WO2020172813A1

WO2020172813A1 - Rate distortion optimization method and apparatus, and computer-readable storage medium

Info

Publication number: WO2020172813A1
Application number: PCT/CN2019/076305
Authority: WO
Inventors: 周益民; 冷龙韬
Original assignee: Oppo广东移动通信有限公司
Priority date: 2019-02-27
Filing date: 2019-02-27
Publication date: 2020-09-03
Also published as: CN112970254B; CN112970254A

Abstract

Provided are a rate distortion optimization method and apparatus, and a computer-readable storage medium. The method can comprise: acquiring a current frame of image to be coded; if the current frame of image to be coded is not the first frame of image to be coded, acquiring an initial Lagrange multiplier, a coding bit rate of a coding unit level of the previous frame of image to be coded and an estimated Lagrange multiplier of the coding unit level of the previous frame of image to be coded; according to the initial Lagrange multiplier, the coding bit rate of the coding unit level of the previous frame of image to be coded, the estimated Lagrange multiplier of the coding unit level of the previous frame of image to be coded, a pre-set coding bit rate and an estimated Lagrange multiplier model, obtaining an estimated Lagrange multiplier of a coding unit level of the current frame of image to be coded; and using the estimated Lagrange multiplier of the coding unit level of the current frame of image to be coded to perform rate distortion processing of the current frame of image to be coded, so as to complete the coding of the current frame of image to be coded.

Description

Rate-distortion optimization method and device, and computer readable storage medium

Technical field

The embodiments of the present application relate to video coding technology, and in particular to a method and device for optimizing rate-distortion, and a computer-readable storage medium.

Background technique

The rate-distortion optimization technique is a technique widely used in video coding. The rate-distortion optimization technique selects a set of parameters in the encoding parameter set so that the encoding result can obtain the smallest image distortion at a limited bit rate.

At present, the current rate-distortion optimization of video coding is performed under the assumption that the coding units are independent, and the Lagrangian multiplier method is used to obtain unlimited coding cost. For example, the cost function: min{J= D+λ·R} to get the coding cost, so as to select the best coding mode among multiple coding modes, and use the best coding mode for video coding; where J represents the coding cost, and λ is the introduced Lager Lange multiplier, D represents the distortion of the reconstructed image, and R represents the coding bit rate.

However, due to the Lagrangian optimization theory, the value of λ is calculated by setting the first-order differential of the cost function to be constant at zero. Under the assumption of high bit rate, the distortion of the image is related to the quantization step size, and the Lagrangian is finally obtained. The daily multiplier λ is a quantity that is only positively correlated with the square of the quantization step. In the coding process of a video image, an image is divided into several equal-sized coding blocks, namely coding units. When the square of the quantization step size is constant, all coding units in an image will share a pre-calculated λ value. This is somewhat different from the actual different coding units corresponding to different Lagrangian multipliers. That is to say, the estimation of the coding cost of different coding units of a frame of image is inaccurate when using the existing λ. The problem leads to inaccurate rate distortion estimation, which affects the coding efficiency of the image.

Summary of the invention

The embodiments of the present application provide a rate-distortion optimization method and device, and a computer-readable storage medium, which can improve coding efficiency.

The technical solutions of the embodiments of the present application are implemented as follows:

An embodiment of the present application provides a rate-distortion optimization method, including:

Acquiring the image to be encoded in the current frame;

If the image to be encoded in the current frame is not the first frame to be encoded, the initial Lagrangian multiplier, the encoding unit-level encoding bit rate of the image to be encoded in the previous frame, and the encoding unit of the image to be encoded in the previous frame are acquired Estimated Lagrange multiplier of the order;

According to the initial Lagrangian multiplier, the coding unit-level coding bit rate of the image to be coded in the previous frame, and the estimated Lagrangian multiplier and pre-coding unit level of the image to be coded in the previous frame Set the coding bit rate and the estimated Lagrangian multiplier model to obtain the estimated Lagrangian multiplier at the coding unit level of the image to be coded in the current frame;

The estimated Lagrangian multiplier at the coding unit level of the image to be encoded in the current frame is used to perform rate-distortion processing of the image to be encoded in the current frame to complete the encoding of the image to be encoded in the current frame.

In the above solution, if the image to be encoded in the current frame is not the first image to be encoded, the initial Lagrangian multiplier, the encoding unit-level encoding bit rate of the image to be encoded in the previous frame, and the previous image are acquired. The estimated Lagrangian multiplier at the coding unit level of the frame to be encoded includes:

If the image to be encoded in the current frame is not the first image to be encoded, the i-th coding unit of the image to be encoded in the current frame, the initial Lagrangian multiplier, and the image to be encoded in the previous frame are acquired. The coding bit rate of i coding units and the estimated Lagrangian multiplier of the i-th coding unit of the image to be coded in the previous frame, where i is a positive integer greater than or equal to 1 and less than or equal to N, and N is a frame to be coded The total number of coding units corresponding to the coded image.

In the above solution, the coding unit-level coding bit rate of the image to be coded in the previous frame, and the coding unit-level estimation of the image to be coded in the previous frame are adjusted according to the initial Lagrangian multiplier, The Grange multiplier, the preset coding bit rate and the estimated Lagrangian multiplier model to obtain the estimated Lagrangian multiplier at the coding unit level of the image to be coded in the current frame includes:

According to the initial Lagrangian multiplier, the coding bit rate of the i-th coding unit of the image to be coded in the previous frame, and the estimated Lagrangian of the i-th coding unit of the image to be coded in the previous frame A multiplier and the preset encoding bit rate and an estimated Lagrangian multiplier model to obtain an estimated Lagrangian multiplier of the i-th coding unit of the image to be encoded in the current frame;

Add 1 to i to obtain the estimated Lagrangian multiplier of the i+1th coding unit of the image to be coded in the current frame until the estimated Lagrangian multiplier of the Nth coding unit of the image to be coded in the current frame is obtained Until the multiplier.

In the above solution, the estimated Lagrangian multiplier at the coding unit level of the image to be encoded in the current frame is used to perform rate-distortion processing of the image to be encoded in the current frame to complete the image to be encoded in the current frame The encoding includes:

Use the estimated Lagrangian multiplier of the first coding unit of the image to be coded in the current frame to the estimated Lagrangian multiplier of the Nth coding unit of the image to be coded in the current frame to perform the current frame The rate-distortion processing of the image to be encoded further completes the encoding of the image to be encoded in the current frame.

In the above solution, the coding unit-level coding bit rate of the image to be coded in the previous frame, and the coding unit-level estimation of the image to be coded in the previous frame are adjusted according to the initial Lagrangian multiplier, Before the Grange multiplier and the preset coding bit rate and the estimated Lagrangian multiplier model are used to obtain the estimated Lagrangian multiplier at the coding unit level of the image to be coded in the current frame, the method further includes:

Obtain the initial coding bit rate and the Lagrangian multiplier model, the preset frame-level coding bit rate model, the preset time-domain proximity approximate coding bit rate model, and the preset time-domain proximity approximate frame-level coding bit rate model;

According to the initial coding bit rate and the Lagrangian multiplier model, the preset frame-level coding bit rate model, the preset time domain proximity approximate coding bit rate model, and the preset time domain proximity approximate frame level The coding bit rate model is used to obtain the preset coding bit rate and the estimated Lagrangian multiplier model.

In the above solution, the preset frame-level coding bit rate model, the preset temporal proximity approximate coding bit rate model, and the pre-coding bit rate model are based on the initial coding bit rate and the Lagrangian multiplier model. Supposing the temporal proximity approximate frame-level coding bit rate model, obtaining the preset coding bit rate and the estimated Lagrangian multiplier model includes:

According to the initial coding bit rate and Lagrangian multiplier model, the theoretical coding bit rate and estimated Lagrangian multiplier model of the coding unit and the actual coding bit rate and estimated Lagrangian multiplier model of the coding unit are derived ；

Obtaining a theoretical optimal Lagrangian multiplier model at the coding unit level according to the theoretical coding bit rate and the estimated Lagrangian multiplier model and the preset frame-level coding bit rate model;

According to the actual encoding bit rate and the estimated Lagrangian multiplier model and the preset frame-level encoding bit rate model, obtain the actual frame-level encoding bit rate model of one frame;

According to the theoretical optimal Lagrangian multiplier model, the preset temporal proximity approximate coding bit rate model, the preset temporal proximity approximate frame-level coding bit rate model, and the actual frame of one frame Level coding bit rate model to obtain the preset coding bit rate and the estimated Lagrangian multiplier model.

In the above solution, after the obtaining the image to be encoded in the current frame, the method further includes:

If the image to be encoded in the current frame is the image to be encoded in the first frame, obtaining an initial Lagrangian multiplier;

The initial Lagrangian multiplier is used to perform rate-distortion processing of the image to be encoded in the current frame, thereby completing the encoding of the image to be encoded in the current frame.

In the above solution, the estimated Lagrangian multiplier at the coding unit level of the image to be encoded in the current frame is used to perform rate-distortion processing of the image to be encoded in the current frame to complete the image to be encoded in the current frame After the encoding, the method further includes:

Enter the rate-distortion optimization process of the next frame to be encoded;

The encoding of the image to be encoded until the last frame of the video sequence to be encoded is processed.

An embodiment of the present application provides a rate-distortion optimization device, including:

A processor, a memory storing a rate-distortion optimization instruction executable by the processor, and a communication bus for connecting the processor and the memory, and when the rate-distortion optimization instruction is executed, the above-mentioned rate distortion is realized Optimization.

An embodiment of the present application provides a computer-readable storage medium on which a rate-distortion optimization instruction is stored, where the rate-distortion optimization instruction is executed by a processor to implement the above-mentioned rate-distortion optimization method.

In the embodiment of the application, the above-mentioned technical implementation scheme is adopted, and the rate-distortion optimization device performs rate-distortion optimization processing on the image to be encoded in the current frame. During the encoding process, the encoding unit level of the image to be encoded in the previous frame that has been encoded can be used. The coding bit rate and the estimated Lagrangian multiplier are combined with the preset coding bit rate and the estimated Lagrangian multiplier model to obtain the estimated Lagrangian multiplier at the coding unit level of the image to be coded in the current frame, and then Realize the rate-distortion optimization processing and encoding, because the rate-distortion optimization device can intervene in the estimated Lagrangian of all coding units of the image to be encoded in the current frame based on the bit rate of the image to be encoded in the previous frame and the estimated Lagrangian multiplier The multiplier of the day, in turn, interferes with the balance of the bit distribution of the coding unit within the frame, thereby improving the coding efficiency.

Description of the drawings

FIG. 1 is a first flowchart of a rate-distortion optimization method provided by an embodiment of this application;

FIG. 2 is a second flowchart of a rate-distortion optimization method provided by an embodiment of the application;

Figure 3 is an exemplary current frame to-be-encoded image provided by an embodiment of the application;

FIG. 4 is a schematic diagram of encoding after turning off the exemplary rate-distortion optimization method provided by the present application;

FIG. 5 is a schematic diagram of encoding after the exemplary rate-distortion optimization method provided by the present application is turned on;

Fig. 6 is an analysis diagram of coding unit division after closing the exemplary rate-distortion optimization method provided by the present application;

FIG. 7 is an analysis diagram of coding unit division after the exemplary rate-distortion optimization method provided by the present application is turned on;

FIG. 8 is a diagram showing the division and analysis of two coding units after the rate-distortion optimization method provided by the present application is turned off;

FIG. 9 is an analysis diagram of the division of two coding units after the rate-distortion optimization method provided by the present application is turned on;

FIG. 10 is a first structural diagram of a rate-distortion optimization device provided by an embodiment of the application;

FIG. 11 is a second structural diagram of a rate-distortion optimization apparatus provided by an embodiment of the application.

detailed description

The technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. It is understandable that the specific embodiments described here are only used to explain the related application, but not to limit the application. In addition, it should be noted that, for ease of description, only the parts related to the relevant application are shown in the drawings.

In the embodiment of the present application, the rate-distortion optimization method is applied in the video encoding process. Video coding methods, including prediction, transformation, quantization and entropy coding. Among them, the prediction method includes intra-frame prediction and inter-frame prediction, and the coding mode, motion vector, and reconstruction data in the prediction information of the current coding unit are used to assist in predicting the subsequent coding unit.

In the embodiments of this application, due to the extensive use of temporal prediction technology, there is a strong correlation between coding units, that is, the coding performance of the current coding unit will affect the encoding of subsequent coding units. Therefore, the embodiments of this application are based on time. The idea of domain proximity is used to obtain the optimal Lagrangian multiplier to realize the rate-distortion optimization method.

The technical solution of the present application will be further elaborated below in conjunction with the drawings and embodiments.

It should be noted that the rate-distortion optimization device provided in the embodiment of the present application may be an encoder, which is not limited in the embodiment of the present application.

The embodiment of the present application provides a rate-distortion optimization method. As shown in FIG. 1, the method may include:

S101. Obtain an image to be encoded in the current frame.

S102. If the image to be encoded in the current frame is not the first frame to be encoded, acquire the initial Lagrangian multiplier, the encoding unit-level encoding bit rate of the image to be encoded in the previous frame, and the encoding unit of the image to be encoded in the previous frame Estimated Lagrange multiplier of the order;

S103. According to the initial Lagrangian multiplier, the coding unit-level coding bit rate of the image to be coded in the previous frame, and the estimated Lagrangian multiplier and the preset coding bit rate at the coding unit level of the image to be coded in the previous frame With the estimated Lagrangian multiplier model, the estimated Lagrangian multiplier of the coding unit level of the image to be encoded in the current frame is obtained.

S104. Using the estimated Lagrangian multiplier at the coding unit level of the image to be encoded in the current frame, perform rate-distortion processing of the image to be encoded in the current frame, and then complete the encoding of the image to be encoded in the current frame.

In the embodiment of the present application, in the process of encoding the video sequence to be encoded, the rate-distortion optimization apparatus may perform rate-distortion optimization processing for the frame-level coding unit, and then select the image of each frame in the video sequence to be encoded. The optimal coding mode of each coding unit adopts the optimal coding mode to encode each coding unit of each frame of the image to be coded, thereby completing the coding of the video sequence to be coded.

It should be noted that, in the embodiment of the present application, the most important process of the rate-distortion optimization device for the rate-distortion optimization of each coding unit is the process of obtaining the Lagrangian multiplier corresponding to each coding unit.

In S101, the rate-distortion optimization apparatus obtains the current frame to be encoded image in the to-be-encoded video sequence, where the current frame to be encoded image may be the first frame to be encoded image or the non-first frame to be encoded image.

It should be noted that, in the embodiment of the present application, during the encoding process of the video sequence, an image is divided into several blocks of equal size. These divided blocks are called coding units, which are the basic units of coding. The coding unit represents the local information of the image. If the spatial correlation between the coding units is not considered, the sum of the cost of all the coding units in an image can be regarded as the cost of the entire image. Therefore, the rate-distortion optimization problem in the embodiment of the present application can be solved at the coding unit level.

In S102, in the embodiment of the present application, the rate-distortion optimization apparatus uses the system Lagrangian multiplier (ie, the initial Lagrangian multiplier) as the first frame to be encoded for the first frame of the image to be encoded in the video sequence to be encoded. Each coding unit in the coded image corresponds to the Lagrangian multiplier, and the rate-distortion optimization device regards the current frame to be coded as a non-first frame to be coded, and uses the coded bits of the image to be coded adjacent to the time domain Rate and estimate the Lagrangian multiplier (the Lagrangian multiplier actually used in the previous frame) to calculate the Lagrangian multiplier corresponding to each coding unit in the image to be encoded in the current frame.

In the embodiment of the present application, when the image to be encoded in the current frame is not the first frame to be encoded, the rate-distortion optimization device obtains the initial Lagrangian multiplier, the encoding unit-level encoding bit rate of the image to be encoded in the previous frame, and The estimated Lagrangian multiplier at the coding unit level of the image to be coded in the previous frame, so that the initial Lagrangian multiplier, the coding unit-level coding bit rate of the image to be coded in the previous frame and the previous frame to be coded are subsequently used The estimated Lagrangian multiplier at the coding unit level of the image acquires the Lagrangian multiplier of the coding unit of the image to be coded in the current frame.

It should be noted that the initial Lagrangian multiplier in the embodiment of the present application can be obtained according to a preset Lagrangian multiplier model.

In some embodiments of the present application, the preset Lagrangian multiplier model is formula (1), as follows:

Among them, c is a constant, taking an empirical value; λ is the initial Lagrangian multiplier, that is, the system Lagrangian multiplier, and q _step represents the quantization step length.

It should be noted that, in the embodiment of the present application, the acquisition process of the preset Lagrangian multiplier model may be as follows:

The embodiment of the application adopts the Lagrangian method. According to the Lagrangian optimization theory, the value of λ is calculated by setting the first-order differential of the cost function to be constant, as shown in formula (2):

Among them, D represents the distortion of the reconstructed image, J represents the cost of encoding, and R represents the encoding bit rate, which can also be understood as the number of bits for encoding a unit.

Since it is impossible to obtain the real bit rate R before encoding, the calculation of formula (2) can only be derived from an empirical mathematical model. Under the assumption of high bit rate, the relationship between image distortion and quantization step size is generally represented by (3), as follows:

Among them, q _step represents the quantization step length, which is uniquely determined by the quantization parameter.

Moreover, the functional relationship between the encoding bit rate R and the distortion D of the reconstructed image is generally expressed by formula (4), as follows:

Among them, δ is used to indicate the intensity of image changes.

In this way, according to formulas (2), (3) and (4), formula (1) of the preset Lagrangian multiplier model can be derived.

In the embodiments of the present application, c can be obtained through experience or experiment. For example, c is 0.85, which is not limited in the embodiments of the present application.

It should be noted that, from the formula (1), under the assumption of high bits, λ is a quantity that is only positively correlated with the square of the quantization step size. Since the quantization step q _step is uniquely determined by the quantization parameter QP, when the quantization parameter QP is given, the value of λ will be directly calculated.

It should be noted that, in the embodiment of the present application, the initial Lagrangian multiplier λ _sys is calculated according to formula (1) when the quantization parameter QP of the coding system is determined.

In the embodiment of this application, when the rate-distortion optimization device is processing the image to be encoded in the current frame, the encoding of the image to be encoded in the previous frame has been completed, and the rate-distortion processing and encoding process are performed in each frame of the image to be encoded. The rate-distortion optimization device can record the encoding bit rate of each coding unit of each frame of the image to be encoded and estimate the Lagrangian multiplier. Therefore, the rate-distortion optimization device can process the image to be encoded in the current frame. , You can directly obtain the coding unit-level coding bit rate of the image to be coded in the previous frame and the estimated Lagrangian multiplier at the coding unit level of the image to be coded in the previous frame.

In some embodiments of the present application, the image to be encoded in the current frame is represented as the image to be encoded at time t, then the encoding unit-level encoding bit rate of the image to be encoded in the previous frame is

Indicates that the estimated Lagrangian multiplier of the coding unit level of the image to be coded in the previous frame adopts

It means that the embodiments of this application are not limited.

In S103, the rate-distortion optimization device obtains the initial Lagrangian multiplier, the coding unit-level coding bit rate of the image to be encoded in the previous frame, and the estimated Lagrangian coefficient at the coding unit level of the image to be encoded in the previous frame. After the multiplier, the parameters obtained above can be combined with the preset encoding bit rate and the estimated Lagrangian multiplier model to obtain the estimated Lagrangian multiplier at the coding unit level of the image to be encoded in the current frame.

In the embodiment of the present application, the preset encoding bit rate and the estimated Lagrangian multiplier model are R-λ models, and the preset encoding bit rate and the estimated Lagrangian multiplier model are the sum of the encoded bits of the frame-level coding unit. Approximate the optimal estimation of the correspondence between Lagrangian multipliers.

In some embodiments of the present application, the preset coding bit rate and the estimated Lagrangian multiplier model can be expressed as formula (5), as follows:

among them,

Represents the estimated Lagrangian multiplier at the coding unit level of the image to be encoded in the current frame, λ _sys represents the initial Lagrangian multiplier,

Indicates the coding unit-level coding bit rate of the image to be coded in the previous frame,

It represents the estimated Lagrangian multiplier at the coding unit level of the image to be coded in the previous frame, and β represents the model parameter, which is determined by data fitting.

It is understandable that the preset coding bit rate and the estimated Lagrangian multiplier model represent the R-λ model between the coding units to be coded adjacent in the time domain. In this way, the rate-distortion optimization device can determine the Lagrangian multiplier of the corresponding coding unit of the current frame through the coded coding unit of the previous frame. Since the Lagrangian multiplier λ is the first derivative of the image distortion and the encoding bit rate, its physical meaning is the slope of the rate-distortion curve (R-D curve) at a given point (R, D). Then, based on the diversity of the source changes of the video image, the shape of the rate-distortion curve of the video sequence to be encoded is closely related to the texture of the source content and the picture motion. Therefore, each video sequence to be coded, each frame of image to be coded, and even each coding unit corresponds to a unique rate-distortion curve. Therefore, different coding units will correspond to different fixed-point slope values, and each coding unit Lagrangian The value of the multiplier λ will vary. The rate-distortion optimization device calculates Lagrangian multipliers for each coding unit of the current frame of the image to be coded by analyzing the bit distribution characteristics of the image to be coded in the previous frame that has been coded, aiming to control the bit consumption of the coding unit in a targeted manner , Improve the overall coding efficiency of the image to be coded in the current frame.

In S104, after the rate-distortion optimization device obtains the estimated Lagrangian multiplier at the coding unit level of the image to be coded in the current frame, the rate-distortion optimization device can use the coding unit-level estimation of the image to be coded in the current frame. The Grange multiplier performs rate-distortion processing of the image to be encoded in the current frame, selects the best encoding mode of the encoding unit, and then uses the best encoding mode to encode the encoding unit to complete the encoding of the image to be encoded in the current frame.

It should be noted that in the embodiment of the present application, the rate-distortion optimization device uses a cost function to perform rate-distortion processing on the image to be encoded in the current frame to obtain the encoding cost, and compare the encoding costs in different encoding modes according to the encoding cost to select the smallest cost The best encoding mode.

In some embodiments of the present application, for a frame of image to be encoded, the cost function can be expressed as formula (6), as follows:

min{J _i ＝D _i +λ _i ·R _i } (6)

Among them, J _i represents the coding cost of the i-th coding unit in a frame of image to be coded, λ _i represents the Lagrangian multiplier of the i-th coding unit in a frame of image to be coded, and D _i represents a frame of image to be coded The distortion of the reconstructed image of the i-th coding unit in, R _i represents the coding bit rate of the i-th coding unit in a frame of image to be coded.

In the embodiment of the present application, the estimated Lagrangian multiplier at the coding unit level of the image to be coded in the current frame is used to perform rate-distortion processing for each coding unit of the image to be coded in the current frame to obtain the coding cost, thereby completing the current frame The encoding of the image to be encoded.

It can be understood that the rate-distortion optimization device performs rate-distortion optimization processing on the image to be encoded in the current frame. During the encoding process, the encoding unit-level encoding bit rate of the image to be encoded in the previous frame that has been encoded can be used to estimate the bit rate. The Grange multiplier combines the preset coding bit rate and the estimated Lagrangian multiplier model to obtain the estimated Lagrangian multiplier at the coding unit level of the image to be coded in the current frame, thereby realizing the rate-distortion optimization processing, Realize encoding, because the rate-distortion optimization device can intervene in the estimated Lagrangian multipliers of all coding units of the image to be encoded in the current frame based on the bit rate of the image to be encoded in the previous frame and the estimated Lagrangian multiplier, thereby intervening in the frame The bit distribution of the inner coding unit tends to be balanced, thereby improving the coding efficiency.

Based on the implementation of the foregoing embodiment, the following describes in detail the process of rate-distortion optimization for each coding unit of the image to be encoded in the current frame. As shown in FIG. 2, the specific implementation process includes: S201-S205. as follows:

S201: Acquire an image to be encoded in the current frame.

S202. If the image to be encoded in the current frame is not the first image to be encoded, obtain the i-th coding unit of the image to be encoded in the current frame, the initial Lagrangian multiplier, and the i-th encoding unit of the image to be encoded in the previous frame. The coding bit rate and the estimated Lagrangian multiplier of the i-th coding unit of the image to be coded in the previous frame, where i is a positive integer greater than or equal to 1 and less than or equal to N, and N is the corresponding to a frame of image to be coded The total number of coding units.

S203. According to the initial Lagrangian multiplier, the coding bit rate of the i-th coding unit of the image to be coded in the previous frame, the estimated Lagrangian multiplier and the pre-coding unit of the i-th coding unit of the image to be coded in the previous frame Assuming the coding bit rate and the estimated Lagrangian multiplier model, the estimated Lagrangian multiplier of the i-th coding unit of the image to be coded in the current frame is obtained.

S204. Add 1 to i to obtain the estimated Lagrangian multiplier of the i+1th coding unit of the image to be coded in the current frame, until the estimated Lagrangian of the Nth coding unit of the image to be coded in the current frame is obtained Longer multiplier.

S205. Use the estimated Lagrangian multiplier of the first coding unit of the image to be coded in the current frame to the estimated Lagrangian multiplier of the Nth coding unit of the image to be coded in the current frame, and perform the calculation of the image to be coded in the current frame. Rate distortion processing, and then complete the encoding of the image to be encoded in the current frame.

In the embodiment of the present application, after the rate-distortion optimization device obtains the image to be encoded in the current frame of the video sequence to be encoded, it can process each coding unit in the image to be encoded in the current frame. Is the image to be encoded in the non-first frame, the rate-distortion optimization device obtains the encoding bit rate of the i-th coding unit of the image to be encoded in the current frame, the initial Lagrangian multiplier, and the encoding bit rate of the i-th encoding unit of the image to be encoded in the previous frame And the estimated Lagrangian multiplier of the i-th coding unit of the image to be coded in the previous frame, based on the relevant parameters of the corresponding i-th coding unit of the image to be coded in the previous frame, to obtain the i-th coding unit of the image to be coded in the current frame The estimated Lagrangian multiplier of a coding unit is used to realize rate-distortion processing and coding.

It should be noted that in the embodiment of the present application, for a frame of image to be encoded, each frame of image to be encoded can be divided into N coding units. Therefore, there may be N coding units in the image to be encoded in the current frame. Starting from the first coding unit and proceeding to the Nth coding unit in sequence, the processing of the coding cost of the image to be coded in the current frame is completed, the best coding mode is selected, and the coding of the image to be coded in the current frame is performed.

In this embodiment of the present application, i is a positive integer greater than or equal to 1 and less than or equal to N, and N is the total number of coding units corresponding to a frame of image to be encoded.

In the embodiment of the present application, the rate-distortion optimization device for the i-th coding unit of the unit to be coded in the current frame is based on the initial Lagrangian multiplier and the coding bit rate of the i-th coding unit of the image to be coded in the previous frame , The estimated Lagrangian multiplier and preset coding bit rate and estimated Lagrangian multiplier model of the i-th coding unit of the image to be encoded in the previous frame, that is, according to formula (5) to obtain the image of the current frame to be encoded The estimated Lagrangian multiplier of the i-th coding unit is obtained. After the estimated Lagrangian multiplier of the i-th coding unit of the image to be encoded in the current frame is obtained, the i-th coding unit can be rate-distorted Processing and encoding, so as to obtain the encoding bit rate of the i-th coding unit of the image to be encoded in the current frame, so that the rate-distortion optimization device can use the encoding bit rate of the i-th encoding unit of the image to be encoded in the current frame and the current frame to be encoded The estimated Lagrangian multiplier of the i-th coding unit of the image is obtained, and the estimated Lagrangian multiplier of the i+1-th coding unit of the image to be coded in the current frame is obtained, that is, i is increased by 1, and the current frame is waiting To obtain the estimated Lagrangian multiplier of the i+1th coding unit of the coded image, the rate-distortion optimization device adopts such a loop implementation method until the estimated Lagrangian of the Nth coding unit of the image to be coded in the current frame is obtained By the date of multiplier, the encoding of the image to be encoded in the current frame is completed. That is, the rate-distortion optimization device uses the estimated Lagrangian multiplier of the first coding unit of the image to be coded in the current frame to the estimated Lagrangian multiplier of the Nth coding unit of the image to be coded in the current frame to perform the current frame waiting The rate-distortion processing of each coding unit of the coded image completes the coding of the image to be coded in the current frame.

It should be noted that when the image to be encoded in the current frame is a non-first frame image, each encoding unit thereof is encoded according to the process of S201-S205.

It is understandable that the rate-distortion optimization device uses the estimated Lagrangian multiplier of the frame-level coding unit to perform rate-distortion processing, and the rate-distortion optimization device can be based on the bit rate of the previous frame to be encoded and the estimated pull The Grange multiplier interferes with the estimated Lagrange multiplier of all coding units of the image to be coded in the current frame, thereby interfering with the bit distribution of the coding unit in the frame to balance, thereby improving the coding efficiency.

In some embodiments of the present application, before S103, a rate-distortion optimization method provided in an embodiment of the present application further includes: S105 and S106. as follows:

S105. Obtain an initial coding bit rate and a Lagrangian multiplier model, a preset frame-level coding bit rate model, a preset time-domain proximity approximate coding bit rate model, and a preset time-domain proximity approximate frame-level coding bit rate model;

S106. According to the initial coding bit rate and the Lagrangian multiplier model, the preset frame-level coding bit rate model, the preset time-domain proximity approximate coding bit rate model, and the preset time-domain proximity approximate frame-level coding bit rate model, obtain Preset encoding bit rate and estimated Lagrangian multiplier model.

In the embodiment of the present application, before S103, the rate-distortion optimization apparatus obtains the preset coding bit rate and the estimated Lagrangian multiplier model by first obtaining the known initial coding bit rate and the Lagrangian multiplier model , Preset frame-level coding bit rate model, preset time-domain proximity approximate coding bit rate model, and preset time-domain proximity approximate frame-level coding bit rate model; then based on the above known model, the preset coding bit rate and estimate Lagrange multiplier model formula (5).

In some embodiments of the present application, the rate-distortion optimization device is based on the initial encoding bit rate and the Lagrangian multiplier model, the preset frame-level encoding bit rate model, the preset temporal proximity approximate encoding bit rate model, and the preset timing. The domain proximity approximates the frame-level coding bit rate model, the process of obtaining the preset coding bit rate and the estimated Lagrangian multiplier model includes: S1061-S1064. as follows:

S1061, according to the initial coding bit rate and Lagrangian multiplier model, deduces the theoretical coding bit rate and estimated Lagrangian multiplier model of the coding unit and the actual coding bit rate and estimated Lagrangian multiplier model of the coding unit ；

S1062, according to the theoretical encoding bit rate and the estimated Lagrangian multiplier model and the preset frame-level encoding bit rate model, obtain the theoretical optimal Lagrangian multiplier model at the coding unit level;

S1063: Obtain the actual frame-level encoding bit rate model of one frame according to the actual encoding bit rate and the estimated Lagrangian multiplier model and the preset frame-level encoding bit rate model;

S1064. According to the theoretical optimal Lagrangian multiplier model, the preset temporal proximity approximate coding bit rate model, the preset temporal proximity approximate frame-level coding bit rate model and the actual frame-level coding bit rate model of one frame, Obtain the preset coding bit rate and the estimated Lagrangian multiplier model.

In the embodiment of the present application, the rate-distortion optimization device can deduce the theoretical encoding bit rate of the coding unit and the estimated Lagrangian multiplier model based on the initial encoding bit rate and the Lagrangian multiplier model.

In the embodiments of this application, the initial encoding bit rate and the Lagrangian multiplier model represent the R-λ model, which is the currently adopted R-λ model, which represents the unity of the initial encoding bit rate and the system Lagrangian multiplier Correspondence, the embodiment of this application does not limit its manifestation.

In some embodiments of the present application, the initial encoding bit rate and the Lagrangian multiplier model can be expressed as formula (7), as follows:

R=α·λ ^β (7)

Among them, α and β are model parameters, which need to be determined by data fitting in the actual use process.

In the embodiment of this application, theoretically, according to formula (7), it can be inferred that at time t, for the i-th coding unit of the image to be coded in the current frame, the optimal Lagrangian multiplier of the i-th coding unit

The ratio β to the system Lagrangian multiplier λ _{sys of the} system configuration is numerically equal to the optimal use of the coding unit i

Bit rate generated by encoding

The ratio of the bit rate R _t,i (λ _sys ) generated by the λ _sys encoding of the current frame to the image to be encoded using the system configuration is the theoretical encoding bit rate of the coding unit and the estimated Lagrangian multiplier model, as shown in formula (8 ) Shows:

Among them, the theoretical coding bit rate of the coding unit and the estimated Lagrangian multiplier model calculates the ratio of the two cases. Without loss of generality, the coding unit i at time t, if the system Lagrangian multiplier λ _{sys is} used for coding, the resulting bit rate is represented by R _t,i (λ _sys ); if the ideal optimal drawing is used Grange Multiplier

Encoding, the resulting bit rate is

Representation, that is, R _t,i (λ _sys ) represents the bit rate of each coding unit of the frame. Optimal Lagrangian multiplier

To seek an approaching goal.

Calculate according to formula (9)

λ _{sys is} known, R _t,i (λ _sys ) and

It cannot be read directly, so R _t,i (λ _sys ) and

Perform modeling.

It should be noted that the embodiment of the present application constructs a hypothetical "frame-level abstract coding unit", which represents the overall coding situation of the entire frame image. The bit rate of the "frame-level abstract coding unit" adopts the actual average bit rate of all coding units of the frame image. Taking the “frame-level abstract coding unit” corresponding to the system Lagrangian multiplier λ _sys as an example, the coding bit rate R _t,i (λ _sys ) of the “frame-level abstract coding unit” is obtained.

It is the average bit rate of the "frame-level abstract coding unit".

That is, in the embodiment of the present application, the preset frame-level coding bit rate model represents the average bit rate of each coding unit of a frame of image to be coded and the system Lagrangian multiplier (initial Lagrangian multiplier) The corresponding relationship.

Exemplarily, it is assumed that the preset frame-level coding bit rate model of the frame to be coded at time t can be expressed as formula (9), as follows:

among them,

It is expressed as the average bit rate of the frame, which is used to directly replace R _t,i (λ _sys ).

In the embodiment of the present application, after the rate-distortion optimization device obtains the theoretical encoding bit rate and the estimated Lagrangian multiplier model and the preset frame-level encoding bit rate model, it can calculate the theoretical encoding bit rate and the estimated Lagrangian The daily multiplier model and the preset frame-level coding bit rate model are used to obtain the theoretical optimal Lagrangian multiplier model at the coding unit level.

The rate-distortion optimization device substitutes formula (9) into formula (8) to obtain the theoretical optimal Lagrangian multiplier model at the coding unit level.

In the embodiment of this application, the theoretical optimal Lagrangian multiplier model can be expressed as formula (10), as follows:

It should be noted,

with

Not available before encoding. Taking into account the time domain correlation between the images to be encoded in each frame, the bit rate generated by the adjacent coding unit in the time domain is very similar to the bit rate generated by the current coding unit. Therefore, the bit rate of the adjacent coding unit in the time domain is adopted.

Instead of the bit rate generated by encoding the current coding unit, as shown in formula (11):

In the embodiment of this application, the coding units adjacent in the time domain are used in the encoding process although they are not the optimal Lagrangian multipliers.

However, when coding adjacent coding units in the time domain, the estimated Lagrangian multiplier actually used in the previous frame is used.

because

Is the optimal Lagrangian multiplier

A better estimate, so it is used to approximate the optimal Lagrangian multiplier

The resulting bit rate result is shown in equation (12):

In this way, the rate-distortion optimization device can obtain the preset time-domain proximity approximate coding bit rate model according to formula (11) and formula (12), as shown in formula (13):

In the embodiment of the present application, the rate-distortion optimization device may also deduce the actual coding bit rate of the coding unit and the estimated Lagrangian multiplier model based on the initial coding bit rate and the Lagrangian multiplier model.

It should be noted that in the encoding process at time t-1, all coding units use independent Lagrangian multipliers

Encoding is different from the λ _sys allocated by the system. In order to characterize the bit rate of the "frame-level abstract coding unit" at time t-1, take coding unit i as an example, according to the Lagrangian multiplier used in the actual coding of the i-th coding unit of the previous frame

And the actual encoding bit rate

Combining the system Lagrangian multiplier λ _sys , based on formula (7), the actual coding bit rate of the coding unit and the estimated Lagrangian multiplier model can be obtained.

Exemplarily, the actual coding bit rate and the estimated Lagrangian multiplier model are shown in formula (14):

In this way, the actual encoding bit rate of the rate-distortion optimization device and the estimated Lagrangian multiplier model and the preset frame-level encoding bit rate model are obtained to obtain the actual frame-level encoding bit rate model of one frame.

In the embodiment of this application, formula (14) is substituted into formula (10), that is, formula (14) is applied to all coding units at t-1, then the "frame-level abstract coding at t-1" can be calculated The bit rate expression of “unit” is the actual frame-level coding bit rate model of one frame, namely formula (15), as follows:

In the embodiment of this application, the bit rate of the "frame-level abstract coding unit" at time t

It cannot be obtained directly. Also because of the time-domain correlation between the video sequences to be encoded, the bit rates of the “frame-level abstract coding units” of neighboring images in the time domain are very close, so the average bit rate at the previous moment is used to approximate the average bit rate at the current moment. The preset time-domain proximity approximate frame-level coding bit rate model can be obtained, as shown in formula (16):

Based on this, the rate-distortion optimization device obtains the theoretical optimal Lagrangian multiplier model formula (10), the preset time-domain proximity approximate coding bit rate model formula (13), and the preset time-domain proximity approximate frame-level coding bits Rate model formula (16) and the actual frame-level coding bit rate model formula (15) of one frame, in this way, the rate-distortion optimization device can be based on the theoretically optimal Lagrangian multiplier model and preset temporal proximity approximate coding The bit rate model, the preset time-domain proximity approximate frame-level coding bit rate model, and the actual frame-level coding bit rate model of one frame, obtain the preset coding bit rate and the estimated Lagrangian multiplier model formula (5).

In the embodiment of this application, the rate-distortion optimization device substitutes formula (15) into formula (16) to obtain

Then

Substituting formula (13) into formula (10), formula (5) is obtained. In this way, the coding unit-level coding bit rate and estimated Lagrangian multiplier of the coding unit level of the previous frame to be coded are obtained, and the estimated Lagrangian multiplier of the coding unit level of the current frame to be coded image is obtained. Preset encoding bitrate and estimated Lagrangian multiplier model.

It is understandable that the rate-distortion optimization device uses the coding unit-level coding bit rate and estimated Lagrangian multiplier of the previously encoded image to be coded, combining the preset coding bit rate and the estimated Lagrangian multiplier. The sub-model is used to obtain the estimated Lagrangian multiplier at the coding unit level of the image to be encoded in the current frame, and then to achieve rate-distortion optimization processing and encoding, because the rate-distortion optimization device can be based on the bit rate of the image to be encoded in the previous frame And estimated Lagrangian multipliers, intervene in the estimated Lagrangian multipliers of all coding units of the image to be coded in the current frame, and further interfere with the bit distribution of the intra-frame coding units tending to balance, thereby improving coding efficiency.

Exemplarily, the encoding process is performed using the encoding image of the current frame as shown in FIG. 3. Figure 4 is a schematic diagram of encoding after the rate-distortion optimization method provided by this application is turned off, and Figure 5 is a schematic diagram of encoding after the rate-distortion optimization method provided by this application is turned on; the lighter the color in the figure, the higher the bit rate of the encoding unit, and vice versa. The deeper it is, the lower the bit rate for encoding the unit. It can be seen from the comparison between FIG. 4 and FIG. 5 that the encoding rate after the rate-distortion optimization method provided by this application is turned on is higher, for example, area 1.

FIG. 6 is an analysis diagram of coding unit division after the rate-distortion optimization method provided by this application is turned off; FIG. 7 is an analysis diagram of coding unit division after the rate-distortion optimization method provided by this application is turned on. Comparing Fig. 6 and Fig. 7, it can be found that the bits of some areas are improved and the division becomes finer, such as area 2. Figure 8 is an analysis diagram of the division of two coding units after the rate-distortion optimization method provided by this application is turned off; Figure 9 is an analysis diagram of the division of two coding units after the rate-distortion optimization method provided by this application is turned on, and the two coding units correspond to The video image content is lawn with complex texture. The coding block division in Fig. 8 is simple and the level is shallow; the coding block division in Fig. 9 becomes finer, the division level is deepened, and a relatively large number of coding bits are used.

It can be seen from the above illustration that the rate-distortion optimization method provided by the embodiment of the present application can improve coding efficiency.

It should be noted that the embodiment of the present application does not directly control the encoding bit rate, but indirectly adjusts the bit distribution by updating the Lagrangian multiplier, intervening in encoding mode selection and encoding quadtree division depth.

Exemplarily, Table 1 shows the coding performance test situation in the low-latency configuration. The video sequence used in the test is the general test video sequence of H.265/HEVC, and the 4 quantization parameter points 22, 27, 32, 37 (ie Class B, Class C, Class D, and Class E) are tested according to the general test standard. . Performance evaluation uses BD-Rate as the evaluation index. BD-Rate represents the increase in bit rate under the same PSNR. When the BD-Rate is negative, it indicates that the encoder performance has been improved. The test uses the open source commercial encoder x265v2.3 version, and the test results of all video sequences are shown in Table 1: When the rate-distortion optimization method provided by the embodiment of this application is introduced, Summary is -, which means that the encoder has achieved performance The improvement can save an average bit rate of 3.08% on Y (brightness), an average bit rate of 3.6% on U (chroma), and an average bit rate of 2.7% on V (concentration).

Table 1

In some embodiments of the present application, after S101, that is, after acquiring the image to be encoded in the current frame, a rate-distortion optimization method provided in an embodiment of the present application further includes: S107 and S108. as follows:

S107: If the image to be encoded in the current frame is the image to be encoded in the first frame, obtain an initial Lagrangian multiplier.

S108: Using the initial Lagrangian multiplier, perform rate-distortion processing of the image to be encoded in the current frame, and then complete the encoding of the image to be encoded in the current frame.

In the embodiment of the present application, if the image to be encoded in the current frame is the first image to be encoded, the rate-distortion optimization device obtains the initial Lagrangian multiplier, and uses the initial Lagrangian multiplier to perform the image processing of the current frame to be encoded Rate distortion processing, and then complete the encoding of the image to be encoded in the current frame.

It should be noted that the image to be encoded in the first frame does not have a temporally adjacent coding unit. Therefore, it is necessary to use the initial Lagrangian multiplier to perform rate-distortion processing of the image to be encoded in the first frame, and then complete the image to be encoded in the first frame. Coded.

In some embodiments of the present application, after S104, the estimated Lagrangian multiplier at the coding unit level of the image to be coded in the current frame is used to perform rate-distortion processing of the image to be coded in the current frame to complete the image to be coded in the current frame After the encoding, the rate-distortion optimization method provided in the embodiment of the present application further includes: S109 and S110. as follows:

S109: Enter the rate-distortion optimization process of the next frame to be encoded.

S110: Encoding of the last frame of the to-be-encoded image of the to-be-encoded video sequence is completed.

In the embodiment of the present application, after the rate-distortion optimization device has processed the image to be encoded in the current frame, it can process the image to be encoded in the next frame, that is, the process of S101-104, that is, S201-S205, is cyclically executed until processing When the encoding of the last frame of the to-be-encoded video sequence is completed, the process ends.

Based on the implementation of the foregoing embodiment, as shown in FIG. 10, an embodiment of the present application also provides a rate-distortion optimization device 1, including:

The acquiring part 10 is configured to acquire the image to be encoded in the current frame; and if the image to be encoded in the current frame is not the first frame to be encoded, acquiring the initial Lagrangian multiplier and the coding unit level of the image to be encoded in the previous frame The coding bit rate and the estimated Lagrangian multiplier at the coding unit level of the image to be coded in the previous frame;

The calculation part 11 is configured to estimate the Lagrange based on the initial Lagrangian multiplier, the coding unit-level coding bit rate of the image to be coded in the previous frame, and the coding unit level of the image to be coded in the previous frame. A Lange multiplier and a preset encoding bit rate and an estimated Lagrangian multiplier model to obtain an estimated Lagrangian multiplier at the coding unit level of the image to be encoded in the current frame;

The processing part 12 is configured to use the estimated Lagrangian multiplier at the coding unit level of the image to be encoded in the current frame to perform rate-distortion processing of the image to be encoded in the current frame, and then to complete the processing of the image to be encoded in the current frame coding.

In some embodiments of the present application, the acquiring part 10 is specifically configured to acquire the i-th coding unit of the current frame to be encoded if the image to be encoded in the current frame is not the first frame to be encoded, The initial Lagrangian multiplier, the coding bit rate of the i-th coding unit of the image to be coded in the previous frame, and the estimated Lagrangian multiplier of the i-th coding unit of the image to be coded in the previous frame, where i It is a positive integer greater than or equal to 1 and less than or equal to N, where N is the total number of coding units corresponding to a frame of image to be encoded.

In some embodiments of the present application, the calculation part 11 is specifically configured to be based on the initial Lagrangian multiplier, the encoding bit rate of the i-th coding unit of the image to be encoded in the previous frame, and the The estimated Lagrangian multiplier of the i-th coding unit of the image to be encoded in the previous frame and the preset encoding bit rate and the estimated Lagrangian multiplier model to obtain the i-th image of the image to be encoded in the current frame The estimated Lagrangian multiplier of the coding unit; and adding 1 to i to obtain the estimated Lagrangian multiplier of the i+1-th coding unit of the image to be encoded in the current frame until the image to be encoded in the current frame is obtained The estimated Lagrangian multiplier of the Nth coding unit.

In some embodiments of the present application, the processing part 12 is specifically configured to use the estimated Lagrangian multiplier of the first coding unit of the image to be encoded in the current frame to the first coding unit of the image to be encoded in the current frame. The estimated Lagrangian multipliers of the N coding units perform rate-distortion processing of the image to be encoded in the current frame, and then complete the encoding of the image to be encoded in the current frame.

In some embodiments of the present application, the acquiring part 10 is further configured to encode the bit rate according to the initial Lagrangian multiplier, the encoding unit level of the image to be encoded in the previous frame, and the The estimated Lagrangian multiplier at the coding unit level of the image to be coded in the previous frame and the preset coding bit rate and the estimated Lagrangian multiplier model to obtain the estimated Lagrangian at the coding unit level of the image to be coded in the current frame Before the Lange multiplier, obtain the initial encoding bit rate and the Lagrangian multiplier model, the preset frame-level encoding bit rate model, the preset temporal proximity approximate encoding bit rate model, and the preset temporal proximity approximate frame-level encoding bits Rate model

The calculation part 11 is further configured to be based on the initial encoding bit rate and the Lagrangian multiplier model, the preset frame-level encoding bit rate model, the preset temporal proximity approximate encoding bit rate model, and the The preset time-domain proximity approximates the frame-level coding bit rate model, and the preset coding bit rate and the estimated Lagrangian multiplier model are obtained.

In some embodiments of the present application, the calculation part 11 is further specifically configured to deduce the theoretical encoding bit rate of the coding unit and the estimated Lagrangian multiplier according to the initial encoding bit rate and the Lagrangian multiplier model. The actual coding bit rate and estimated Lagrangian multiplier model of the sub-model and coding unit; and according to the theoretical coding bit rate and the estimated Lagrangian multiplier model and the preset frame-level coding bit rate model, obtain The theoretical optimal Lagrangian multiplier model at the coding unit level; and according to the actual encoding bit rate and the estimated Lagrangian multiplier model and the preset frame-level encoding bit rate model, the actual frame level of the previous frame is obtained Coding bit rate model; and according to the theoretical optimal Lagrangian multiplier model, the preset time domain proximity approximate coding bit rate model, the preset time domain proximity approximate frame-level coding bit rate model, and the The actual frame-level coding bit rate model of the previous frame is used to obtain the preset coding bit rate and the estimated Lagrangian multiplier model.

In some embodiments of the present application, the acquiring part 10 is further configured to acquire the initial Lagrangian multiplier after acquiring the image to be encoded in the current frame, if the image to be encoded in the current frame is the first image to be encoded child;

The processing part 12 is further configured to use the initial Lagrangian multiplier to perform rate-distortion processing of the image to be encoded in the current frame, and then to complete the encoding of the image to be encoded in the current frame.

In some embodiments of the present application, the processing part 12 is further configured to use the estimated Lagrangian multiplier at the coding unit level of the image to be encoded in the current frame to perform the calculation of the image to be encoded in the current frame. Rate-distortion processing, and after completing the encoding of the image to be encoded in the current frame, enter the rate-distortion optimization process of the image to be encoded in the next frame; the encoding of the image to be encoded in the last frame of the video sequence to be encoded is completed.

As shown in FIG. 11, an embodiment of the present application also provides a rate-distortion optimization device, including:

The processor 13, the memory 14 storing the rate-distortion optimization instruction executable by the processor 13, and the communication bus 15 for connecting the processor 13 and the memory 14, when the rate-distortion optimization instruction is executed , To achieve the above-mentioned rate-distortion optimization method.

In the embodiment of the present application, the aforementioned processor 13 may be an Application Specific Integrated Circuit (ASIC), a digital signal processor (Digital Signal Processor, DSP), a digital signal processing device (Digital Signal Processing Device, DSPD). ), programmable logic device (ProgRAMmable Logic Device, PLD), field programmable gate array (Field ProgRAMmable Gate Array, FPGA), central processing unit (Central Processing Unit, CPU), controller, microcontroller, microprocessor At least one of. It is understandable that, for different devices, the electronic devices used to implement the above-mentioned processor functions may also be other, which is not specifically limited in the embodiment of the present application. The intra-frame prediction apparatus may also include a memory 14, which may be connected to the processor 13, wherein the memory 14 is used to store executable program code, the program code includes computer operation instructions, and the memory 14 may be a volatile memory ( Volatile memory, such as random access memory (Random-Access Memory, RAM); or non-volatile memory (non-volatile memory), such as read-only memory (ROM), flash memory (flash memory) ), a hard disk (Hard Disk Drive, HDD) or a solid-state drive (Solid-State Drive, SSD); or a combination of the foregoing types of memories, and provides instructions and data to the processor 13.

In the embodiment of the present application, the communication bus 15 is used to connect the processor 13 and the memory 14 and the mutual communication between these devices.

In addition, the functional modules in the embodiments of the present application may be integrated into one processing unit, or each unit may exist alone physically, or two or more units may be integrated into one unit. The above-mentioned integrated unit can be realized in the form of hardware or software function module.

If the integrated unit is implemented in the form of a software function module and is not sold or used as an independent product, it can be stored in a computer readable storage medium. Based on this understanding, the technical solution of this embodiment is essentially or correct The part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions to enable a computer device (which can be a personal A computer, a server, or a network device, etc.) or a processor (processor) execute all or part of the steps of the method in this embodiment. The aforementioned storage media include: U disk, mobile hard disk, read only memory (Read Only Memory, ROM), random access memory (Random Access Memory, RAM), magnetic disk or optical disk and other media that can store program codes.

Those skilled in the art should understand that the embodiments of the present application can be provided as methods, systems, or computer program products. Therefore, this application may adopt the form of hardware embodiments, software embodiments, or embodiments combining software and hardware. Moreover, this application may adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, optical storage, etc.) containing computer-usable program codes.

This application is described with reference to the schematic diagrams and/or block diagrams of the implementation process of the method, equipment (system), and computer program product according to the embodiments of the application. It should be understood that computer program instructions can be used to implement each process and/or block in the schematic flow diagram and/or block diagram, and to implement a combination of processes and/or blocks in the schematic flow diagram and/or block diagram. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing equipment to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing equipment are generated A device for realizing the functions specified in one or more processes in the schematic flow chart and/or one block or multiple blocks in the block diagram.

These computer program instructions can also be stored in a computer-readable memory that can guide a computer or other programmable data processing equipment to work in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture including the instruction device. The device realizes the functions specified in one or more processes in the schematic diagram and/or one block or more in the block diagram.

These computer program instructions can also be loaded on a computer or other programmable data processing equipment, so that a series of operation steps are executed on the computer or other programmable equipment to produce computer-implemented processing, so as to execute on the computer or other programmable equipment. The instructions provide steps for implementing functions specified in one or more processes in the schematic diagram and/or one block or more in the block diagram.

The above are only the preferred embodiments of the present application, and are not used to limit the protection scope of the present application.

Industrial applicability

The embodiments of the present application provide a rate-distortion optimization method and device, and a computer-readable storage medium. The rate-distortion optimization device performs rate-distortion optimization processing on the image to be encoded in the current frame. During the encoding process, the encoding process can be completed by encoding. The coding unit-level coding bit rate and estimated Lagrangian multiplier of a frame of image to be coded are combined with the preset coding bit rate and the estimated Lagrangian multiplier model to obtain the coding unit level of the image to be coded in the current frame Estimate the Lagrangian multiplier, and then realize the rate-distortion optimization process and realize the encoding. Because the rate-distortion optimization device can intervene in the current frame of the image to be encoded based on the bit rate of the image to be encoded in the previous frame and the estimated Lagrange multiplier The estimated Lagrangian multipliers of all coding units in the frame, and then interfere with the bit distribution of the intra-frame coding units tend to be balanced, thereby improving the coding efficiency.

Claims

A rate-distortion optimization method, characterized in that it includes:

Acquiring the image to be encoded in the current frame;

If the image to be encoded in the current frame is not the first frame to be encoded, the initial Lagrangian multiplier, the encoding unit-level encoding bit rate of the image to be encoded in the previous frame, and the encoding unit of the image to be encoded in the previous frame are acquired Estimated Lagrange multiplier of the order;

According to the initial Lagrangian multiplier, the coding unit-level coding bit rate of the image to be coded in the previous frame, and the estimated Lagrangian multiplier and pre-coding unit level of the image to be coded in the previous frame Set the coding bit rate and the estimated Lagrangian multiplier model to obtain the estimated Lagrangian multiplier at the coding unit level of the image to be coded in the current frame;

The estimated Lagrangian multiplier at the coding unit level of the image to be encoded in the current frame is used to perform rate-distortion processing of the image to be encoded in the current frame to complete the encoding of the image to be encoded in the current frame.
The method according to claim 1, wherein if the image to be encoded in the current frame is not the image to be encoded in the first frame, the initial Lagrangian multiplier and the encoding unit of the image to be encoded in the previous frame are acquired The coding bit rate of the previous frame and the estimated Lagrangian multiplier of the coding unit of the image to be coded in the previous frame include:

If the image to be encoded in the current frame is not the first image to be encoded, the i-th coding unit of the image to be encoded in the current frame, the initial Lagrangian multiplier, and the image to be encoded in the previous frame are acquired. The coding bit rate of i coding units and the estimated Lagrangian multiplier of the i-th coding unit of the image to be coded in the previous frame, where i is a positive integer greater than or equal to 1 and less than or equal to N, and N is a frame to be coded The total number of coding units corresponding to the coded image.
The method according to claim 2, wherein the encoding bit rate according to the initial Lagrangian multiplier, the coding unit level of the previous frame to be encoded, and the previous frame to be encoded The estimated Lagrangian multiplier at the coding unit level of the image and the preset coding bit rate and the estimated Lagrangian multiplier model to obtain the estimated Lagrangian multiplier at the coding unit level of the image to be coded in the current frame, include:

According to the initial Lagrangian multiplier, the coding bit rate of the i-th coding unit of the image to be coded in the previous frame, and the estimated Lagrangian of the i-th coding unit of the image to be coded in the previous frame A multiplier and the preset encoding bit rate and an estimated Lagrangian multiplier model to obtain an estimated Lagrangian multiplier of the i-th coding unit of the image to be encoded in the current frame;

Add 1 to i to obtain the estimated Lagrangian multiplier of the i+1th coding unit of the image to be coded in the current frame until the estimated Lagrangian multiplier of the Nth coding unit of the image to be coded in the current frame is obtained Until the multiplier.
The method according to claim 3, wherein the estimated Lagrangian multiplier at the coding unit level of the image to be encoded in the current frame is used to perform rate-distortion processing of the image to be encoded in the current frame, and then Completing the encoding of the image to be encoded in the current frame includes:

Use the estimated Lagrangian multiplier of the first coding unit of the image to be coded in the current frame to the estimated Lagrangian multiplier of the Nth coding unit of the image to be coded in the current frame to perform the current frame The rate-distortion processing of the image to be encoded further completes the encoding of the image to be encoded in the current frame.
The method according to any one of claims 1 to 4, wherein the encoding bit rate according to the initial Lagrangian multiplier, the encoding unit level of the image to be encoded in the previous frame, and the The estimated Lagrangian multiplier at the coding unit level of the image to be coded in the previous frame and the preset coding bit rate and the estimated Lagrangian multiplier model to obtain the estimated Lagrangian at the coding unit level of the image to be coded in the current frame Before the Lange multiplier, the method further includes:

Obtain the initial coding bit rate and the Lagrangian multiplier model, the preset frame-level coding bit rate model, the preset time-domain proximity approximate coding bit rate model, and the preset time-domain proximity approximate frame-level coding bit rate model;

According to the initial coding bit rate and the Lagrangian multiplier model, the preset frame-level coding bit rate model, the preset time domain proximity approximate coding bit rate model, and the preset time domain proximity approximate frame level The coding bit rate model is used to obtain the preset coding bit rate and the estimated Lagrangian multiplier model.
The method according to claim 5, characterized in that, according to the initial coding bit rate and the Lagrangian multiplier model, the preset frame-level coding bit rate model, the preset time domain proximity approximation The coding bit rate model and the preset time-domain proximity approximate frame-level coding bit rate model to obtain the preset coding bit rate and the estimated Lagrangian multiplier model include:

According to the initial coding bit rate and Lagrangian multiplier model, the theoretical coding bit rate and estimated Lagrangian multiplier model of the coding unit and the actual coding bit rate and estimated Lagrangian multiplier model of the coding unit are derived ；

Obtaining a theoretical optimal Lagrangian multiplier model at the coding unit level according to the theoretical coding bit rate and the estimated Lagrangian multiplier model and the preset frame-level coding bit rate model;

According to the actual encoding bit rate and the estimated Lagrangian multiplier model and the preset frame-level encoding bit rate model, obtain the actual frame-level encoding bit rate model of one frame;

According to the theoretical optimal Lagrangian multiplier model, the preset temporal proximity approximate coding bit rate model, the preset temporal proximity approximate frame-level coding bit rate model, and the actual frame of one frame Level coding bit rate model to obtain the preset coding bit rate and the estimated Lagrangian multiplier model.
The method according to claim 1, characterized in that, after the obtaining the image to be coded in the current frame, the method further comprises:

If the image to be encoded in the current frame is the image to be encoded in the first frame, obtaining an initial Lagrangian multiplier;

The initial Lagrangian multiplier is used to perform rate-distortion processing of the image to be encoded in the current frame, thereby completing the encoding of the image to be encoded in the current frame.
The method according to claim 1, wherein the estimated Lagrangian multiplier at the coding unit level of the image to be encoded in the current frame is used to perform rate-distortion processing on the image to be encoded in the current frame, and then After the encoding of the image to be encoded in the current frame is completed, the method further includes:

Enter the rate-distortion optimization process of the next frame to be encoded;

The encoding of the image to be encoded until the last frame of the video sequence to be encoded is processed.
A rate-distortion optimization device, characterized by comprising:

A processor, a memory storing a rate-distortion optimization instruction executable by the processor, and a communication bus for connecting the processor and the memory, and when the rate-distortion optimization instruction is executed, the implementation is as claimed in claim 1. -8 The method described in any one.
A computer-readable storage medium, characterized in that a rate-distortion optimization instruction is stored thereon, wherein, when the rate-distortion optimization instruction is executed by a processor, the method according to any one of claims 1-8 is implemented.