CN101867807B - Device and method for selecting an in-frame prediction mode - Google Patents

Abstract

The present invention discloses a device and method for selecting a prediction mode within a picture frame, wherein the method comprises: first, receiving a compression configuration of picture frame data and receiving a picture resolution of the picture frame data; then, selecting a corresponding plurality of prediction modes according to the compression configuration and the picture resolution, and scheduling the selected plurality of prediction modes to sequentially calculate and obtain a corresponding plurality of value functions; and finally, selecting one of the plurality of prediction modes according to the plurality of comparison value functions as the prediction mode of the picture frame data.

Description

Device and method for selecting an in-frame prediction mode

technical field

The present invention relates to a selection device and method for an intra-frame prediction mode, in particular to a selection device and method for an H.264 format intra-frame prediction mode.

Background technique

In recent years, with the advancement of electronic technology, there have been corresponding breakthroughs in image display technology. Especially in the display of animated images, in order to make the display can effectively improve the performance of its animated images, a variety of image compression standards have been proposed one after another. Among them, the H.264 codec standard is the most representative high-end Compressed digital video codec standard.

In the H.264 encoding and decoding standard, when determining the intraprediction mode to be selected, it is necessary to calculate the cost function (cost function) for all possible intra-frame modes (intra mode). The intra-frame prediction modes include four types: I4 bulk mode, I8 bulk mode, I16 bulk mode, and chroma (chroma) mode. The four types of in-frame prediction modes also include 26 sub-modes that need to be calculated. In addition, the I 4 bulk mode and the I 8 bulk mode require so-called closed-loop prediction, so it can be seen that it takes a considerable amount of time to determine the in-frame prediction mode to be selected. Hardware resources and the bandwidth used to perform calculations. Therefore, it is very important to schedule the predictions for the I4 bulk mode, the I8 bulk mode, the I16 bulk mode, and the chroma mode.

Please refer to FIG. 1 below. FIG. 1 shows a scheduling method of an existing in-frame prediction mode. The illustration in FIG. 1 is mainly for the I4 bulk mode, the I16 block mode and the chroma mode to perform prediction sequentially. To put it simply, the prediction 111-112 of the I4 bulk mode and the reconstruction 121-122 of the I4 bulk mode are performed sequentially for different blocks of the frame data, and then sequentially performed for different blocks of the frame data I 16 The prediction 131-132 of the bulk mode, and finally the prediction 141 of the chroma mode. Next, please refer to FIG. 2 . FIG. 2 illustrates an existing interleaved scheduling method of the intra-frame prediction mode. The difference from the scheduling approach of FIG. 1 is that in which the predictions 231, 232 for the I16 bulk mode are inserted in the gap times during the execution of the reconstructions 221, 222 for the I4 bulk mode, respectively. In this way, the extra prediction time for executing the I16 bulk mode can be effectively saved to save the bandwidth of the operation.

Due to the scheduling methods shown in Fig. 1 and Fig. 2, the I8 bulk mode is not considered. Therefore, there is an existing high profile simplified intra-frame prediction mode scheduling method as shown in FIG. 3 . Please refer to Figure 3. The scheduling method shown in Figure 3 ignores the part of the I16 block mode and uses open-loop (open~loop) operations (prediction is only performed for the I4 and I8 block modes, not rebuild) to reduce the hardware resources and bandwidth required for computing.

Contents of the invention

An object of the present invention is to provide a method for selecting a prediction mode within a frame, which schedules several prediction modes of frame data to generate several value functions in sequence.

Another object of the present invention is to provide a device for selecting an intra-frame prediction mode, which is used for scheduling several prediction modes of frame data to sequentially generate several value functions.

To achieve the above object, the present invention proposes a method for selecting an in-frame prediction mode, including: firstly, receiving the compressed configuration of the frame data, and receiving the frame resolution of the frame data. Next, select corresponding several prediction modes according to the compression configuration and the picture resolution, and schedule the selected several prediction modes to sequentially calculate and obtain corresponding several value functions . Finally, according to the plurality of comparative value functions, one of the plurality of prediction modes is selected as the prediction mode of the picture frame data.

In an embodiment of the present invention, the above-mentioned compressed configuration includes a high configuration, a main configuration or a basic configuration.

In an embodiment of the present invention, the aforementioned image resolution includes high resolution or normal resolution.

In an embodiment of the present invention, the aforementioned prediction modes include I4 macroblock mode, I8 macroblock mode, I16 macroblock mode and chroma mode.

In an embodiment of the present invention, the aforementioned selection of the corresponding prediction mode according to the compression configuration and picture resolution includes: when the compression configuration is the main configuration, and the picture resolution is general resolution or high resolution When the brightness is high, select I4 bulk mode, I16 bulk mode and chroma mode. If the compression configuration is the basic configuration, and the screen resolution is normal resolution or high resolution, also select I4 bulk mode, I16 bulk mode and chroma mode. In addition, when the compression configuration is the high configuration and the image resolution is the normal resolution, the I4 bulk mode, the I16 bulk mode and the chroma mode are also selected. The only difference is that when the compression configuration is high and the picture resolution is high resolution, I8 bulk mode, I16 bulk mode and chroma mode are selected.

In one embodiment of the present invention, the aforementioned scheduling for the prediction mode includes: when the selected prediction mode is the I4 macroblock mode, the I16 macroblock mode and the chroma mode, performing the prediction of the I4 macroblock mode Prediction of the I16 bulk mode is performed during the prediction gap when reconstruction is performed, and prediction of the chroma mode is performed during the reconstruction gap time when reconstruction of the I4 bulk mode is performed. In addition, when the selected prediction mode is I8 macroblock mode, I16 macroblock mode and chroma mode, the prediction of I8 macroblock mode is performed first, then the prediction of I16 macroblock mode is performed, and then the prediction of chroma mode is performed, And at the end perform a rebuild of i8 bulk mode.

To achieve the above object, the present invention also proposes a device for selecting an in-frame prediction mode, which includes a schedule controller, a value function generator, and a value function comparison circuit. Wherein, the scheduling controller receives the compression configuration and picture resolution of the picture frame data, selects corresponding several prediction modes according to the compression configuration and picture resolution, and performs scheduling for the several prediction modes. The value function generator is connected to the schedule controller, and the value function generator executes the prediction of the prediction modes sequentially according to the scheduling of the several prediction modes by the schedule controller, and generates corresponding value functions. The cost function comparison circuit is connected to the cost function generator, and the cost function comparison circuit compares the several cost functions to select one of the several prediction modes as the prediction mode of the frame data.

To sum up, the present invention selects the corresponding prediction mode through the compressed configuration of the frame data and the frame resolution, and schedules the prediction mode for the corresponding selected inner frame. In this way, only the necessary prediction modes will be predicted, which can effectively reduce the number of clocks or hardware resources for performing prediction operations, and through the results of scheduling, several prediction modes can be more efficient when making predictions application system resources, reducing cost and power waste.

In order to make the above-mentioned features and advantages of the present invention more comprehensible, the following specific embodiments are described in detail together with the accompanying drawings.

Description of drawings

Figures 1 to 3 respectively illustrate the scheduling methods of three different existing in-frame prediction modes;

FIG. 4 shows an apparatus 400 for selecting an intra-frame prediction mode according to an embodiment of the present invention;

FIG. 5 shows an implementation manner of scheduling performed by the scheduling controller 410 in the embodiment of the present invention;

FIG. 6 shows another implementation of scheduling by the scheduling controller 410 in the embodiment of the present invention;

FIG. 7 is a flowchart of a method for selecting an intra-frame prediction mode according to an embodiment of the present invention.

Explanation of symbols of main components in the drawings

111~112, 511: Prediction of I4 block mode;

121~122, 221, 222, 512: Reconstruction of I4 block mode;

131~132, 231, 232, 513, 612, 652: prediction of I16 block mode;

611, 651: Prediction of I8 bulk mode;

614: Reconstruction of I8 bulk mode;

141, 514, 613, 653: prediction of chroma mode;

400: a device for selecting an in-frame prediction mode;

410: schedule controller;

420: value function generator;

430: value function comparison circuit;

510~550, 610~650: block prediction of frame data;

560, 660: comparing value functions;

S710-S740: Steps of the selection method of the intra-frame prediction mode.

Detailed ways

Please refer to FIG. 4 below. FIG. 4 illustrates an apparatus 400 for selecting an intra-frame prediction mode according to an embodiment of the present invention. The device 400 for selecting an intra-frame prediction mode includes a schedule controller 410 , a cost function generator 420 and a cost function comparison circuit 430 . The scheduling controller 410 receives the compression configuration and frame resolution of the frame data to be processed, selects several corresponding prediction modes according to the compression configuration and frame resolution, and performs scheduling for these selected prediction modes. The so-called compression profile includes high profile (high profile), main profile (main profile) or basic profile (base profile), and image resolution includes high-resolution FULL HD and general resolution CIF . In addition, the above three configurations and two resolutions are well known to those skilled in the art in the field of the H.264 codec standard, so no further description is given here.

In addition, the prediction modes used in the H.264 codec standard generally include I4 macroblock mode, I8 macroblock mode, I16 macroblock mode and chroma mode. The so-called I4 block mode is to use the 4×4 block in the frame data as a unit to predict the picture in the frame. Similarly, the I8 bulk mode and the I16 bulk mode are for predicting the picture in the frame in units of 8×8 and 16×16 blocks in the frame data respectively.

As for the scheduling controller 410 selecting several forecasting modes to be used, the scheduling controller 410 selects according to Table 1 in the following example. Table 1 of which is as follows:

Table 1:

compression configuration screen resolution Selected forecasting model basic configuration CIF I4 bulk mode, I16 bulk mode and chroma mode basic configuration FULL HD I4 bulk mode, I16 bulk mode and chroma mode main configuration CIF I4 bulk mode, I16 bulk mode and chroma mode main configuration FULL HD I4 bulk mode, I16 bulk mode and chroma mode High configuration CIF I4 bulk mode, I16 bulk mode and chroma mode High configuration FULL HD I8 bulk mode, I16 bulk mode and chroma mode

It can be clearly seen from Table 1 that the scheduling controller 410 can classify various types with different compression configurations and picture resolutions, and select an applicable prediction mode for prediction, which can save computing bandwidth.

After the scheduling controller 410 selects the forecasting mode to be predicted, the scheduling controller 410 schedules the selected forecasting mode. Among them, the scheduling controller 410 will perform different scheduling methods according to the selected forecasting modes, and different implementation methods will be proposed below for different scheduling methods to illustrate, and it is hoped that those skilled in the art can understand and understand to implement.

Please refer to FIG. 4 and FIG. 5 at the same time. FIG. 5 shows an implementation manner of scheduling performed by the scheduling controller 410 in the embodiment of the present invention. Wherein when the prediction mode selected by the scheduling controller 410 is the I4 bulk mode, the I16 bulk mode and the chroma mode, the scheduling controller 410 performs the scheduling as shown in FIG. 5 . That is, first perform prediction 510 on block 0 of the frame data, including performing prediction and reconstruction in I4 bulk mode, prediction in I16 bulk mode, and prediction in chroma mode. To further illustrate, the prediction 511 of the I4 bulk mode is performed first, and then the reconstruction 512 of the I4 bulk mode is performed. The execution of the I16 macroblock mode prediction 513 is interleaved with the prediction gap time generated when the I4 macroblock mode prediction 511 is executed. In contrast, the chroma mode prediction 514 is interleaved with the reconstruction gap generated when the I4 bulk mode reconstruction 512 is performed.

After the prediction of block 0 of the picture frame data is completed, the prediction of blocks 1 to 15 of the picture frame data is performed sequentially, and after the prediction 550 of block 15 of the picture frame data is completed, the comparison value is performed Act 560 of the function.

Note here that the generation of the value function follows the execution of various prediction modes. Please refer to FIG. 4 and FIG. 5 at the same time, the value function generator 420 generates the value function according to the prediction sequence of the prediction mode scheduled by the scheduling controller 410 . In other words, when all the prediction modes are executed, the value functions corresponding to these prediction modes are also generated.

Therefore, when all the prediction modes are executed, the value function comparing circuit 430 performs comparison according to the value function, and selects the one with the lowest value function as the selected prediction mode. To put it simply, if the value function corresponding to the I4 large block mode is lower than the value function corresponding to the I16 large block mode, the I4 large block mode is selected as the prediction mode for the frame data; on the contrary, if the I4 large block mode is When the cost function corresponding to the block mode is higher than the value function corresponding to the I16 bulk mode, the I16 bulk mode is selected as the prediction mode for the frame data.

Please refer to FIG. 4 and FIG. 6 at the same time. FIG. 6 shows another implementation of scheduling by the scheduling controller 410 in the embodiment of the present invention. Wherein when the prediction mode selected by the scheduling controller 410 is the I8 bulk mode, the I16 bulk mode and the chroma mode, the scheduling controller 410 performs the scheduling as shown in FIG. 6 . That is, first perform prediction 610 on block 0 of the frame data, including performing prediction and reconstruction in I8 bulk mode, prediction in I16 bulk mode, and prediction in chroma mode. To further explain, it is to first perform the prediction 611 of the I8 bulk mode, and then perform the prediction 612 of the I16 bulk mode, then perform the prediction 613 of the chroma mode, and finally perform the reconstruction 614 of the I8 bulk mode .

After the prediction for block 0 is completed, the prediction actions for blocks 1-15 are performed in sequence. It is worth mentioning that, in the prediction 650 for block 15, since it is the last block, after the prediction 651 of the I8 bulk mode and the prediction of the I16 bulk mode are performed in sequence After 652 and chroma mode prediction 653, there is no need to rebuild I8 bulk mode. The reason is that block 15 is already the last block, so there is no need to rebuild and provide information to provide prediction actions for the next block.

Same as the embodiment shown in FIG. 5 , and after the prediction 650 of the block 15 of the frame data is completed, the value function comparison circuit 430 performs an action 660 of comparing the value functions, so as to select a suitable prediction mode. Here, the operation of how to select a suitable prediction mode according to the cost function is the same as that shown in FIG. 5 , and will not be repeated here.

An embodiment of the method for selecting the intra-frame prediction mode of the present invention is presented below for illustration.

Please refer to FIG. 7 below. FIG. 7 is a flow chart of a method for selecting an intra-frame prediction mode according to an embodiment of the present invention. The steps include: firstly, respectively receiving the compressed configuration and the picture resolution of the picture frame data (steps S710, S720). Among them, in the codec standard of H.264, the compression configuration includes main configuration, basic configuration and advanced configuration. The screen resolution can be divided into high-resolution FULL HD and general-resolution CIF. Next, several corresponding prediction modes are selected according to the compression configuration and the picture resolution, and scheduling is performed for these prediction modes to sequentially calculate and obtain corresponding cost functions (step S730 ).

Next, compare according to several value functions calculated in step S730, and select one of several prediction modes as the prediction mode of the frame data (step S740). The action of comparing several value functions is, for example, to find out the smallest of the value functions, and select the corresponding prediction mode as the prediction mode of the frame data.

To sum up, the present invention selects the corresponding prediction mode according to the compression module and picture resolution of the picture frame data, and performs scheduling for the selected prediction mode to sequentially generate the corresponding value function. It effectively saves hardware resources and effectively saves computing bandwidth.

Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that: it still Modifications or equivalent replacements can be made to the technical solutions of the present invention, and these modifications or equivalent replacements cannot make the modified technical solutions deviate from the spirit and scope of the technical solutions of the present invention.

Claims (2)

1. the system of selection of an intra-frame prediction mode is characterized in that, comprising:

Receive a compressed configuration of a picture frame data, described compressed configuration comprises high configuration, main configuration or basic configuration;

Receive a picture resolution of described picture frame data, described picture resolution comprises high-res or General Analytical degree;

Select several corresponding predictive modes according to described compressed configuration and described picture resolution, and carry out scheduling sequentially calculate to obtain several corresponding cost functions for described several predictive modes, wherein, described several predictive modes comprise the huge block mode of an I4, the huge block mode of an I8, the huge block mode of an I16 and a colourity pattern, wherein, select the step of corresponding described predictive mode to comprise according to described compressed configuration and described picture resolution:

When described compressed configuration is main configuration, and described picture resolution is high-res or General Analytical when spending, and selects the huge block mode of described I4, the huge block mode of described I16 and described chroma mode;

When described compressed configuration is basic configuration, and described picture resolution is selected the huge block mode of described I4, the huge block mode of described I16 and described chroma mode when being General Analytical degree or high-res;

When described compressed configuration is high configuration, and described picture resolution is General Analytical when spending, and selects the huge block mode of described I4, the huge block mode of described I16 and described chroma mode;

And

When described compressed configuration is high configuration, and described picture resolution is selected the huge block mode of described I8, the huge block mode of described I16 and described chroma mode when being high-res;

Wherein, carrying out scheduling for described several predictive modes comprises sequentially to calculate the step that obtains corresponding described cost function:

When described several predictive modes of selecting are described I 4 huge block modes, the huge block mode of described I16 and described chroma mode, wherein carry out the prediction of the huge block mode of described I16 in the prediction time dead when carrying out the prediction of the huge block mode of described I4, and the reconstruction time dead when carrying out the reconstruction of the huge block mode of described I4 is carried out the prediction of described chroma mode; And

When described several predictive modes of selecting are the huge block mode of described I8, the huge block mode of described I16 and described chroma mode, carry out first the prediction of the huge block mode of described I8, carry out again the prediction of the huge block mode of described I16, then carry out again the prediction of described chroma mode, and in the end carry out the reconstruction of the huge block mode of described I8; And

Come predictive mode as described picture frame data according to more described several cost functions with one of them that select described several predictive modes.

2. the choice device of an intra-frame prediction mode is characterized in that, comprising:

One scheduling controller, receive a compressed configuration and a picture resolution of a picture frame data, select several corresponding predictive modes according to described compressed configuration and described picture resolution, and carry out scheduling for described several predictive modes, described compressed configuration comprises high configuration, main configuration or basic configuration, described picture resolution comprises high-res or General Analytical degree, and described several predictive modes comprise the huge block mode of an I4, the huge block mode of an I8, the huge block mode of an I16 and a colourity (chroma) pattern;

One cost function generator connects described scheduling controller, sequentially carries out the prediction of described several predictive modes according to the scheduling that described scheduling controller carries out described several predictive modes, and produces several corresponding cost functions; And

One cost function comparison circuit connects described cost function generator, and more described several cost functions come predictive mode as described picture frame data with one of them that select described several predictive modes,

When described compressed configuration is main configuration, and described picture resolution is when being General Analytical degree or high-res, described scheduling controller is selected the huge block mode of described I4, the huge block mode of described I16 and described chroma mode, when described compressed configuration is basic configuration, and described picture resolution is when being General Analytical degree or high-res, described scheduling controller is selected the huge block mode of described I4, the huge block mode of described I16 and described chroma mode, when described compressed configuration is high configuration, and described picture resolution is that General Analytical is when spending, described scheduling controller is selected the huge block mode of described I4, the huge block mode of described I16 and described chroma mode, when described compressed configuration is high configuration, and when described picture resolution is high-res, described scheduling controller is selected the huge block mode of described I8, the huge block mode of described I16 and described chroma mode, when described several predictive modes of selecting are the huge block mode of described I4, when the huge block mode of described I16 and described chroma mode, described scheduling controller makes the prediction of carrying out the huge block mode of described I16 in the prediction time dead of described cost function generator when carrying out the prediction of the huge block mode of described I4, and the reconstruction time dead when described cost function generator is carried out the reconstruction of the huge block mode of described I4 is carried out the prediction of described chroma mode, when the described predictive mode of selecting is the huge block mode of described I8, when the huge block mode of described I16 and described chroma mode, described scheduling controller makes described cost function generator carry out first the prediction of the huge block mode of described I8, carry out again the prediction of the huge block mode of described I16, then carry out again the prediction of described chroma mode, and in the end carry out the reconstruction of the huge block mode of described I8.

Images