KR100991138B1 - Encoder and method for selecting candidate mode types - Google Patents

Abstract

Disclosed are an encoder and a method of determining candidate modes of an encoder. The method of determining a candidate mode includes receiving a spatiotemporal context including a mode type index of a mode type for a reference macroblock in a current frame and a mode type for b reference macroblock in one or more previous frames; And determining a candidate mode type by using one or more of the sum of the mode type indexes included in the spatiotemporal context, the number of intra mode types, and the number of P8x8 mode types. By the present invention, mode selection and encoding processing of the AVC / H.264 standard can be made quickly.

Encoder, encoding, mode

Description

{Encoder and method for selecting candidate mode types}

The present invention relates to an encoder, and more particularly, to an encoder and a method of determining a candidate mode of an encoder capable of quickly determining an encoding mode type of a macroblock.

MPEG-4 AVC / H.264, the latest video coding standard, is known to provide maximum performance in compression ratios among existing video coding standards. AVC / H.264 video encoder that compresses block-based video data reduces residual data by using a predictive encoding method, and residual data and mode in this process. Information, motion vector information, and the like are made into a compressed data stream by a process such as block DCT, quantization, entropy coding, or the like.

The prediction encoding method is divided into inter prediction and intra prediction. Inter prediction is a method of using a temporal correlation of a previous image, and intra prediction is a method of using a spatial correlation. In a video encoder such as an H.264 / AVC scheme, for example, intra prediction is performed on an I-frame, and both inter prediction and intra prediction are performed on a P-frame or a B-frame.

Among the various compression methods, the mode selection process for the macroblock (MB, macroblock) of the AVC / H.264 JM encoder is performed by evaluating a plurality of mode types for each macroblock and then performing the highest performance mode among them. This is the process of selecting the type as the mode type of the macro block.

Table 1 below lists the macroblock mode types used in the AVC / H.264 baseline profile (BP) standard, and FIG. 1 shows a process for selecting an optimal mode type in AVC / H.264 JM 11.0 according to the prior art. Drawing.

Table 1. Macroblock Mode Types of the AVC / H.264 BP Standard

As shown in FIG. 1, in order to determine a best mode type for encoding a macroblock, inter prediction and intra prediction are sequentially performed for each mode type shown in Table 1 (step 110 and After step 120), the rate-distortion cost for each mode is calculated (step 130), so that the mode type showing the minimum bit rate-distortion cost is selected as the optimal mode type (step 140) and compression is performed. To perform.

For reference, two factors, such as coding bits (rate) and image quality (distortion), may be considered in determining an optimal mode type of the encoder. The AVC / H.264 standard supports three modes in determining the mode type.

First, as the "RDO = ON" mode, the RDO function is used to measure the cost of each mode type. In this case, the RDO function calculates the cost incurred when one mode type is applied in consideration of the bit rate and image distortion.

Second, as the "RDO = ON and earlySkip" mode, the RDO function is used in the same manner as the "RDO = ON" mode to determine the mode type, but it is determined early whether the macroblock type is the skip mode. In case of skip mode, other mode types are not evaluated. Through this, there is an advantage that can significantly reduce the complexity compared to the "RDO = ON" mode.

Third, the "RDO = OFF" mode is a method of evaluating all the mode types by reducing the computational complexity by measuring only some bit rates and image distortions.

As described above, the method of determining one mode type from among numerous macroblock mode types for encoding every macroblock has enormous complexity. Enormous complexity also degrades the performance of the encoder, which can cause problems with real-time video data processing.

The present invention uses mode and temporal correlation among various mode types to select some candidate mode types (CMT, candidate mode types) by selecting the mode and encoding process of the AVC / H.264 standard. It is to provide an encoder and a method of determining a candidate mode of an encoder so that the operation can be performed quickly.

An object of the present invention is to provide a method of determining an encoder and a candidate mode of an encoder which can minimize the complexity of an encoder by simplifying a mode selection method using spatial and temporal correlation.

In addition, the present invention is to provide an encoder and a method of determining a candidate mode of the encoder is more effective in the real-time video processing is maximized due to the reduction of the complexity.

Another object of the present invention is to provide a method of determining a candidate mode of an encoder and an encoder to minimize video encoding time in all devices using the MPEG-4 AVC / H.264 encoder.

Other objects of the present invention will be readily understood through the following description.

According to an aspect of the present invention, there is provided a recording medium in which a candidate mode type determination method and a program for performing the method are recorded.

According to an embodiment of the present invention, a method of determining a candidate mode type (CMT) for encoding a macroblock performed in an encoder, the mode type for a reference macroblock in a current frame and b references in one or more previous frames Receiving a spatio-temporal context including a mode type index of a mode type for the macroblock; And determining a candidate mode type by referring to the mode type index included in the spatiotemporal context. Here, a and b may each be any natural number.

One or more of a sum value of the mode type index, the number of intra mode types, and the number of P8x8 mode types may be used to determine the candidate mode type.

If the sum of the mode type indexes included in the spatiotemporal context is 0, the SKIP mode and the 16x16 mode may be determined as candidate mode types.

If the sum of the mode type indexes included in the spatiotemporal context is 1 or more and less than n, the SKIP mode, the 16x16 mode, the 16x8 mode, and the 8x16 mode may be determined as candidate mode types. Here, n may be a predetermined natural number. For example, n may be 4.

If the sum of the mode type indexes included in the spatiotemporal context is not less than n and the number of intra mode types is 0, the SKIP mode, the 16x16 mode, the 16x8 mode, the 8x16 mode, and the P8x8 mode may be determined as candidate mode types.

If the sum of the mode type indexes included in the spatiotemporal context is not less than n, the number of intra mode types is not 0, and the number of P8x8 modes is 0, SKIP mode, 16x16 mode, 16x8 mode, 8x16 mode, I4MB mode, and I16MB mode May be determined as a candidate mode type.

If the sum of the mode type indexes included in the spatiotemporal context is not less than n, the number of intra mode types is not 0, and the number of P8x8 modes is not 0, SKIP mode, 16x16 mode, 16x8 mode, 8x16 mode, P8x8 mode, The I4MB mode and the I16MB mode may be determined as candidate mode types.

The two reference macroblocks in the current frame may be macroblocks located respectively above and to the left of the macroblock to be encoded. The b reference macroblocks may include a macroblock at the same position as the macroblock in a t-1th frame that is a frame immediately before the tth frame that is a current frame, and the macroblock in a t-2th frame that is a previous frame. It may include a macroblock in the same position.

According to another embodiment of the present invention, a method of determining a candidate mode type (CMT) for encoding a macroblock performed at an encoder, the mode type for a reference macroblock in a current frame and b references in at least one previous frame Receiving a spatiotemporal context including a mode type index of the mode type for the macroblock; And determining a candidate mode type according to a mode type index included in the spatiotemporal context based on prestored occurrence probability data. Here, the occurrence probability data is data in which occurrence probability values for the mode types of each macroblock for each context type are described, and a and b may each be any natural numbers.

The determining of the candidate mode type may include summing occurrence probability values in order of a mode type in which occurrence probability values are high based on the occurrence probability data until a preset frequency threshold is satisfied; And determining a mode type of the summed occurrence probability value as a candidate mode type.

The two reference macroblocks in the current frame may be macroblocks located respectively above and to the left of the macroblock to be encoded.

The b reference macroblocks are in the same position as the macroblock in the t-1th frame that is the immediately preceding frame of the tth frame that is the current frame and in the same position as the macroblock in the t-2th frame that is the previous frame. It may include a macroblock.

According to another embodiment of the present invention, there is provided a method of determining a candidate mode type (CMT) for encoding a macroblock performed in an encoder, the method comprising: storing a CMT table in which candidate mode types are designated for each of the input spatiotemporal contexts; Receiving a construction context that includes a mode type for a reference macroblock in a current frame and a mode type index of a mode type for b reference macroblocks in one or more previous frames; And selecting a candidate mode type designated to match the input spatiotemporal context with reference to the CMT table. Wherein a and b may each be any natural number.

The two reference macroblocks in the current frame may be macroblocks located above and to the left of the macroblock to be encoded, and the b reference macroblocks are the t-1 th frames that are immediately preceding the t th frame that is the current frame. The macroblock may include a macroblock at the same position as the macroblock in a frame and a macroblock at the same position as the macroblock in a t-2th frame, which is a previous frame.

According to another aspect of the present invention, an encoder to which the candidate mode type determination method is applied is provided.

An encoder according to an embodiment of the present invention uses a spatiotemporal context that includes a mode type index of a mode type for a reference macroblock in a current frame and a mode type for b reference macroblock in one or more previous frames. A candidate mode determination unit for determining a candidate mode type (CMT); A mode prediction unit that performs one or more of inter prediction and intra prediction for the candidate mode type; And a cost calculation unit for determining an optimal mode type according to one or more measurements of rate and distortion with respect to the mode type in which the mode prediction is performed. A and b may each be any natural number.

The candidate mode determination unit may determine a candidate mode type by using one or more of the sum of the mode type indexes included in the spatiotemporal context, the number of intra mode types, and the number of P8x8 mode types.

If the sum of the mode type indexes included in the spatiotemporal context satisfies the first condition of 0, the SKIP mode and the 16x16 mode are determined as candidate mode types, and the sum of the mode type indexes included in the spatiotemporal context is 1 or more and less than n. If the second condition is satisfied, the SKIP mode, the 16x16 mode, the 16x8 mode, and the 8x16 mode are determined as candidate mode types, and the sum of the mode type indexes included in the spatiotemporal context is less than n, and the number of intra mode types is zero. 3 If the condition is satisfied, SKIP mode, 16x16 mode, 16x8 mode, 8x16, and P8x8 mode are determined as candidate mode types, and the sum of the mode type indexes included in the spatiotemporal context is less than n, and the number of intra mode types is 0. If the fourth condition where the number of P8x8 modes is 0 is satisfied, SKIP mode, 16x16 mode, 16x8 mode, 8x16, I4MB mode, and I16MB mode are candidate mode types. When the first condition to the fourth condition is not satisfied, the SKIP mode, 16x16 mode, 16x8 mode, 8x16 mode, P8x8 mode, I4MB mode, and I16MB mode may be determined as candidate mode types. Here, n may be a predetermined natural number, for example, n may be 4.

The candidate mode determination unit may determine a candidate mode type according to a mode type index included in the spatiotemporal context based on prestored occurrence probability data. Here, the occurrence probability data may be data in which an occurrence probability value of a mode type of each macroblock for each context type is described.

The determining of the candidate mode type may include summing occurrence probability values in order of a mode type in which occurrence probability values are high based on the occurrence probability data until a preset frequency threshold is satisfied; And determining a mode type of the summed occurrence probability value as a candidate mode type.

According to an embodiment of the present invention, the candidate mode types (CMT) are selected by using temporal and temporal correlation among various mode types. There is an effect that the mode selection and encoding process can be made quickly.

In addition, by simplifying the mode selection method using spatial and temporal correlation, the complexity of the encoder can be minimized.

The reduced complexity also maximizes the performance of the encoder and has the advantage of being more effective for real-time video processing.

It also has the effect of minimizing video encoding time in all devices using MPEG-4 AVC / H.264 encoder.

As the invention allows for various changes and numerous embodiments, particular embodiments will be illustrated in the drawings and described in detail in the written description. However, this is not intended to limit the present invention to specific embodiments, it should be understood to include all transformations, equivalents, and substitutes included in the spirit and scope of the present invention. In the following description of the present invention, if it is determined that the detailed description of the related known technology may obscure the gist of the present invention, the detailed description thereof will be omitted.

The terms first, second, etc. may be used to describe various components, but the components should not be limited by the terms. The terms are used only for the purpose of distinguishing one component from another.

The terminology used herein is for the purpose of describing particular example embodiments only and is not intended to be limiting of the present invention. Singular expressions include plural expressions unless the context clearly indicates otherwise. In this application, the terms "comprise" or "have" are intended to indicate that there is a feature, number, step, operation, component, part, or combination thereof described in the specification, and one or more other features. It is to be understood that the present invention does not exclude the possibility of the presence or the addition of numbers, steps, operations, components, components, or a combination thereof. Also, it is to be understood that the symbol "/" includes "and" and "or" and is to be interpreted in either sense in the sentence or both.

Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings.

2 is a block diagram illustrating a configuration of a fast mode selector according to an embodiment of the present invention, and FIG. 3 is a process of determining a candidate mode type of a mode decision model (MDM) according to an embodiment of the present invention. Is a flow chart showing.

Referring to FIG. 2, the fast mode selection unit includes a candidate mode determination unit 210, a mode prediction unit 220, and a cost calculation unit 230. One or more components included in the fast mode selector or the fast mode selector may be implemented in a hardware configuration or in the form of a software algorithm.

The candidate mode determination unit 210 receives four macroblock type information (T ₁ , T ₂ , T ₃ , T ₄ ) and uses candidate mode types CMTs, using a mode decision model (MDM). candidate mode types) are selected and information about the selected posterior mode types is provided to the mode prediction unit 220.

Four macroblock type information may be recognized using temporal and temporal correlation with respect to a macroblock for determining a current encoding mode in three frames.

For example, as the spatial correlation, the type information of the top macroblock and the left macroblock of the macroblock in the current frame (t th frame) is immediately preceding the temporal correlation. Type information of the macroblock at the same position in the frame (t-1st frame) and the previous frame (t-2th frame) may be used. Here, the candidate mode type determination method according to the present embodiment may be used when t is a natural number of 3 or more.

 In the above example, the temporal and temporal correlation for the type of the macroblock at the position (i, j) in the t-th frame may be expressed by Equation 1 below.

(T1,T2,T3,T4) = ()

(T1, T2, T3, T4) = ()

은 t번째 프레임의 (i, j) 위치의 매크로블록 타입을 나타낸다.here,

Denotes a macroblock type at position (i, j) of the t-th frame.

Of course, the positions of the four macroblocks referred to are not limited to the above example, and may be arbitrarily designated as long as the mode type is determined in advance of the macroblock in time. In addition, it is obvious that the number of macroblocks for reference is not limited to the four macroblocks shown in the above example.

The method of notifying candidate mode types (CMTs) selected for the macroblock type at the position (i, j) of the t-th frame using the four MB types information according to Equation 1 described above is shown in Equation 2 below. May be exemplified.

CMT (T1, T2, T3, T4) = b ₀ b ₁ ... b _n

Here, b _k = 1 when the mode type k is selected, and b _k = 0 otherwise. The mode type k can be found in the mode type index of Table 1 described above. That is, the CMT may be expressed in the form of 7 bits, and each bit indicates whether each mode type is selected in the order of the mode type indexes described in Table 1. Of course, the notation or order of the CMT may be variously designated.

For example, in the current AVC / H.264 JM encoder, since all mode types are always activated in every macroblock, the CMT may be expressed as “1111111” when referring to Equation 2 and FIG. 2. However, if only SKIP and 16x16 mode types are selected as candidate mode types, the CMT may be indicated as “1100000”. Likewise, if only SKIP, 16x16, 16x8 and 8x16 mode types are selected as candidate mode types, the CMT may be indicated as "1111000".

The mode decision model (MDM) of the candidate mode decision unit 210 may include a spatio-temporal context such as information about candidate mode types CMTs (for example, (T1, T2, T3, T4). temporal context)) to receive candidate mode types more effectively.

The type of mode types may be set to five as shown in Table 2 below so that the mode decision model may determine candidate mode types more quickly. Of course, the type of mode types may be set in more detail.

Table 2. CMT format based on mode type

Table 2 lists the mode types and the mode types supported by each CMT. The mode decision model determines the CMT type using the spatiotemporal context of each macroblock. In this case, examples of the contexts used are as follows.

First, there may be a sum of contexts (Context_sum = T1 + T2 + T3 + T4) that is a sum of T1, T2, T3, and T4. As shown in Equation 1, if (T1, T2, T3, T4) = (0, 0, 0, 0), all four reference macroblocks are in SKIP mode according to Table 1, the sum (Context_sum) ) Becomes 0. However, if (T1, T2, T3, T4) = (0, 0, 0, 1), then, according to Table 1, three reference macroblocks are in SKIP mode and the other reference macroblock is 16x16 mode. Context_sum) is 1.

Next, there may be a number of intra mode types (Intra_cnt) in the context. If (T1, T2, T3, T4) = (0, 0, 0, 0), all four reference macroblocks are in inter mode according to Table 1, and Intra_cnt is 0.

Next, there may be a number P8x8_cnt of P8x8 mode types in the context. If (T1, T2, T3, T4) = (0, 0, 0, 0), according to Table 1, all four reference macroblocks are SKIP modes among the inter modes, and P8x8_cnt becomes 0.

Hereinafter, a method of determining a CMT type using a spatiotemporal context into which a mode determination model is input will be described with reference to FIG. 3.

In operation 310, the mode determination model determines whether the sum of contexts Context_sum is zero. If the sum of contexts is 0, it is determined that the CMT 1 type, and 1100000 is output as the CMT information.

However, if the sum of contexts is not zero, the mode decision model determines whether the sum of contexts (Context_sum) is less than n, which is a natural number, in step 320. However, it is assumed here that n is 4. This is because the sum of the probability of occurrence where n is less than 4 reaches 45% when referring to the probability of occurrence of the context type and the macroblock illustrated in Table 3 to be described later. Of course, the value of n may be different according to the probability accuracy setting according to the analysis of the designer or the user.

If the sum of the contexts is less than 4, it is determined that the CMT 2 type, and 111111 is output as the CMT information.

However, when Context_sum is 4 or more, the existence of intra mode types or P8x8 mode types is an important variable as described below.

That is, if the sum of the contexts is not less than 4, the mode determination model determines whether the number of intra mode types Intra_cnt is 0 in step 330. If the number of intra mode types is 0, it is determined to be a CMT 3 type, and 1111100 is output as CMT information. This is because if there are no intra mode types in the context, there is a high probability that the intra mode types do not exist in the current macroblock CMT.

However, if there are intra mode types, the CMT type is determined according to the presence or absence of the P8x8 mode type as described below.

That is, if the number of intra mode types is not zero, the mode determination model determines whether the number of P8x8 mode types (P8x8_cnt) is zero at step 340. If the number of P8x8 mode types (P8x8_cnt) is 0, it is determined to be a CMT 4 type, and 1111011 is output as CMT information. However, if the number of P8x8 mode types (P8x8_cnt) is not 0, it is determined that the CMT 5 type and outputs 1111111 as CMT information.

As discussed above, having the mode decision model determine the CMT using the spatiotemporal context can significantly reduce the complexity caused by evaluating all or multiple mode types.

The mode prediction unit 220 performs inter prediction and / or intra prediction sequentially or in parallel using information on the candidate mode type input from the candidate mode determination unit 210.

The cost calculating unit 230 calculates a rate-distortion cost for the modes according to the prediction result of the mode prediction unit 220, thereby converting the mode type showing the minimum bit rate-distortion cost into an optimal mode type. Choose. The encoding will be performed with the selected optimal mode type.

The cost calculating unit 230 is a separate mode for selecting the optimum mode type, the "RDO = ON" mode, the "RDO = ON and earlySkip" mode, the "RDO = OFF" mode, and other modes for determining the optimum mode type. It may be able to perform one or more of any of the modes.

4 is a flowchart illustrating a candidate mode type determination process of a mode decision model (MDM) according to another embodiment of the present invention.

The mode decision model may select the top several mode types having a high occurrence frequency for each input spatiotemporal context as candidate mode types.

Table 3 below shows the probability of a context type that can occur in all macroblocks and the probability of occurrence of macroblocks having the context type.

Table 3. Probability of occurrence of context type and macroblock

Table 3 shows data for the top 10 contexts out of a total of 2,401 context types. As shown in Table 3, 10 contexts account for nearly 50% of the probability of generating 2,401 contexts. Also. In the case of all the contexts (0, 0, 0, 0), the SKIP mode accounts for about 34% of the total probability of generating 2,401 contexts. In addition, in the case of context (0, 0, 0, 0), the candidate mode for the current macroblock is 0, that is, in case of the SKIP mode, there is a 96% probability of occurrence.

The mode determination model may determine the candidate mode type based on the above probability data. Hereinafter, a candidate mode type determination method of the mode determination model will be described with reference to FIG. 4.

Referring to FIG. 4, a frequency threshold is set in step 410. The frequency threshold may be set as a probability value such as, for example, 95% as an occurrence rate reference value, and may be set as a default or / and modified by a user.

If a spatio-temporal context is input in step 420, the mode determination model selects a candidate mode type using occurrence probability data for the mode type of the current macroblock according to the spatio-temporal context input in step 430.

In step 410, if the frequency threshold is set to 95%, and the input spatiotemporal context is (0, 0, 0, 0), 0 (SKIP mode) having a probability of occurrence of 96% based on the probability of occurrence data is a candidate mode type. Is selected. This is because the frequency threshold is satisfied only by the occurrence probability for the SKIP mode, so other mode types need not be considered.

However, if the spatiotemporal context input for the same frequency threshold is (8,8,8,8), then the candidate mode type is 8 (P8x8) with 70% probability in order of occurrence frequency, 1 (with 11% probability). 16x16), 2 (16x8) with 9% probability, 3 (8x16) with 8% probability, 9 (I4MB) with 2% probability, and mode types 8 (P8x8) that satisfy the frequency threshold. ), 1 (16x16), 2 (16x8), and 3 (8x16-8) may be selected as candidate mode types. This is because the sum of the occurrence probabilities of the mode types selected as candidate mode types satisfies the frequency threshold (70% + 11% + 9% + 8% = 98%). If the frequency threshold is specified as 90% in this case, only 8 (P8x8), 1 (16x16) and 2 (16x8) will be selected as candidate mode types for the same reason (70% + 11% + 9% = 90%). ).

As described above, the mode decision model may adjust the number of candidate mode types at a set frequency threshold. This means that the compression rate and complexity of the entire encoding can be adjusted.

The mode decision model outputs information (CMT information) about candidate mode types selected in step 440.

In addition, according to another embodiment of the present invention, a mode decision model for extracting candidate mode type information using a prestored CMT table according to a spatiotemporal context may be provided. That is, the mode type for n (arbitrary natural numbers) reference macroblocks in the current frame and k in one or more previous frames, with the CMT table pre-designated candidate mode types according to all kinds of spatiotemporal contexts in advance. When the spatiotemporal context including the mode type index of the mode type for the reference macroblock is input, CMT information may be generated by extracting candidate mode types predetermined to match the spatiotemporal context input from the pre-configured CMT table. It may be. In this case, the CMT table may be variously configured according to the needs or design intentions of a user or a designer.

5 to 8 are graphs illustrating measurement of encoding processing performance according to an embodiment of the present invention.

Table 4 below shows the average encoding time per sequence in the "RDO = ON" mode described above. According to the fast mode selection with MDM (FMS-MDM) according to an embodiment of the present invention, about 25% in CIF and SIF sequences, and in full HD sequences, compared to the conventional AVC / H.264 JM encoder It's about 20% faster. In addition, in all cases, it can be seen that the mode selection method according to the embodiment shows a faster encoding speed.

Table 4. Average Encoding Time by Sequence in RDO = ON Mode

Table 5 below shows the average encoding time of the entire sequence for each RDO mode. As shown in Table 5, the mode selection method according to the embodiment shows excellent performance in all RDO modes, and shows relatively best performance in the "RDO = ON" mode.

Table 5. Average Encoding Time of Entire Sequence by RDO Mode

In addition, as shown in the measurement graphs of FIGS. 5 to 8, the compression performance of the mode selection method (FMS-MDM) according to the embodiment is identical to that of the AVC / H.264 JM encoder in all RDO modes. For reference, FIG. 5 is a measurement graph for mobile, FIG. 6 is foreman, FIG. 7 is pedestrian, and FIG. 8 is rush-hour.

In conclusion, the proposed technique has the advantage of significantly reducing the encoding time at the same compression performance.

It is apparent that the above-described candidate mode determination method may be performed by an automated procedure in time series order by a software program or the like embedded in the encoder. The codes and code segments that make up the program can be easily deduced by a computer programmer in the field. In addition, the program is stored in a computer readable media, and read and executed by a computer to implement the method. The information storage medium includes a magnetic recording medium, an optical recording medium and a carrier wave medium.

Although the above has been described with reference to embodiments of the present invention, those skilled in the art may variously modify the present invention without departing from the spirit and scope of the present invention as set forth in the claims below. And can be changed.

1 is a view showing an optimal mode type selection process in AVC / H.264 JM 11.0 according to the prior art.

2 is a block diagram showing a configuration of a fast mode selection unit according to an embodiment of the present invention.

3 is a flowchart illustrating a process of determining a candidate mode type of a mode decision model (MDM) according to an embodiment of the present invention.

4 is a flowchart illustrating a process of determining a candidate mode type of a mode decision model (MDM) according to another embodiment of the present invention.

5 to 8 are graphs showing measurement performance of encoding processing according to an embodiment of the present invention.

Claims (21)

A method of determining a candidate mode type (CMT) for encoding a macroblock performed at an encoder, A mode type for a reference macroblock in a t-th frame that is a current frame containing a macroblock to be encoded, and the macroblock to be encoded in each of a t-1 th frame and a t-2 th frame that are previous frames of the current frame Receiving a spatio-temporal context including mode type indexes of the mode type for the reference macroblock at the same position as the first macroblock; Determining a sum value of the mode type indexes included in the spatiotemporal context, the number of intra mode types among the mode types, and the number of P8x8 mode types among the mode types; And Selecting a group type of any one candidate mode type from among preset group types of candidate mode types by using the result information of the determination, Wherein a is any natural number and t is any natural number of 3 or more. delete The method of claim 1, And when the result information indicates that the sum of the mode type index is zero, the group type of the selected candidate mode type includes a SKIP mode and a 16x16 mode as candidate modes. . The method of claim 3, If the result information indicates that the sum of the mode type index is 1 or more and less than n, the group type of the selected candidate mode type includes a SKIP mode, a 16x16 mode, a 16x8 mode, and an 8x16 mode as candidate modes. And n is a predetermined natural number set in advance. The method of claim 4, wherein When the result information indicates that the sum of the mode type index is not less than n and the number of the intra mode types is zero, the group type of the selected candidate mode type is SKIP mode, 16x16 mode, 16x8 mode. And a 8x16 mode and a P8x8 mode as candidate modes. The method of claim 5, If the result information indicates that the sum of the mode type index is not less than n, the number of the intra mode types is not 0, and the number of the P8x8 modes is 0, the group type of the selected candidate mode type is SKIP. 12. A method for determining a candidate mode type comprising a mode, a 16x16 mode, a 16x8 mode, an 8x16 mode, an I4MB mode, and an I16MB mode. The method of claim 6, If the result information indicates that the sum of the mode type index is not less than n, the number of the intra mode types is not 0, and the number of the P8x8 modes is not 0, the group type of the selected candidate mode type is SKIP. 12. A method for determining a candidate mode type comprising a mode, a 16x16 mode, a 16x8 mode, an 8x16 mode, a P8x8 mode, an I4MB mode, and an I16MB mode. The method of claim 4, wherein Wherein n is 4; The method of claim 1, A reference macroblocks in the current frame are macroblocks positioned above and to the left of the macroblock to be encoded, respectively. A program of instructions that can be executed by a digital processing apparatus for performing the method of determining a candidate mode type according to any one of claims 1 to 3 is tangibly implemented and can be read by the digital processing apparatus. Recording medium that records the program. In the encoder, A mode type for a reference macroblock in a t-th frame that is a current frame containing a macroblock to be encoded, and the macroblock to be encoded in each of a t-1 th frame and a t-2 th frame that are previous frames of the current frame Receiving a spatio-temporal context including mode type indexes of the mode type for the reference macroblock which is the same position as, and a sum value of the mode type indexes included in the spatiotemporal context, Determining a candidate mode for selecting a group type of any one candidate mode type among preset group types of candidate mode types by using the result of determining the number of intra mode types among the mode types and the number of P8x8 mode types among the mode types. unit; A mode prediction unit that performs one or more of inter prediction and intra prediction on a candidate mode included in the group type of the selected candidate mode type; And A cost calculation unit for determining an optimal mode type according to one or more measurements of bit rate and image distortion for the mode type on which the mode prediction has been performed, Wherein a is any natural number and t is any natural number of 3 or more. The method of claim 11, And if the result information indicates that the sum of the mode type index is zero, the group type of the selected candidate mode type includes a SKIP mode and a 16x16 mode as candidate modes. The method of claim 12, If the result information indicates that the sum of the mode type index is 1 or more and less than n, the group type of the selected candidate mode type includes a SKIP mode, a 16x16 mode, a 16x8 mode, and an 8x16 mode as candidate modes. And n is a predetermined natural number. The method of claim 13, When the result information indicates that the sum of the mode type index is not less than n and the number of the intra mode types is zero, the group type of the selected candidate mode type is SKIP mode, 16x16 mode, 16x8 mode. And an 8x16 mode and a P8x8 mode as candidate modes. The method of claim 14, If the result information indicates that the sum of the mode type index is not less than n, the number of the intra mode types is not 0, and the number of the P8x8 modes is 0, the group type of the selected candidate mode type is SKIP. And a candidate mode including a mode, a 16x16 mode, a 16x8 mode, an 8x16 mode, an I4MB mode, and an I16MB mode. The method of claim 15, If the result information indicates that the sum of the mode type index is not less than n, the number of the intra mode types is not 0, and the number of the P8x8 modes is not 0, the group type of the selected candidate mode type is SKIP. And a candidate mode including a mode, a 16x16 mode, a 16x8 mode, an 8x16 mode, a P8x8 mode, an I4MB mode, and an I16MB mode. delete delete delete delete delete

Images