, for the pixels considered in the texture component CTi, whether the corresponding depth values are correlated (value set to 1) or not ( value set to 0) to the values of these pixels;
- CM confidence map
- the generation of the summary data is implemented from a threshold which is for example a distortion value SAD (“Sum of Absolute Differences”) which is representative of the quality of the setting.
- SAD Sud of Absolute Differences
- stereoscopic correspondence for example or even more generally on the basis of any other indicator making it possible to establish a degree of confidence in each pixel, for the synthesis data item to be generated for this pixel.
- the summary data map is only filled for the pixels whose confidence is greater than a given threshold.
- the summary data map is filled for all the pixels of the image, but is accompanied either by a confidence map at each point, or by the value SAD which must be compared to the threshold , at each point.
- the summary data map is incomplete, but is nevertheless accompanied either by a confidence map at each existing point, or by the value SAD which must be compared with the threshold, at each point. existing point.
- the iterative synthesis method according to the invention continues with a calculation S2o by the calculation module M2 of FIG. 2 of an image of a synthesized view ISo, from the synthesis data DSo generated in S1o and at minus one image IVj among the images IVi to IVN.
- the image IVj can be the image IVi or another image among N.
- the S2o calculation is implemented by a conventional synthesis algorithm, such as for example of the RVS type (“Reference View Synthesizer” in English), VVS (“Versatile View Synthesizer” in English), etc.
- An example ISo synthesized image is shown in Figure 4A.
- the ISo synthesized image is analyzed against a CS synthesis performance criterion, using the M3 analysis module in Figure 2.
- - CS2 size of the largest area to be filled in the ISo synthesized image below a threshold, such a criterion signifying that an inpainting algorithm will succeed in reconstructing the missing data of this image efficiently;
- the ISo synthesized image is considered valid and is delivered to be stored waiting to be displayed by a terminal of a user who has requested this image and / or directly displayed by the terminal.
- the analysis module M3 transmits to the module M1 information INFo on the pixels / areas of pixels of the synthesized image ISo which have not been synthesized or which were not considered to have been correctly synthesized at the end of the PO image processing phase.
- INFo information is contained in a dedicated message or request. More specifically, the information INFo is represented in the form of the coordinates of these pixels not or poorly synthesized according to the aforementioned criteria or even in the form of a percentage of non-synthesized pixels.
- the M3 analysis module also transmits in S4o the OMi occupancy card or the CMi trust card generated in S1o.
- the SYNT synthesis device then triggers a new image processing phase P1, in order to improve / refine the ISo synthesized image which was obtained previously.
- the synthesis data generation module M1 generates at again synthetic data DSi from texture data (pixels) of at least one of the images IVi to IVN. It could be for example:
- the synthesis data DSi can also be generated from at least one item of information MD associated with the sequence of multi-view images IVi to IVN and / or from one or more depth maps CPi to CPN available. , associated respectively with the texture components CTi to CTN of the images IVi to IVN. It can also be depth data, partial depth maps, contour maps, etc. used individually or in combination, the choice of the type of this data is not necessarily the same as that used in S1o.
- step S11 there is obtained a synthesis data map DSi generated for these texture data.
- a synthesis data map DSi generated for these texture data.
- the selection of the image (s) to generate the synthesis data DSi can be implemented in different ways.
- the synthesis data generation module M1 can be configured to select in S11 the two texture components CT3 and C ⁇ 4, respectively. closest images IV3 and IV4 to images IV1 and IV2.
- the module M1 can be configured to use in S11 the texture components CTi and CT3 or else CT2 and CT3 to generate the DS-i synthesis data.
- the module M1 can simultaneously use several texture components, for example four texture components CT3, CI4, CTs, CTe.
- the module M1 can use, in addition to or instead of the texture components CTi to CTN, one or more images previously synthesized (s). For example, in S11, the module M1 can use the synthesized image ISo to generate the synthesis data DSi.
- the choice in S11 of the texture data and / or P and / or Tk is conditioned by the information INFo. If, for example, the information INFo is representative of the location of one or more occlusion zones in the synthesized image ISo, the module M1 is configured to select in S11 one or more texture components of images more distant than that (s) used in S1o to fill these areas. Such a selection can be guided by the information MD corresponding for example to camera parameters, such as for example the angle of the camera and its position in the scene.
- DSi summary data is generated in a similar manner to DSo data.
- the synthesis data DSi are generated by modifying all or part of the synthesis data DSo with the texture data Ti and / or T′i and / or Tk.
- intermediate synthesis data DS int i is generated in a similar manner to synthesis data DSo. This intermediate summary data is then combined / merged with the DSo summary data, providing the DS-i summary data.
- the summary data missing in a summary data card for example the card containing the summary data DS int i, are recovered from the existing summary data DSo of the other card, or else Conversely.
- the value of SAD can be calculated by a stereoscopic matching algorithm which performs a match between the image IVi or IVk used in S1i to generate the synthesis data and another image among N.
- the vectors found to match these two images come from a motion estimate which returns a level of correlation SAD which is considered as an example of a degree of confidence.
- these DSo synthesis data are replaced by these DSi synthesis data, so as to keep the information of a depth plane first (in this example, the foreground of the scene).
- the image processing phase P1 continues with a calculation S2i, by the calculation module M2 of FIG. 2, of a new image of a synthesized view IS-i.
- calculation S2i is implemented by a conventional synthesis algorithm, such as for example RVS or VVS.
- An example of an ISi synthesized image is shown in Figure 4B.
- the calculation module M2 modifies the synthesized image ISo calculated in S2o, at least from the synthesis data DSi generated in S11. At least one image among the IVi to IVN images can also be used when making this change.
- the calculation module M2 calculates an intermediate synthesized image IS int i from the synthesis data DSi obtained in S1i and optionally from one or more images IVi to IVN, then combines the intermediate synthesized image IS int i with the synthesized image ISo previously calculated in S2o to generate a synthesized image IS-i.
- the synthesized image ISi is obtained by recovering the missing pixels in a synthesized image among the images IS int i and ISo, for example the image IS int i, among the pixels existing in the image ISo .
- the synthetic image ISi is obtained by selecting from among the images IS int i and ISo, the common synthesized pixel which is associated with the highest degree of confidence using the aforementioned confidence map CM.
- the synthetic image ISi is obtained by selecting from among the images IS int i and ISo, the common synthesized pixel which is associated with a priority depth plane, the foreground of the scene for example.
- the image processing phase P1 continues with an analysis, in S3i, of the synthesized image ISi against a synthesis performance criterion CS, by means of the analysis module M3 in FIG. 2.
- the performance criteria used in the image processing phase P1 may be the same as the criteria CS1 to CS3 above. Two other following criteria can also be considered in a non-exhaustive way:
- the value of certain synthesized pixels in the ISi synthesized image is equal or substantially equal to that of certain synthesized pixels of the same respective position in the ISo synthesized image, resulting in two similar synthesized images.
- the synthesized image ISi is considered valid and is delivered to be stored while waiting to be displayed by a terminal of a user who has requested this image and / or directly displayed by the terminal.
- the analysis module M3 transmits in S4i to the module M1 information INF1. If the S3i analysis of the synthesized image IS1 uses the synthesis performance criterion CS4, the information INF1 contains the coordinates of the pixels of the zone (s) of the synthesized image IS1 considered to lack temporal consistency with the pixels of corresponding position in the synthesized image ISo. If the S3i analysis of the synthesized image IS1 uses the synthesis performance criterion CS5, the information IN Fi contains the coordinates of the pixels of the zone (s) of the synthesized image IS1 considered to lack pixel coherence with the pixels of corresponding position in the ISo synthesized image.
- the SYNT synthesis device then triggers a new image processing phase P2 similar to the image processing phase P1, in order to further improve / refine / complete the IS1 synthesized image that was obtained previously.
- an image processing phase can be iterated as long as one or more synthesis performance criteria are not met.
- a current image processing phase Pc takes place as follows:
- the synthesis data generation module M1 again generates DSc synthesis data from texture data (pixels) of at least one of the images IV1 to IVN. It could be for example:
- T’i texture data Tk already used in the previous summary data generation steps, and / or
- the synthesis data DSc can also be generated from at least one item of information MD associated with the sequence of multi-view images IV1 to IVN and / or from one or more depth maps CPi to CPN available, associated respectively with the texture components CTi to CTN of images IV1 to IVN. It can also be depth data, partial depth maps, contour maps, etc. used individually or in combination, the choice of the type of this data not necessarily being the same as that used in the preceding summary data generation steps S1 o, S11, etc.
- the selection of the image (s) to generate the synthesis data DSc is carried out in the same manner as in the aforementioned embodiments.
- the synthesis data generation module M1 can be configured to select in SIC one or more texture components of the images IVi to IVN which, depending on the result of the analysis of the synthesized images carried out in the processing phases of previous image, can be closer or further from the images already used in the previous image processing phases.
- the module M1 can use, in addition to or instead of the texture components CTi to CTN, one or more images ISo, ISi, ... already synthesized during the preceding processing phases.
- the choice in S1 c of the texture data can be conditioned by the information INF c -i, this choice can also be supplemented by the information INFo, INF-i, ... generated at the end steps of calculating synthetic images ISo, ISi, ... respectively.
- the selection of the new texture data will be based on the criteria of the table Table2 mentioned above, considering the image l c -i synthesized during the previous image processing phase Pc-1, with regard to at least one of the other ISo, ISi, ... images already synthesized during the previous processing phases.
- the synthesis data DSc can be generated in Sic by modifying, using the texture data Ti and / or T'i and / or Tk and / or Tm used at the input of the module M1, all or part of the synthesis data DSc-i generated in the previous image processing phase P c -i, but also the synthesis data DSo, DS-i, ... before the phase P c - 1.
- the generation Sic of the synthesis data may be implemented from a threshold which is for example a distortion value SAD .
- the SAD distortion value can be the same at each iteration. However, such an SAD value may vary.
- the required SAD value is for example as low as possible and is increased in the following processing phases, for example from the processing phase current image Pc.
- the SAD value can be gradually increased at each iteration of an image processing phase.
- the current image processing phase Pc continues with a calculation S2 C , by the calculation module M2 of FIG. 2, of a new image of a synthetic view ISc.
- the calculation S2 C is implemented by a conventional synthesis algorithm, such as for example RVS or VVS.
- the calculation module M2 modifies the synthesized image ISc-i obtained during the previous image processing phase P c -i, at less from the DSc summary data generated in Sic. At least one image among the images IVi to IVN can also be used during this modification.
- the calculation module M2 calculates an intermediate synthesized image IS int c from the synthesis data DSc obtained in Sic and optionally from one or more images IVi to IVN, then combines the intermediate synthesized image IS int c with the synthesized image ISc-i calculated previously in the previous image processing phase Pc-1 and possibly with one or more ISo images,
- the Pc image processing phase continues with an analysis, in S3c, of the synthesized image ISc against a CS synthesis performance criterion, using the M3 analysis module in Figure 2.
- the performance criteria used in the image processing phase Pc are the aforementioned criteria CS1 to CS5. If one or more of these criteria is / are met, the synthesized image ISc is considered valid and is delivered to be stored waiting to be displayed by a terminal of a user who has requested this image and / or directly displayed by the terminal. If one or more of these criteria is / are not met, the analysis module M3 transmits in S4 C to the module M1 information INF C. A new image processing phase Pc + 1 is then triggered.
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Multimedia (AREA)
- Signal Processing (AREA)
- General Physics & Mathematics (AREA)
- Computer Graphics (AREA)
- Computing Systems (AREA)
- Geometry (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Image Processing (AREA)
Abstract
Description
Claims
Applications Claiming Priority (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| FR2000015A FR3106014A1 (fr) | 2020-01-02 | 2020-01-02 | Synthèse itérative de vues à partir de données d’une vidéo multi-vues |
| PCT/FR2020/052454 WO2021136895A1 (fr) | 2020-01-02 | 2020-12-15 | Synthese iterative de vues a partir de donnees d'une video multi-vues |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| EP4085613A1 true EP4085613A1 (fr) | 2022-11-09 |
Family
ID=70456900
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| EP20845681.4A Pending EP4085613A1 (fr) | 2020-01-02 | 2020-12-15 | Synthese iterative de vues a partir de donnees d'une video multi-vues |
Country Status (4)
| Country | Link |
|---|---|
| US (1) | US12062129B2 (fr) |
| EP (1) | EP4085613A1 (fr) |
| FR (1) | FR3106014A1 (fr) |
| WO (1) | WO2021136895A1 (fr) |
Family Cites Families (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7408986B2 (en) * | 2003-06-13 | 2008-08-05 | Microsoft Corporation | Increasing motion smoothness using frame interpolation with motion analysis |
| US8144161B2 (en) * | 2006-12-08 | 2012-03-27 | Fraunhofer-Gesellschaft Zur Foerderung Der Angewandten Forschung E.V. | Texture synthesis |
| WO2013068457A1 (fr) * | 2011-11-11 | 2013-05-16 | Fraunhofer-Gesellschaft zur Förderung der angewandten Forschung e.V. | Concept pour déterminer une mesure d'un changement de distorsion dans une vue synthétisée dû à des modifications de carte de profondeur |
| WO2013068491A1 (fr) * | 2011-11-11 | 2013-05-16 | Fraunhofer-Gesellschaft zur Förderung der angewandten Forschung e.V. | Codage multivue avec exploitation de sections pouvant être rendues |
| EP2777256B1 (fr) * | 2011-11-11 | 2017-03-29 | GE Video Compression, LLC | Codage multivue avec gestion effective de sections pouvant être rendues |
| US20130271565A1 (en) * | 2012-04-16 | 2013-10-17 | Qualcomm Incorporated | View synthesis based on asymmetric texture and depth resolutions |
| US9571809B2 (en) * | 2013-04-12 | 2017-02-14 | Intel Corporation | Simplified depth coding with modified intra-coding for 3D video coding |
| GB201313680D0 (en) * | 2013-07-31 | 2014-01-08 | Mbda Uk Ltd | Image processing |
| US20160050440A1 (en) * | 2014-08-15 | 2016-02-18 | Ying Liu | Low-complexity depth map encoder with quad-tree partitioned compressed sensing |
| CN108475330B (zh) * | 2015-11-09 | 2022-04-08 | 港大科桥有限公司 | 用于有伪像感知的视图合成的辅助数据 |
| US9710898B2 (en) * | 2015-11-18 | 2017-07-18 | Adobe Systems Incorporated | Image synthesis utilizing an active mask |
| US10165258B2 (en) * | 2016-04-06 | 2018-12-25 | Facebook, Inc. | Efficient determination of optical flow between images |
| US10217265B2 (en) * | 2016-07-07 | 2019-02-26 | Disney Enterprises, Inc. | Methods and systems of generating a parametric eye model |
| EP3649618A1 (fr) * | 2017-07-03 | 2020-05-13 | Artomatix Ltd. | Systèmes et procédés de fourniture de synthèse de texture non paramétrique de forme arbitraire et/ou de données de matériau dans un cadre unifié |
-
2020
- 2020-01-02 FR FR2000015A patent/FR3106014A1/fr not_active Ceased
- 2020-12-15 EP EP20845681.4A patent/EP4085613A1/fr active Pending
- 2020-12-15 US US17/790,301 patent/US12062129B2/en active Active
- 2020-12-15 WO PCT/FR2020/052454 patent/WO2021136895A1/fr not_active Ceased
Also Published As
| Publication number | Publication date |
|---|---|
| WO2021136895A1 (fr) | 2021-07-08 |
| US20230053005A1 (en) | 2023-02-16 |
| FR3106014A1 (fr) | 2021-07-09 |
| US12062129B2 (en) | 2024-08-13 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| EP3878170B1 (fr) | Synthèse de vues | |
| WO2010007285A1 (fr) | Procede et dispositif de remplissage des zones d'occultation d'une carte de profondeur ou de disparites estimee a partir d'au moins deux images | |
| WO2019211541A2 (fr) | Procédé et dispositif de décodage d'une vidéo multi-vue, et procédé et dispositif de traitement d'images | |
| EP1473944A2 (fr) | Procédé de tatouage d'une séquence vidéo à sélection adaptative de la zone d'insertion du tatouage, procédé de détection, dispositifs, support de données et programmes d'ordinateur correspondants. | |
| EP3939304B1 (fr) | Procédés et dispositifs de codage et de décodage d'une séquence vidéo multi-vues | |
| WO2010043809A1 (fr) | Prediction d'une image par compensation en mouvement en avant | |
| EP3449634A1 (fr) | Procédé de composition contextuelle d'une représentation vidéo intermédiaire | |
| FR2959636A1 (fr) | Procede d'acces a une partie spatio-temporelle d'une sequence video d'images | |
| WO2020260034A1 (fr) | Procede et dispositif de traitement de donnees de video multi-vues | |
| EP4140136A1 (fr) | Procédés et dispositifs de codage et de décodage d'une séquence vidéo multi-vues | |
| WO2019115899A1 (fr) | Procédés et dispositifs de codage et de décodage d'une séquence vidéo multi-vues représentative d'une vidéo omnidirectionnelle | |
| EP4085613A1 (fr) | Synthese iterative de vues a partir de donnees d'une video multi-vues | |
| WO2020070409A1 (fr) | Codage et décodage d'une vidéo omnidirectionnelle | |
| FR2856548A1 (fr) | Procede de representation d'une sequence d'images par modeles 3d, signal et dispositifs correspondants | |
| EP4360319A1 (fr) | Procédé de construction d'une image de profondeur d'une vidéo multi-vues, procédé de décodage d'un flux de données représentatif d'une vidéo multi-vues, procédé de codage, dispositifs, système, équipement terminal, signal et programmes d'ordinateur correspondants | |
| EP3158749B1 (fr) | Procédé de codage et de décodage d'images, dispositif de codage et de décodage d'images et programmes d'ordinateur correspondants | |
| FR2934453A1 (fr) | Procede et dispositif de masquage d'erreurs | |
| EP4104446A1 (fr) | Procédé et dispositif de traitement de données de vidéo multi-vues | |
| WO2022069809A1 (fr) | Codage et decodage d'une video multi-vues | |
| EP3649786A1 (fr) | Procédé de codage et décodage d'images, dispositif de codage et décodage et programmes d'ordinateur correspondants | |
| EP4544779A1 (fr) | Procede de segmentation d'une pluralite de donnees, procede de codage, procede de decodage, dispositifs, systemes et programme d'ordinateur correspondants | |
| FR3036016A1 (fr) | Codage/decodage predictif d'images comportant un artefact, utilisant une extrapolation | |
| Dumont et al. | Plane Sweeping in Eye-Gaze Corrected, Teleimmersive 3D Videoconferencing | |
| FR2873213A1 (fr) | Procede et dispositif d'obtention d'un signal stereoscopique |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| STAA | Information on the status of an ep patent application or granted ep patent |
Free format text: STATUS: UNKNOWN |
|
| STAA | Information on the status of an ep patent application or granted ep patent |
Free format text: STATUS: THE INTERNATIONAL PUBLICATION HAS BEEN MADE |
|
| 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 |
|
| STAA | Information on the status of an ep patent application or granted ep patent |
Free format text: STATUS: REQUEST FOR EXAMINATION WAS MADE |
|
| 17P | Request for examination filed |
Effective date: 20220712 |
|
| AK | Designated contracting states |
Kind code of ref document: A1 Designated state(s): AL AT BE BG CH CY CZ DE DK EE ES FI FR GB GR HR HU IE IS IT LI LT LU LV MC MK MT NL NO PL PT RO RS SE SI SK SM TR |
|
| DAV | Request for validation of the european patent (deleted) | ||
| DAX | Request for extension of the european patent (deleted) | ||
| RAP3 | Party data changed (applicant data changed or rights of an application transferred) |
Owner name: ORANGE |
|
| STAA | Information on the status of an ep patent application or granted ep patent |
Free format text: STATUS: EXAMINATION IS IN PROGRESS |
|
| 17Q | First examination report despatched |
Effective date: 20240830 |