Concept of multi component image or video coding

Abstract

To provide a decoder and a method for enhancing a coding efficiency of a multi component image.SOLUTION: In a decoder 10, a coding efficiency of a multi component image or a video coding concept is improved by reconstructing a third component signal 30 in regard to a module 28 for a third component of a multi component video using an interval component prediction from a first component signal 22 reconstructed from a data stream 12 and a second component signal 26 reconstructed.SELECTED DRAWING: Figure 1

Description

[Technical field]

[0001] The present invention relates to multi-component image coding, e.g., color image or video coding. Or video coding.

[0002] Colored images and videos are expressed in three or four aspects, so-called color spaces. The three components are also called components. An example of a color space is the R'G'B' color space. In this color space, the primary colors red (R'), green ( Blue (G'), and Blue (B') form the basis of a three-dimensional color cube. However, signal processing applications In the application space, image and video compression is often driven by power consumption, implementation cost, or compression cost. Minimizing the correlation between different components to achieve high efficiency in terms of efficiency is It is important to note that the R'G'B' signal is therefore divided into two components: the first component called luminance (Y ') and the remaining two components called chroma elements (C b C r ) in Y'C b C r For a while In contrast to the R'G'B' color space, the Y'C b C r The chroma component of the primary blue or red. Therefore, chroma energy is often In this case, the conversion cost is reduced, thus resulting in a higher compression efficiency. However, a certain conversion, i.e. Y'C b C r Due to the application of R'G'B' to R' conversion, the resulting output is often The possibility of overcoming such limitations is limited by the color components. Such predictions are often applied to local predictions between components. It is called Inter-Color Prediction (ICP). ICP is a function of R'G'B' and Y'C b C r Apply to both signals In the first case, the ICP reduces the energy due to the chroma element. and therefore it can be treated as an alternative to external color space conversion. In the second case, the ICP approach introduces further decorrelation steps between the different color components. The ICP approach can therefore be treated as a This results in high compression efficiency. For simplicity, we use the remaining The description of the first component and the main component and the first chroma or second chroma element are shown. For each of these, b or C r to indicate the two remaining chroma elements. or Y'. Processing instructions may occur sequentially for some applications. It is important to note that the order of chroma components can be important. be. Summary of the Invention [Problem to be solved by the invention]

[0003] However, for example, it is not possible to predict one chroma element based on the other chroma elements. By doing the same for the metric elements, ICP is applied in pairs, if at all. However, it would be advantageous to be able to increase the coding efficiency even further.

[0004] The object of the present invention is to provide a method for coding multi-component images or video signals that allows for higher coding efficiency. The objective of this paper is to provide a concept for video encoding.

[0005] This object is achieved by the subject matter of the independent claims of the present application. [Means for solving the problem]

[0006] The concept of multi-component image coding efficiency or video coding has been reconstructed. The component is extracted from the reconstructed first component signal and the reconstructed second component signal. Inter-component prediction is used to estimate the third component of a multi-component video. The present invention relates to a method for reconstructing the component signals of the This is a fundamental discovery.

[0007] According to an embodiment of the present invention, different or multiple ICP encoding modes can be used. Within the ICP mode portion of a component image or video data stream The transmission of multiple source ICP signals is used to switch between ICP encoding modes. , including a plurality of source ICP coding modes and at least one non-ICP coding mode. ICP mode part of a multi-component image or multiple source ICP coding mode In a video where For this purpose, the above is a spatially corresponding portion of the reconstructed first component signal or Prediction from the spatially corresponding parts of the constructed second component signal, or a combination of both The signal transmission indicates for each subpart whether there is inter-component prediction to be performed. In at least one non-ICP coding mode, The spatially corresponding portions of the reconstructed first component signal and the reconstructed second component signal are This kind of combination or alteration between spatially corresponding portions of the two component signals is Not available. ICP is fully supported in ICP mode parts of non-ICP coding modes. The ICP may simply switch off the first and second component signals. The ICP mode section may be used for a single image. , a single image sequence or a single slice or consecutively coded images, an image sequence It may be a can or a slice. Involved in parameterizing the sub-picture granularity of multi-source ICP according to an embodiment of the present invention At least one non-ICP coding mode that can transmit any additional signal. Overheads and coding modes Multiple source ICP juxtaposition and non-ICP coding In the case of multiple source ICPs, the coding is performed in relation to the encoding mode. You can restrict the number of ICP encoding modes to at least one non-ICP encoding mode and multiple source ICP encoding modes. By juxtaposing the coding modes, any additional signaling overhead is now eliminated. Involved in parameterizing the sub-picture granularity of multi-source ICP according to an embodiment of the invention and it can be used to encode multiple source ICP codes associated with non-ICP coding modes. This additional signaling overhead can be increased in the case of the multi-mode The code obtained by multiple component sources can be used for inter-component prediction. The increase in coding efficiency may be limited to the portion of the video that is overcompensated.

[0008] According to an embodiment, the first component is luminance and the second component is the first The first component is the chroma component of the first pixel, and the third component is the chroma component of the second pixel. In this case, the combination of the second chroma component based on the luminance component and the further first chroma component is The inter-component prediction is coded, for example, using transform and entropy coding. Improve the decorrelation of the residuals.

[0009] According to an embodiment, the explicit ICP source signaling in the data stream is reconstructed. Using inter-component prediction from the spatially corresponding parts of the first component signal, The third component signal in the current image of a multi-component image or video and reconstructing the spatially corresponding second component signal. Multi-component image or video using inter-component prediction from the corresponding parts During reconstruction of the third component signal in the current image, sub-picture granularity Despite the overhead of sending signals, this method The available prediction improvements therefore overcompensate for the signal transmission overhead. This leads to higher coding efficiency. Alternatively, implicit signaling can be used. Regardless of the type of signal transmission used, the sub-picture granularity is The reconstruction of the three component signals allows switching between spatial, temporal, and / or visual The third component signal may correspond to the granularity of switching between inter-point prediction modes. Reconstructing the number of events to allow for spatial, temporal and / or inter-predictive modes of interaction switching and operating ICP and second component based on the first component signal The switching between signals can be done on a prediction block basis or in another way. This can be done using a coding unit. Alternatively, using a transform block. can be done.

[0010] ICP can be implemented using linear prediction, where the IC of the data stream The transmission of the signal of the P prediction parameters is performed by the reconstructed first component signal and the reconstructed The second component signal linearly participates in inter-component prediction at subpicture granularity. It is possible to use two weightings to change the weights of the two The weighted sum of both is used to obtain the inter-component prediction, with sub-picture granularity. in the data stream, one of which is the reconstructed Weighting for the first component signal and the reconstructed second component signal One transmits the signaling of the other weights. Regardless, linear prediction is used to improve the ICP prediction parameters while still achieving sufficient decorrelation. Meter signaling overhead can be kept at a reasonable cost. The prediction block unit is used to find the signaling or block in the data stream. The values ​​available for weighting may be distributed around zero. It may also contain zeros.

[0011] According to an embodiment, the ICP prediction parameter signaling outlined immediately above is This measure can include the transmission of conditional signals of the ICP prediction parameters. The signaling overhead associated with transmitting data signals is further reduced.

[0012] The sign of the weights is then coded using context modeling according to the sign. The signal of the ICP prediction parameters is first coded with the absolute value of the weighting contains in a sense an encoding of the weights. With this treatment, if the absolute value is The possibility of using a certain entropy for encoding / decoding is set depending on the code. and thereby improve the accuracy of the estimation due to a closer match of the probabilities used. It all comes down to compression ratio.

[0013] According to an embodiment of the present invention, the inter-component prediction is a third method of the inter-component method. Spatial, temporal and / or inter-view predictions used in reconstructing the component signals In other words, inter-component prediction is applied to the prediction of the third component. A kind of third-order prediction compared to spatial, temporal and / or inter-component prediction of the input signal. It represents a two-stage prediction. Similarly, the reconstructed first component signal and the reconstructed The second component signal is the first and second components of a multi-component image or video signal. and a second component for spatial, temporal and / or can represent the prediction residual of inter-view prediction.

[0014] Inter-component prediction can be performed in the spatial or spectral domain.

[0015] According to another embodiment, the transmission of an explicit ICP source signal in the data stream is It is used to switch between available ICP sources with subpicture granularity of 1. On the other hand, ICP is a predictive model, which is implemented with respect to the signaling ICP source. Signaling of ICP prediction parameters in the data stream is used to adjust , where the entropy coding / decoding of the signal transmission of the explicit ICP source is The prediction performance of the prediction model is adjusted according to the transmission of the ICP prediction parameter signal at the second granularity. This measurement includes the use of a context model that depends on both The correlation between the transmission of signals, i.e. signals ICP source and signals ICP prediction parameters To reduce the signaling overhead by transmitting It can be used for.

[0016] According to a further embodiment, the reconstructed third component signal and the reconstructed first component signal are Swap signals in the data stream to swap two component signals For example, within an image, one sub-block may transmit the second chrominance. Available luminance and first chrominance components for inter-component prediction For example, within one image, the transmission of the swap signal is Varying the order at subpicture granularity in the second and third components. The other sub-blocks are used to predict the first chroma component. It has a luminance component and a second chrominance component.

[0017] According to an embodiment, encoding of an ICP source indicator for a particular block of an image of ICP prediction parameters for the third and second components For the transmission of the signal, the difference between the predicted parameters of the previously coded ICP is used. If the difference exceeds a certain limit, the ICP source indicator The ICP source indicator is not present for each block. The second component reconfigured as an ICP source for the three component signal In other words, the signal sent for the current block The predicted parameters of ICP are as follows: If the components are similar enough, the component signals of all the components will be ICP Source Indicator and Highly Indicative of Criteria for ICP Components It can be assumed that the ICP prediction parameters are similar. , use the first component as the basis instead of the ICP for the second component. The second component is based on ICP while the third component is based on ICP. The first case is more likely to occur in the case of RGB color components. and the second, more likely case occurs for YCC color components. This measurement shows that the coding efficiency is improved by adding side information for ICP signaling. It can be increased by lowering the head.

[0018] Advantageous implementations of embodiments of the invention are the subject matter of the dependent claims. Preferred embodiments of the invention are described below with reference to the drawings. [Brief description of the drawings]

[0019] [Figure 1] FIG. 1 shows a block diagram of a decoder according to an embodiment of the present application. [Diagram 2] FIG. 2 shows a block diagram of a decoder according to an embodiment of the present application, applying ICP to the inter-component prediction residual. [Diagram 3] FIG. 3 shows a block diagram of an encoder compatible with the decoder of FIG. 1 according to an embodiment. [Figure 4] FIG. 4 shows a block diagram of an encoder compatible with the decoder of FIG. 2 according to an embodiment. [Figure 5A] FIG. 5A shows a schematic diagram of an embodiment in which the transmission of a signal of a multiple source ICP can switch the multiple source ICP on / off. [Figure 5B] FIG. 5B illustrates a block diagram of an embodiment of a multi-source ICP module according to an embodiment in which the ICP signaling includes an ICP prediction mode parameter in the form of weightings and an ICP source indicator. [Figure 5C] FIG. 5C illustrates a block diagram of an embodiment of a multi-source ICP module according to an embodiment in which the ICP signaling includes an ICP prediction mode parameter in the form of a weighting per available ICP source. [Figure 5D] FIG. 5D shows a block diagram of an embodiment for a multi-source ICP module, with multi-source ICP being performed in the spatial domain according to FIGS. 5B and 5C. [Figure 5E] FIG. 5E shows a block diagram of an embodiment for a multi-source ICP module, with multi-source ICP being performed in the spatial domain according to FIGS. 5B and 5C. [Figure 5F] FIG. 5F shows a block diagram of an embodiment for a multi-source ICP module, with multi-source ICP being performed in the spatial domain according to FIGS. 5B and 5C. [Figure 5G] FIG. 5G shows a block diagram of an embodiment for a multi-source ICP module, with multi-source ICP being performed in the spatial domain according to FIGS. 5B and 5C. [Figure 6] FIG. 6 shows a flow diagram of processing of ICP parameter data in a decoder and encoder according to an embodiment in which the ICP parameter data for the third component includes weightings and source flags. [Figure 7]FIG. 7 shows a flow diagram of processing of ICP parameter data for a third component according to an embodiment with weighting of a weighted sum of ICP sources for which ICP parameter data is available. [Figure 8] FIG. 8 shows a block diagram illustrating the processing of ICP parameter data for the second and third components when using a swap component to improve ICP efficiency. [Figure 9] FIG. 9 shows a schematic diagram of an image to illustrate the possibility of signaling source flags and weightings with different granularity. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0020] FIG. 1 shows a block diagram of a decoder according to an embodiment of the present application. The different components from the spatial data stream are shown with reference numeral 10. The present invention is configured to decode multi-component video that samples a scene. In particular, in the decoder 10, a data stream 12 is The input is a multi-component video signal that is reconstructed by decoding. FIG. 1 shows an example in which the time axis extends horizontally and the other axis extends vertically. By illustrating the images 18 belonging to one time, as shown in the figure above, Thus, for each time instant, one image 18 is included for each of the three components A, B, and C. However, the IEEE 802.11b standard does not include multi-component video. It should be noted that there may be several components, as outlined above and further below. Thus, components A, B and C are R, G and B or one luminance and two color components. Alternatively, however, the components may be color components such as Roma components. As components related to other wavelengths that exist outside the visible wavelength range, for example: Components may also relate to other characteristics of the scene. The components are not limited to frequencies or fractions of frequencies within a , at least in part due to other physical parameters, such as for example reflectivity, etc. FIG. 1 shows that the images 18 of the different components are the same size, It is suggested to sample the scene accordingly with the same spatial resolution, but The time resolution may differ between components. The latter situation could in principle also be That is, the image rate can be adjusted to different resolutions for components A, B, C, and D. If not different, the term "image" is instead generally used to refer to a single It can be used to show samples of all components A through C at a time.

[0021] Internally, the decoder 10 derives multi-component video 1 from the data stream 12. for reconstructing the first component signal 22 with respect to the first component A of 6 A first module 20, which also extracts multi-component Reconstructing a second component signal 26 for a second component B of the video 16 a second module 24 for converting the multi-component video 16 into a third component a module 28 for reconstructing a third component signal 30 for the component C; The latter module 28 receives the first component signal and the third component signal. A second component signal 30 is reconstructed from the components 22 and 26. To use inter-bit prediction, a first and a second module 20 and 24 are connected to their outputs. The module 28 has a third component for component C. To reconstruct the signal, the first and second components A and B of the signal 22 and 26 to improve the inter-component prediction used by module 28 This allows the decoder 10 of FIG. This increases the achievable coding efficiency.

[0022] As will be outlined in more detail below, modules 24, 20 and 28 are actually , there exists one such hybrid system decoding branch for each component A, B and C. In FIG. 1, , coding branch 32 A The system includes a module 20 for receiving a first component from a data stream 12. Component A is reconstructed by coding branch 32. B contains module 24 , and configured to reconstruct component B from data stream 12, Gi 32 C includes a module 28 for regenerating a component C from the data stream 12. Such a decoding branch is shown by the dotted line 32. A , 32 B and 32 C To In particular, modules 20 and 24 are illustrated, as outlined in more detail below. and the first and second components are reconstructed and used for inter-component prediction in module 28. The two component signals 22 and 26 may in fact represent prediction residuals. The constructed first component signal 22 is, for example, a branch 32 A To decrypt The prediction residual of the inter-component prediction performed by The computation performed by the module 20 based on the already reconstructed part of the image 18 The prediction residual of inter-component prediction can be expressed as the spatial and temporal Similarly, the second component signal 26 may be a multi-view and / or cross-view predicted signal. Decode branch 32 B The prediction residual for cross-component prediction, which is done by running That is, it can be understood that the already reconstructed image 18 of component B is simply We can express the prediction residual based only on the part of the model that has been trained. The inter-prediction may be spatial, temporal, and / or inter-view prediction. The resulting third component signal 30 is fed to a decoding branch 32. C By carrying out It can mean the prediction residual of inter-component prediction. That is, it can simply mean the It is possible to represent the prediction residuals based only on the already reconstructed parts of the image 18 of the first C. where this inter-component prediction also includes spatial, temporal and / or Inter-view prediction may be used, i.e. spatial, temporal and / or inter-view prediction may be used. Regardless of whether the cross-view prediction is performed in modules 20, 24 and / or 28, the prediction For example, the prediction mode may vary depending on the coding device or In other words, the prediction block, i.e., each coding device or data A sub-portion of an image 18 having a prediction mode used for signaling in the stream 12 The prediction modes are typically set for components A, B, and C, or or, separately, or even more, on the one hand, a first component A and, separately, on the other hand For the second and third components B and C, the signal is added to the data stream 12. The signaling of this inter-component prediction can be performed in the data stream 12. In addition, modules 20, 24 and 28 may be configured to receive component specific signals. Receive prediction residual data. The prediction residual data may be in the form of transform coefficients of a spectral decomposition, or in the spatial domain, in the form of residual samples, and then send them to the data stream in the transform domain. When signals are sent in the spectral domain, inter-component prediction can be performed in the spatial domain. It may also be performed in the temporal and / or spectral domain. The 12 ICP region indicators vary the region over which inter-component prediction is performed. For example, in units of the prediction blocks, or in units of the spectrum The ICP domain indicators are, for example, in units of the transform blocks for which the decomposition is performed. For example, you can vary the regions in an image sequence, in a single image, or with sub-picture granularity. can.

[0023] Module 24 performs inter-component prediction, i.e., the prediction of the output of module 20 to module The first reconstructed component is shown by the dashed branch leading to the input of controller 24. It is also possible to use inter-component prediction based on the component signal 20, as will be described later. That is, the module 24 performs inter-component prediction within the range of component B. The first and second components, i.e., I of A to B, are input to receive the prediction residual for The residual signal sent by the signal can be received as the prediction residual signal of the CP prediction. P prediction is based on the first component signal 22 and is The second component signal 26 used by the module 28 for ICP is Based on the residual signal sent to component B, which is either combined or not yet combined This may be the result of a restructuring of the module 24.

[0024] Also, as outlined in more detail below, the prediction module 28 of the ICP can be parameterized (and optionally by module 24) This parameterization may be, for example, the first and second component signals 22 and 26. Based on this, the first component signal 22, the second component signal 23, and the Regarding the selection among the output signals 26 and their combinations, for example, the third component The first and second component signals 22 and 23 are used to estimate the ICP of a signal or the like. 6 contributes to modify the ICP of module 28 in the form of a weighted sum. In addition, ICP signaling of multiple sources in the data stream is achieved by the ICP in module 28. In addition to the first component signal 22 for Such a multiple source IC can signal availability or unavailability. The scope of P signaling can be a complete video, a single image or time, or an image or time instance. It may also relate to a slice or an array of slices or a sequence of slices.

[0025] How the ICP of module 28 can be determined in data stream 12, respectively How can it be parameterized and changed by sending signals? The various embodiments of which are discussed further below. For ease of understanding, FIG. 2 shows an example of a decoder that illustrates a more specific embodiment of the decoder of FIG. In the embodiment, the constituent signals 22, 26 and 30 are fed to individual decoding branches 32. A ~3 2 C This is about the prediction residual of the inter-component prediction performed in the 3 and 4 show an embodiment of an encoder that is compatible with the embodiment of FIGS. 1 and 2. Briefly presented are various implementations of adjusting the embodiment of Figs. Details of example embodiments are described with respect to subsequent figures.

[0026] FIG. 2 shows modules 20, 24 and 28, as well as inter-component prediction. vessel 34 A , 34 B and 34 C Three decoding branches 32 A ~32 C and, respectively, A combiner 36 that combines the component signals 22, 26 and 30 A , 36 B and 36 C And, predictor 34 A , 34 B and 34 C The output of the inter-component prediction signal by the module 1 and outputs via the control signals 20, 24 and 28. Specifically, the decryption branch 32 A Inter-component predictor 34 A In order , regarding the reconstructed component A, i.e., the component A of the image of the video, The first component signal 22 is reconstructed to provide a sample value. Na36 A The inter-component prediction signal 40 is combined A Components A, i.e. 38 A The inter-component prediction parameters are controlled for the module. As mentioned above, module 20 can perform spatial or First component residual data 42 sent in the spectral domain A By Similarly, inter-component prediction 34 B Is that it? This is then combined into a reconstructed second component signal 26 output by module 24. A second inter-component predicted signal 40 which is in turn combined by the B To determine Then, in data stream 12, component B, i.e., 38 B Inter-component forecast The module 24 is controlled by the measurement parameters. A Locally, like in the form of residual samples of the transformation applied to the spatial The second component can be present in the data stream 12 in the region Residual data 42 B However, according to the embodiment shown in FIG. , module 24 is applied in module 24 based on first component signal 22 In order to change the inter-component predictions to be performed, Wachi, 44 B , for the second component, via the ICP parameter data, This option can be optionally excluded, as described above. Then, the module 24 combines the signal 22 and the data to obtain a second component signal 26. Ta42 B the residual signal from, i.e. the residual signal fed with the signal for component B Then, it is the inter-component prediction for component B. and then, accordingly, obtain a sample of the second component of the image of the video. As you can see, combiner 26 B The second inter-component predicted signal 40 B Combined with do.

[0027] Decryption branch 32 B Similarly, a third component that decodes the decode branch 32C Inter-predictor 34 C performs inter-component prediction to obtain the inter-component predicted signal 40 C The inter-component prediction parameters present in the data stream 12 are used to extract 44 C 3, which in turn is controlled by a third component signal 30. Nbaina 26 C The third component in data stream 12 is Based on the residual data 42C, the third component, i.e., 44 of 12 C For components controlled via parameter data The first and second component signals are predicted by the inter-component prediction of the module 28. As an intra-prediction using inter-component prediction based on Nos. 22 and 26, The filter 28 produces a third component signal 30. Other combinations, such as multiplication, may be possible. The combiner 36 can be used in a similar manner. A ~36 C The combination is illustrated in FIG. This can be implemented additionally, so as to

[0028] As presented above, Figs. 3 and 4 are the same as the corresponding encoders, i.e. Figs. 1 and 2. 3 shows an embodiment of an encoder that is compatible with the decoder shown in FIG. The encoder 100 corresponds to the decoder 10. The encoder 100 is shown in FIG. Then, different components A, B and C, and receives the video 16. The encoder 100 is configured to correspond to FIG. The 00 elements simply have the 100 added to the respective reference numbers of the corresponding elements in the decoder. using reference numbers different from those used for the decoder only in that Therefore, the encoder 100 of FIG. For each of components A, B and C, a respective module, i.e. Module 120 includes data for encoding a first component signal 22 into a stream 12, The code 124 is used to encode the second component signal 26 into the data stream 12. and a module 128 adds a third component signal to the data stream 12. 30, where the module 128 encodes the first and second codes Inter-component prediction is used from the component signals 22 and 26. As mentioned above, the signals 22, 26 and 30 are residual signals, i.e., component specific Coding branch 132 A , 132 B and 132 C The residual of the inter-component prediction performed within The difference signals can each be represented.

[0029] FIG. 4 thus shows a more specific embodiment of the encoder of FIG. 3 corresponding to the decoder of FIG. Correspondingly, the encoder 100 of FIG. Code branch 132 A , 132 B and 132 C Each of these components includes Inter-participant prediction 134 A , 134 B , 134 C and the residual former 136 A , 134 B , 134 C And that Each of the modules 120, 124, and 128 includes an inter-component prediction 134. A ~1 34 C As far as A ~34 C acts similarly to They are the inter-component prediction parameters38 A ~34 C The only difference is that the components of Video 16 Based on the already coded parts, each inter-component predicted signal 40 A , 40 B oh and 40 C The predictor 34 is generated. A ~34 C While is controlled by the latter, e.g. , predictor 134 A ~134C Same thing in the rate / distortion optimization sense or anything else of that nature. Then, adjust the residual shaping 136 A ~136 C are the modules 120, respectively. , 124 and 128, which are lossy encoded into the data stream 12 To obtain the inter-component prediction signals 120, 124 and 130 of the first instructions, , Inter-component prediction signal 40 A ~40 C and Video 16 and their respective components A prediction residual is formed between the co-located parts. The difference signal 122, i.e. free from the coding losses introduced in module 120, The reconstructed version of the first command prediction residual signal 122 is added to the data stream 12 as a prediction residual. To encode the signal 126, the module 126 is then used as the basis for inter-component prediction. 22, which is used by module 124. As a result of the encoding, module 120 , the first component residual data 42 A Then it generates the data Similarly, the module 124 receives an inbound residual signal 126 As a result of encoding the second component residual data 42B and the ICP parameters Data 44 B which is then inserted into the data stream 12. Based on the latter data, a reconstructable version of the residual signal 126, i.e. together with the residual signal 22 to a first instruction to encode the residual signal 130 into a data stream. The reconstructable version used by the module 128 for inter-component prediction The module 26 can be derived as a result of encoding the module 128. , module 128 is ICP parameter data 44 C and the respective residual data 42 C Output do.

[0030] After describing various embodiments of the decoder and encoder with reference to FIGS. 1-4, the modules The method of adapting the two-source ICP prediction performed by the methods 28 and 128, respectively, is Regarding this, various embodiments are described that are related to various possibilities. A quick detour is to use H.264 as an example for a hybrid video compression scheme. 65 / HEVC, and, however, benefits arising from the two source ICP predictions. For motivation, the entire discussion outlined below is limited to H.265 / HEVC extensions. Rather, the following explanations are merely exemplary and may be used to clarify the It only mentions H.265 / HEVC.

[0031] In the case of hybrid video compression methods such as H.265 / HEVC, the prediction signal is temporally This occurs using spatial or inter-view prediction (predictive coding), and then the result is The resulting prediction (called the residual) is then transformed, quantized, and sent to the decoder (transcoding). Some entry points for inter-component prediction (ICP) are For example, ICP can be applied on the first sample value. For the embodiment of -4, this requires a source of ICP (i.e., signals 22 and 26). This means that the residual of inter-component prediction is not necessarily acceptable. , they may relate directly to the first source signal, i.e. the video 16. In the spatial domain, or in the spectral domain, ICP is the original signal, i.e., video 16 , i.e. on the transform coefficients of the spectral decomposition, e.g. the transform block of an image Transform coefficients such as DCT that are applied locally in units of , or wavelet transform coefficients In terms of numbers, the same is done for implementation and application efficiency purposes, i.e. To improve, however, the ICP may be described, for example, with reference to Figs. 2 and 4 above. In this way, the inter-component prediction can be performed on the residual signal, as exemplified by For the method, affine prediction can be used for complex reasons. In the latter case, the only gradient of the prediction is ICP Parameter Data 44 B and 44 C Prediction parameters in the bitstream as part of However, in addition, as outlined below, The embodiment can be easily extended to affine transformations or more complex predictions.

[0032] The ICP approach for residuals using a linear predictor is usually as follows: I can state this with certainty. r'(x,y)=r B (x,y)+α×r0(x,y)

[0033] In the above formula, r'(x,y) is the spatial position (x,y),r B Component B of (x,y) In the above equation, r'(x,y) is the spatial position position(x,y),r Bis the final residual sample value for component B of (x,y), That is, the prediction parameters for component B included in bitstream 44B Decoding / signaling for component B obtained from α and r0(x,y) Then, component A is placed at the same spatial location in the prediction source. The reconstructed residual sample of x,y is r0(x,y) B (x,y), i.e., signaling provides the residual of the computed component A. Note that α can be a floating point value. So lame.

[0034] See, for example, module 24 in the decoder embodiment. Module 24 , using the residual signal 22 as the prediction source r0, and the ICP parameter data 44 B Among them, Sig Multiply the same with the nulled alpha to get component B, i.e. 26, That is, to extract the inter-component prediction residual for r', we use the residual data 42 B , Sand Wachi, r B Module 24 in that it adds a signaled residual signal derived from can apply ICP.

[0035] For purposes of implementation, floating point values ​​can be mapped to integer values. For example, The floating-point value α may be allowed to range between −1 and 1, inclusive. Using three bits of precision, the formula can be rewritten as: r'(x,y)=r B (x,y)+(α×r0(x,y))≫3

[0036] The right shift operation is a division by 2 to a power of 3. Thus, the floating-point α can take the following integer values: α∈[0,±1,±2,±3,±4,±5,±6,±7,±8]

[0037] Also, for practical purposes, α can be limited to the following value: α∈[0,±1,±2,±4,±8]

[0038] That is, the final mapping allows, for example, a decoder or more specifically a module Rule 24 is alpha d By using The bit stream, i.e., the ICP parameter data of data stream 12, This is the decoded value from that contained in 44B. α=1≪α d

[0039] The ICP approach described above as representative for module 24 is e.g. The first facet is The first facet specifies the prediction source, the second facet specifies the prediction model, and the third and final facet specifies the prediction source. The template specifies the ICP prediction parameters or, more specifically, the prediction model parameters. Considering the above formula, the prediction source is r0, and then the prediction model is is a linear predictor using the spatial location of the same source components as The parameter is α.

[0040] There is a strong dependency between the second and third facets / modules. An example of dependency is again the input, which has another component, the prediction source. In this example, a linear prediction using the same spatial location would be used. Due to the limitations on the measurements, only one prediction (model) parameter is needed, so this is a single Only the prediction parameters of α, i.e., α, need to be transmitted in the bitstream. There is also an interaction between the source and the prediction model. Usually, the luminance residual is adopted as the prediction source. This is the case, for example, when luminance is used as component A in module 24. and for the first chroma component as component B. However, all available components are used as a forecast source. uses the appropriate prediction, i.e., a prediction that includes samples from both components This is the case for module 28. This module is available Two components, i.e., components A and B, are used as two forecast sources, i.e. That is, for components A and B, the luminance component and the first chrominance component are the same example. For example, linear prediction requires two model parameters instead. To be more accurate, the linear prediction that can be used for module 28 is as follows: It can be defined as follows: r''(x,y)=r C (x,y)+α0×r0(x,y)+α1×r1(x,y)

[0041] Thus, as far as module 28 is concerned, two prediction sources are available: Signals 22 and 26 are r0 and r1, respectively. That is, the module 28 for ICP is a co-located portion of the signals 22 and 26. With this available, ICP now predicts the signal 30, i.e., the r' portion. 42 remaining data C , i.e., rC As can be seen from the above, the module 28 The parts of the signal weighted by α0 and the signal weighted by α1 are located at the same position. The weighted sum of signals 26 weighted by the signal is modified by the addition of the signaled residual. The weighted sum is used.

[0042] As far as the relationship between modules 28 and 24 is concerned, the following is noted: As shown by ICP prediction and transmitted second component residual data 42 B In combination with ICP can be used on the basis of the second component signal that is obtained. That is, r' is the basis of the ICP performed by module 28, i.e. one prediction That is, r' can be used as r1. However, as described above, according to an alternative, module 24 does not use ICP. In that case, in Figs. 2 and 4, on the one hand, the output of module 20 and module 2 4, on the other hand, is excluded (or between modules 120 and 124). In that case, r1 is r B In a further embodiment, r1 is based on component A. Module 26 using ICP B The point that it is chosen because it is equal to 2 and 4 can be modified. In that case, in FIG. The arrows leading from module 24 to module 28 indicate the outputs other than those leading to combiner 36B. It starts from the further output of module 24.

[0043] Thus, according to the embodiment outlined above, the possibility exists to create multiple predictive solutions for ICP. In order to utilize the ICP technique, the performance characteristics of the ICP technique are improved by providing a higher compression efficiency. In the following, various details are described in detail. A general multiple source ICP is presented. The following description first considers multiple It deals with global signaling, which may be source ICP. Then there are various According to such an embodiment, local signaling of multiple source ICPs is addressed. Additional dependencies for multi-source ICP and image and video compression applications Further techniques are described that allow for high efficiency for

[0044] A kind of global signaling of multiple sources for ICP is multiple source IC Multiple Source ICP is permitted using a global flag sent at the appropriate level, even if If so, a further bit or series of bits is transmitted, because each transforms each predictor or encoder and signals its use. The global flag indicates that the transmission of a type of multiple source ICP signal in the data stream is A set of ICP coding modes including a source ICP coding mode and at least one non-ICP coding mode This shows the switching between the various ICP coding modes. There are also various source ICP coding modes possible. For example, ICP They differ in domain, i.e. spatial or spectral domain, and other adjustment powers. Therefore, the global flags further restrict the forecast model to one or more forecast parameters. "The further information transmitted for each A bit or sequence of bits is input to each of the blocks or to each of the predictors or encoders. " and in case of multiple source ICPs allowed, change the ICP parameters as described above. Metadata 44 C Examples for are possible.

[0045] Global flags are used to indicate whether parameters are set in H.265 / HEVC or Set the sequence parameters such as the image parameters to be set It can be transmitted in the sequence or even in the slice header. depends on the use of additional tools, e.g. for tile or wavefront processing, Even different levels of signaling for parallel processing of different types are possible. The source ICP can be enabled for the entire video sequence, but only for the There may be invalidation of certain content within the service. Consider such an example: The global flags in the sequence parameter set are, for example, This will overwrite the flag.

[0046] Thus, in summary, according to this embodiment, the decoder 10 is a multi-component For different ICP mode portions 202 of source video 16, multiple source ICP encoding modes are A set of ICP coding modes including ICP and non-ICP coding modes. The encoder 100 inserts the bit sequence 100 into the data stream 12 so as to switch between It may be configured to respond to the signal transmission 200 of multiple source ICPs.

[0047] According to the multi-source ICP coding mode, inter-component prediction is performed when the signal transmission is decoded. In Data Stream 12, explicitly or implicitly, the first and second The signals of spatially corresponding portions 204 and 206 of the component signals 22 and 26 are Based on the nulled or multiple source ICP encoding modes is calculated based on a combination, such as a weighted sum, of the spatially corresponding portions 204 and 206. The third component signal 30 of the component C is predicted by the inter-component prediction. For greater accuracy, if multiple source ICP coding modes are available, 5A Multiple Source ICP Signaling 200 for the mid-ICP mode portion of Video 16 In that case, for example, modules 28 and 128 are respectively Then, ICP for each block of component C can be executed. And, It may be, for example, a predicted block or the like. Metadata 44 C Each block 208 is either a spatially corresponding portion 204 or an empty Whether or not it is inter-component prediction is determined from the inter-corresponding parts 206. 208. Alternatively, the blocks of the image within the intermediate portion 202 may be At block 208, the third component signal 30 is a combination of both portions 204 and 206. Then, each portion 208 is predicted from the combination of the respective blocks 20 The weighted sum of parts 204 and 206 gives the inter-component prediction for 8. With respect to the weighting factors α0 and α1, which indicate the weighting factor, the co-located part 204 and 206, a method for component-to-component prediction from ICP parameter data 44 C can be shown.

[0048] In non-ICP coding modes, inter-component prediction is performed by each ICP mode unit 20 2. Therefore, for such ICP mode parts, ICP Parameter data 44 C does not need to be present in the data stream 12. Alternatively, For such an ICP mode unit 202, inter-component prediction is performed for component A. and B, by modules 28 and 128, respectively. For example, such a block of the non-ICP encoding mode unit 202 may be Block 208 is the I predicted from block 204 located at the same position in component A. Alternatively, the entire block 208 may be a CP. However, the prediction between blocks 208 may be inter-component prediction from block 206. That is, all of the blocks 20 within the portion 202 are not switched. 8 is also an inter-component prediction from block 204 of component A. Or, All blocks 208 within portion 202 are in the same location in component B. , and the inter-component prediction from block 206 located at

[0049] As already mentioned above, the ICP mode unit 202 can process image sequences, individual images, or a slice, or a series of slices. As mentioned above, multiple source ICP For ICP mode section 202 where the coding mode is enabled, the ICP parameter data is 44 C includes, for example, two weighting factors α0 and α1 for portion 208. On the other hand, for the ICP mode unit 202 in operation, the non-ICP encoding mode Parameter data 44 C There are no ICP parameters for block 208. Data signal 44 C simply means that only one of the weighting factors α0 and α1 is It is set to zero for all blocks 208 in this portion 202 .

[0050] Several possibilities exist for multiple source local signaling for ICP. For example, when multiple source ICPs are used, the decision 210 for the forecast source may be As can be seen in Figure 5B, the ICP data C It can be sent locally as part of the For example, components A, B and C may be, for example, luma, first chroma and second Then, the ICP parameter data 44 C The components Whether the inter-prediction is for the luma, i.e. A, or the first chroma component, i.e. B, Each block 208 includes a flag 210 indicating It may be a coding block, a prediction block or a transformation block. For example, the parameter 38 of the inter-component prediction C However, there is a switch between spatial and temporal prediction. The predicted block is a block that is replaced by the inter-component prediction parameter 38. C But, The type of prediction and the relationship that are signaled for the coding block of which each prediction block is a part. It may also be a block that signals or changes associated prediction parameters. For example, , if spatial prediction is signaled for the coding block, The prediction parameters for the coding block are A block is spatially predicted from adjacent already decoded / encoded parts of minponent C. The signal can be expressed along the spatial direction in which the signal is to be transmitted. The temporal prediction mode signaled for the coding block to which the prediction block belongs is generated. In this case, the prediction block can signal a motion vector as a prediction parameter. The transform block performs inter-component prediction for component C on the spectral In this case, the source for prediction can be either luminance or primary A flag 210 indicating whether the chroma components of the Could the ICP prediction parameters 212, e.g., the prediction weighting α, be independent? It is important to note that the components that make up the ICP forecast source For component A, α0 is used, and for component B, which forms the ICP forecast source, In this case, α1 is used. The forecast source is always weighted in this way. When the parameter(s) are signaled, for example, the dependency is given. On the other hand, the source always has a flag 210 for a prediction block or a coding block. However, when the ICP prediction parameter 2 is 12 are transmitted per transform block.

[0051] Considering the above example, a block in component C, such as flag 210, which indicates the forecast source, 208, for example to convert a block of the second chroma component. The ICP prediction source may be transmitted only for the purpose of reducing the weight α for block 208. While not being equal to, the ICP parameter data 44C is within the range of I In the case of CP prediction only, as part of the ICP parameter data 44C, block 208 The ICP prediction can then be signaled to the block of component C. This means that the condition for multiple sources must be met for the check 208. There are two situations where the condition is given. The first component, say the luminance component, and the second When each transform block of a component, e.g. the first chroma component, inherits the residual The first situation occurs when the first chroma component contains only zero-valued residual values ​​while the luma transformation block The second situation is given when the block inherits the residual. However, the luminance is This is predicted from the

[0052] Due to flag 210 indicating the ICP prediction source, the fixed context model is not used. Dependence on the circumstances under which multiple sources occur can also be used in two different contexts. Furthermore, other I models such as weighting are possible. Depending on the CP prediction parameter(s) 212, e.g., negative α may result in a different context model than positive α. Another possibility is that Dependence on the absolute value. In this example, an absolute value of α greater than 2 means that α is equal to or less than 2. This results in a different usage of the context model than if

[0053] Therefore, for example, as shown in the embodiment illustrated in FIG. is conditionally determined by the ICP source index 210 as an ICP parameter 212. Component C. 6, in response to modules 28 and 128, respectively, in step 220, P parameter data 44 C Read / write α to / from α. And α is actually zero. If it is found that α is equal to zero, check whether α is equal to zero. CP parameter data 44 C are further ICP parameters for this block 208 As shown in 224, the ICP contains either component A or B. However, if α is not zero, then ICP parameter data 44C is the ICP parameter data in step 226 or 228. ICP prediction source as source flag 212 read from or written to 44C Or includes a signal for block 208.

[0054] For illustrative purposes, FIG. 6 now illustrates whether α in step 220 meets a predetermined criterion. Depending on the source flags, different contexts are used for entropy coding / decoding. This indicates the possibility of being used for the purpose, and the situation is checked in 229. Then, if the absolute value of α is greater than a predetermined value, for example, 2, or if α is negative or equal, The criterion may be met regardless of whether α is zero or not. However, as already mentioned above, The dependency of the context used to decode / encode the base flag is merely optional. Accordingly, steps 229 and 228 may be omitted according to an alternative embodiment. Regardless of whether steps 226 or 228 apply. For α different from zero, in the case of ICP prediction, α0 is the first component A. and ICP is performed using α in step 220, or the ICP prediction software is In the case of the source, α1 is the source flag and is the second component B of step 230. This shows that.

[0055] Thus, summarizing FIG. 6, the ICP parameter data 44C is 8, i.e., 212, block 208 (i.e., 21 2) may have one weight α that is encoded into the The quantization may require spatial prediction based on the α of neighboring blocks. The parameter data 44C is determined by the value of α for this block 208. 8 may have a source flag 210 encoded in the data stream 12, The encoding of may include entropy encoding, and a particular context-adaptive entropy encoding may be used. Then, we encode the current block 208 using the α-dependent context. Therefore, steps 226 and 228 are performed to implement a context-adaptive encoding. Step 230 may include tropy encoding / decoding of the above-mentioned The application of a confirmed method is required. The other of α0 and α1 are set to zero, for example. will be done.

[0056] The extended ICP technique is described next. In particular, multiple sources of component C are When available, as in the case of the ICP prediction, the possible ICP predictions are Both components A and B can be accepted. In this case, the ICP prediction parameters 212 are the result of the linear combination of each component as described above. Depending on the result, the weighting of the two ICP predictions must be specified.

[0057] Figures 5c and 7 show that the ICP parameter data 44C is calculated according to the above confirmed equation, respectively. The same applies in step 236 according to the steps 232 and 234. Each block 208 to be read / written contains two weights α0 and α1. 3 illustrates the possibility of transmitting a signal.

[0058] For the sake of completeness, Figures 5D to 5G already show the above-mentioned situation. The ICP of Joule 28 has a residual signal at 42C in the spectral domain or In the spatial domain by signaled components C, i.e., the transform coefficients, e.g. For example, in the form of DCT coefficients or other transform coefficients of a transform block or the like. Performing spatial domain multiple source ICP can be performed as shown in FIGS. 5B and 5C. 5D and 5E, which use the module 28 of the embodiment of As shown, signals 24 and 22 are spatially The samples arrive in a corresponding manner, i.e., in the same position. are multiplied by weights α0 and α1 and then summed together, i.e. The samples are then concatenated into transform coefficients 42 to give the signal 30. C Apply to the top of By performing multiple source ICP in the spectral domain, and, respectively, again using the example modules of Figs. 5B and 5C as a basis. As shown therein, signals 24 and 25 are and 22 can be reached in a manner that is spatially coincident, i.e., located at the same position. Then, the samples are shown to be in any case, subject to the conversion 217 (before), To obtain the same spectral resolution, i.e., change the coefficients. and 24 already arrive in the transformation domain. The transformation 217 is, for example, In case of misalignment of transformation blocks between C and B and C, respectively, it is necessary The coefficients of signals 20 and 24 are multiplied by weights α0 and α1, respectively, to obtain And each corresponds to a spectrum, i.e., C You can get it from The summed coefficients are then summed to give the signal 30. To obtain the result, the signal undergoes an inverse transformation 216.

[0059] A further approach to improving compression efficiency is component swapping for ICP. According to this method, syntax elements are signaled within the ICP parameter data 44C. Then, we can determine the correct order of the residuals. In this example, we have three transformation blocks. A block, for example, has spatially corresponding parts, one for each component A, B, and C. Then, the ICP can be used. It is also possible to exchange components or even the first and third components. This adaptive component is especially meaningful in combination with ICP and multiple source ICP. The switch allows for lower prelude costs and reduced energy. For example, In this example, the application allows switching between two chroma components B and C. In the embodiment, two chroma components B and C from the luminance, i.e., component A Prediction results in the same cost, but the second prediction using the first chroma component Chroma prediction requires more overhead but uses the second chroma component. Prediction of the first chroma component requires fewer bits and results in lower cost.

[0060] This is again shown in Figure 8, which shows the first, second and third components Residual data of 42 A , 42 B and 42 C are steps 240, 242 and 2 44 indicates that it is being read / written. In this case, module 24 and ICP with 28, 124 and 128 respectively, i.e. 246 swapped signals 248, the weight for ICP performed on the basis of the first component, i.e., α Includes 44 ICP parameter data B and 44 C And it is controlled by , ICP parameter data for the second component 44 B And the second and third The weights used for ICP between the three components, i.e., for the third component ICP parameter data for 44 C α1 is obtained from the swap signal In response, the transmitted residual signal 42 A ~42 CFollow the solid or dashed lines to ICP246 As can be seen, the second component signal 26 is coupled to the ICP. Therefore, the swap signaling can be based on the first or third component. Similarly, a third component signal 30 is obtained in response to the first component signal 30 by the ICP. It is obtained on the basis of the first component or the second component. In the non-exchanged case, the ICP parameter data weights α and α1 are That is, they remain the same as the second component signal 26 and the third component signal 27. However, the ICP predictions of the component signals 30 are not shown. In the exchanged case, the ICP parameter data weights α and α1 are, respectively, The first and second stage ICP predictions are reinterpreted to control for ICP. The weights α of the P parameter data are referred to the first stage ICP, and the ICP parameters The weight α1 of the metadata is referred to the first stage ICP. If it is not exchanged, The interpretation does not change anything. However, in the swapped case, the third component signal 3 0 is an inter-component prediction that is predicted before the second component signal 26. As a result, the ICP parameter data weight α1 is effectively In other words, swapped In this state, as far as inter-component prediction is concerned, the second component B is It takes on the role of the third component C and vice versa.

[0061] Data 44 C The weight α1 of the ICP parameter data is calculated based on further data, e.g., the ICP parameter The parameter data weighting α0 or source indicator 212 may be included. Note that the embodiment of FIG. 8 can be combined with the previous one. C refers to second stage ICP, i.e. the second component where the signal undergoes ICP. For the input, this is reinterpreted as signal 26 or 30. is available in both cases, regardless of the swap signaling 248. And, ICP The parameter data weight α0 is based on component A in both cases and the ICP is It can be used to control the components A and B in the non-exchanged case. and B, and between components A and C of the replaced case, respectively. Alternatively, the source indicator can be interpreted.

[0062] Further, in a method similar to that of FIG. 8, the switching flag is changed to a color space change, i.e., It is possible to turn the switching between components into signaling flags. For each component, the ICP prediction parameter α0 or α1 is transmitted. Then, as described above, In addition, the color region change flag indicates whether the color region is to be changed. If it does not change, the second and third components B and C, for example, reveals two saturation components, α0 or Component A is predicted separately from A using α1. Then, component A is predicted just before However, for example, color changes may occur. If the two are predicted individually from A using α0 or α1, then They are then generally subjected to a transformation to a change of color space, which may For example, converting the ABC color space linearly to the YCC color space. It is only when a change occurs between A and B that it is not predicted from A, for example, when a color space change occurs. It is effectively predicted from cases A and C. Alternatively, the signaled color space ABC can be converted to YCC Two or more color transformations that map to a color space may be available. In this way, the signaling is switched locally in various ways. In this method, the color space in which the ABC components are transmitted may vary between two or more color spaces. and, as signaled by transmitting a signal 248, The change in the color is not a change between the color transformation and the ICP prediction parameters. The data signal is transmitted in a data stream of a coding device that is larger than the unit at which the data signal is transmitted. And for example, it transforms the block. Thus, α0 or α1 is For components B and C, the first block is transmitted as a transform block. Residuals sent for B B is r B '=r B +α0·r A For component A according to Residuals sent to A And similarly, the residual r C teeth, r B '=r B +α1·r A Follow A The prediction residual signal (22, 26, 30) is combined with For simplicity, we denote it as T, or directly as the YCC prediction residual signal, i.e., (r Y ,r C1,r C2 ) T =T (r A ,r B ',rC') T or (r Y ,r C1 ,r C2 ) T =(r A , r B ',r C ') T By using the residual of either It is obtained by a color transformation, which can be similar to the RGB transformation. Whether or not the signal is transmitted is controlled via transmission 248 of a signal. TIFF2025016507000002.tif15142

[0063] The aforementioned data 44 B or 44 C ICP prediction parameter α or α0 or α1 in The encoding is done as above using integer nominator signalling with small weights. It can be encoded or even further improved. ICP is R'G'B' When applied in color space, the value of α is typically α, α0, or α1. It is used for either, mainly positive relatively large values, i.e., with floating-point precision, 1 or 0.5. On the other hand, Y'C b C r Therefore, the value of α is often set to zero. In the center, it is relatively small. Asymmetric mapping and α regulation can provide further efficiency in terms of compression ratio. can be used to improve the α mapping: It is. α∈{0,±1,±2,+4,+8} or α∈{0,±1,+2,+4,+8}

[0064] Note that it identifies the maximum allowed value of α in the order, and the sign of α is In a symmetric mapping, if the instructions are unrelated, then all That is, the symbol can be transmitted after transmission of the absolute value of α. Furthermore, the absolute value of α is , to account for different probability distributions for the frequency of occurrence of positive and negative values ​​of α, Entropy coding using context modeling accordingly It could be.

[0065] Some of the embodiments are described in more detail below.

[0066] According to an embodiment, for example, in the case of two prediction sources, it is the case of the embodiment shown above. Since this was a case where the ICP prediction parameters 212 and the flag 210 were After that, which forecast source should be used is determined by flag 210 (source in Figure 6 above). For the purposes of this embodiment, the predicted source flag 2 is transmitted. 10 is only necessary when predictions have to be applied. And it is IC P can be derived from the prediction parameters.

[0067] In a further aspect, each block 208, e.g., each of the prediction blocks From the specific ICP prediction parameters 212 of the nested transformation block, The source flag 210 indicates which prediction source should be used for each prediction unit. In this preferred embodiment, I such as α of the predicted block level is always transmitted. It should be noted that it is also possible to transmit CP prediction parameters 212 .

[0068] In another preferred embodiment of the present invention, nested transformations of each coding block are A flag indicating which forecast source is used for a particular forecast parameter in the conversion block. The source (i.e. source flag) must always be sent for each coding block. In this preferred embodiment, the prediction block level or the coding block level It should be noted that it is also possible to transmit ICP prediction parameters.

[0069] FIG. 9 illustrates the possibilities just outlined as an example. In particular, as described above, ICP The parameter data 44C includes two components, namely, ICP prediction parameters 2 12, for example, weight α and unconditionally encoded ICP source flag 210. As already shown in FIG. 6, steps 220 and 222 are executed as outlined above. Steps 226, 228 and 229 are rearranged to lie between In this case, steps 228 and 229 may even be ignored. In 9, the ICP parameter data 44C is the accuracy signaled by the ICP prediction source. This illustrates the possibility that the accuracy of the ICP prediction parameter α may be coarser than the accuracy with which it is transmitted. The ICP source flags shown by the continuous lines in FIG. 9 are sent in units of blocks 250. The ICP prediction parameter α is transmitted to the ICP prediction parameter data 44C. , subblock 252, i.e., a subdivision of block 250. 1 illustrates blocks 250 and 252 of a rectangular image 18, and also illustrates blocks of other shapes. It should be noted that the q can be used in a similar manner.

[0070] In a further embodiment, the source flags 210 indicating the prediction source are respectively The ICP prediction parameters 212 can be linked to the weights, i.e., the current The ICP prediction parameters are set for the block or the current prediction block or the current code. In this embodiment, the source flag 2 indicates the prediction source. For linear prediction, the only prediction parameter is the range between 0.25 and +0.25 (or or the range -2 and +2 for α with integer values, or to be more precise An integer is encoded only if its nominator is evaluated, but all-inclusively. For example, see FIG. 6, where ICP parameters or data 44C are , it is necessary to decode the source flag only if α is different from zero. Alternatively, the source flags can be set to a specific range of values ​​for α (e.g., -0.25 to +0.25). In case of the above mentioned cases, both are inclusive and specific blogs. For locking, ICP parameter data 44 C If α is If it is out of range, the source flag is not transmitted explicitly in the data stream. The first component A is not estimated, for example, so that it can be called an ICP prediction source. For example, component A is the luminance component, and components B and C are the two If it is a chroma component, then a large absolute value of α will result in a prediction source for the second chroma component C. is the luminance component rather than the other first chroma component B. Thus, In Figure 6, a further check is made to see if α needs to transmit / read the source flag. It checks if it is within the range of values ​​and lies between 222 and 229. If not, the source flags are not read / written, otherwise step 2 29 is accessed.

[0071] In another embodiment, the source flag indicating the prediction source is the IC of the first chroma component. In this example, the ICP parameter data 44B, i.e. That is, in the case of linear prediction for the co-located blocks of α, only one ICP predictor is used. The measured parameters are between -0.25 and +0.25 or -0.2 and +0.2. When the value ranges may be chosen differently, both are inclusive and predictive. ICP parameter data showing the measurement source 44 C The source flags are, for example, For example, a check can be added to Figure 7. According to ICP parameter data 44 B are the blocks placed at the same position The inter-component prediction of component B from the sigma-based ... It is checked whether to display for the block placed in the position, and its size If the size exceeds some predefined value, for example, Components A, B, and C indicate that the third component C must probably be an ICP. In this case, the second component, i.e., the first chroma component of the estimate is used as the basis. In this case, when the source flag is estimated, the second component is set to I Used as an indication that a color space such as YCC is supported to identify CP prediction sources. However, the size of α for co-locating objects is If so, parameter data 44 B is smaller than a predefined value. As shown by the above, the source flags are used to indicate the current block as part of the ICP parameter data. Signal 44 for lock C will be sent.

[0072] According to an embodiment, the context model for the flag indicating the prediction source is IC P is independent of prediction parameters, e.g. weights. Hence, one fixed context This is, for example, in relation to the description of FIG. can be ignored by the yes pass of check 222 which directly leads to 226 This means that.

[0073] However, in another embodiment, the context model for the source flags is It depends on the ICP prediction parameters, e.g. weighting. When linear prediction methods are used, only For one prediction parameter, for example, different context models can be used: , the weights are, as shown above, for prediction parameters with integer values, e.g., -2 and Both #1 and #2 are inclusive ranges.

[0074] Furthermore, in the case of linear prediction methods, if the only ICP prediction parameter is negative, The following context models can be used, as also mentioned above:

[0075] The use of multiple prediction sources can be achieved, for example, by setting a sequence parameter set. It can be turned on or off using a flag sent with As discussed above with respect to FIG. 5. The transmission, however, in the picture parameter set Or it could occur in the slice header.

[0076] The usage of multiple forecast sources can be specified at different hierarchical levels. For example, it is important to consider the sequence parameter set and the image parameter set. Since the image parameter set flag is low, this preferred embodiment Possibility to disable multiple source ICP for an image or frame of a video sequence In a further aspect, the prediction source, i.e., the source flag, is indicated. The flags are the ICP prediction parameters of the current and previous components, For example, the weights of the ICP prediction parameters, i.e., the relative differences in weightings, may depend on If the relative difference is greater than a predefined limit, the flag is set to, for example, 1. In other words, in the case of Figure 6, the additional check is Ream check 222 is applied, where ICP parameter data 44 B By The weight associated with the co-located block is greater than a predetermined amount. For the current block, ICP parameter data 44 C Whether it is different from α of Regardless, it is checked with this additional check. If yes, this is Components A through C are related to each other as in the YCC color space. And, therefore, The source flag specifies a specific first chroma component, i.e., component B. The implication is that it can be inferred that the ICP source for C's ICP can be identified, This can be interpreted otherwise as read / write of source flags. In other words, data44 C In response to transmitting 212 the ICP prediction parameter signal Thus, the third component prediction parameter α can vary with sub-picture granularity. Similarly, the second component signal 44 B ICP prediction parameter signal for The component prediction parameter α varies with sub-picture granularity. Thus, the weighting is ICP for component B and ICP for component C. There may be local differences between the multi-source ICPs. Thus, the check is component and third component predictor parameters differ beyond a predetermined limit In this case, for example, the difference or index of alpha values ​​can be used as the standard of deviation. For locations that exceed the above limits, the ICP source indicator 210 , otherwise, one might infer, but for places where they do not differ beyond a limit To this end, the ICP source indicator 210 C may be present in

[0077] If multiple forecast sources are available, further According to an embodiment, the prediction parameter(s) of the ICP are the sum of the two prediction sources. Predictions that support weighted combinations can be identified and transmitted.

[0078] According to a further embodiment, linear prediction is used for multiple source ICP, IC In this embodiment, the only absolute ICP prediction parameter is The data is binarized and encoded using an abbreviated singleton code, and the sign The signature may be encoded first, and the only prediction parameter is Depending on the sign of the data, different context models are used for the binary decomposition bins. According to a further embodiment, the signature is first encoded. A code signature for the desired implementation, i.e., the parameters are executed differently. According to the quantization of the prediction, the maximum allowed prediction parameters are between 0.25 and 1, both All of this could be included in the scope.

[0079] According to an embodiment, the flag indicates whether the second and third components are swapped, This was outlined above with respect to FIG. 8. In this embodiment, the first The residual constructed for the transform block associated with the chroma component is the residual of the second component can be treated as a difference and vice versa.

[0080] With respect to the above description of specific embodiments, the same is applicable to the coding of multi-component images. Note that this is easily transferable.

[0081] Although some aspects have been described in the context of an apparatus, these aspects may also be implemented in block diagram form. The method description corresponds to a method step or a method step feature. Similarly, aspects described in the context of method steps are also Means a description of a block or item or a feature of a corresponding device. Part or all of the software may be implemented using hardware devices such as microprocessors. Thus, a programmable computer or electronic circuit can be implemented. In some embodiments, one or more of the most important method steps may be performed by such an apparatus. It can be executed.

[0082] The coded image or video signal of the invention can be stored on a digital storage medium. or a transmission medium, e.g. a wireless transmission medium, or a wired transmission medium, e.g. It can be sent over the Internet.

[0083] Depending on the particular implementation requirements, embodiments of the invention may be implemented in hardware or in software. Implementation may be by electronically readable control stored thereon. Digital storage media with control signals, such as floppy disks, DVDs, Blu-rays, Using CD, ROM, PROM, EPROM, EEPROM or FLASH memory It can be programmed to execute each method. cooperates (or is capable of cooperating) with a computer system capable of The digital storage medium may be computer readable.

[0084] In some embodiments of the present invention, a method for carrying out any of the methods described herein is provided. An electronically readable memory capable of cooperating with a programmable computer system. A data storage medium having a control signal is included.

[0085] Generally, embodiments of the present invention are implemented as a computer program product having program code. The program code may be implemented to perform one of the methods. The computer program product is implemented when it is run on a computer. The program code may for example be stored on a machine-readable carrier.

[0086] Other embodiments include any of the methods described herein and stored on a machine readable carrier. The present invention includes a computer program for carrying out one of the methods described above.

[0087] In other words, embodiments of the inventive method are therefore the methods described herein. is a computer program having a program code for executing one of At that time, the computer program runs on the computer.

[0088] Thus, further embodiments of the inventive method are disclosed herein in a method as set forth herein. A data carrier (or digital storage medium) containing a computer program for carrying out one of the methods. Data storage media, digital storage media or computer readable media. Storage media or recorded media are typically tangible and / or non-transitive.

[0089] A further embodiment of the inventive method therefore comprises carrying out one of the methods described herein. A data stream or series of signals representing a computer program for A data stream or a sequence of signals may be transmitted, for example, over the Internet. For example, the configuration may be via a data communication connection.

[0090] Further embodiments may be implemented, for example, in a computer or programmable logic device. A processor configured or adapted to perform any of the methods described herein. This includes the means for processing.

[0091] In a further embodiment, a computer for carrying out any of the methods described herein is provided. This includes computers on which the program is installed.

[0092] A further embodiment of the present invention is a computer program product for carrying out one of the methods described herein. configured to transfer (e.g., electronically or optically) a data program to a receiver. The receiver may be, for example, a computer, a mobile device, The device or system may be, for example, a computer to a receiver. It may include a file server for transporting computer programs.

[0093] In some embodiments, some of the functionality of the methods described herein may be achieved. A programmable logic device (e.g., a field programmable logic device) may be used to perform all or part of the In some embodiments, a gate array as described herein may be used. The method of claim 1, wherein the first and second electrodes are field programmable to perform one of the methods described herein. Such a gate array can cooperate with a microprocessor. Usually, the method is as follows: The present invention is preferably implemented by a hardware apparatus comprising:

[0094] The devices described herein may use hardware devices or may be computer-based. Use a computer or a combination of a hardware device and a computer This can be implemented.

[0095] The methods described herein may be implemented using hardware devices or Use a computer or a combination of a hardware device and a computer It can be executed.

[0096] The above described embodiments are merely illustrative for the principles of the present invention. It is understood that variations and details of modifications and arrangements will be apparent to those skilled in the art. Therefore, the scope of the present invention is not limited to the scope of the claims, but rather to the examples of the present invention. It is intended that limitations be placed on the specific details presented for purposes of description and illustration.

Claims

1. A first input of a multi-component image or video (16) from a data stream (12). reconstructing the first component signal (22) in terms of the components (A) of The multi-component image or video ( reconstructing the second component signal (26) with respect to the second component (B) of Build The reconstructed first component signal (22) and the reconstructed second component signal (23) are The multi-component signal (26) is subjected to inter-component prediction to obtain the multi-component signal (26). A third component (C) relating to the image or the third component (16) of said video. By reconstructing the input signal (30), Sampling the scene for spatially distinct components and a decoder configured to decode a component image or said video (16).

2. The decoder detects in the data stream (12) a set of ICP coding modes Among the ICP encoding modes, different ICP mode portions of the multi-component image or transmitting (200) multiple source ICP signals to be switched for video (16); configured to respond to A plurality of source ICP coding modes; At least one of the non-ICP coding mode and the fixed one source ICP coding mode. Also includes one and The multiple source ICP coding modes are determined by the decoder according to the signaled: Alternatively, a spatially corresponding portion of the reconstructed first component signal and the reconstructed The inter-component prediction is performed from a combination of the second component signal and the second component signal. a current sub-portion of a current image of said multi-component image or video, configured to reconstruct three component signals (30); The non-ICP coding modes do not allow the decoder to use any inter-component prediction. of the current sub-portion of the current image of the multi-component image or video that is not configured to reconstruct the third component signal (30); The fixed single source ICP coding mode is used by the decoder to the fixed first component signal and the reconstructed second component signal. The multi-component prediction method uses the inter-component prediction from one spatially corresponding part obtained by the multi-component prediction method. The third component of the current sub-portion of the current image of a component image or video. and a decoder configured to reconstruct a plurality of source component signals, the decoder comprising: In response to the transmission of an ICP signal, the encoding mode is switched to the fixed one source ICP encoding mode. The multi-component image or video is fixed within each of the portions of the multi-component image or video. Item 2. A decoder according to item 1.

3. The transmission of the multiple source ICP signals is performed when the ICP mode portion (202) captures a single image, Signaling in the data stream (12) as being an image sequence or slice.

3. The decoder of claim 2, wherein the decoder is programmed to generate a first decoder.

4. The first component (A) is luminance and the second component (B) is the first and the third component (C) is the second chroma component. A decoder according to any one of claims 1 to 3.

5. The decoder extracts spatially corresponding components of the reconstructed first component signal (22). a current image of the multi-component image using inter-component prediction from a portion of the current image; or reconstructing said third component signal (30) of said video (16); A component signal is extracted from the spatially corresponding portion of the reconstructed second component signal (26). The third prediction in the current picture of the multi-component video using inter-component prediction. Switching between reconstructing the component signals (30) at a first sub-image granularity A decoder according to any one of claims 1 to 4, configured to:

6. The decoder extracts spatially corresponding components of the reconstructed first component signal (22). said multi-component image or video using inter-component prediction from a portion of said reconstructing a third component signal (30) in the current image of (16); In response to transmitting (210) an explicit ICP source signal in the data stream (12), , a component signal (26) from a spatially corresponding portion of the reconstructed second component signal (26). Current image of a multi-component image or video (16) using inter-component prediction During the reconstruction of the third component signal (30) in The method according to any one of claims 1 to 5, wherein the switching is performed at an image granularity. decoder.

7. The decoder performs spatial, temporal and / or inter-view prediction on a prediction block basis. By changing between the modes, the spatial, temporal and / or inter-view prediction is used. Regarding the third component (C) of the multi-component image or video and configured to reconstruct the third component signal (30), the decoder comprising: The first sub-image granularity is configured to subdivide the current image in units of prediction blocks. A decoder according to claim 5 or claim 6, comprising:

8. The decoder receives the third component of the multi-component image or video. using spatial, temporal and / or inter-view prediction for the Inverse transforming the residual prediction by spatial, temporal and / or inter-view prediction on a per-block basis. and configured to perform a reconstruction of the third component signal (30) by and the decoder converts the first sub-image granularity of the current image into a transform block granularity.

7. A decoder according to claim 5 or 6, adapted to subdivide Yes.

9. The decoder converts the reconstructed first component into a first component using linear prediction. from the first component signal (22) and the reconstructed second component signal (26). configured to perform the inter-component prediction and to a component signal (22) and said reconstructed second component signal (26); is linearly related to the inter-component prediction at the second sub-image granularity by weighting , in response to transmitting (212) an ICP prediction parameter signal in said data stream (12). A decoder according to any one of claims 1 to 8, adapted to:

10. The decoder performs spatial, temporal and / or inter-view prediction on a prediction block basis. by switching between the spatial, temporal and / or inter-view predictions. a third component (C) of said multi-component image or video configured to perform a reconstruction of the third component signal (30), The first sub-image granularity is set so as to re-divide the current image in units of prediction blocks. The decoder according to any one of claims 1 to 9, wherein the decoder is configured as follows:

11. The decoder uses spatial, temporal and / or inter-view prediction and transforms Predicting the residual by spatial, temporal and / or inter-view prediction on a block-by-block basis a third component of the multi-component image or video by inverse transforming the to perform a reconstruction of the third component signal (30) with respect to the component (C). and the decoder is configured to:

11. The method according to claim 9, further comprising the steps of: Decoder.

12. The decoder weights the data with non-uniform quantization that is asymmetric about zero. The ICP prediction parameter signal is extracted from the data stream (212). A decoder according to any one of claims 9 to 11.

13. The decoder filters the data stream with zeros belonging to a set of possible values ​​of weight. (12) extracting said weights from the transmission of the ICP prediction parameter signal (212) (220) configured to: Whether the multi-component image is zero or not due to the transmission of a parameter signal For each block of a current image of an image or video (16), the decoder checks and if not, the data stream (12) extracting the ICP source index (210) for each block (208) 226, if possible, to output the ICP source for each block. Extraction of the indices is suppressed (224), and the decoder performs If the weighting is not zero, then the reconstructed first component signal (22) For each block using linear prediction from the spatially corresponding block (204), and dividing the third component signal (30) according to the respective non-zero weightings. or to reconstruct the second component signal (26) Each of said blocks (24) is subjected to linear prediction from a corresponding block (24). 08), in accordance with the respective non-zero weightings, To reconstruct the signal (30), the ICP source index ( The method according to any one of claims 9 to 12, further comprising: Decoder included.

14. The decoder extracts zeros from the data stream that belong to a set of possible values ​​of the weights. (12) extracting the weighting from the transmission of the ICP prediction parameter signal (212) and transmitting (212) said ICP prediction parameter signal. Whether the weighting for each block (208) satisfies a predetermined criterion is determined. of a block of the current image of said multi-component image or video, regardless of whether For each one, the decoder is checked, and if not, the data For each block from the stream, and if possible, the IC P source index (210) and extract the ICP source index for each block. The ICP source index is set to each of the blocks according to a predetermined standard. and weighting the reconstructed image according to the weights for the respective blocks. A linear prediction is then used to estimate the spatially corresponding block of the first component signal. to reconstruct the third component signal of each of the blocks, or the reconstructed second block weighted according to the weight for each block; Each of said components is subjected to linear prediction from spatially corresponding blocks of the component signals. and reconstructing the third component signal of each of the blocks. The decoder is configured to respond to the ICP source indication for a A decoder according to any one of claims 9 to 13.

15. The decoder first decodes the sign of the weighting and generates a computation signal according to the sign. in the data stream by decoding its absolute value using text modeling. configured to derive said weightings from said transmission of said ICP prediction parameter signal (212); A decoder according to any one of claims 9 to 14,

16. The decoder may perform spatial, temporal and / or inter-view prediction on top of residual prediction to the covariance matrix. By applying inter-component prediction, the spatial, temporal and / or inter-view prediction is The third component of the multi-component image or video is measured using a to perform a reconstruction of the third component signal (30) for the component (C). The decoder according to any one of claims 9 to 14, configured as follows:

17. The decoder separates the first and second components of the multi-component audio signal. For each component, intra-component spatial, temporal and / or cross-view prediction (34 A, 34B), and the reconstructed first component is configured to execute The first component signal and the reconstructed second component signal are With respect to said first and second components of the audio signal, said spatial, temporal and / or or a prediction residual of inter-view prediction.

2. A decoder according to claim 1 .

18. The decoder receives a third component or components of the multi-component image. using intra-component spatial, temporal and / or inter-view prediction (34C); and , from the spectral domain to the spatial domain, spatial, temporal and / or Alternatively, in inter-view prediction, the residual prediction is inversely transformed (216) to obtain a previous performing the reconstruction of the third component signal and then subtracting the residual in the spatial domain; configured to apply the inter-component prediction onto the prediction or the spectral region. A decoder according to any one of claims 1 to 17,

19. The decoder performs intra-component spatial, temporal and / or viewpoint prediction (34A , 34B) to perform the reconstruction of the first and second component signals. A decoder according to any one of claims 1 to 18, configured as follows:

20. The decoder receives spatially corresponding portions of the reconstructed first component signal. (204) of the multi-component image using the inter-component prediction. reconstructing said third component signal in the current image or video; and and from the spatially corresponding portion (206) of the reconstructed second component signal. A current image or a multi-component image using the inter-component prediction during reconstruction of the third component signal of the video, the reconstructed first component signal being configured to switch at an image granularity; and a predetermined prediction model according to which the reconstructed second component is different. and detecting prediction parameters of a predictive model at a second sub-image granularity in the data stream using the In response to the meter signal, i.e., the ICP prediction parameter signal (212), the component 2. Perform inter-component prediction to transmit an explicit ICP source signal in the data stream ( 10) performing switching at the first sub-image granularity in response to the first sub-image granularity to obtain a prediction performance of a prediction model; Explicit extraction from data streams using context modeling that supports The method of claim 1, further comprising the steps of: decoding a transmission of an ICP source signal; 2. A decoder according to claim 1 .

21. The decoder uses a prediction model controlled by gradients as the prediction parameters.

20. A decoder as claimed in claim 19, configured to use linear prediction.

22. The decoder may select the gradient for the subportion regardless of whether the gradient is positive or negative. First, for a current sub-portion of a current image of said multi-component image or video, and performing the context modeling in response to the check (229).

22. A decoder according to claim 19 or claim 21, configured as follows:

23. The decoder determines whether the gradient for that subportion is within a predetermined interval. Check for the current sub-portion of the current picture for a component image or video and performing context modeling in response to said checking. A decoder according to claim 19 or claim 22.

24. The decoder receives a predetermined signal from the current image of the multi-component image or video. and extracting first and second prediction parameters from said data stream (12) for the selected sub-portion. A reconstructed prediction parameter is obtained using a predetermined prediction model and the first prediction parameter. and performing inter-component prediction from the first component signal (22) to obtain the predetermined A reconstructed second component using the constructed prediction model and the second prediction parameters. A first linearly reconstructed signal (26) is obtained by performing the inter-component prediction from the first linearly reconstructed signal (26). The inter-component prediction from the component signal and the reconstructed second component and combining inter-component prediction from a component signal. A decoder according to any one of claims 1 to 23.

25. The decoder outputs the reconstructed third component signal at a third sub-image granularity. and the reconstructed second component signal.

25. The method according to claim 1, wherein the swapping unit is responsive to a swap signal transmission (248) in response to the swapping unit. Decoder included.

26. an IC for the third component signal (30) of the data stream (12); In response to transmitting (212) the P prediction parameter signal, a first iteration of the prediction model is performed at a sub-image granularity. By switching between three component prediction parameters, a predefined prediction model can be used. The reconstructed first component signal (22) and the reconstructed second component signal (23) The inter-component prediction is performed from two component signals (26). The coda is composed, and deriving the inter-component prediction from the reconstructed first component signal (22); The second component (B) or video of the multi-component image being used performing a reconstruction of the second component signal (26) with respect to the digital signal; ICP prediction parameters for the second component signal of the data stream (12). In response to transmitting the signal, a component signal prediction parameter of the prediction model at a sub-image granularity is The reconstructed first prediction model using the predetermined prediction model by switching A decoder for performing the inter-component prediction from the component signals (22). Da is constructed, and The prediction parameters of the second component and the third component are predetermined. If the difference is more than the specified limit, and if it is not due to location, For each location from the data stream, an ICP source index (21 0), and if yes, suppress the extraction of the ICP source indicator (210); Set the ICP source index (210) at each location according to a predetermined standard, and The coder generates an ICP source index (210) for each location for each third component. The reconstructed first component is weighted by a prediction parameter. In each case, linear prediction is used to estimate the spatially corresponding position of the component signal. and adding an ICP source index (210) to the ICP source index (210) to reconstruct the third component signal. or (each third component predictive parameter the spatially corresponding components of the reconstructed second component signal are weighted by the Reconstructing the third component signal at each location using linear prediction from the location where the third component signal is to be reproduced. A decoder according to any preceding claim, adapted to construct

27. Regarding the third component (C) of the multi-component image or video: When reconstructing the third component signal (30), A residual signal for the component-to-component division is extracted from the data stream, thereby A decoder according to any one of claims 1 to 26, adapted to correct the prediction. Yes.

28. Sampling the scene for spatially distinct components and creating multi-component 1. A method for decoding a digital image or video (16), comprising: The method comprises: The multi-component image or video ( 16) reconstructing the first component signal (22) with respect to the first component (A) of The process of building The multi-component image or video ( reconstructing the second component signal (26) with respect to the second component (B) of and The reconstructed first component signal (22) and the reconstructed second component signal (23). The multi-component signal (26) is obtained by using inter-component prediction from the component signal (26). a third component (C) of said first image or said video (16); reconstructing a component signal (30).

29. A first input of a multi-component image or video (16) into a data stream (12). encoding a first component signal (22) with respect to a component (A); The multi-component image or video (1) is added to the data stream (12). 6) encoding the second component signal (26) with respect to the second component (B) of To do, The encoded first component signal (22) and the encoded second component signal (23) The multi-component signal (26) is subjected to inter-component prediction to obtain the multi-component signal (26). A third component (C) relating to the image or the third component (16) of said video. By encoding the input signal (30), Sampling the scene for spatially distinct components and an encoder configured to encode a component image or the video (16).

30. Sampling the scene for spatially distinct components and creating multi-component 1. A method for encoding a digital image or video (16), comprising: A first component of a multi-component image or video (16) is added to the data stream (12). encoding a first component signal (22) with respect to one component (A); 、 The multi-component image or video (1) is added to the data stream (12). 6) encoding the second component signal (26) with respect to the second component (B) of and The encoded first component signal (22) and the encoded second component signal (23) The multi-component signal (26) is subjected to inter-component prediction to obtain the multi-component signal (26). A third component (C) relating to the image or the third component (16) of said video. The method includes the step of encoding an input signal (30).

31. A computer program is a program that, when running on a computer, 31. The method of claim 28 or claim 30, wherein the gram code is

Images