Intra-frame prediction method and device, electronic equipment and computer program product

Abstract

The invention provides an intra-frame prediction method and device, electronic equipment and a computer program product, and relates to the technical field of video coding. The method comprises: obtaining an intra-frame prediction coding logic corresponding to a pixel to be predicted, and determining a component to be operated of a target instruction set in the intra-frame prediction coding logic, the component to be operated comprising one or more of an initial summation component, a unit accumulation component and a change operation component; obtaining a pre-constructed pixel filtering reference table, and obtaining reference data corresponding to the to-be-calculated component from the pixel filtering reference table; and performing an intra-frame coding operation based on the intra-frame prediction coding logic and the reference data to obtain a prediction pixel value. According to the invention, acceleration of the intra-frame prediction mode is realized based on the coding characteristics of the target instruction set, and the overall coding speed of all prediction modes is greatly improved.

Description

TECHNICAL FIELD

[0001] The present disclosure relates to the technical field of video coding, and particularly relates to an intra prediction method, an intra prediction device, an electronic device and a computer program product. BACKGROUND

[0002] With the rapid growth of high-definition video and ultra-high-definition video coding demand, the requirement of video coding standards such as High Efficiency Video Coding (HEVC) for computing resources is increasing. Among them, intra prediction (Intra Prediction) as a key link in video coding, undertakes the important task of reducing redundant information by using spatial correlation. Intra prediction predicts the target block by referring to the neighboring pixels in the current frame, thereby greatly improving the coding efficiency and video quality.

[0003] HEVC supports 35 intra prediction modes, including: DC (mode 0); Planar (mode 1); 33 angle modes (modes 2-34). However, the computational complexity of intra prediction is high, especially in high-resolution video and complex prediction mode, which becomes one of the performance bottlenecks of video encoder. With the increasing demand for low-power and high-performance video processing of mobile devices and embedded systems, how to accelerate the intra prediction algorithm under limited computing resources has become a problem to be solved. SUMMARY

[0004] The present disclosure provides an intra prediction method, an intra prediction device, an electronic device, a computer readable storage medium and a computer program product to at least solve the problem that the computational complexity of intra prediction operation in related technologies is high, and part of the calculation has redundancy, resulting in low calculation efficiency. The technical solutions of the present disclosure are as follows:

[0005] According to a first aspect of an embodiment of the present disclosure, an intra prediction method is provided, comprising: obtaining an intra prediction coding logic corresponding to a to-be-predicted pixel, determining a to-be-operated component of the target instruction set in the intra prediction coding logic, the to-be-operated component including one or more of an initial summation component, a unit accumulation component and a change operation component; obtaining a pre-constructed pixel filtering reference table, and obtaining reference data corresponding to the to-be-operated component from the pixel filtering reference table; performing an intra prediction operation based on the intra prediction coding logic and the reference data to obtain a predicted pixel value.

[0006] In an example implementation of the present disclosure, the obtaining the intra prediction coding logic corresponding to the pixel to be predicted, and determining the operation component to be operated in the target instruction set in the intra prediction coding logic, comprises: obtaining the intra prediction coding logic corresponding to the pixel to be predicted in a specified angle mode, the specified angle mode comprising a planar mode; and determining the operation component to be operated in the intra prediction coding logic based on the coding characteristics of the target instruction set.

[0007] In an example implementation of the present disclosure, the obtaining the intra prediction coding logic corresponding to the pixel to be predicted in a specified angle mode, comprises: determining a current to-be-predicted block in which the pixel to be predicted is located, obtaining block size information of the current to-be-predicted block and row and column information of the pixel to be predicted, the block size information comprising a block height and a block width; determining reference pixels corresponding to the pixel to be predicted based on the row and column information, the reference pixels comprising an upper reference pixel and a left reference pixel; and generating the intra prediction coding logic according to the block height, the block width, the row and column information, and the reference pixels, the intra prediction coding logic being used to determine a predicted pixel value of the pixel to be predicted.

[0008] In an example implementation of the present disclosure, the determining the operation component to be operated in the intra prediction coding logic based on the coding characteristics of the target instruction set, comprises: determining an operation constant corresponding to the intra prediction coding logic based on the coding characteristics of the target instruction set; determining a row calculation component corresponding to a row prediction pixel in the intra prediction coding logic based on the operation constant; determining an initial summation component and a unit accumulation component corresponding to the row prediction pixel according to the row calculation component; and determining the variable operation component based on the intra prediction coding logic, the initial summation component, and the unit accumulation component.

[0009] In an example implementation of the present disclosure, the pixel filtering reference table comprises a filtering coefficient table, a reference pixel projection table, and a reference pixel expansion table; and the obtaining the pre-constructed pixel filtering reference table comprises: determining angle modes corresponding to an intra prediction operation, and configuring corresponding filtering coefficients for each of the angle modes; constructing the filtering coefficient table according to the filtering coefficients corresponding to the plurality of angle modes; determining reference pixel projection positions corresponding to each row pixel in a prediction block in a horizontal prediction mode; constructing the reference pixel projection table based on a deviation of the reference pixel projection positions and a current row number; determining expansion reference pixels corresponding to the intra prediction coding in a negative angle mode, and constructing the reference pixel expansion table based on the expansion reference pixels, a number of expansion pixels and positions of the expansion pixels of the expansion reference pixels being determined based on an angle value of the negative angle mode.

[0010] In an example implementation of the present disclosure, the obtaining the reference data corresponding to the to-be-operated component from the pixel filter reference table comprises: determining a target angle mode corresponding to the to-be-predicted pixel; and obtaining the reference data from the pixel filter reference table based on the target angle mode, the reference data comprising one or more of a target filter coefficient, a target reference pixel, and a target extended reference pixel.

[0011] In an example implementation of the present disclosure, the obtaining the reference data from the pixel filter reference table based on the target angle mode comprises: when the target angle mode is a vertical type mode, obtaining a first filter coefficient of a row in which the target angle mode is located from the filter coefficient table as a target filter coefficient; and copying the target filter coefficient to a vector register for an intra prediction operation; and when the target angle mode is a horizontal type mode, obtaining filter coefficients of the row in which the target angle mode is located from the filter coefficient table as target filter coefficients.

[0012] In an example implementation of the present disclosure, the obtaining the reference data from the pixel filter reference table based on the target angle mode comprises: when the target angle mode is a horizontal type mode, obtaining a reference pixel projection deviation corresponding to the target angle mode based on the reference pixel projection table; determining a target reference pixel corresponding to the to-be-predicted pixel according to the reference pixel projection deviation; and when the target angle mode is a vertical type mode and a reference pixel change event in an intra prediction process is detected, determining the target reference pixel corresponding to a current row based on the reference pixel projection table.

[0013] In an example implementation of the present disclosure, the obtaining the reference data from the pixel filter reference table based on the target angle mode comprises: when the target angle mode is a negative angle mode, obtaining a target extended reference pixel from the reference pixel extension table; determining a main reference pixel corresponding to the to-be-predicted pixel; and storing the target extended reference pixel and the main reference pixel into a stack together to read pixel data from the stack through a load instruction.

[0014] According to a second aspect of the present disclosure, an intra prediction device is provided, comprising: a to-be-operated component determination module configured to obtain an intra prediction coding logic corresponding to a to-be-predicted pixel, and determine a to-be-operated component of a target instruction set in the intra prediction coding logic, the to-be-operated component comprising one or more of an initial summation component, a unit accumulation component, and a change operation component; a reference data obtaining module configured to obtain a pixel filter reference table constructed in advance, and obtain reference data corresponding to the to-be-operated component from the pixel filter reference table; and an intra prediction module configured to perform an intra prediction operation based on the intra prediction coding logic and the reference data, and obtain a predicted pixel value.

[0015] In an example implementation of the present disclosure, the operation component determination module comprises an operation component determination unit configured to: obtain an intra prediction coding logic corresponding to the to-be-predicted pixel in a specified angle mode, the specified angle mode comprising a planar mode; and determine a to-be-operated component corresponding to the intra prediction coding logic based on the encoding characteristics of the target instruction set.

[0016] In an example implementation of the present disclosure, the operation component determination unit comprises a target logic determination sub-unit configured to: determine a current to-be-predicted block in which the to-be-predicted pixel is located, obtain block size information of the current to-be-predicted block and row and column information of the to-be-predicted pixel, the block size information comprising a block height and a block width; determine a reference pixel corresponding to the to-be-predicted pixel based on the row and column information, the reference pixel comprising an upper reference pixel and a left reference pixel; and generate the intra prediction coding logic based on the block height, the block width, the row and column information, and the reference pixel, the intra prediction coding logic being used to determine a predicted pixel value of the to-be-predicted pixel.

[0017] In an example implementation of the present disclosure, the operation component determination unit comprises an operation component determination sub-unit configured to: determine an operation constant corresponding to the intra prediction coding logic based on the encoding characteristics of the target instruction set; determine a row calculation component corresponding to a row prediction pixel in the intra prediction coding logic based on the operation constant; determine an initial summation component and a unit accumulation component corresponding to the row prediction pixel based on the row calculation component; and determine the change operation component based on the intra prediction coding logic, the initial summation component, and the unit accumulation component.

[0018] In an example implementation of the present disclosure, the pixel filtering reference table comprises a filtering coefficient table, a reference pixel projection table, and a reference pixel expansion table; and the reference data obtaining module comprises a reference table construction unit configured to: determine an angle mode corresponding to an intra prediction operation, and configure a corresponding filtering coefficient for each of the angle modes; construct the filtering coefficient table based on the filtering coefficients corresponding to the plurality of angle modes; determine a reference pixel projection position corresponding to each row pixel in a prediction block in a horizontal prediction mode; construct the reference pixel projection table based on a deviation of the reference pixel projection position and a current row number; determine an expansion reference pixel corresponding to the intra prediction coding in a negative angle mode, and construct the reference pixel expansion table based on the expansion reference pixel, a number and a position of expansion pixels of the expansion reference pixel being determined based on an angle value of the negative angle mode.

[0019] In an example implementation of the present disclosure, the reference data obtaining module comprises a reference data obtaining unit configured to: determine a target angle mode corresponding to the pixel to be predicted; and obtain the reference data from the pixel filter reference table based on the target angle mode, the reference data comprising one or more of a target filter coefficient, a target reference pixel, and a target extended reference pixel.

[0020] In an example implementation of the present disclosure, the reference data obtaining unit comprises a first data obtaining sub-unit configured to: when the target angle mode is a vertical type mode, obtain a first filter coefficient of a line in which the target angle mode is located from the filter coefficient table as the target filter coefficient; and copy the target filter coefficient to a vector register for an intra prediction operation; and when the target angle mode is a horizontal type mode, obtain filter coefficients of the line in which the target angle mode is located from the filter coefficient table as the target filter coefficient.

[0021] In an example implementation of the present disclosure, the reference data obtaining unit comprises a second data obtaining sub-unit configured to: when the target angle mode is a horizontal type mode, obtain a reference pixel projection deviation corresponding to the target angle mode based on the reference pixel projection table; and determine a target reference pixel corresponding to the pixel to be predicted according to the reference pixel projection deviation; and when the target angle mode is a vertical type mode and a reference pixel change event in an intra prediction process is detected, determine the target reference pixel corresponding to a current line based on the reference pixel projection table.

[0022] In an example implementation of the present disclosure, the reference data obtaining unit comprises a third data obtaining sub-unit configured to: when the target angle mode is a negative angle mode, obtain a target extended reference pixel from the reference pixel extension table; and determine a main reference pixel corresponding to the pixel to be predicted, and store the target extended reference pixel and the main reference pixel in a stack together, so as to read pixel data from the stack by a load instruction.

[0023] According to a third aspect of the present disclosure, an electronic device is provided, comprising: a processor; a memory for storing processor-executable instructions; wherein the processor is configured to execute the instructions to implement the intra prediction method according to any one of the preceding aspects.

[0024] According to a fourth aspect of the present disclosure, a computer-readable storage medium is provided, when instructions in the computer-readable storage medium are executed by a processor of an electronic device, the electronic device is enabled to perform the intra prediction method according to any one of the preceding aspects.

[0025] According to a fifth aspect of the embodiments of the present disclosure, a computer program product is provided, comprising a computer program which, when executed by a processor, implements the intra prediction method according to any one of the preceding aspects.

[0026] The embodiments of the present disclosure at least have the following beneficial effects: on the one hand, the to-be-operated components are determined by the mathematical common component extraction, and the encoding processing is performed based on different types of to-be-operated components, so that the operation amount of intra prediction can be effectively reduced. On the other hand, the reference data used for the intra prediction operation is obtained from the pre-constructed reference table, so that the parallel calculation of the reference data can be realized, and the overall encoding speed is further improved.

[0027] It should be understood that the foregoing general description and the following detailed description are only exemplary and explanatory, and cannot limit the present disclosure. BRIEF DESCRIPTION OF DRAWINGS

[0028] The accompanying drawings, which are incorporated into and form part of the specification, illustrate embodiments consistent with the present disclosure and, together with the description, serve to explain the principles of the present disclosure, and do not constitute an improper limitation on the present disclosure.

[0029] Figure 1 is a schematic diagram of reference pixels corresponding to predicted pixels in a Planar mode in a related scheme.

[0030] Figure 2 is a flowchart of an intra prediction method according to an exemplary embodiment.

[0031] Figure 3 is a schematic diagram of intra prediction modes supported by HEVC according to an exemplary embodiment.

[0032] Figure 4 is a schematic diagram of reference pixel projection in an intra prediction operation according to an exemplary embodiment.

[0033] Figure 5 is a block diagram of an intra prediction device according to an exemplary embodiment.

[0034] Figure 6 A block diagram of an electronic device according to an exemplary embodiment of the present disclosure is schematically shown. DETAILED DESCRIPTION

[0035] In order for those skilled in the art to better understand the technical solutions of the present disclosure, the technical solutions in the embodiments of the present disclosure will be described clearly and completely below with reference to the drawings.

[0036] It is to be noted that the terms "first", "second", and the like in the description and in the claims of the present disclosure and the above-described drawings are used to distinguish similar objects, and do not necessarily indicate a specific order or sequence. It should be understood that the data thus used can be interchanged, where appropriate, so that the embodiments of the present disclosure described herein can be carried out in other sequences than those illustrated or described herein. The implementations described in the following exemplary embodiments do not represent all implementations consistent with the present disclosure. Instead, they are merely examples of apparatuses and methods consistent with some aspects of the present disclosure as detailed in the appended claims.

[0037] The Advanced RISC Machine (ARM) architecture, as the mainstream processor platform of mobile devices and embedded markets, provides a rich Single Instruction Multiple Data (SIMD) instruction set for data parallel processing, such as the NEON extension architecture, with a good performance-power ratio. Manual optimization using ARM assembly language can fully exploit the hardware potential and significantly improve the calculation efficiency of intra prediction.

[0038] In the prior art, most video encoders still mainly rely on compiler automatic vectorization for intra prediction optimization, and it is difficult to achieve the highest performance of the hardware. How to design an efficient intra prediction implementation combined with the characteristic instruction set of ARM assembly to reduce memory access and instruction pipeline bottlenecks is an important direction to improve the overall performance of the video encoder. The intra prediction modules in the commonly used HEVC open source encoders such as x265 and libhevc either only provide simple NEON assembly support for the main prediction mode, or the assembly code is large and complex, with high maintenance costs; and the optimization of memory access and data alignment is insufficient, which affects the parallel computing efficiency.

[0039] Taking the x265 assembly implementation as an example, the implementation logic of the main prediction mode is as follows. For the Planar mode, the reference Figure 1 , Figure 1 is a schematic diagram of the reference pixels corresponding to the predicted pixels in the Planar mode in the related scheme. The output of the Planar mode is a to-be-predicted block with a width of W and a height of H; the input has refTop (upper reference pixel) and refLeft (left reference pixel), and the intra prediction operation is used to calculate the predicted value of any point (X, Y).

[0040] For the angle mode, the following steps are included: (1) judging whether the current angle needs to be expanded (11-17 and 19-25), if not, jumping to step (3); (2) looking up the corresponding anti-angle parameters of the current angle, and deciding the number and position of the pixels to be expanded according to the anti-angle parameters, and filling the corresponding pixels into the reference pixel list; (3) calculating the projection of each pixel on the reference pixel list according to the current angle, the projection point is usually a non-integer position, which needs to be interpolated; (4) for the non-integer position, 1D interpolation is performed, assuming that the projection point falls between reference pixels A and B, the distance is d, then: P(x, y) = (1-d) A + d B. The ARM assembly implementation of the angle mode in X265 is manually calculated according to the offset of the angle mode to calculate the reference pixel position and reference pixel weight corresponding to each row of pixels.

[0041] However, the above scheme has the following defects: (1) the Planar mode contains multiple multiplication and addition operations, the calculation complexity is high, and part of the calculation is redundant, the calculation efficiency is not high; (2) the angle mode calculates the projection position and reference pixel weight of each row of pixels, which is not easy to use SIMD characteristics for parallel optimization; (3) in x265, the angle mode other than horizontal and vertical diagonal is divided into four large areas for calculation: 3-9, 11-17, 19-25, 27-33, the implementation code of the prediction function of each area is large and complex, and the maintenance cost is high.

[0042] Based on this, according to the embodiments of the present disclosure, an intra prediction method, an intra prediction device, an electronic device, a computer readable storage medium and a computer program product are provided.

[0043] Figure 2 is a flow chart of an intra prediction method according to an example embodiment, as Figure 2 As shown in the figure, the intra prediction method can be used in a computer device, wherein the computer device described in the present disclosure can include mobile terminal devices such as mobile phones, tablet computers, notebook computers, palm computers, personal digital assistants (Personal Digital Assistant, PDA), and fixed terminal devices such as desktop computers. The present example embodiment is illustrated by applying the method to a computer device, it can be understood that the method can also be applied to a server, and can also be applied to a system including a computer device and a server, and is realized through the interaction of the computer device and the server. Specifically, the following steps are included.

[0044] In step S210, the intra prediction coding logic corresponding to the to-be-predicted pixel is obtained, and the to-be-operated components in the target instruction set in the intra prediction coding logic are determined, the to-be-operated components including one or more of an initial summation component, a unit accumulation component and a change operation component;

[0045] In step S220, a pixel filter reference table is acquired, and reference data corresponding to the to-be-operated component is acquired from the pixel filter reference table.

[0046] In step S230, intra-frame prediction operation is performed based on the intra-frame prediction coding logic and the reference data, and a predicted pixel value is obtained.

[0047] According to the intra-frame prediction method in the example embodiment, on one hand, the to-be-operated component is determined through mathematical common component extraction, and the encoding processing is performed based on different types of to-be-operated components, which can effectively reduce the operation amount of intra-frame prediction. On the other hand, the reference data used for intra-frame prediction operation is acquired from the pre-constructed reference table, which can realize parallel calculation of the reference data and further improve the overall encoding speed.

[0048] In the following, the intra-frame prediction method in the example embodiment will be further described.

[0049] In an example embodiment of the disclosure, for step S210, the intra-frame prediction coding logic corresponding to the to-be-predicted pixel is acquired, and the to-be-operated component of the target instruction set in the intra-frame prediction coding logic is determined, including: acquiring the intra-frame prediction coding logic corresponding to the to-be-predicted pixel in a specified angle mode, and the specified angle mode includes a planar mode; and determining the to-be-operated component corresponding to the intra-frame prediction coding logic based on the encoding characteristics of the target instruction set.

[0050] The to-be-predicted pixel can be a pixel point to be subjected to intra-frame prediction. The specified angle mode can be a pre-specified angle mode, for example, the specified angle mode can be a Planar mode. The intra-frame prediction coding logic can be a pixel prediction value calculation logic for intra-frame prediction coding in the specified angle mode. The target instruction set can be an instruction set used for intra-frame prediction operation. The to-be-operated component can be a component that needs to be operated in the intra-frame prediction process.

[0051] The video coding standard such as HEVC supports multiple intra-frame prediction modes, for example, for the Planar mode, the calculation method of the prediction value for any point (X, Y) is configured with corresponding intra-frame prediction coding logic. Due to the characteristics of SIMD, the values related to the horizontal coordinate X can be quickly read and calculated through assembly instructions, and according to the characteristics of SIMD, the to-be-operated component that needs to be calculated in the intra-frame prediction coding logic can be determined. According to the encoding characteristics of the target instruction set, the corresponding to-be-operated component is determined, and in the subsequent encoding process, different types of calculation components can be processed accordingly to reduce the number of operations.

[0052] In an example embodiment of the present disclosure, the determination of the intra-prediction coding logic corresponding to the pixel to be predicted in the specified angle mode includes: determining a current to-be-predicted block in which the pixel to be predicted is located, obtaining block size information of the current to-be-predicted block and row and column information of the pixel to be predicted, the block size information including a block height and a block width; determining reference pixels corresponding to the pixel to be predicted based on the row and column information, the reference pixels including an upper reference pixel and a left reference pixel; and generating the intra-prediction coding logic according to the block height, the block width, the row and column information, and the reference pixels, the intra-prediction coding logic being used to determine a predicted pixel value of the pixel to be predicted.

[0053] The current to-be-predicted block can be a current intra-prediction coding block. The block size information can be size information of a block corresponding to the current to-be-predicted block, and can include a block height and a block width, etc. For example, the block size can include but is not limited to 4x4, 8x8, 16x16, 32x32, 64x64, etc. The row and column information of the pixel to be predicted can be a row coordinate and a column coordinate corresponding to the pixel to be predicted in the current prediction. The reference pixels can be coded pixels referenced for coding prediction of the pixel to be predicted. The upper reference pixel can be a pixel located above the pixel to be predicted. The left reference pixel can be a pixel located to the left of the pixel to be predicted.

[0054] When using the Planar mode for intra-prediction, the encoder can first determine a current to-be-predicted block in which intra-prediction is currently being performed, and determine block size information corresponding to the current to-be-predicted block, including a block height H and a block width W, which will be used in the prediction process of the pixel based on the numerical values of the length and width of the current to-be-predicted block. Since the encoder performs coding prediction line by line, in the intra-prediction process, the row and column information of the current pixel to be predicted can be determined, i.e., the horizontal coordinate X and the vertical coordinate Y corresponding to the pixel to be predicted.

[0055] According to the determined horizontal coordinate X and vertical coordinate Y of the pixel to be predicted, the reference pixels corresponding to the pixel to be predicted can be determined, respectively. The reference pixels usually include an upper reference pixel (refTop) and a left reference pixel (refLeft), and the reference Figure 1 As shown in the formula, the upper reference pixel and the left reference pixel are pixels in the refTop array and the refLeft array, respectively.

[0056] After determining the calculation process of the pixel to be predicted and using various types of information required, the intra-prediction coding logic can be generated based on the above information, as shown in formula 1. The predicted pixel value of the pixel to be predicted will be determined based on the intra-prediction coding logic. The intra-prediction coding logic provides a specific calculation method of the pixel to be predicted.

[0057] pred(X, Y) = ((W - 1 - X) * refLeft[Y] + (X + 1) * refTop[W] + ((H - 1 - Y) * refTop[X] + (Y + 1) * refLeft(H) + W) / (W * 2) (Equation 1)

[0058] Wherein, pred(X, Y) can be a specific pixel value corresponding to a pixel to be predicted; X can be a horizontal coordinate corresponding to the pixel to be predicted; Y can be a vertical coordinate corresponding to the pixel to be predicted; W can be a width of a current block to be predicted; H can be a height of the current block to be predicted; refTop[X] can be an upper reference pixel of the pixel to be predicted in a vertical direction; refTop[W] can be a reference pixel corresponding to the pixel to be predicted in the refTop array with index W (the starting index is 0); refLeft[Y] can be a left reference pixel of the pixel to be predicted in a horizontal direction; refTop[H] can be a reference pixel corresponding to the pixel to be predicted in the refLeft array with index H (the starting index is 0). Based on the above steps, the intra prediction encoding logic corresponding to the pixel to be predicted is determined, and subsequent intra prediction operations can be performed.

[0059] In an exemplary embodiment of the present disclosure, based on the encoding characteristics of the target instruction set, the operation component corresponding to the intra prediction encoding logic is determined, comprising: based on the encoding characteristics of the target instruction set, determining an operation constant corresponding to the intra prediction encoding logic; based on the operation constant, determining a row calculation component corresponding to a row prediction pixel in the intra prediction encoding logic; according to the row calculation component, determining an initial summation component and a unit accumulation component corresponding to the row prediction pixel; based on the intra prediction encoding logic, the initial summation component and the unit accumulation component, determining a change operation component.

[0060] Wherein, the operation constant can be a component whose value remains unchanged during the operation of the intra prediction encoding logic. The row calculation component can be a component in the row-by-row encoding during the intra prediction process. The row prediction pixel can be a whole row of pixels to be predicted determined according to the row-by-row encoding characteristics of the target instruction set. The initial summation component can be an initial summation value initialized in the operation process of the pixel to be predicted. The unit accumulation component can be a component that needs to be accumulated in the row-by-row pixel operation process. The change operation component can be a component that changes and needs to be operated when calculating the specific pixel value of each pixel point.

[0061] The encoding characteristic of SIMD mainly reflects in data parallelism, that is, processing multiple data at the same time through single instruction. The core characteristics include: single instruction stream, all processing units execute the same instruction; multiple data stream, the instruction acts on multiple data; parallel processing, multiple data execute the same operation at the same time, which improves efficiency. According to the characteristics of SIMD, the value related to the horizontal coordinate X can be quickly read and calculated through the assembly instruction, so all refTop[X] pixels can be extracted at one time, that is, a whole row of upper reference pixels.

[0062] Due to the single instruction stream and multiple data stream characteristics of SIMD, all refTop[X] pixels are extracted at one time, and when a whole row of pixels is predicted, the value of (X+1)*refTop[W] can be calculated at the same time for the to-be-predicted pixels at different positions in the same row, so (X+1)*refTop[W]+W in formula 1 is equivalent to a constant, which does not need to be repeatedly calculated in the prediction process of each row of pixels, and only the part related to the vertical coordinate Y needs to be concerned, that is, (W-1-X)*refLeft[Y]+(H-1-Y)*refTop[X]+(Y+1)*refLeft(H). In addition, for the remaining part, it can be observed that (H-1-Y)*refTop[X]+(Y+1)*refLeft(H) can actually be regarded as the accumulation of the invariant refLeft(H)-refTop[X] in the calculation process of each row, so an initial summation value, that is, an initial summation component Acc, is initialized, which is specifically shown in formula 2.

[0063] Acc=(X+1)*refTop[W]+(H-1)*refTop[X]+(1)*refLeft(H)+W (formula 2)

[0064] And calculate the unit accumulation value, that is, the unit accumulation component Add, which is specifically shown in formula 3.

[0065] Add=*refLeft(H)-refTop[X] (formula 3)

[0066] In the pixel prediction process, when moving down one row each time, only Add needs to be accumulated on Acc, so that the value of (X+1)*refTop[W]+(H-1-Y)*refTop[X]+(Y+1)*refLeft(H)+W of the current row can be obtained, and only one component of (W-1-X)*refLeft[Y] needs to be multiplied and added, so the number of operations required is greatly reduced.

[0067] To make the calculation process more suitable for SIMD optimization, the disclosure proposes a method of obtaining the reference pixel projection position and the corresponding filter coefficient of each row of pixels by using a lookup table, and the reference pixel of the negative angle mode expansion. After obtaining the above information by lookup table for each angle mode, the remaining task is only the filter calculation of the reference pixel (only contains multiplication and addition operation), which is suitable for SIMD optimization.

[0068] In an exemplary embodiment of the disclosure, for step S220, a pre-constructed pixel filter reference table is obtained, including: determining the angle mode corresponding to the intra prediction operation, configuring the corresponding filter coefficient for each angle mode; constructing a filter coefficient table according to the filter coefficients corresponding to the plurality of angle modes; determining the reference pixel projection position corresponding to each row of pixels in the prediction block in the horizontal prediction mode; constructing a reference pixel projection table based on the deviation of the reference pixel projection position and the current row number; determining the expansion reference pixel corresponding to the negative angle mode of the intra prediction encoding, constructing a reference pixel expansion table based on the expansion reference pixel, and the number and position of the expansion pixels of the expansion reference pixel are determined based on the angle value of the negative angle mode.

[0069] The filter coefficient table can be a reference table composed of filter coefficients. The reference pixel projection table can be a reference table composed of the projection positions of all possible reference pixels in the intra prediction process. The reference pixel expansion table can be a reference table used for expansion operation of the reference pixel. The filter coefficient is mainly used to process the boundary pixel, smooth the prediction signal by a specific mathematical operation, reduce the noise and maintain the boundary continuity. The reference pixel projection position can be the projection position of the reference pixel corresponding to the pixel to be predicted. The expansion reference pixel can be the pixel selected from the reference pixel on the other side for expansion of the reference list in the negative angle mode. The number of expansion pixels can be the specific number of expansion pixels. The expansion pixel position can be the position of the expansion pixel.

[0070] Reference Figure 3 , Figure 3 is a schematic diagram of the HEVC supported intra prediction mode according to an exemplary embodiment. Figure 3 The prediction directions of the angle modes 2-34 are shown in Figure 3 The angle mode 27 in corresponds to the vertical direction, and the angle mode 10 corresponds to the horizontal direction; the angle mode 18 corresponds to the prediction direction of 45° to the upper left, and the angle mode 34 corresponds to the prediction direction of 45° to the upper right. Taking the size of the prediction unit (PU) as 32x32 as an example, the SIMD instruction will predict 32 to-be-predicted pixels at a time, so there are 32 filter coefficients in each row of the filter coefficient table.

[0071] For the above angle modes 2-34, excluding the horizontal, vertical and diagonal angle modes, the remaining angle modes can be divided into vertical type and horizontal type, such as angle modes 19-33 are vertical type, because the vertical or nearly vertical direction accounts for the majority of these directions; angle modes 3-17 are horizontal type, because the horizontal or nearly horizontal direction accounts for the majority of these directions. Taking the vertical direction class as the main research object, the prediction of the horizontal direction class can be converted to the prediction of the vertical direction class. The corresponding filter coefficients are configured for the above two types of angle modes. In this scheme, 7 types of angle mode corresponding filter coefficients are configured, please refer to Table 1 for details.

[0072] Table 1 filter coefficient table

[0073]

[0074] Each row in Table 1 is a filter coefficient of an angle mode, and each row has 32 filter coefficient values. Each angle mode has its own corresponding filter coefficient value.

[0075] In Table 1, from top to bottom, they correspond to angle modes 9-3, angle modes 11-17, angle modes 25-19, and angle modes 27-33, respectively. That is, angle modes 9, 11, 25 and 27 correspond to the filter coefficient values of the first row; angle modes 8, 12, 24 and 28 correspond to the filter coefficient values of the second row; and so on, angle modes 3, 17, 19 and 33 correspond to the filter coefficient values of the seventh row. The filter coefficient values corresponding to each angle mode can be set according to the angle corresponding to the angle mode and the position of the pixel to be predicted in this row.

[0076] In the intra prediction process, the reference pixels of each pixel to be predicted can be obtained for subsequent pixel prediction process. Figure 4 , Figure 4 FIG. 1 is a schematic diagram of reference pixel projection in intra prediction operation according to an exemplary embodiment. For the reference array in angle prediction, the "angle" of each angle mode, i.e. the number of offsets, can be determined. In the pixel projection process, it may be exactly projected into an element in the array, or it may be projected into the position between two elements in the array.

[0077] For the horizontal type mode, the projection positions of different pixels in each row on the left reference pixel are different. According to the above characteristics, a reference pixel projection table can be configured for the horizontal type mode, as shown in Table 2.

[0078] Table 2 reference pixel projection table

[0079]

[0080] Each row in Table 2 is the deviation of the reference pixel projection of an angle mode compared with the current row number, and each row has 32 reference pixel projection position values. In Table 2, from top to bottom, they correspond to angle mode 9-3, angle mode 11-17 respectively. After obtaining the reference pixels and filter coefficients by looking up the table, the horizontal type mode can perform filter calculation.

[0081] According to Figure 3 It can be seen that there are positive angle modes and negative angle modes in the angle mode. In the vertical direction type, greater than mode 26, the "angle" is +; less than mode 26, the "angle" is -. For the mode with negative angle value (angle mode 11-25), the reference list needs to be expanded by selecting pixels from the other side reference pixels. The number and position of the expanded pixels depend on the angle size. In order to facilitate SIMD calculation, the disclosure introduces a reference pixel expansion table for determining the expanded reference pixels corresponding to different angle modes. The specific table is shown in Table 3.

[0082] Table 3 Reference pixel expansion table

[0083]

[0084]

[0085] Each row in Table 3 is the position of the expanded reference pixel of an angle mode in the reference pixel list on the other side. From top to bottom, they correspond to angle mode 11-17, angle mode 25-19 respectively. By constructing the reference table corresponding to the prediction pixel, the reference data corresponding to each row of pixels can be obtained by looking up the table in the subsequent process, which can reduce the calculation of SIMD and is more conducive to the optimization of SIMD.

[0086] In an exemplary embodiment of the disclosure, for step S220, obtaining the reference data corresponding to the to-be-operated component from the pixel filter reference table includes: determining a target angle mode corresponding to the to-be-predicted pixel; and based on the target angle mode, obtaining the reference data from the pixel filter reference table, the reference data including one or more of a target filter coefficient, a target reference pixel, and a target expanded reference pixel.

[0087] The target angle mode can be an angle mode corresponding to the to-be-predicted pixel. The target filter coefficient can be a coefficient used for filter processing of the to-be-predicted pixel. The target reference pixel can be a reference pixel corresponding to the to-be-predicted pixel. The target expanded reference pixel can be an expanded reference pixel corresponding to the to-be-predicted pixel.

[0088] In the intra prediction of the to-be-predicted pixel, a target angle mode corresponding to the to-be-predicted pixel can be determined first, for example, whether the target angle mode is a vertical mode or a horizontal mode, or whether the target angle mode is a positive angle mode or a negative angle mode. After the target angle mode is determined, reference data can be obtained from the pixel filter reference table based on the target angle mode.

[0089] Specifically, since the pixel filter reference table provides three different types of reference tables, and the reference data corresponding to different angle modes is defined in the reference table, the target filter coefficient corresponding to the to-be-predicted pixel, the target reference pixel, and the target extended reference pixel, and other data can be directly obtained according to the target angle mode. Then, the pixel prediction processing can be performed according to the above-mentioned reference data, which can be beneficial to the optimization of SIMD.

[0090] In an exemplary embodiment of the present disclosure, the reference data is obtained from the pixel filter reference table based on the target angle mode, including: when the target angle mode is a vertical mode, the first filter coefficient of the row in which the target angle mode is located is obtained from the filter coefficient table as the target filter coefficient; the target filter coefficient is copied to the vector register for intra prediction operation; when the target angle mode is a horizontal mode, the filter coefficient of the row in which the target angle mode is located is obtained from the filter coefficient table as the target filter coefficient.

[0091] When the target angle mode is a vertical mode, for the vertical mode (19-33), the filter coefficients of all pixels in each row are the same. The first filter coefficient of the row in which the corresponding angle mode is located is read first, and the first filter coefficient can be the first filter coefficient of the row in which the target angle mode is located in the filter coefficient table. Then, the obtained first filter coefficient is copied to each element of the vector register for calculation. Each time a row is calculated, the next coefficient in the lookup table is read.

[0092] When the target angle mode is a horizontal mode, for the horizontal mode (3-17), the filter coefficients of the pixels in different columns in each row are obtained from the row in which the angle mode is located in the lookup table, for the subsequent filter processing of the to-be-predicted pixel. The filter coefficients are directly obtained from the filter coefficient table, and the remaining tasks are only the filter calculation of the reference pixel (only including multiplication and addition operations), which is suitable for SIMD optimization.

[0093] In an example embodiment of the present disclosure, obtaining reference data from a pixel filter reference table based on a target angle mode comprises: when the target angle mode is a horizontal type mode, obtaining a reference pixel projection deviation corresponding to the target angle mode based on a reference pixel projection table; and determining a target reference pixel corresponding to the pixel to be predicted according to the reference pixel projection deviation; when the target angle mode is a vertical type mode, and a reference pixel change event in an intra prediction process is detected, determining a target reference pixel corresponding to a current row based on the reference pixel projection table.

[0094] The reference pixel projection deviation can be a deviation of the reference pixel projection from a current row number (i.e., a row where the pixel is located). The reference pixel change event can be an event of a change in the reference pixel of the pixel to be predicted.

[0095] Since the deviation of the reference pixel projection of each angle mode in Table 2 from the current row number is from top to bottom, it corresponds to angle mode 9-3 and angle mode 11-17, respectively. When the target angle mode is a horizontal type mode, the horizontal type mode can obtain the reference pixel and the filter coefficient by table lookup, and then perform filter calculation.

[0096] When the target angle mode is a vertical type mode, the projection deviation of each row of pixels on the reference pixel above is the same, so in the calculation process from top to bottom, only the reference pixel needs to be re-read when the reference pixel changes. If the reference pixel used is unchanged, the reading process of the reference pixel can be omitted.

[0097] In an example embodiment of the present disclosure, obtaining reference data from a pixel filter reference table based on a target angle mode comprises: when the target angle mode is a negative angle mode, obtaining a target extended reference pixel from an extended reference pixel table; determining a main reference pixel corresponding to the pixel to be predicted, and storing the target extended reference pixel and the main reference pixel in a stack together to read pixel data from the stack through a loading instruction.

[0098] The negative angle mode can be a mode with a negative angle value, such as angle mode 11-25. The target extended reference pixel can be an extended reference pixel corresponding to the pixel to be predicted. The main reference pixel can be a reference pixel corresponding to the pixel to be predicted determined from the coded pixels. The pixel data can be pixel data stored in the stack, which can include the reference pixel and the extended reference pixel.

[0099] When the target angle mode is a negative angle mode, the reference pixels need to be expanded. In this embodiment, all expanded pixels can be obtained by looking up a table, and then the expanded reference pixels and the main reference pixels are stored in a stack together, so that the reference pixel data can be read by using a load instruction. For the load instruction, the NEON instruction indicates loading a one-dimensional array or a two-dimensional array by using the mnemonics "ld1" and "ld2". In this disclosure, the expanded reference pixels and the main reference pixels are stored in the stack together, and then can be read by using the ld1 instruction. After the above optimization, the complexity of the assembly function body is greatly reduced, the loop body is arranged more reasonably, and the parallel computing efficiency is greatly improved.

[0100] Based on the above operation, the disclosure implements an implementation architecture for accelerating HEVC intra prediction mode by using an ARMv8 SIMD instruction set on a client. Compared with the current open source implementation, under the same test conditions (including 8x8, 16x16 and 32x32 block sizes), the planar mode is accelerated by 54%, and the overall encoding speed of all prediction modes is improved by 41.9%.

[0101] In summary, the intra prediction method of the disclosure obtains the intra prediction coding logic corresponding to the to-be-predicted pixel, determines the to-be-operated component in the target instruction set in the intra prediction coding logic, the to-be-operated component includes one or more of the initial summation component, the unit accumulation component and the change operation component; obtains the pre-constructed pixel filtering reference table, and obtains the reference data corresponding to the to-be-operated component from the pixel filtering reference table; and performs an intra prediction operation based on the intra prediction coding logic and the reference data to obtain a predicted pixel value. On the one hand, the to-be-operated component is determined by extracting the mathematical common component, and the encoding processing is performed based on different types of to-be-operated components, which can effectively reduce the operation amount of intra prediction. On the other hand, the reference data used for the intra prediction operation is obtained from the pre-constructed reference table, which can realize parallel computing of the reference data and further improve the overall encoding speed. On the other hand, the reference pixel projection position and the corresponding filtering coefficient corresponding to each row of pixels are obtained by looking up the table, and the reference pixels of the negative angle mode are expanded, so that the remaining task is only the filtering calculation of the reference pixels, which is suitable for the optimization of the SIMD.

[0102] Figure 5 is a block diagram of an intra prediction device according to an exemplary embodiment. Referring to Figure 5 The intra prediction device 500 includes an operation component determination module 510, a reference data acquisition module 520, and an intra prediction module 530.

[0103] Specifically, the operation component determination module 510 is configured to acquire an intra prediction coding logic corresponding to the to-be-predicted pixel, determine a to-be-operated component of the target instruction set in the intra prediction coding logic, and the to-be-operated component includes one or more of an initial summation component, a unit accumulation component, and a change operation component; the reference data acquisition module 520 is configured to acquire a pixel filtering reference table constructed in advance, and acquire reference data corresponding to the to-be-operated component from the pixel filtering reference table; and the intra prediction module 530 is configured to perform an intra prediction operation based on the intra prediction coding logic and the reference data to obtain a predicted pixel value.

[0104] In an example embodiment of the present disclosure, the operation component determination module 510 includes an operation component determination unit configured to: acquire an intra prediction coding logic corresponding to the to-be-predicted pixel in a specified angle mode, the specified angle mode including a planar mode; and determine the to-be-operated component corresponding to the intra prediction coding logic based on the encoding characteristics of the target instruction set.

[0105] In an example embodiment of the present disclosure, the operation component determination unit includes a target logic determination subunit configured to: determine a current to-be-predicted block in which the to-be-predicted pixel is located, acquire block size information of the current to-be-predicted block and row and column information of the to-be-predicted pixel, the block size information including a block height and a block width; determine a reference pixel corresponding to the to-be-predicted pixel based on the row and column information, the reference pixel including an upper reference pixel and a left reference pixel; and generate the intra prediction coding logic according to the block height, the block width, the row and column information, and the reference pixel, the intra prediction coding logic being used to determine a predicted pixel value of the to-be-predicted pixel.

[0106] In an example embodiment of the present disclosure, the operation component determination unit includes an operation component determination subunit configured to: determine an operation constant corresponding to the intra prediction coding logic based on the encoding characteristics of the target instruction set; determine a row calculation component corresponding to a row predicted pixel in the intra prediction coding logic based on the operation constant; determine an initial summation component and a unit accumulation component corresponding to the row predicted pixel according to the row calculation component; and determine a change operation component based on the intra prediction coding logic, the initial summation component, and the unit accumulation component.

[0107] In an example implementation of the present disclosure, the pixel filter reference table includes a filter coefficient table, a reference pixel projection table, and a reference pixel extension table; the reference data obtaining module 520 includes a reference table construction unit configured to: determine an angle mode corresponding to the intra prediction operation, and configure a corresponding filter coefficient for each angle mode; construct the filter coefficient table according to the filter coefficients corresponding to the plurality of angle modes; determine a reference pixel projection position corresponding to each row pixel in the prediction block in the horizontal prediction mode; construct the reference pixel projection table based on the deviation of the reference pixel projection position and the current row number; determine an extension reference pixel corresponding to the negative angle mode in the intra prediction coding, and construct the reference pixel extension table based on the extension reference pixel, wherein the number and position of the extension pixels of the extension reference pixel are determined based on the angle value of the negative angle mode.

[0108] In an example implementation of the present disclosure, the reference data obtaining module 520 includes a reference data obtaining unit configured to: determine a target angle mode corresponding to the pixel to be predicted; and obtain reference data from the pixel filter reference table based on the target angle mode, wherein the reference data includes one or more of a target filter coefficient, a target reference pixel, and a target extension reference pixel.

[0109] In an example implementation of the present disclosure, the reference data obtaining unit includes a first data obtaining subunit configured to: when the target angle mode is a vertical type mode, obtain a first filter coefficient of a row in which the target angle mode is located in the filter coefficient table as the target filter coefficient; and copy the target filter coefficient to a vector register for use in the intra prediction operation; and when the target angle mode is a horizontal type mode, obtain the filter coefficient of the row in which the target angle mode is located in the filter coefficient table as the target filter coefficient.

[0110] In an example implementation of the present disclosure, the reference data obtaining unit includes a second data obtaining subunit configured to: when the target angle mode is a horizontal type mode, obtain a reference pixel projection deviation corresponding to the target angle mode based on the reference pixel projection table; and determine a target reference pixel corresponding to the pixel to be predicted according to the reference pixel projection deviation; and when the target angle mode is a vertical type mode and a reference pixel change event in the intra prediction process is detected, determine a target reference pixel corresponding to the current row based on the reference pixel projection table.

[0111] In an example implementation of the present disclosure, the reference data obtaining unit includes a third data obtaining subunit configured to: when the target angle mode is a negative angle mode, obtain a target extension reference pixel from the reference pixel extension table; determine a main reference pixel corresponding to the pixel to be predicted; and store the target extension reference pixel and the main reference pixel in a stack together, so as to read pixel data from the stack through a load instruction.

[0112] With regard to the apparatus in the above-described embodiments, in which the specific manner in which each module performs operations has been described in detail in the embodiments related to the method, no detailed elaboration will be made here.

[0113] The electronic device 600 according to this embodiment of the present disclosure will be described below with reference to Figure 6 Figure 6 The displayed electronic device 600 is merely an example and should not bring any limitation to the function and use range of the embodiments of the present disclosure.

[0114] As Figure 6 shown, the electronic device 600 is in the form of a general computing device. The components of the electronic device 600 can include, but are not limited to, the above-mentioned at least one processing unit 610, the above-mentioned at least one storage unit 620, a bus 630 connecting different system components, including the storage unit 620 and the processing unit 610, and a display unit 640.

[0115] The storage unit stores program codes that can be executed by the processing unit 610, so that the processing unit 610 performs the steps according to various exemplary embodiments of the present disclosure described in the above "Exemplary Method" section of the present specification.

[0116] The storage unit 620 can include a readable medium in the form of a volatile storage unit, such as a random access memory (RAM) 621 and / or a cache memory 622, and can further include a read-only memory (ROM) 623.

[0117] The storage unit 620 can include program / utilities 624 having a set of (at least one) program modules 625, such as an operating system, one or more application programs, other program modules, and program data, each of which or some combination of which can include the implementation of a network environment.

[0118] The bus 630 can represent one or more of several types of bus structures, including a storage unit bus or storage unit controller, a peripheral bus, a graphics acceleration port, a processing unit, or a local bus using any of a variety of bus structures.

[0119] ​The electronic device 600 can also communicate with one or more external devices 670 such as a keyboard or pointing device, a Bluetooth device, or a database, and / or one or more devices that enable a user to interact with the electronic device 600 and / or one or more devices (e.g., a router, a modem, a server, etc.) that enable the electronic device 600 to communicate with one or more other computing devices. Such communication can occur via an input / output (I / O) interface 650. Still yet, the electronic device 600 can communicate with one or more networks, such as a local area network (LAN), a wide area network (WAN), and / or the Internet, through a network adapter 660. As depicted, the network adapter 660 communicates with the other components of the electronic device 600 via the bus 630. It should be appreciated that the network adapter 660 and / or the bus 630 can be implemented using one or more types of technology that are now known or later developed. For example, the network adapter 660 and / or the bus 630 can be implemented using any type of hardware, software, and / or firmware, including but not limited to a set of one or more microprocessors, a set of one or more micro-controllers, a set of one or more programmable logic devices, a set of one or more application-specific integrated circuits, and / or any other set of one or more hardware and / or software components.

[0120] In exemplary embodiments, a computer readable storage medium including instructions, such as a memory including instructions, is also provided. The instructions can be executed by a processor of an apparatus to perform the above-mentioned intra prediction method. Optionally, the computer readable storage medium can be a ROM, a random access memory (RAM), a CD-ROM, a magnetic tape, a floppy disc, and an optical data storage device, etc.

[0121] In exemplary embodiments, a computer program product including a computer program is also provided. The computer program, when executed by a processor, implements any of the above-mentioned intra prediction methods.

[0122] Other embodiments of the disclosure will be apparent to those skilled in the art from consideration of the specification and practice of the features of the disclosure disclosed herein. It is intended that the specification and examples be considered as exemplary only, with a true scope and spirit of the disclosure being indicated by the following claims.

[0123] It is to be understood that the present disclosure is not limited to the precise construction described and as shown in the attached drawings, and that various modifications and changes can be made by those skilled in the art without departing from the scope of the present disclosure. The scope of the present disclosure is limited only by the claims that follow.

Claims

1. An intra prediction method, characterized by, include: Obtain the intra-frame prediction coding logic corresponding to the pixel to be predicted, and determine the components to be operated on in the intra-frame prediction coding logic of the target instruction set. The components to be operated on include one or more of the initial summation component, the unit accumulation component, and the change operation component. Obtain a pre-constructed pixel filter reference table, and obtain reference data corresponding to the component to be processed from the pixel filter reference table; Intra-frame prediction is performed based on the intra-frame prediction coding logic and the reference data to obtain the predicted pixel value.

2. The method according to claim 1, characterized in that, The step of obtaining the intra-frame predictive coding logic corresponding to the pixel to be predicted and determining the component to be computed in the intra-frame predictive coding logic of the target instruction set includes: Obtain the intra-frame prediction coding logic corresponding to the pixel to be predicted in a specified angle mode, wherein the specified angle mode includes a planar mode; Based on the coding characteristics of the target instruction set, the component to be computed corresponding to the intra-frame predictive coding logic is determined.

3. The method according to claim 2, characterized in that, The step of obtaining the intra-frame prediction coding logic corresponding to the pixel to be predicted in the specified angle mode includes: Determine the current block to be predicted where the pixel to be predicted is located, and obtain the block size information of the current block to be predicted and the row and column information of the pixel to be predicted. The block size information includes the block height and the block width. Based on the row and column information, a reference pixel corresponding to the pixel to be predicted is determined, and the reference pixel includes an upper reference pixel and a left reference pixel; The intra-frame predictive coding logic is generated based on the block height, the block width, the row and column information, and the reference pixel. The intra-frame predictive coding logic is used to determine the predicted pixel value of the pixel to be predicted.

4. The method according to claim 2, characterized in that, The step of determining the component to be computed corresponding to the intra-frame predictive coding logic based on the coding characteristics of the target instruction set includes: Based on the coding characteristics of the target instruction set, the operational constants corresponding to the intra-frame predictive coding logic are determined; Based on the operational constant, the row computation component corresponding to the row prediction pixel in the intra-frame predictive coding logic is determined; Based on the row calculation components, determine the initial summation component and the unit accumulation component corresponding to the row prediction pixel; The change operation component is determined based on the intra-frame predictive coding logic, the initial summation component, and the unit accumulation component.

5. The method according to claim 1, characterized in that, The pixel filtering reference table includes a filtering coefficient table, a reference pixel projection table, and a reference pixel expansion table; The step of obtaining the pre-built pixel filtering reference table includes: Determine the angle mode corresponding to the intra-frame prediction operation, and configure corresponding filter coefficients for each angle mode; The filter coefficient table is constructed based on the filter coefficients corresponding to the multiple angle modes. Determine the reference pixel projection position for each row pixel in the prediction block under horizontal prediction mode; Based on the deviation between the reference pixel projection position and the current row number, the reference pixel projection table is constructed; Determine the extended reference pixels corresponding to the intra-frame predictive coding in the negative angle mode, construct the reference pixel extension table based on the extended reference pixels, and determine the number of extended pixels and the position of the extended pixels based on the angle value of the negative angle mode.

6. The method according to claim 5, characterized in that, The step of obtaining the reference data corresponding to the component to be processed from the pixel filtering reference table includes: Determine the target angle pattern corresponding to the pixel to be predicted; Based on the target angle pattern, the reference data is obtained from the pixel filter reference table. The reference data includes one or more of the target filter coefficients, target reference pixels, and target extended reference pixels.

7. The method according to claim 6, characterized in that, The step of obtaining reference data from the pixel filtering reference table based on the target angle pattern includes: When the target angle pattern is a vertical pattern, the first filter coefficient in the row containing the target angle pattern is obtained from the filter coefficient table and used as the target filter coefficient. The target filter coefficients are copied to the vector register for use in intra-frame prediction operations; When the target angle pattern is a horizontal pattern, the filter coefficient of the row containing the target angle pattern is obtained from the filter coefficient table and used as the target filter coefficient.

8. The method according to claim 6, characterized in that, The step of obtaining reference data from the pixel filtering reference table based on the target angle pattern includes: When the target angle mode is a horizontal mode, the reference pixel projection deviation corresponding to the target angle mode is obtained based on the reference pixel projection table. Based on the projection deviation of the reference pixel, the target reference pixel corresponding to the pixel to be predicted is determined; When the target angle mode is a vertical mode and a reference pixel change event is detected during the intra-frame prediction process, the target reference pixel corresponding to the current row is determined based on the reference pixel projection table.

9. The method according to claim 6, characterized in that, The step of obtaining reference data from the pixel filtering reference table based on the target angle pattern includes: When the target angle mode is a negative angle mode, the target expanded reference pixel is obtained from the reference pixel expansion table; The primary reference pixel corresponding to the pixel to be predicted is determined, and the target extended reference pixel and the primary reference pixel are stored together in a stack so that pixel data can be read from the stack through a load instruction.

10. An intra-frame prediction device, characterized in that, include: The computational component determination module is used to obtain the intra-frame prediction coding logic corresponding to the pixel to be predicted, and determine the computational component of the target instruction set in the intra-frame prediction coding logic. The computational component includes one or more of the initial summation component, the unit accumulation component, and the change computational component. The reference data acquisition module is used to acquire a pre-built pixel filter reference table and obtain reference data corresponding to the component to be processed from the pixel filter reference table; The intra-frame prediction module is used to perform intra-frame prediction operations based on the intra-frame prediction coding logic and the reference data to obtain predicted pixel values.

11. An electronic device, characterized in that, include: processor; Memory used to store the processor's executable instructions; The processor is configured to execute the instructions to implement the intra-frame prediction method as described in any one of claims 1 to 9.

12. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by the processor, it implements the intra-frame prediction method according to any one of claims 1 to 9.

Images