Video image component prediction method and apparatus, and computer storage medium
Patent Information
- Application Number
- KR1020267027341
- Authority / Receiving Office
- KR · KR
- Patent Type
- Applications
- Current Assignee / Owner
- Filing Date
- 2018-08-09
- Publication Date
- 2026-09-04
Smart Images
Figure PAT00064_ABST
Abstract
Description
Technology Field
[0001] The embodiments of the present application relate to intra-frame prediction technology in the field of video coding, and in particular to a method and apparatus for predicting video image elements and a computer storage medium. Background Technology
[0002] In next-generation video coding standards (H.266) or Versatile Video Coding (VVC), prediction of chromaticity values or between chromaticity values from luminance values can be implemented through the Cross-component Linear Model Prediction (CCLM) method. Specifically, by using a linear regression method to perform linear modeling on adjacent luminance parameters and chromaticity parameters corresponding to the current coding block, a chromaticity prediction value can be calculated based on the linear model and the reconstructed luminance value.
[0003] However, when modeling a linear model, the factors currently considered are not sufficiently comprehensive. For example, when predicting chromaticity values based on luminance values, most factors considered are those of adjacent reference luminance and adjacent reference chromaticity values; similarly, when performing predictions between chromaticity values, most factors considered are those of adjacent reference chromaticity elements. Therefore, if there is a relatively large deviation between the adjacent reference luminance value and the parameter corresponding to the current coding block, or between the adjacent reference chromaticity element and the parameter corresponding to the current coding block, a deviation is generated between the linear model obtained through calculation and the desired model. Consequently, this reduces the prediction precision of the chromaticity prediction value of the current coding block, causing a relatively large deviation between the predicted chromaticity value and the actual chromaticity value. The problem to be solved
[0004] To solve the above technical problem, the embodiments of the present application aim to provide a method and apparatus for predicting video image elements and a computer storage medium that can effectively improve the prediction accuracy of a second image element prediction value so that the second image element prediction value approaches an actual element value. means of solving the problem
[0005] The technical method of the embodiment of the present application is implemented as follows.
[0006] In a method for predicting video image elements, the above method is,
[0007] A step of obtaining a first image element adjacent reference value and a first image element reconstruction value corresponding to a current coding block - the first image element adjacent reference value is intended to represent a first image element parameter corresponding to an adjacent reference point of the current coding block, and the first image element reconstruction value is intended to represent a reconstruction parameter of one or more first image elements corresponding to the current coding block - ;
[0008] A step of determining an association coefficient according to the first image element adjacent reference value and the first image element reconstruction value - the association coefficient is intended to indicate the degree of deviation of the image element between the current coding block and the adjacent reference point - ;
[0009] A step of inputting the above association coefficient into a preset weight calculation model to obtain a weight coefficient corresponding to the above adjacent reference point;
[0010] A step of determining a proportionality factor according to the above weighting coefficient; and
[0011] It includes the step of obtaining a second image element prediction value corresponding to the current coding block through the proportional factor above. Effects of the invention
[0012] Embodiments of the present application provide a method and apparatus for predicting video image elements, a computer storage medium, and a prediction apparatus, wherein a first image element adjacent reference value and a first image element reconstruction value corresponding to a current coding block are obtained; wherein the first image element adjacent reference value is intended to represent a first image element parameter corresponding to an adjacent reference point of the current coding block, and the first image element reconstruction value is intended to represent one or more reconstruction parameters of first image elements corresponding to the current coding block; wherein an association coefficient is determined according to the first image element adjacent reference value and the first image element reconstruction value; wherein the association coefficient is intended to represent the degree of image element deviation between the current coding block and the adjacent reference point; wherein the association coefficient is input into a preset weight calculation model to obtain a weighting coefficient corresponding to the adjacent reference point; wherein a proportionality factor is determined according to the weighting coefficient; and wherein a second image element prediction value corresponding to the current coding block is obtained through the proportionality factor.As can be seen from this, in an embodiment of the present application, the prediction device can determine an association coefficient based on a first image element adjacent reference value and a first image element reconstruction value corresponding to the current coding block, thereby enabling the distribution of different weighting coefficients for different adjacent reference points according to the association of element parameters between the adjacent reference point and the current coding block. By modeling a linear model that is more closely aligned with the desired model, defects in the deviation between the linear model and the desired model can be effectively overcome when there is a relatively large deviation between the first image element adjacent reference value and the element parameter corresponding to the current coding block, or when there is a relatively large deviation between the third image element adjacent reference value and the element parameter corresponding to the current coding block. Consequently, when predicting the elements of the current coding block according to the linear model, the prediction accuracy of the video image element prediction value can be significantly improved, thereby allowing the video image element prediction value to approach the actual video image element value more closely. Brief explanation of the drawing
[0013] Figure 1 is an example diagram of the video coding flow. Figure 2 is an example diagram of the video decoding flow. Figure 3 is an example of the location of adjacent reference sampling, Figure 1. Figure 4 is an example of the location of adjacent reference samples, Figure 2. Figure 5 is an example of determining a linear model in the prior art. Figure 6 is an example of Figure 2 for determining a linear model in the prior art. FIG. 7 is an example flowchart of the implementation of a method for predicting video image elements provided in an embodiment of the present application. FIG. 8 is an example diagram showing the removal of interference points in an embodiment of the present application. FIG. 9 is an example diagram 1 of the configuration structure of a prediction device provided in an embodiment of the present application. FIG. 10 is an example diagram 2 of the configuration structure of a prediction device provided in an embodiment of the present application. Specific details for implementing the invention
[0014] Hereinafter, the technical solution of the embodiments of the present application is clearly and completely explained in conjunction with the attached drawings of the embodiments of the present application. It should be understood that the specific embodiments described herein are merely for the purpose of interpreting the associated application and are not a limitation to said application. Furthermore, for the convenience of explanation, only the parts related to the associated application are illustrated in the drawings.
[0015] In a video image, a coding block is typically represented using a first image element, a second image element, and a third image element; wherein the first image element, the second image element, and the third image element may include one luminance element and two chromaticity elements. Specifically, the luminance element is typically represented using the symbol Y, and the chromaticity elements are typically represented using the symbols Cb and Cr, where Cb represents a blue chromaticity element and Cr represents a red chromaticity element.
[0016] It should be explained that in the embodiments of the present application, the first image element, the second image element, and the third image element may each be a luminance element (Y), a blue chromaticity element (Cb), and a red chromaticity element (Cr), for example, the first image element may be a luminance element (Y), the second image element may be a red chromaticity element (Cr), and the third image element may be a blue chromaticity element (Cb), and the embodiments of the present application are not specifically limited thereto.
[0017] Furthermore, in an embodiment of the present application, the sampling format represented by a conventional luminance element and a chromaticity element, respectively, is also referred to as a YCbCr format, wherein the YCbCr format may include a 4:4:4 format; a 4:2:2 format and a 4:2:0 format.
[0018] When a video image uses the YCbCr 4:2:0 format, if the luminance element of the video image is a coding block of size 2N×2N, the corresponding chromaticity element is a coding block of size N×N, where N is the length of the side of the coding block. In the embodiment of the present invention, the 4:2:0 format is used as an example for explanation, but the technical method of the embodiment of the present invention is similarly applicable to other sampling formats.
[0019] In an embodiment of the present application, the method for predicting video image elements can be applied to an intra-frame prediction portion of a video coding mixed framework, and specifically, the method for predicting video image elements can be applied simultaneously to a coding end and a decoding end. For example, FIG. 1 is an example diagram of a video coding flow, and as shown in FIG. 1, video coding may include a plurality of specific steps such as intra-frame estimation, intra-frame prediction, and motion compensation, where the method for predicting video image elements provided in the present application can be applied to the intra-frame prediction portion; FIG. 2 is an example diagram of a video decoding flow, and as shown in FIG. 2, video decoding may include a plurality of specific steps such as carrier, intra-frame prediction, and motion compensation, where the method for predicting video image elements provided in the present application can be applied to the intra-frame prediction portion.
[0020] In H.266, to further improve coding performance and efficiency, Cross-component Prediction (CCP) has been extended and improved, and a Cross-component Linear Model Prediction (CCLM) method is proposed. In H.266, CCLM implements prediction from a first image element to a second image element, from a first image element to a third image element, and between a second image element and a third image element. Although the prediction from a first image element to a second image element is described below as an example, the technical method of the embodiments of this application may be similarly applied to the prediction of other image elements.
[0021] Specifically, in the prior art, when the CCLM method implements prediction from a luminance element to a chromaticity element, a cross-component linear model prediction mode is used in a decoder of a next-generation video coding standard, such as the H.266 / VVC Joint Exploration Model (JEM) or VVC Test Model (VTM), to reduce overlap between the luminance element and the chromaticity element and between different chromaticity elements. For example, according to formula (1), a predicted chromaticity value is generated using the reconstructed luminance value of the same coding block.
[0022] (1)
[0023] Here, represents the location coordinates of the sampling point among the coding blocks, and It indicates the horizontal direction, and represents the vertical direction, and The position coordinates among the coding blocks It represents the predicted value of the second image element at the sampling point, The position coordinates in the same coding block (through downsampling) Represents the first image element reconstruction value of the sampling point, and is a proportional factor of the linear model and can be derived by minimizing the regression error of the first image element adjacent reference value and the second image element adjacent reference value, as shown in the formula (2) below.
[0024] (2)
[0025] Here, represents the first image element adjacent reference values (e.g., left and top) through downsampling, and represents the adjacent reference values of the second image element (e.g., left and top), and N is the number of adjacent reference values of the second image element. FIG. 3 is an example of the location of adjacent reference sampling, FIG. 4 is an example of the location of adjacent reference sampling, FIG. 4 is an example of FIG. 3 and FIG. 4, and as shown in FIG. 3 and FIG. 4, and is an adjacent reference pixel point, and N is the number of adjacent reference values of the second image element, where, for a video image in 4:2:0 format, a first image element coding block of size 2N×2N is as shown in FIG. 3, and the size of the corresponding second image element is N×N, as shown in FIG. 4. For a rectangular coding block, these two formulas can be applied directly. For a non-rectangular coding block, adjacent sampling of a relatively long edge is downsampled first to obtain the same number of samplings as the number of samplings of a relatively short edge. and There is no need for transmission, and it is obtained by calculating through formula (2) in the decoder; in the embodiments of the present application, this is not specifically limited.
[0026] FIG. 5 is an example of a linear model determining a prior art, and as shown in FIG. 5, a, b, and c are first image element adjacent reference values, A, B, and C are second image element adjacent reference values, e is the first image element reconstruction value of the current coding block, and E is the second image element prediction value of the current coding block; where, all first image element adjacent reference values of the current coding block ( ) and the adjacent reference value of the second image element ( Using ), according to Equation (2) and It can be obtained by calculating; and the first image element reconstruction value (e) of the current coding block can be input into a linear model according to Equation (1) to calculate and obtain the second image element prediction value (E) of the current coding block.
[0027] Specifically, in the prior art, the CCLM prediction mode includes a method for predicting a chromaticity element with a luminance element, i.e., predicting a second image element with a first image element or predicting a third image element with a first image element, in addition to including a prediction method between two chromaticity elements, i.e., a prediction method between a second image element and a third image element, wherein in the embodiment of the present application, not only can a Cr element be predicted from a Cb element, but a Cb element can also be predicted from a Cr element.
[0028] It should be explained that, in the embodiments of the present application, the prediction between the chromaticity elements in the CCLM, that is, the prediction between the second image element and the third image element, can be applied to the residual domain. Taking the prediction of the Cr element as an example, the Cr residual can be predicted using the Cb residual. The final predicted value of the Cr element is obtained by adding a reconstructed Cb residual with weights to the conventional intra-frame predicted value of the Cr element, as shown in Formula (3).
[0029] (3)
[0030] Here, The position coordinates among the above current coding blocks are It represents the final predicted value of the Cr component at the sampling point, is the predicted residual of the reconstructed Cb component, and the proportionality factor ( The calculation method of ) and the calculation method of prediction model parameters from the luminance component to the chromaticity component in CCLM are identical, and the only difference is the default in the error function. The proportional factor obtained by adding the regression cost associated with the value ( ) can be set to a default value biased toward -0.5, and specifically, the proportional factor can be calculated through formula (4).
[0031] (4)
[0032] Here, represents the adjacent reference Cb value of the current coding block above, and represents the adjacent reference Cr value of the current coding block above, and λ may be an empirical value, for example, λ=∑(Cb(n) Cb(n))≫9.
[0033] However, when the conventional technology calculates a computational linear model by predicting a second image element or a third image element based on a first image element, it considers only the first image element adjacent reference value, the second image element adjacent reference value, or the third image element adjacent reference value, and, for example, as shown in FIG. 5 above, all adjacent first image element adjacent reference values ( ) and all second image element adjacent reference values( Using the proportional factor ( and ) is calculated and obtained. Next, the first image element reconstruction value of the current coding block is substituted into a linear model to obtain the second image element prediction value of the current coding block. In this process, the correlation between the first image element reconstruction value of the current coding block and the first image element adjacent reference value is not considered, and if the linear model is modeled using a first image element adjacent reference value that deviates significantly from the current luminance value, it may result in the second image element prediction value or the third image element prediction value of the current coding block deviating significantly from the actual image element value, thereby reducing the prediction accuracy. For example, FIG. 6 is an example of determining a linear model among the prior art, and as shown in FIG. 6, pixel points (b, c) deviate from the reconstructed luminance value of the current coding block, and the linear relationship between the reconstructed luminance value and the corresponding chromatic value deviate from the current coding block and the corresponding chromatic value follows the calculation method of the prior art, and the two pixel points b and c can also be used in the calculation of the parameters of the linear model, so that the derived linear model deviates from the desired linear model.
[0034] In an embodiment of the present application, a linear model calculation method based on element association is proposed as a method for predicting video image elements. Specifically, when calculating the proportionality factor of the linear model, not only are the first image element adjacent reference value, the second image element adjacent reference value, or the third image element adjacent reference value considered, but the association and degree of similarity between the first image element reconstruction value of the current coding block and the first image element adjacent reference value are also considered. Then, by performing weighting or selection on adjacent reference pixel points according to the degree of similarity, the linear model parameters are made to approach the current first image element reconstruction value more closely, and furthermore, the second image element prediction value or the third image element prediction value of the current coding block is made more accurate.
[0035] The technical method in the embodiments of the present application is described below in conjunction with the drawings in the embodiments of the present application, clearly and completely. In each of the embodiments below, the first image element may be a luminance element (Y), the second image element may be a red chromaticity element (Cr), and the third image element may be a blue chromaticity element (Cb), and the embodiments of the present application are not specifically limited thereto.
[0036] In Example 1,
[0037] An embodiment of the present application provides a method for predicting video image elements, and FIG. 7 is an example flowchart of the implementation of the method for predicting video image elements provided in an embodiment of the present application. As illustrated in FIG. 7, in an embodiment of the present application, the method of the prediction device predicting video image elements may include the following steps.
[0038] In step 101, a first image element adjacent reference value and a first image element reconstruction value corresponding to the current coding block are obtained; wherein the first image element adjacent reference value is intended to represent a first image element parameter corresponding to an adjacent reference point of the current coding block, and the first image element reconstruction value is intended to represent one or more reconstruction parameters of the first image elements corresponding to the current coding block.
[0039] In an embodiment of the present application, the prediction device may first obtain a first image element adjacent reference value and a first image element reconstruction value corresponding to the current coding block. It is to be explained that in an embodiment of the present application, the current coding block is divided by the prediction device and is a coding block including at least one pixel point.
[0040] It needs to be explained that, in an embodiment of the present application, the first image element adjacent reference value is intended to represent a first image element parameter corresponding to an adjacent reference point of the current coding block. Here, in the case of a downsampling method, if the current coding block contains one pixel point, the corresponding adjacent reference point may be each pixel point above and to the left of the one pixel point; and if the current coding block contains multiple pixel points, the corresponding adjacent reference point may be multiple pixel points above and to the left of the multiple pixel points.
[0041] Furthermore, in an embodiment of the present application, the adjacent reference point is a pixel point adjacent to the current coding block, and specifically, in an embodiment of the present application, in the case of a downsampling method, the adjacent reference point may be an adjacent pixel point located to the left and above the current coding block.
[0042] It needs to be explained that, in an embodiment of the present application, when the prediction device obtains the first image element adjacent reference value, it may obtain all first image element adjacent reference values corresponding to the adjacent reference point. Specifically, in an embodiment of the present application, in the case of one first image element adjacent reference value corresponding to one adjacent reference point, that is, when the adjacent reference point is a plurality of pixel points, the prediction device may obtain a plurality of first image element adjacent reference values corresponding to the plurality of pixel points.
[0043] Furthermore, in an embodiment of the present application, the first image element reconstruction value is intended to represent one or more reconstruction parameters of the first image element corresponding to the current coding block; specifically, in an embodiment of the present application, if the current coding block contains one pixel point, the reconstruction parameter of the corresponding first image element is one; and if the current coding block contains multiple pixel points, the reconstruction parameter of the corresponding first image element is multiple.
[0044] It must be explained that, in an embodiment of the present application, when the prediction device obtains the first image element reconstruction value, it may obtain all first image element reconstruction values corresponding to the current coding block. Specifically, in an embodiment of the present application, one pixel point of the current coding block corresponds to one first image element reconstruction value.
[0045] In step 102, an association coefficient is determined according to the first image element adjacent reference value and the first image element reconstruction value; where the association coefficient is intended to indicate the degree of image element deviation between the current coding block and the adjacent reference point.
[0046] In an embodiment of the present application, the prediction device may obtain the first image element adjacent reference value and the first image element reconstructed value corresponding to the current coding block, and then additionally determine an association coefficient according to the first image element adjacent reference value and the first image element reconstructed value.
[0047] It should be explained that, in an embodiment of the present application, the association coefficient may be used to indicate the degree of image element deviation between the current coding block and the adjacent reference point, and specifically, the association coefficient may be used to indicate the association between the first image element reconstruction value corresponding to the current coding block and the first image element adjacent reference value corresponding to the adjacent reference point.
[0048] Furthermore, in an embodiment of the present application, when the prediction device determines the association coefficient according to the first image element adjacent reference value and the first image element reconstructed value, it can perform a difference value calculation for the first image element adjacent reference value and the first image element reconstructed value, thereby additionally determining the association coefficient according to the difference value result; and simultaneously, when the prediction device determines the association coefficient according to the first image element adjacent reference value and the first image element reconstructed value, it can perform a matrix multiplication calculation for the matrix corresponding to the first image element adjacent reference value and the matrix corresponding to the first image element reconstructed value, thereby additionally determining the association coefficient according to the multiplication result.
[0049] In step 103, the association coefficient is input into a pre-set weight calculation model to obtain the weight coefficient corresponding to the adjacent reference point.
[0050] In an embodiment of the present application, the prediction device can determine the association coefficient based on the first image element adjacent reference value and the first image element reconstructed value, and then input the association coefficient into a preset weight calculation model, thereby obtaining a weighting coefficient corresponding to the adjacent reference point.
[0051] What needs to be explained is that in the embodiments of the present application, the pre-set weight calculation model is a calculation model that is pre-set by the prediction device and is weighted and assigned according to the association coefficient.
[0052] Furthermore, in an embodiment of the present application, the weighting factor may be a weight calculated based on the association between the adjacent reference point and the current coding block when predicting a second image element prediction value corresponding to the current coding block.
[0053] It must be explained that, in the embodiments of the present application, the greater the association between the reference point at the adjacent reference point and the current coding block, the greater the corresponding weight corresponding to the reference point.
[0054] It needs to be explained that in an embodiment of the present application, when calculating the second image element prediction value, any one of the adjacent reference points has a corresponding weighting coefficient.
[0055] Furthermore, in an embodiment of the present application, when the prediction device determines the weighting coefficient, it can calculate it through the preset weighting calculation model, and specifically, in an embodiment of the present application, the weighting calculation model is as shown in formula (5).
[0056] (5)
[0057] Here, represents the adjacent reference values of the left and upper first image elements of downsampling, and is the current first image element reconstruction value, and σ is a value associated with the quantization parameter during coding.
[0058] In step 104, the proportionality factor is determined according to the weighting coefficient.
[0059] In an embodiment of the present application, the prediction device inputs the association coefficient into a preset weight calculation model to obtain the weighting coefficient corresponding to the adjacent reference point, and then determines a proportionality factor according to the weighting coefficient.
[0060] It needs to be explained that, in an embodiment of the present application, the proportionality factor may be a coefficient in a linear model obtained when the prediction device predicts video image elements.
[0061] Furthermore, in an embodiment of the present application, for different prediction modes, the proportionality factor is a different coefficient, and for example, when the preset device predicts a second image element or a third image element through a first image element, the proportionality factor is in the formula (1). and It may be; and when the above-mentioned preset device performs a prediction between the second image element and the third image element, the proportionality factor is in the above-mentioned formula (3). It could be.
[0062] In step 105, a second image element prediction value corresponding to the current coding block is obtained through a proportional factor.
[0063] In an embodiment of the present application, the prediction device may determine the proportionality factor according to the weighting coefficient and then additionally obtain a second image element prediction value corresponding to the current coding block through the proportionality factor. Here, the second image element prediction value is a prediction result obtained when the prediction device performs a second image element prediction for the current coding block.
[0064] It needs to be explained that, in an embodiment of the present application, when the prediction device obtains the second image element prediction value according to the proportionality factor, a linear model for predicting video image elements according to the proportionality factor can first be constructed.
[0065] Furthermore, in an embodiment of the present application, for different prediction modes, the linear model constructed by the prediction device is also different; that is, the linear model constructed when predicting a second image element or a third image element through a first image element and the linear model constructed when performing a prediction between the second image element and the third image element are not the same.
[0066] It must be explained that, in an embodiment of the present application, when predicting a second image element or a third image element through a first image element, the second image element prediction value obtained is a second image element or a third image element corresponding to the current coding block, i.e., a Cb element or a Cr element; and when performing a prediction between the second image element and the third image element, the second image element prediction value obtained is a second image element or a third image element corresponding to the current coding block, i.e., a Cb element or a Cr element.
[0067] Furthermore, in an embodiment of the present application, the prediction device can obtain the second image element prediction value by constructing the linear model according to the proportionality factor and then performing video image element prediction for the current coding block according to the linear model.
[0068] Furthermore, in an embodiment of the present application, there are currently two prediction modes of CCLM, namely, one is a prediction mode of a single-model CCLM; and the other is a prediction mode of a multiple-model CCLM (MMLM), also referred to as a prediction mode of MMLM. As the names suggest, the prediction mode of a single-model CCLM is one in which a single linear model performs prediction from a first image element to a second image element or to a third image element; and the prediction mode of MMLM is one in which various linear models perform prediction from a first image element to a second image element or to a third image element. For example, in the prediction mode of MMLM, the first image element adjacent reference value and the second image element adjacent reference value of the current coding block are divided into two groups, and each group can be used as a training set to independently derive linear model parameters, that is, each group can derive a proportional factor of one group. Accordingly, the method for predicting video image elements provided in the embodiment of the present application may be applied to the prediction mode of MMLM, and similarly, when there is a relatively large deviation between the first image element adjacent reference value and the parameter corresponding to the current coding block, or when there is a relatively large deviation between the third image element adjacent reference value and the parameter corresponding to the current coding block, the defect of the linear model and the desired model deviation can be overcome, thereby significantly improving the prediction accuracy of the second image element prediction value when predicting the second image element of the current coding block according to the linear model.
[0069] As can be seen from this, in an embodiment of the present application, the prediction device can determine an association coefficient based on a first image element adjacent reference value and a first image element reconstruction value corresponding to the current coding block, thereby enabling the distribution of different weighting coefficients for different adjacent reference points according to the association of element parameters between the adjacent reference point and the current coding block. By modeling a linear model that is more closely aligned with the desired model, defects in the deviation between the linear model and the desired model can be effectively overcome when there is a relatively large deviation between the first image element adjacent reference value and the element parameter corresponding to the current coding block, or when there is a relatively large deviation between the third image element adjacent reference value and the element parameter corresponding to the current coding block. Consequently, when predicting the elements of the current coding block according to the linear model, the prediction accuracy of the video image element prediction value can be significantly improved, thereby allowing the video image element prediction value to approach the actual video image element value more closely.
[0070] In Example 2,
[0071] Based on Example 1 above, in an embodiment of the present application, a method for determining an association coefficient according to the first image element adjacent reference value and the first image element reconstruction value of the prediction device may include the following steps.
[0072] In step 102a, a difference value operation is performed on each first image element reconstruction value using one of the first image element adjacent reference values, and an element difference value corresponding to one of the reference values is obtained; wherein one first image element adjacent reference value and one first image element reconstruction value correspond to one difference value.
[0073] In an embodiment of the present application, the prediction device obtains the first image element adjacent reference value and the first image element reconstructed value corresponding to the current coding block, and then performs a difference value calculation for each of the reference value among the first image element adjacent reference values and the first image element reconstructed value to obtain an element difference value corresponding to the one reference value.
[0074] It needs to be explained that in an embodiment of the present application, one reference value among the first image element adjacent reference values and one first image element reconstruction value among the first image element reconstruction values correspond to one difference value.
[0075] Furthermore, in an embodiment of the present application, the prediction device performs a difference value calculation for each of the first image element reconstruction values using any one of the first image element adjacent reference values, and this can indicate the degree of association of video image elements between the adjacent reference point and the current coding module.
[0076] It needs to be explained that in the embodiments of the present application, the larger the difference value of the element, the smaller the degree of association between the adjacent reference point and the current coding module can be considered.
[0077] In step 102b, the smallest difference value among the factor difference values is determined as the association coefficient.
[0078] In an embodiment of the present application, the prediction device may perform a difference value calculation for each first image element reconstruction value using any one of the first image element adjacent reference values, obtain the element difference value corresponding to any one of the reference values, and then determine the smallest difference value among the element difference values as the association coefficient.
[0079] It should be explained that, in an embodiment of the present application, the prediction device can determine the smallest difference value from the factor difference value and further determine the association coefficient according to the smallest difference value.
[0080] Furthermore, in an embodiment of the present application, since the degree of association between the adjacent reference point and the current coding module can be considered greater as the factor difference value is smaller, the prediction device can determine the association factor according to the smallest difference value among the factor difference values.
[0081] Furthermore, in an embodiment of the present application, the method of the prediction device predicting a video image element may further include the following steps, after the prediction device performs a difference value calculation for each of the reference values among the adjacent reference values of the first image element and the reconstruction value of the first image element to obtain an element difference value corresponding to any one of the reference values, and before determining a proportionality factor according to the weighting coefficient.
[0082] In step 106, if the difference values of the elements corresponding to any one reference value are all greater than the preset difference value threshold, the weighting coefficient of the adjacent reference point corresponding to any one reference value is set to 0.
[0083] In an embodiment of the present application, the prediction device performs a difference value calculation for each of the reference values among the first image element adjacent reference values and the first image element reconstruction value, and obtains the element difference value corresponding to the one reference value, and before determining the proportionality factor according to the weighting coefficient, if the element difference value corresponding to the one reference value is all greater than a preset difference value threshold, the prediction device may set the weighting coefficient of the adjacent reference point corresponding to the one reference value to 0.
[0084] What needs to be explained is that in an embodiment of the present invention, the prediction device can obtain the element difference value corresponding to any one reference value and then compare the element difference value with the preset difference value threshold. If all of the element difference values corresponding to any one reference value are greater than the preset difference value threshold, it can be considered that there is a relatively large deviation between any one first image element adjacent reference value and the current coding block. Therefore, the prediction device can remove the adjacent reference point corresponding to any one first image element adjacent reference value, that is, the weighting coefficient of the adjacent reference point corresponding to any one reference value can be set to 0.
[0085] Furthermore, in an embodiment of the present invention, the prediction device may consider the correlation between the first image element reconstruction value of the current coding block and the first image element adjacent reference value before calculating a linear model. That is, the first image element adjacent reference value is selected based on the difference between the first image element adjacent reference value and the first image element reconstruction value.
[0086] Furthermore, in an embodiment of the present invention, the prediction device may pre-set a threshold value, namely the preset difference value threshold, and then iterate through each reference value among the first image element adjacent reference values to obtain the difference value between each of the first image element reconstruction values in the current coding block. If the minimum value of the difference between a specific first image element adjacent reference value and each of the first image element reconstruction values is greater than the preset difference value threshold, the prediction device may consider the adjacent reference point as an interference point for the linear model that calculates precision, and thus can remove it from the training samples of the linear model. In this way, the model parameters can be calculated by using the first image element adjacent reference value after the removal of the interference point as the training samples of the linear model.
[0087] In an embodiment of the present invention, FIG. 8 is an example diagram of removing interference points among embodiments of the present application, and as shown in FIG. 8, when some first image element values among the first image element adjacent reference values of the current coding block deviate from the first image element reconstruction values of the current coding block, when calculating a linear model, said points are removed and not used as training samples for the linear model, so the linear model calculated at this time is more accurate.
[0088] As can be seen from this, in an embodiment of the present application, the prediction device can determine an association coefficient based on a first image element adjacent reference value and a first image element reconstruction value corresponding to the current coding block, thereby enabling the distribution of different weighting coefficients for different adjacent reference points according to the association of element parameters between the adjacent reference point and the current coding block. By modeling a linear model that is more closely aligned with the desired model, defects in the deviation between the linear model and the desired model can be effectively overcome when there is a relatively large deviation between the first image element adjacent reference value and the element parameter corresponding to the current coding block, or when there is a relatively large deviation between the third image element adjacent reference value and the element parameter corresponding to the current coding block. Consequently, when predicting the elements of the current coding block according to the linear model, the prediction accuracy of the video image element prediction value can be significantly improved, thereby allowing the video image element prediction value to approach the actual video image element value more closely.
[0089] In Example 3,
[0090] Based on Example 1 above, in an embodiment of the present application, the method for the prediction device to predict video image elements before the prediction device determines a proportionality factor according to the weighting coefficient, i.e., before step 104, may further include the following steps.
[0091] In step 107, a second image element adjacent reference value corresponding to the current coding block is obtained; where the second image element adjacent reference value corresponds to an adjacent reference point and is a second image element parameter different from the first image element parameter.
[0092] In an embodiment of the present application, the prediction device may first obtain a second image element adjacent reference value corresponding to the current coding block before determining the proportionality factor according to the weighting coefficient.
[0093] It needs to be explained that, in an embodiment of the present application, the second image element adjacent reference value corresponds to the adjacent reference point and is a second image element parameter different from the first image element parameter.
[0094] In an embodiment of the present application, the proportionality factor includes a first proportionality parameter and a second proportionality parameter, and the method of the prediction device determining the proportionality factor according to the weighting coefficient may include the following steps.
[0095] In step 104a, the weighting coefficient, the first image element adjacent reference value, and the second image element adjacent reference value are input into the first preset factor calculation model to obtain the first proportional parameter.
[0096] In an embodiment of the present invention, the prediction device inputs the association coefficient into a preset weight calculation model to obtain the weighting coefficient corresponding to the first image element adjacent reference value, and then inputs the weighting coefficient, the first image element adjacent reference value, and the second image element adjacent reference value into a first preset factor calculation model to obtain the first proportional parameter.
[0097] It needs to be explained that, in an embodiment of the present invention, when the preset device predicts a second image element through a first image element, the proportionality factor may include the first proportionality parameter and the second proportionality parameter, wherein the first proportionality parameter and the second proportionality parameter are intended to model the second image element prediction value according to the first image element reconstruction value.
[0098] Furthermore, in an embodiment of the present application, when the prediction device additionally determines the calculation of the first proportional parameter according to the first image element adjacent reference value and the second image element adjacent reference value, the available first pre-set factor calculation model is as shown in formula (6).
[0099] (6)
[0100] Here, is the first proportionality parameter, and represents the adjacent reference values of the left and upper first image elements of downsampling, and represents the adjacent reference values of the left and upper second image elements, and is a weighting coefficient corresponding to the adjacent reference value of each first image element.
[0101] In step 104b, the weighting coefficient, the first proportionality parameter, the first image element adjacent reference value, and the second image element adjacent reference value are input into the second preset factor calculation model to obtain the second proportionality parameter.
[0102] In an embodiment of the present application, the prediction device inputs the weighting coefficient, the first image element adjacent reference value, and the second image element adjacent reference value into the first preset factor calculation model to obtain the first proportional parameter, and then can continue to input the weighting coefficient, the first proportional parameter, the first image element adjacent reference value, and the second image element adjacent reference value into the second preset factor calculation model to obtain the second proportional parameter.
[0103] Furthermore, in an embodiment of the present application, when the prediction device additionally determines the calculation of the second proportional parameter according to the first image element adjacent reference value and the second image element adjacent reference value, the available second pre-set factor calculation model is as shown in formula (7).
[0104] (7)
[0105] Here, is the second proportionality parameter, and represents the adjacent reference values of the left and upper first image elements of downsampling, and represents the adjacent reference values of the left and upper second image elements, and is the corresponding weighting coefficient for each adjacent reference point.
[0106] In an embodiment of the present application, furthermore, before the prediction device determines a proportionality factor according to the weighting coefficient, i.e., before step 104, the method of the prediction device predicting video image elements may further include the following steps.
[0107] In step 108, a second image element adjacent reference value and a third image element adjacent reference value corresponding to the current coding block are obtained; wherein the second image element adjacent reference value and the third image element adjacent reference value represent the second image element parameter and the third image element parameter of the adjacent reference point, respectively.
[0108] In an embodiment of the present application, the prediction device may first obtain a second image element adjacent reference value and a third image element adjacent reference value corresponding to the current coding block before determining the proportionality factor according to the weighting coefficient.
[0109] It must be explained that, in the embodiments of the present application, the second image element adjacent reference value and the third image element adjacent reference value represent the second image element parameter and the third image element parameter of the adjacent reference point, respectively. Specifically, the third image element adjacent reference value and the second image element adjacent reference value can be used for prediction between identical elements.
[0110] Furthermore, in an embodiment of the present application, the third image element adjacent reference value and the second image element adjacent reference value may each be an adjacent reference Cb value and an adjacent reference Cr value corresponding to the adjacent reference point.
[0111] In an embodiment of the present application, furthermore, before the prediction device obtains a second image element prediction value corresponding to the current coding block through the proportional factor, i.e., before step 105, the method of the prediction device predicting a video image element may further include the following steps.
[0112] In step 109, a second image element estimate and a third image element reconstruction residual corresponding to the current coding block are obtained.
[0113] In an embodiment of the present application, the prediction device may first obtain a second image element estimate and a third image element reconstruction residual corresponding to the current coding block according to the proportionality factor before obtaining the second image element prediction value corresponding to the current coding block.
[0114] It must be explained that, in an embodiment of the present application, the second image element estimate is obtained by performing a conventional element prediction according to the second image element corresponding to the current coding block; and the third image element reconstruction residual is intended to represent the prediction residual of the third image element corresponding to the current coding block.
[0115] In an embodiment of the present application, furthermore, the proportionality factor includes a third proportionality parameter, and the method of the prediction device determining the proportionality factor according to the weighting coefficient may include the following steps.
[0116] In step 104c, the weighting coefficient, the third image element adjacent reference value, and the second image element adjacent reference value are input into the third preset factor calculation model to obtain the third proportional parameter.
[0117] In an embodiment of the present invention, the prediction device inputs the association coefficient into a preset weight calculation model to obtain the weight coefficient corresponding to the adjacent reference point, and then inputs the weight coefficient, the third image element adjacent reference value, and the second image element adjacent reference value into a third preset factor calculation model to obtain the third proportional parameter.
[0118] It needs to be explained that, in an embodiment of the present invention, when the preset device performs a prediction between a second image element and a third image element, the proportionality factor may include the third proportionality parameter.
[0119] Furthermore, in an embodiment of the present application, when the prediction device additionally determines the calculation of the third proportional parameter according to the third image element adjacent reference value and the second image element adjacent reference value, the available third pre-set factor calculation model is as shown in formula (8).
[0120] (8)
[0121] Here, is the third proportionality parameter, and represents the adjacent reference Cb value of the current coding block, i.e., the adjacent reference value of the third image element; represents the adjacent reference Cr value of the current coding block, i.e., the adjacent reference value of the second image element; and λ may be a single empirical value, is the corresponding weighting coefficient for each adjacent reference point.
[0122] As can be seen from this, in an embodiment of the present application, the prediction device can determine an association coefficient based on a first image element adjacent reference value and a first image element reconstruction value corresponding to the current coding block, thereby enabling the distribution of different weighting coefficients for different adjacent reference points according to the association of element parameters between the adjacent reference point and the current coding block. By modeling a linear model that is more closely aligned with the desired model, defects in the deviation between the linear model and the desired model can be effectively overcome when there is a relatively large deviation between the first image element adjacent reference value and the element parameter corresponding to the current coding block, or when there is a relatively large deviation between the third image element adjacent reference value and the element parameter corresponding to the current coding block. Consequently, when predicting the elements of the current coding block according to the linear model, the prediction accuracy of the video image element prediction value can be significantly improved, thereby allowing the video image element prediction value to approach the actual video image element value more closely.
[0123] Example 4
[0124] Based on Example 3 above, in an embodiment of the present application, the method for the prediction device to obtain a second image element prediction value corresponding to the current coding block through the proportional factor may include the following steps.
[0125] In step 105a, a second image element prediction value is obtained according to the first proportional parameter, the second proportional parameter, and the first image element reconstruction value.
[0126] In an embodiment of the present application, the prediction device determines the proportionality factor according to the weighting coefficient, and then further obtains the second image element prediction value according to the first proportionality parameter, the proportionality factor including the second proportionality parameter, and the first image element reconstruction value.
[0127] It needs to be explained that, in an embodiment of the present application, the prediction device can obtain a proportional factor including the first proportional parameter and the second proportional parameter, and then construct a linear model for predicting video image elements according to the first proportional parameter, the second proportional parameter and the first image element reconstruction value, thereby modeling the second image element of the current coding block according to the linear model and obtaining the second image element prediction value.
[0128] Furthermore, in an embodiment of the present application, the prediction device can determine the second image element prediction value through the formula (1), that is, the prediction device can calculate and obtain the second image element prediction value according to the first proportional parameter, the second proportional parameter, and the first image element reconstruction value.
[0129] In an embodiment of the present application, the method of the prediction device obtaining a second image element prediction value corresponding to the current coding block through the proportional factor may further include the following steps.
[0130] In step 105b, a second image element prediction value is obtained based on the third proportional parameter, the second image element estimate value, and the third image element reconstruction residual.
[0131] In an embodiment of the present application, the prediction device determines the proportionality factor according to the weighting coefficient, and then further obtains the second image element prediction value according to the proportionality factor including the third proportionality parameter, the second image element estimate value, and the third image element reconstruction residual.
[0132] It needs to be explained that, in an embodiment of the present application, the prediction device can obtain a proportional factor including the third proportional parameter, and then construct a linear model for video image element prediction according to the third proportional parameter, the second image element estimate value, and the third image element reconstruction residual, thereby modeling the second image element of the current coding block according to the linear model to obtain the second image element prediction value.
[0133] Furthermore, in an embodiment of the present application, the prediction device can determine the second image element prediction value through the formula (3), that is, the prediction device can obtain the second image element prediction value by calculating according to the third proportional parameter, the second image element estimate value, and the third image element reconstruction residual.
[0134] As can be seen from this, in an embodiment of the present application, the prediction device can determine an association coefficient based on a first image element adjacent reference value and a first image element reconstruction value corresponding to the current coding block, thereby enabling the distribution of different weighting coefficients for different adjacent reference points according to the association of element parameters between the adjacent reference point and the current coding block. By modeling a linear model that is more closely aligned with the desired model, defects in the deviation between the linear model and the desired model can be effectively overcome when there is a relatively large deviation between the first image element adjacent reference value and the element parameter corresponding to the current coding block, or when there is a relatively large deviation between the third image element adjacent reference value and the element parameter corresponding to the current coding block. Consequently, when predicting the elements of the current coding block according to the linear model, the prediction accuracy of the video image element prediction value can be significantly improved, thereby allowing the video image element prediction value to approach the actual video image element value more closely.
[0135] Example 5
[0136] Based on the same inventive concept of Examples 1 to 4 above, FIG. 9 is an example diagram 1 of the configuration structure of a prediction device provided in the embodiment of the present application, and as shown in FIG. 9, the prediction device (1) provided in the embodiment of the present application may include an acquisition part (11), a determination part (12), and a setting part (13).
[0137] The acquisition portion (11) is configured to acquire a first image element adjacent reference value and a first image element reconstruction value corresponding to the current coding block; wherein the first image element adjacent reference value is intended to represent a first image element parameter corresponding to an adjacent reference point of the current coding block, and the first image element reconstruction value is intended to represent one or more reconstruction parameters of first image elements corresponding to the current coding block.
[0138] The determination part (12) is configured to determine an association coefficient according to the first image element adjacent reference value and the first image element reconstructed value after the acquisition part (11) acquires the first image element adjacent reference value and the first image element reconstructed value corresponding to the current coding block; wherein the association coefficient is intended to indicate the degree of deviation of the image element between the current coding block and the adjacent reference point.
[0139] The acquisition part (11) is also configured such that the determination part (12) determines an association coefficient based on the first image element adjacent reference value and the first image element reconstruction value, and then inputs the association coefficient into a preset weight calculation model to obtain a weighting coefficient corresponding to the adjacent reference point.
[0140] The determination part (12) is also configured such that the acquisition part (11) inputs the association coefficient into a preset weight calculation model to obtain a weighting coefficient corresponding to the adjacent reference point, and then determines a proportionality factor according to the weighting coefficient.
[0141] The acquisition part (11) is also configured such that the determination part (12) determines a proportionality factor according to the weighting factor, and then obtains a second image element prediction value corresponding to the current coding block through the proportionality factor.
[0142] Furthermore, in an embodiment of the present application, the determination part (12) is specifically configured to perform a difference value calculation for each of the reference values among the first image element adjacent reference values and the first image element reconstruction value to obtain an element difference value corresponding to the one reference value - one first image element adjacent reference value and one first image element reconstruction value correspond to one difference value -; and to determine the smallest difference value among the element difference values as the association coefficient.
[0143] Furthermore, in an embodiment of the present application, the acquisition portion (11) is also configured to acquire a second image element adjacent reference value corresponding to the current coding block before determining a proportionality factor according to the weighting coefficient; wherein the second image element adjacent reference value corresponds to the adjacent reference point and is a second image element parameter different from the first image element parameter.
[0144] Furthermore, in an embodiment of the present application, the proportionality factor includes a first proportionality parameter and a second proportionality parameter, and the determination part (12) is also specifically configured to obtain the first proportionality parameter by inputting the weighting coefficient, the first image element adjacent reference value, and the second image element adjacent reference value into a first preset factor calculation model; and to obtain the second proportionality parameter by inputting the weighting coefficient, the first proportionality parameter, the first image element adjacent reference value, and the second image element adjacent reference value into a second preset factor calculation model.
[0145] Furthermore, in an embodiment of the present application, the acquisition portion (11) is also configured to acquire a third image element adjacent reference value and a second image element adjacent reference value corresponding to the current coding block before determining a proportionality factor according to the weighting coefficient; wherein the second image element adjacent reference value and the third image element adjacent reference value represent the second image element parameter and the third image element parameter of the adjacent reference point, respectively.
[0146] Furthermore, in an embodiment of the present application, the acquisition portion (11) is also configured to acquire a second image element estimate and a third image element reconstruction residual corresponding to the current coding block before acquiring a second image element prediction value corresponding to the current coding block through the proportional factor; wherein the second image element estimate is obtained by performing element prediction according to the second image element corresponding to the current coding block; and the third image element reconstruction residual is intended to represent a prediction residual corresponding to the current coding block.
[0147] Furthermore, in an embodiment of the present application, the proportionality factor includes a third proportionality parameter, and the determination part (12) is also specifically configured to obtain the third proportionality parameter by inputting the weighting coefficient, the second image element adjacent reference value, and the third image element adjacent reference value into a third preset factor calculation model.
[0148] Furthermore, in an embodiment of the present application, the acquisition portion (11) is specifically configured to acquire the second image element prediction value according to the first proportional parameter, the second proportional parameter, and the first image element reconstruction value.
[0149] Furthermore, in an embodiment of the present application, the acquisition portion (11) is also specifically configured to acquire the second image element prediction value according to the third proportional parameter, the second image element estimate value, and the third image element reconstruction residual.
[0150] Furthermore, in an embodiment of the present application, the setting portion (13) is configured to perform a difference value calculation for each of the reference values among the first image element adjacent reference values and the first image element reconstruction value, and to obtain an element difference value corresponding to the one reference value, and then, before determining a proportionality factor according to the weighting coefficient, if the element difference value corresponding to the one reference value is all greater than a preset difference value threshold, the weighting coefficient of the adjacent reference point corresponding to the one reference value is set to 0.
[0151] FIG. 10 is an example diagram 2 of the configuration structure of a prediction device provided in an embodiment of the present application, and as shown in FIG. 10, the prediction device (1) provided in an embodiment of the present application may further include a processor (14), a memory (15) that stores instructions executable by the processor (14), a communication interface (16), and a bus (17) connecting the processor (14), the memory (15), and the communication interface (16).
[0152] In an embodiment of the present application, the processor (14) may be at least one of an Application Specific Integrated Circuit (ASIC), a Digital Signal Processor (DSP), a Digital Signal Processing Device (DSPD), a Programmable Logic Device (PLD), a Field Programmable Gate Array (FPGA), a Central Processing Unit (CPU), a controller, a microcontroller, and a microprocessor. It is understood that for different devices, the electronic components for implementing the processor function may be different, and the embodiments of the present application are not specifically limited. The device (1) may further include memory (15), and the memory (15) may be connected to a processor (14), wherein the memory (15) is for storing executable program code, and the program code includes computer operation instructions, and the memory (15) may include high-speed RAM memory, may include non-volatile memory, and, for example, may include at least two magnetic disk memories.
[0153] In an embodiment of the present application, the bus (17) is for connecting the communication interface (16), the processor (14), and the memory (15) and the mutual communication between these elements.
[0154] In an embodiment of the present application, the memory (15) is for storing instructions and data.
[0155] Furthermore, in an embodiment of the present application, the processor (14) obtains a first image element adjacent reference value and a first image element reconstruction value corresponding to a current coding block - the first image element adjacent reference value is intended to represent a first image element parameter corresponding to an adjacent reference point of the current coding block, and the first image element reconstruction value is intended to represent one or more reconstruction parameters of first image elements corresponding to the current coding block -; determines an association coefficient according to the first image element adjacent reference value and the first image element reconstruction value - the association coefficient is intended to represent the degree of image element deviation between the current coding block and the adjacent reference point -; inputs the association coefficient into a preset weight calculation model to obtain a weighting coefficient corresponding to an adjacent reference point; determines a proportionality factor according to the weighting coefficient; and obtains a second image element prediction value corresponding to the current coding block through the proportionality factor.
[0156] In actual application, the memory (15) may be a volatile memory such as Random-Access Memory (RAM); or a non-volatile memory such as Read-Only Memory (ROM), flash memory, Hard Disk Drive (HDD), or Solid-State Drive (SSD); or a combination of the first memory of the above categories, and provides instructions and data to the processor (14).
[0157] Additionally, each functional module in the present embodiment may be integrated into a single process unit, or each unit may be an independent physical entity, or two or more units may be integrated into a single unit. The integrated unit may be implemented using hardware or may be implemented using software functional modules.
[0158] When the integrated unit is implemented in the form of a software function module rather than being sold or used as an independent product, it may be stored on a single computer-readable storage medium. Based on this understanding, the technical solution of the present embodiment, or parts that contribute to existing technology, or all or part of the said technical solution, may be implemented in the form of a software product. The computer software product is stored on a single storage medium and includes several instructions that cause a computer device (such as a personal computer, server, or network device) or processor to execute all or part of the steps of the method of the present embodiment. The aforementioned storage medium includes various media capable of storing program code, such as U disks, mobile disks, Read-Only Memory (ROM), Random Access Memory (RAM), magnetic disks, or optical disks.
[0159] In the device provided in the embodiment of the present application, the prediction device can determine an association coefficient based on a first image element adjacent reference value and a first image element reconstruction value corresponding to the current coding block. This enables the distribution of different weighting coefficients for different adjacent reference points according to the association of element parameters between the adjacent reference point and the current coding block. By modeling a linear model that is more closely aligned with the desired model, defects in the deviation between the linear model and the desired model can be effectively overcome when there is a relatively large deviation between the first image element adjacent reference value and the element parameter corresponding to the current coding block, or when there is a relatively large deviation between the third image element adjacent reference value and the element parameter corresponding to the current coding block. Consequently, when predicting the elements of the current coding block according to the linear model, the prediction accuracy of the video image element prediction value can be significantly improved, thereby allowing the video image element prediction value to approach the actual video image element value more closely.
[0160] An embodiment of the present application provides a first computer-readable storage medium in which a program is stored, and when said program is executed by a processor, implements a method such as embodiments 1 to 4.
[0161] Specifically, the program instruction corresponding to the method for predicting video image elements in the present embodiment may be stored on a storage medium such as an optical disk, a hard disk, or a U disk, and when the program corresponding to the method for predicting video image elements on the storage medium is read, accessed, or executed by an electronic device,
[0162] A step of obtaining a first image element adjacent reference value and a first image element reconstruction value corresponding to a current coding block - the first image element adjacent reference value is intended to represent a first image element parameter corresponding to an adjacent reference point of the current coding block, and the first image element reconstruction value is intended to represent a reconstruction parameter of one or more first image elements corresponding to the current coding block - ;
[0163] A step of determining an association coefficient according to the first image element adjacent reference value and the first image element reconstruction value - the association coefficient is intended to indicate the degree of deviation of the image element between the current coding block and the adjacent reference point - ;
[0164] A step of inputting the above association coefficient into a preset weight calculation model to obtain a weight coefficient corresponding to the above adjacent reference point;
[0165] A step of determining a proportionality factor according to the above weighting coefficient;
[0166] It includes the step of obtaining a second image element prediction value corresponding to the current coding block through the proportional factor above.
[0167] A person skilled in the art should know that the embodiments of the present application may be provided as methods, systems, or computer program products. Accordingly, the present application may adopt a form of hardware embodiments, software embodiments, or a combination of embodiments in terms of software and hardware. Additionally, the present application may adopt a form of a computer program product executed on one or more computer-usable storage media (such as magnetic disk storage devices and optical memory) containing computer-usable program code.
[0168] The present application is described according to at least one of the flowchart and frame diagram of the method, apparatus (system), computer program product of the embodiment of the present application. At least one of each flow and frame in the flowchart and frame diagram, and at least one of the combination of flows and frames in the flowchart and frame diagram can be implemented by computer program instructions. To create a machine, such computer program instructions can be provided to a processor of a general-purpose computer, a dedicated computer, an embedded processor, or other programmable data processing device, so that instructions executed through the processor of the computer or other programmable data processing device create to implement at least one flow of the flowchart or multiple flows and at least one frame of the frame diagram.
[0169] These computer program instructions can be stored in computer-readable memory capable of inducing a computer or other programmable data processing device to operate in a specific manner, so that instructions within the instructions stored in said computer-readable memory produce a manufactured product including an instruction device, said instruction device implements a designated function among one flow or multiple flows of a flow example diagram and one frame or multiple frames of a frame diagram.
[0170] These computer program instructions are loaded into a computer or other programmable data processing device and, by performing a series of operations on the computer or other programmable device to implement processing of the computer implementation, the instructions executed on the computer or other programmable device provide steps for implementing a specified function in at least one of one flow or multiple flows of a flow example diagram and one frame of a frame diagram.
[0171] The foregoing description is merely a preferred embodiment of the present application and is not intended to limit the scope of protection of the present application. Industrial applicability
[0172] Embodiments of the present application provide a method and apparatus for predicting video image elements and a computer storage medium. The prediction apparatus can determine an association coefficient based on a first image element adjacent reference value and a first image element reconstruction value corresponding to a current coding block. This enables the distribution of different weighting coefficients for different adjacent reference points according to the association of element parameters between an adjacent reference point and a current coding block. By modeling a linear model that is more closely aligned with the desired model, defects in the deviation between the linear model and the desired model can be effectively overcome when there is a relatively large deviation between the first image element adjacent reference value and the element parameter corresponding to the current coding block, or when there is a relatively large deviation between the third image element adjacent reference value and the element parameter corresponding to the current coding block. Consequently, when predicting the elements of the current coding block according to the linear model, the prediction accuracy of the video image element prediction value can be significantly improved, thereby allowing the video image element prediction value to approach the actual video image element value more closely.
Claims
Claim 1 A method for predicting video image elements, applied to an encoder, comprising the step of obtaining a first image element adjacent reference value and a first image element reconstructed value corresponding to a current coding block - wherein the first image element adjacent reference value is intended to represent a first image element parameter corresponding to an adjacent reference point of the current coding block, and the first image element reconstructed value is intended to represent a reconstructed value of one or more first image elements corresponding to the current coding block -; the step of determining a weighting coefficient corresponding to an adjacent reference point according to the current coding block and the adjacent reference point; the step of determining a proportional factor using a preset factor calculation model according to the first image element adjacent reference value, the first image element reconstructed value, and the weighting coefficient; and the step of obtaining a second image element predicted value corresponding to the current coding block based on a linear model corresponding to the proportional factor; wherein the method for predicting video image elements comprises the step of performing selection processing on the adjacent reference point; and the step of determining a partial reference point by performing downsampling processing on the adjacent reference point. A method for predicting video image elements characterized by including at least one more of the following. Claim 2 A method for predicting video image elements according to claim 1, wherein the step of determining a weighting coefficient corresponding to an adjacent reference point according to the current coding block and the adjacent reference point comprises the step of determining a weighting coefficient corresponding to the adjacent reference point according to the association between the current coding block and the adjacent reference point—the association is intended to represent a positional association between the current coding block and the adjacent reference point. Claim 3 A method for predicting a video image element according to claim 1, wherein, before determining a proportionality factor according to the first image element adjacent reference value, the first image element reconstructed value, and the weighting coefficient, the method further comprises the step of obtaining a second image element adjacent reference value corresponding to the current coding block - wherein the second image element adjacent reference value corresponds to the adjacent reference point and is a second image element parameter different from the first image element parameter. Claim 4 A method for predicting video image elements according to claim 3, wherein the step of determining a proportionality factor using a preset factor calculation model according to the first image element adjacent reference value, the first image element reconstructed value, and the weighting coefficient includes the step of inputting the weighting coefficient, the first image element adjacent reference value, and the second image element adjacent reference value into the preset factor calculation model to obtain the proportionality factor. Claim 5 In claim 3, the first image element adjacent reference value and the first image element reconstructed value corresponding to the current coding block include at least two groups of first image element adjacent reference values and first image element reconstructed values, and the method for predicting video image elements further comprises: a step of determining a weighting coefficient for each group of first image element adjacent reference values and first image element reconstructed values; a step of inputting the weighting coefficient, the first image element adjacent reference value, and the second image element adjacent reference value into a preset factor calculation model to obtain at least two groups of proportional factors corresponding to at least two groups of first image element adjacent reference values and first image element reconstructed values; and a step of selecting one group of proportional factors from the at least two groups of proportional factors. Claim 6 A method for predicting video image elements according to claim 5, wherein the method further comprises the step of calculating the rate distortion cost result of each group of proportional factors in at least two groups of proportional factors; and the step of selecting one group of proportional factors corresponding to the minimum rate distortion cost result. Claim 7 A method for predicting video image elements according to claim 3, wherein, before obtaining a second image element prediction value corresponding to the current coding block based on a linear model corresponding to the proportional factor, the method for predicting video image elements further comprises the step of obtaining a second image element estimate value corresponding to the current coding block—the second image element estimate value being obtained by performing element prediction according to the second image element corresponding to the current coding block. Claim 8 A method for predicting video image elements according to claim 4 or 5, wherein the step of obtaining a second image element prediction value corresponding to the current coding block based on a linear model corresponding to the proportional factor comprises the step of obtaining the second image element prediction value according to the linear model corresponding to the proportional factor and the first image element reconstruction value. Claim 9 A method for predicting video image elements according to claim 7, wherein the step of obtaining a second image element prediction value corresponding to the current coding block based on a linear model corresponding to the proportional factor comprises the step of obtaining the second image element prediction value according to the proportional factor and the second image element estimate value. Claim 10 A method for predicting a video image element according to paragraph 2, wherein the method for predicting the image element further comprises the step of performing preprocessing on the adjacent reference point. Claim 11 A method for predicting a video image element according to claim 10, further comprising the step of setting the weighting coefficient of a partial reference point corresponding to a partial adjacent reference value to 0—where the positional association between the partial reference point and the current coding block does not satisfy a preset threshold value. Claim 12 A method for predicting a video image element according to claim 10, wherein the method further comprises the step of removing a partial reference point from the adjacent reference point—that is, the positional association between the partial reference point and the current coding block does not satisfy a preset threshold. Claim 13 A method for predicting a video image element according to claim 1, wherein the adjacent reference point includes at least one of the upper pixel point of the current coding block, the upper left pixel point of the current coding block, and the left pixel point of the current coding block. Claim 14 A method for predicting video image elements according to claim 13, wherein the adjacent reference point further includes at least one of the upper right pixel point of the current coding block and the lower left pixel point of the current coding block. Claim 15 A method for predicting video image elements according to claim 1, wherein the proportionality factor is a coefficient of the linear model, and the linear model is for predicting the second image element according to the first image element. Claim 16 A method for predicting video image elements, applied to a decoder, comprising the step of obtaining a first image element adjacent reference value and a first image element reconstructed value corresponding to a current block - wherein the first image element adjacent reference value is intended to represent a first image element parameter corresponding to an adjacent reference point of the current block, and the first image element reconstructed value is intended to represent one or more first image element reconstructed values corresponding to the current block -; the step of determining a weighting coefficient corresponding to an adjacent reference point according to the current block and the adjacent reference point; the step of determining a proportional factor using a preset factor calculation model according to the first image element adjacent reference value, the first image element reconstructed value, and the weighting coefficient; and the step of obtaining a second image element prediction value corresponding to the current block based on a linear model corresponding to the proportional factor, wherein the method for predicting video image elements comprises the step of performing selection processing on the adjacent reference point; and the step of determining a partial reference point by performing downsampling processing on the adjacent reference point. A method for predicting video image elements characterized by including at least one more of the following. Claim 17 A method for predicting video image elements according to claim 16, wherein the step of determining a weighting coefficient corresponding to the adjacent reference point according to the current block and the adjacent reference point further comprises the step of determining a weighting coefficient corresponding to the adjacent reference point according to the association between the current block and the adjacent reference point—wherein the association is intended to represent the positional association between the current block and the adjacent reference point. Claim 18 In claim 16, the method for predicting a video image element, prior to determining a proportionality factor according to the first image element adjacent reference value, the first image element reconstructed value, and the weighting coefficient, further comprises the step of obtaining a second image element adjacent reference value corresponding to the current block - wherein the second image element adjacent reference value corresponds to the adjacent reference point and is a second image element parameter different from the first image element parameter. Claim 19 A method for predicting video image elements according to claim 18, wherein the step of determining a proportionality factor using a preset factor calculation model according to the first image element adjacent reference value, the first image element reconstructed value, and the weighting coefficient comprises the step of inputting the weighting coefficient, the first image element adjacent reference value, and the second image element adjacent reference value into the preset factor calculation model to obtain the proportionality factor. Claim 20 In claim 18, the method for predicting video image elements further comprises, before obtaining a second image element prediction value corresponding to the current block based on a linear model corresponding to the proportional factor, a step of obtaining a second image element estimate value corresponding to the current block—the second image element estimate value being obtained by performing element prediction according to the second image element corresponding to the current block. Claim 21 A method for predicting video image elements according to claim 19, wherein the step of obtaining a second image element prediction value corresponding to the current block based on a linear model corresponding to the proportional factor comprises the step of obtaining the second image element prediction value according to the linear model corresponding to the proportional factor and the first image element reconstruction value. Claim 22 A method for predicting video image elements according to claim 20, wherein the step of obtaining a second image element prediction value corresponding to the current block based on a linear model corresponding to the proportional factor comprises the step of obtaining the second image element prediction value according to the proportional factor and the second image element estimate value. Claim 23 A method for predicting video image elements according to claim 17, characterized in that the method further comprises the step of performing preprocessing on the adjacent reference point. Claim 24 A method for predicting a video image element according to claim 23, characterized in that the method comprises the step of setting the weighting coefficient of a partial reference point corresponding to a partial adjacent reference value to 0 - wherein the positional association between the partial reference point and the current block does not satisfy a preset threshold value. Claim 25 A method for predicting a video image element according to claim 23, wherein the method further comprises the step of removing a partial reference point from the adjacent reference point—that is, the positional association between the partial reference point and the current block does not satisfy a preset threshold. Claim 26 A method for predicting video image elements according to claim 16, wherein the adjacent reference point comprises at least one of the upper pixel point of the current block, the upper left pixel point of the current block, and the left pixel point of the current block. Claim 27 A method for predicting video image elements according to claim 26, wherein the adjacent reference point further includes at least one of the upper right pixel point of the current block and the lower left pixel point of the current block. Claim 28 A method for predicting video image elements according to claim 16, characterized in that the proportionality factor is a coefficient of the linear model, and the linear model is for predicting the second image element according to the first image element. Claim 29 A method for transmitting a bit stream, characterized by generating a bit stream by executing a method for predicting video image elements according to any one of claims 1 to 15, and transmitting said bit stream.