KR101009740B1 - Apparatus and method for efficient implementation of the rate-distortion optimized mode decision algorithm - Google Patents

Abstract

The bit rate estimation method includes a binarization step, a context grouping step, and a bit rate determination step. The binarization step receives an input signal and performs binarization to output a bit string that is a sequence of bits having a value of 0 or 1. The context grouping step generates the context group by receiving the bit string and grouping data using the same context. The bit rate determining step determines the estimated bit rate for each context group using first and second bit rate estimation tables and adds the estimated bit rates for each context group to provide a bit rate output.

Description

Apparatus and Method for Bit Rate Estimation for Efficient Implementation of Rate-Distortion Optimized Mode Decision Algorithm {APPARATUS AND METHOD FOR EFFICIENT IMPLEMENTATION OF THE RATE-DISTORTION OPTIMIZED MODE DECISION ALGORITHM}

The present invention relates to a bit rate estimation apparatus and method, and more particularly, to a bit rate estimation method and method for efficient implementation of a rate-distortion optimized mode decision algorithm. will be.

With the development of image compression technology, many complex algorithms have been added to the encoder, thereby increasing the number of encoding modes that can be selected per coding unit (eg, macroblock). In order to select one mode among many encoding modes, a specific criterion must be calculated. The best performance among them is the rate-distortion cost function of Equation 1 below.

[Equation 1]

J = D + λ R

Here, λ denotes a Lagrange multiplier, D denotes how much the encoded image is lost compared to the original image, and R denotes a bit that is actually transmitted after the encoding process is completed for the candidate coding unit. Represents a bit-rate. A method of determining a mode of a coding unit to minimize the rate-distortion cost function is called a rate-distortion optimized mode decision algorithm. When the algorithm is used, the size of data to obtain the same image quality is reduced, but the complexity of the encoder is increased, compared to the case of using the low complexity mode decision algorithm commonly used in H.264 / AVC.

That is, in order to obtain a bit-rate (R) in the rate-distortion optimized mode determination algorithm, since the actual number of bits must be obtained by performing encoding on the candidate coding unit, the candidate for determining the mode of one coding unit. If there are 10, the mode for one coding unit can be determined only after coding 10 coding units, thereby increasing the complexity of the encoder and slowing down the mode decision speed.

Accordingly, there is a need for a method and apparatus for rapidly estimating a bit rate for a candidate coding unit while reducing the complexity of the encoder.

There are two entropy coding algorithms adopted as the H.264 / AVC standard: Context-Adaptive Variable Length Coding (CAVLC) and Context-Adaptive Binary Arithmetic Coding (CABAC). Of these, CABAC is more complex than CAVLC but has better compression efficiency. Can be obtained. Up to now, CAVLC-based encoder system has been used and research on how to calculate bit rate quickly has been conducted on encoder system based on CAVLC. However, as the demand for high-performance encoder increases, CABAC-based There is also a need for a method and apparatus for quickly calculating a bit rate in order to improve the speed of an encoder while reducing the complexity of the encoder.

Accordingly, an object of the present invention is to provide an apparatus capable of effectively estimating a bit rate in a system using CABAC as an entropy encoder.

One object of the present invention is to provide a method for effectively estimating a bit rate in a system using CABAC as an entropy encoder.

In order to achieve the above object of the present invention, the bit rate estimation apparatus according to the first embodiment of the present invention includes a binarization unit, a context loader, and a table reading unit. The binarizer receives an input signal, performs binarization, and provides a bit stream. The context loader receives the bit string to determine the context used by the bit string and provides context data associated with the context. The table reader receives the bit string and the context data and provides a bit rate output using a bit rate estimation table.

The bit rate estimating apparatus may further include a memory unit configured to store the context data and the bit rate estimation table.

In example embodiments, the context data may include a maximum probability symbol (MPS) indicating a symbol having a higher probability of occurrence, and a probability state indicator indicating a probability of occurrence of the MPS.

In an embodiment, the probability state indicator has an integer value between 0 and 62, and the probability of occurrence of the MPS may increase as the value of the probability state indicator increases.

In an embodiment, the bit rate estimation table is provided for each value of the probability state indicator, respectively, when a bit included in the bit string matches the MPS and a bit included in the bit string does not match the MPS. The estimated bit rate can be stored.

In an embodiment, the estimated bit rate is probabilistic for each value of the probability state indicator by modeling a range, which is an internal variable of a CABAC algorithm, as a random variable having an integer value of 256 or more and 510 or less. Can be determined.

는 상기 range가 A인 경우 구간세분화(interval subdivision)한 이후의 변경된 range를 나타내고,

(k)|_σ,α는 주어진 σ 및 α에 대해 상기

가 특정 값 k가 될 확률을 나타내고, b(k)는 상기

가 특정 값 k일 경우 재정규화(renormalization) 과정을 통해 생성되는 비트의 수를 나타낸다).In an embodiment, the estimated bit rate may be generated using Equation 1 below, where σ represents the probability state indicator, α represents each bit included in the bit string, and B ( σ, α) represents the estimated bit rate produced probabilistic for a given σ and α,

Represents a changed range after interval subdivision when the range is A,

(k) | _{σ, α} is for the given σ and α

Denotes the probability of being a specific value k, and b (k) is

Is a specific value k, it represents the number of bits generated through the renormalization process).

[Equation 1]

(k)|_σ,α는 아래의 [수학식 2]에 의해 생성될 수 있다(여기서, p_A는 상기 range의 확률분포를 나타내고, δ는 두 인자가 서로 같을 경우에는 1을 갖고 두 인자가 서로 다를 경우에는 0을 갖는 함수를 나타내고,

|_σ,α는 상기 range가 a인 경우에 주어진 σ 및 α에 대해 구간세분화한 결과를 나타낸다). In an embodiment, the

(k) | _{σ, α} can be generated by Equation 2 below, where p _A represents the probability distribution of the range, δ is 1 when the two factors are equal to each other, and Represents a function with zero,

| _{σ, α} represents the result of interval segmentation for σ and α given when the range is a).

[Equation 2]

In an embodiment, the bit rate output may be generated by Equation 3 below (where x represents specific coding unit data, R _x represents an estimated bit rate of the specific coding unit data, and α _k Denotes the k-th bit included in the bit string, σ _k denotes the value of the probability state indicator loaded in the k-th bit processing included in the bit string, and B (σ _k , α _k ) denotes the given σ _k and the estimated bit rate read from the bit rate estimation table for α _k ).

&Quot; (3) "

In order to achieve the above object of the present invention, the bit rate estimation method according to the first embodiment of the present invention includes a binarization step, a context loading step and a table reading step. The binarization step receives an input signal, performs binarization, and provides a bit stream. The context loading step receives the bit string to determine the context used by the bit string and provides context data associated with the context. The table reading step receives the bit string and the context data and provides a bit rate output using a bit-rate estimation table.

In example embodiments, the context data may include a maximum probability symbol (MPS) indicating a symbol having a higher probability of occurrence, and a probability state indicator indicating a probability of occurrence of the MPS.

In an embodiment, the bit rate estimation table may be configured for each value of the probability state indicator when the bits included in the bit string match the MPS and when the bits included in the bit string do not match the MPS. It can store the estimated bit rate for.

The reading of the table may include determining whether a bit included in the bit string matches the MPS, and if the bit included in the bit string matches the MPS, the probability state from the bit rate estimation table. The estimated bit rate for the case where the bit included in the bit string matches the MPS is read with respect to the current value of the indicator, and the probability is obtained from the bit rate estimation table if the bit included in the bit string does not match the MPS. Reading an estimated bit rate for a case where a bit included in the bit string does not match the MPS with respect to a current value of a status indicator, and summing the read estimated bit rates for each bit included in the bit string; Providing an output.

In order to achieve the above object of the present invention, the bit rate estimation method according to the second embodiment of the present invention includes a binarization step, a context grouping step, and a bit rate determination step. The binarization step receives an input signal, performs binarization, and provides a bit stream. The context grouping step generates the context group by receiving the bit string and grouping data using the same context. The bit rate determining step determines the estimated bit rate for each context group by using first and second bit-rate estimation tables and adds the estimated bit rates for each context group to provide a bit rate output.

The bit rate determining step may include: a maximum probability symbol (MPS) indicating a symbol having a higher probability of occurrence among 0 and 1 and a probability state indicator indicating a probability of occurrence of the MPS for each of the context groups. A context loading step of loading, for each of the context groups, the MPS and the least probable symbol (LPS) included in the context group are separated, and the first number and the lowest probability symbols, which are the number of the MPSs, are separated. Symbol separation step of determining a second number which is the number of (Least Probable Symbol, LPS), for each of the context group using the first bit rate estimation table to determine a first estimated bit rate for the MPS, the second A table reading step of determining a second estimated bit rate for the LPS using a bit rate estimation table, the first estimated bit rate for each of the context groups The second may comprise a group-specific bit-rate determining step estimated by summing a bit rate for determining the group context-dependent estimated bit rate and the bit rate sum providing a bit rate output by adding the context, the group-specific bit-rate estimation.

In an embodiment, the probability state indicator has an integer value between 0 and 62, and the probability of occurrence of the MPS may increase as the value of the probability state indicator increases.

In example embodiments, the first bit rate estimation table may, for each value of the probability state indicator, indicate that the bits included in the bit string match the MPS and that the bits included in the bit string do not match the MPS. In this case, an estimated bit rate for each can be stored.

In an embodiment, the estimated bit rate stored in the first bit rate estimation table may be determined by modeling a range, which is an internal variable of a Context-Adaptive Binary Arithmetic Coding (CABAC) algorithm, as a random variable having an integer value of 256 or more and 510 or less. Probability can be determined for each value.

는 상기 range가 A인 경우 구간세분화(interval subdivision)한 이후의 변경된 range를 나타내고,

(k)|_σ,α는 주어진 σ 및 α에 대해 상기

가 특정 값 k가 될 확률을 나타내고, b(k)는 상기

가 특정 값 k일 경우 재정규화(renormalization) 과정을 통해 생성되는 비트의 수를 나타낸다).In an embodiment, the estimated bit rate stored in the first bit rate estimation table may be generated using Equation 4 below (where σ represents the probability state indicator and α is each included in the bit string). Denotes the bit of, and B (σ, α) denotes the estimated bit rate generated probabilistic for a given σ and α,

Represents a changed range after interval subdivision when the range is A,

(k) | _{σ, α} is for the given σ and α

Denotes the probability of being a specific value k, and b (k) is

Is a specific value k, it represents the number of bits generated through the renormalization process).

&Quot; (4) "

(k)|_σ,α는 아래의 [수학식 5]에 의해 생성될 수 있다(여기서, p_A는 상기 range의 확률분포를 나타내고, δ는 두 인자가 서로 같을 경우에는 1을 갖고 두 인자가 서로 다를 경우에는 0을 갖는 함수를 나타내고,

|_σ,α는 상기 range가 a인 경우에 주어진 σ 및 α에 대해 구간세분화한 결과를 나타낸다). In an embodiment, the

(k) | _{σ, α} can be generated by Equation 5 below, where p _A represents the probability distribution of the range, δ is 1 when the two factors are equal to each other, and Represents a function with zero,

| _{σ, α} represents the result of interval segmentation for σ and α given when the range is a).

[Equation 5]

In an embodiment, the first estimated bit rate of each of the context groups may be determined by Equation 6 and Equation 7 below, where B (S _MPS ) represents the first estimated bit rate. , B (σ _M , MPS) represents an estimated bit rate read from the first bit rate estimation table for the case where the probability state indicator is σ _M and MPS is input, and n _MPS represents the first number, [sigma] _old represents a value of a probability state indicator before the first estimated bit rate calculation, and [] represents a Gaussian symbol).

&Quot; (6) "

[Equation 7]

In an embodiment, the second bit rate estimation table classifies values of the probability state indicators of 0 to 62 into 14 probability state indicator groups having group numbers of 0 to 13. For example, an estimated bit rate for a case where a bit included in the bit string matches the LPS may be stored.

In example embodiments, the probability state indicator groups may have the same number of consecutive LPSs required to make the probability state indicator zero for each value of the probability state indicator in a context-adaptive binary arithmetic coding (CABAC) algorithm. The probability state indicators are grouped and the probability state indicator group number may indicate the number of consecutive LPSs required to make the probability state indicators zero.

)는 그룹

의 추정 비트율을 나타내고, B(s,LPS)는 상기 확률상태 지시자가 s이고 LPS가 입력되었을 경우 에 대해 상기 제 1 비트율 추정 테이블로부터 판독한 추정 비트율을 나타내고, p_σ(s)는 CABAC 부호화 과정 중에 상기 확률상태 지시자가 특정 값 s를 가질 확률을 나타낸다).In an embodiment, the estimated bit rate stored in the second bit rate estimation table may be generated using Equation 8 below (where, B _G (

) Group

B (s, LPS) denotes an estimated bit rate of, and B (s, LPS) denotes an estimated bit rate read from the first bit rate estimation table for the case where the probability state indicator is s and LPS is input, and p _σ (s) is a CABAC encoding process. The probability state indicator has a specific value s).

[Equation 8]

In an embodiment, p _σ (s) may be generated by Equation 9 below, where q is a real number greater than 0 and less than 1.

[Equation 9]

In an embodiment, the second estimated bit rate of each of the context groups may be determined by Equations 10, 11, and 12 below, where σ _MPS is the first estimate. Represents the value of the probability state indicator after bit rate calculation, σ _old represents the value of the probability state indicator before the first estimated bit rate calculation, n _MPS represents the first number, and n _LPS represents the second number G _num (σ) represents the probability state indicator group number to which the probability state indicator σ belongs, B (S _LPS ) represents the second estimated bit rate, and B _G (G _M ) represents the probability state indicator group. And an estimated bit rate read from the second bit rate estimation table for the case where the group number of is G _M ).

[Equation 10]

[Equation 11]

[Equation 12]

_, if G_num(σ_MPS) ≥ n_LPS

_, if G _num (σ _MPS ) ≥ n _LPS

_, if G_num(σ_MPS) < n_LPS

_, if G _num (σ _MPS ) <n _LPS

In example embodiments, the estimated bit rate for each context group may be generated by Equation 13 below, and the bit rate output may be generated by Equation 14 below, where x is specific coding unit data. Where R _{x, j} represents the estimated bit rate for each context group for the j-th context group in coding unit data x, and B _{x , j} (S _MPS ) represents the for the j-th context group in coding unit data x. Represents a first estimated bit rate, B _{x , j} (S _LPS ) represents the second estimated bit rate for the j-th context group in coding unit data x, and R _x represents the bit rate output for coding unit data x , m _x represents the number of contexts used by the coding unit data x).

[Equation 13]

[Equation 14]

In order to achieve the above object of the present invention, a bit rate estimation apparatus according to a second embodiment of the present invention includes a binarization unit, a context grouping unit, a bit number calculating unit, a switch, a context loading unit, and a plurality of group rate estimation units. And an accumulator. The binarizer receives an input signal, performs binarization, and provides a bit stream. The context grouper generates the context group by receiving the bit string and grouping data using the same context. The bit number calculator receives the bit string and provides the number of bits belonging to the bit string. The switch selectively connects the output of the binarizer to the context grouper or the bit number calculator according to a logic level of a bypass signal. The context loading unit receives the context group to identify a context used by the context group, and indicates that a maximum probability symbol (MPS) and a MPS indicating a symbol having a higher probability of occurrence among 0 and 1 for the context are determined. Provides a probability state indicator indicating the probability of occurrence. The bit rate estimator for each group determines the estimated bit rate for each context group by receiving the MPS and the probability state indicator for the context group and the context group, respectively. The accumulation unit accumulates and sums outputs of the plurality of group bit rate estimation units and outputs of the bit number calculator to provide a bit rate output.

Hereinafter, a bit rate estimation apparatus and method for an efficient implementation of a rate-distortion optimized mode decision algorithm according to embodiments of the present invention with reference to the accompanying drawings. Although described in detail, the present invention is not limited to the following embodiments, and those skilled in the art may implement the present invention in various other forms without departing from the technical spirit of the present invention. will be.

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 "having" are intended to indicate that there is a feature, number, step, action, component, part, or combination thereof that is described, and that one or more other features or numbers are present. It should be understood that it does not exclude in advance the possibility of the presence or addition of steps, actions, components, parts or combinations thereof. Unless defined otherwise, all terms used herein, including technical or scientific terms, have the same meaning as commonly understood by one of ordinary skill in the art. Terms such as those defined in the commonly used dictionaries should be construed as having meanings consistent with the meanings in the context of the related art, and, unless expressly defined in this application, are construed in ideal or excessively formal meanings. It doesn't work.

On the other hand, when an embodiment is otherwise implemented, a function or operation specified in a specific block may occur out of the order specified in the flowchart. For example, two consecutive blocks may actually be performed substantially simultaneously, and the blocks may be performed upside down depending on the function or operation involved. Like reference numerals in the drawings denote like elements.

First embodiment

1 is a block diagram illustrating a bit rate estimation apparatus according to a first embodiment of the present invention.

Referring to FIG. 1, the bit rate estimating apparatus 100 includes a binarizer 110, a context loader 120, and a table reader 130.

The binarization unit 110 receives an input signal and performs binarization to output a bit stream that is a sequence of bits having a value of 0 or 1. The context loader 120 receives the bit string to determine the context used by the bit string and provides context data related to the context. The table reader 130 receives the bit string and the context data and provides a bit-rate output using a bit-rate estimation table.

The bit rate estimation apparatus 100 may further include a memory unit 140 that stores the context data and the bit rate estimation table.

The context data includes a maximum probability symbol (MPS) representing a symbol having a higher probability of occurrence of 0 and 1 in data using the context and a probability state indicator (σ) indicating a probability of occurrence of the MPS. . Specifically, if the MPS is 1, it means that 1 is more likely to occur in the data using the context, and the probability state indicator (σ) is equal to or greater than 0 as defined in the CABAC standard context-adaptive algorithm. As the probability state indicator (σ) is close to 0 with an integer value of 62 or less, the probability of occurrence of MPS is close to 0.5, and the probability of MPS occurrence is close to 1.0 as the probability state indicator (σ) is closer to 62. In addition, the MPS and probability state indicator [sigma] is updated as defined in the CABAC standard algorithm after reading the estimated bit rate from the bit rate estimation table for each bit.

The bit rate estimation table includes, for each value of the probability state indicator σ, the number of probable bits that will occur when the MPS is input and the probability bits that will occur when the least probability symbol (LPS) is input. Store the number of. Specifically, the bit rate estimation table stores estimated bit rates for the case where the bits included in the bit string are MPS and the LPS for each value of the probability state indicator (σ) of 0 or more and 62 or less.

2 shows an example of the bit rate estimation table. In FIG. 2, σ represents the probability state indicator, α represents each bit included in the bit string, B (σ, α) represents the estimated bit rate, which is generated probabilistic for a given σ and α, W (σ, α) indicates that B (σ, α) is quantized and expressed as a weight of an integer value, such that 1.00 corresponds to 8 and 0.50 corresponds to 4. W (σ, α) is an integer of the estimated bit rate to facilitate hardware implementation, and the final bit rate output may be generated by shifting the bit rate output to the right three bits in the bit rate output step. In FIG. 2, 1.00 is quantized to correspond to 8, but another method of quantization may be applied. For example, 1.00 may be quantized to correspond to 16. In this case, the final bitrate output may be generated by shifting the bitrate output 4 bits to the right.

Hereinafter, an exemplary method for determining each estimated bit rate stored in the bit rate estimation table will be described in detail. In this embodiment, each value of the probability state indicator σ is modeled by modeling a range (hereinafter referred to as A), which is an internal variable of the CABAC algorithm, as a random variable having an integer value of 256 or more and 510 or less. Probably determine an estimated bit rate for.

를 구하고, 재정규화 과정에서는

이 2⁸=256보다 작으면

≥256이 만족될 때까지

를 두 배로 만든다. 이 과정을 거쳐 나온 값은 다음 입력 비트를 처리할 때에 새로운 A값으로 사용된다. 여기서

이 두 배 될 때마다 한 비트가 출력된다. 예를 들어,

가 128≤

<256을 만족할 경우 한 비트가 출력되고,

가 62≤

<128을 만족할 경우 두 비트가 출력된다. 즉,

를 알면 몇 비트가 생성될지 알 수 있다. 여기서,

값을 생성되는 비트의 수로 대응시켜주는 함수 b(

)를 정의한다. In more detail, CABAC goes through two processes: interval subdivision and renormalization to encode one bit. In the segmentation process, a new operation is performed by performing a specific operation on the internal variable A according to the input bit.

In the process of fiscal regulation

Is less than 2 ⁸ = 256

Until ≥256 is satisfied

Double The resulting value is used as the new A value when processing the next input bit. here

Each time it doubles, one bit is output. E.g,

Is 128≤

If <256 is satisfied, one bit is outputted.

Is 62≤

If <128 is satisfied, two bits are output. In other words,

Knowing how many bits will be generated. here,

Function b that maps values to the number of bits generated

).

의 분포를 구할 수 있고, 각각의

값이 발생할 확률에 b(

)를 곱한 것을 전부 더하면 입력 비트 α 및 확률상태 지시자 σ가 정해졌을 때 확률적으로 생성되는 비트율을 구할 수 있다. 상기 과정을 수학식으로 나타내면 아래의 [수학식 1]과 같다.Since CABAC internal variable A has a value of 256 or more and 510 or less, model A as a random variable having an integer value of 256 or more and 510 or less. If the probability distribution of A is defined, the input bit α and the probability state indicator σ are determined.

We can find the distribution of, and each

The probability of a value being b (

By multiplying by), we can find the bit rate that is generated stoch- gerously when the input bit α and the probability state indicator σ are determined. When the process is represented by an equation, it is the same as Equation 1 below.

[Equation 1]

(k)|_σ,α는 주어진 σ 및 α에 대해 상기

가 특정 값 k가 될 확률을 나타낸다.

(k)|_σ,α는 A의 확률분포 p_A로부터 연역적으로 도출될 수 있으며 그 과정은 아래의 [수학식 2]와 같다.Σ denotes the probability state indicator, α denotes a bit included in the bit string, B (σ, α) denotes the estimated bit rate randomly generated for a given σ and α,

(k) | _{σ, α} is for the given σ and α

Represents the probability of becoming a specific value k.

(k) | _{σ, α} can be deduced from the probability distribution p _A of _A and the process is shown in Equation 2 below.

[Equation 2]

|_σ,α는 상기 A가 a인 경우에 주어진 σ 및 α에 대해 구간세분화한 결과를 나타낸다.Where δ represents a function having 1 when the two factors are equal to each other and 0 when the two factors are different from each other,

| _{σ, α} represents the result of section segmentation for σ and α given when A is a.

The probability distribution p _A of _A can be defined in various ways. The probability distribution of an exemplary A p _A may be defined by [Equation 3] to [formula 5] below.

&Quot; (3) &quot;

&Quot; (4) &quot;

[Equation 5]

|_{s, LPS}는 A=a이고 확률상태 지시자가 특정값 s인 경우에 LPS가 입력된 이후의 구간세분화 결과를 나타내고,

|_{s, MPS}는 A=a이고 확률상태 지시자가 특정값 s인 경우에 MPS가 입력된 이후의 구간세분화 결과를 나타내고, N(

)는

가 재정규화를 거친 후의 결과를 나타내고, q는 0초과 1미만의 실수를 나타낸다. 여기서, p_σ(s)는 기하분포(geometric distribution)를 따른다고 가정하였다. 상기 q값을 조절함에 따라 A의 확률분포 p_A를 조절할 수 있다. 도 2의 테이블은 q=0.07의 값을 이용하여 작성된 것이다.Here, the p _A ⁽ⁿ⁾ (k) is the probability the A to n when the second bit processing including the bit column indicates the probability that a particular value k, p _σ (s) is the probability state indicator to be a specific value s P _LPS (s) represents the probability that the LPS is input when the probability state indicator is a specific value s,

| _{s, LPS} represents the result of interval segmentation after LPS is input when A = a and the probability state indicator is a specific value s,

| _{s and MPS} represent the result of interval segmentation after MPS is input when A = a and the probability state indicator is a specific value s.

)

Represents the result after renormalization, and q represents a real number greater than 0 and less than 1. Here, it is assumed that p _σ (s) follows the geometric distribution. The probability distribution p _A of _A may be adjusted by adjusting the q value. The table in FIG. 2 is created using a value of q = 0.07.

Through the above-described process, a bit rate estimation table can be obtained. Meanwhile, in order to increase the accuracy of the bit rate estimation, the bit rate estimation table may be updated in real time. For example, the bit rate estimation table may be changed according to the quantization parameter (QP) value of the coding unit.

If a bit rate estimation table is provided, the bit rate can be estimated by reading the bit rate estimation table without directly encoding the bit string by a CABAC algorithm to determine the bit rate. Specifically, the bit rate output of the bit rate estimating apparatus 100 may be generated by Equation 6 below.

&Quot; (6) &quot;

Here, x represents specific coding unit data, R _x represents an estimated bit rate of the specific coding unit data, α _k represents a k-th bit included in the bit string, and σ _k represents a k-th bit included in the bit string. Represents the value of the probability state indicator loaded at the time of processing, and B (σ _k , α _k ) represents the estimated bit rate read from the bit rate estimation table for a given σ _k and α _k .

3 is a flowchart illustrating a bit rate estimation method according to a first embodiment of the present invention.

Referring to FIG. 3, the bit rate estimation method includes a binarization step S310, a context loading step S320, and a table reading step S330.

The binarization step S310 receives an input signal and performs binarization to output a bit string that is a sequence of bits having a value of 0 or 1. The context loading step (S320) receives the bit string to determine the context used by the bit string and provides context data related to the context. The table reading step S330 receives the bit string and the context data and provides a bit rate output using a bit rate estimation table.

The context data includes an MPS and a probability state indicator. Since the MPS and the probability state indicator have been described above with reference to FIGS. 1 and 2, a detailed description thereof will be omitted. The bit string estimation table is the same as the bit string estimation table described with reference to FIGS. 1 and 2, and a method of obtaining the bit string estimation table has been described above with reference to FIGS. 1 and 2, and thus a detailed description thereof will be omitted.

4 is a flowchart showing the table reading step S330 in detail.

The table reading step S330 may include determining whether a bit included in the bit string coincides with the MPS (S331), and when the bit included in the bit string coincides with the MPS, the probability state from the bit rate estimation table. The estimated bit rate for the case where the bit included in the bit string matches the MPS is read with respect to the current value of the indicator, and the probability is obtained from the bit rate estimation table if the bit included in the bit string does not match the MPS. Reading an estimated bit rate for a case where a bit included in the bit string does not match the MPS with respect to a current value of a status indicator (S333) and the read estimated bit rates for each bit included in the bit string; Summing to provide the bit rate output (S335). The table reading step S330 may be performed by Equation 6 described above, and since Equation 6 has been described above with reference to FIGS. 1 and 2, a detailed description thereof will be omitted.

As described above, according to the first embodiment, the final bit rate is obtained by summing values obtained by reading the estimated bit rates for each bit of the coding unit from a previously provided bit rate estimation table without actually performing encoding on the coding units. Since it can be estimated, the complexity of the encoder can be reduced.

Second embodiment

5 is a flowchart illustrating a bit rate estimation method according to a second embodiment of the present invention.

Referring to FIG. 5, the bit rate estimation method includes a binarization step S510, a context grouping step S520, and a bit rate determining step S530.

The binarization step S510 receives an input signal and performs binarization to output a bit string that is a sequence of bits having a value of 0 or 1. In the context grouping step (S520), the bit string is received to group the data using the same context to generate a context group. The bit rate determining step S530 determines an estimated bit rate for each context group using first and second bit rate estimation tables, and adds an estimated bit rate for each context group to provide a bit rate output.

The CABAC standard algorithm divides the data to be encoded into several parts by various conditions, and then encodes highly correlated data using the same context. As described above, in this embodiment, the context group is generated by grouping data using the same context, the estimated bit rate is determined for each context group using the same context, and the estimated bit rate for each group is summed to obtain a bit rate output. to provide.

Although the value of the probability state indicator used for encoding each bit included in the bit string is updated each time a bit is input, the same context is used as described above, although the input bit string must be sequentially processed. As described in detail below, the reason for determining the estimated bit rate for each context group is that the present embodiment models the range, which is an internal variable of the CABAC algorithm, as a random variable, as in the first embodiment.

6 is a flowchart showing the bit rate determination step S530 in detail.

Bit rate determination step (S530) is a context loading step (S531) for loading a probability state indicator (σ) indicating the probability of occurrence of MPS and MPS representing a symbol having a higher probability of occurrence between 0 and 1 for each of the context groups Symbol separation step (S533) for determining a first number, which is the number of MPSs and a second number, of LPSs, for each of the context groups, and the first bit rate estimation for each of the context groups. A table reading step (S535) of determining a first estimated bit rate for MPS using a table and a second estimated bit rate for LPS using the second bit rate estimation table, the first estimation for each of the context groups A bit rate determination step (S537) for determining the estimated bit rate for each context group by adding a bit rate and the second estimated bit rate, and the estimated bit rate for each context group The acid includes a bit rate summing step (S539) for providing the output bit rate.

The probability state indicator has an integer value of 0 or more and 62 or less as defined by the CABAC algorithm. As the probability state indicator is close to 0, the probability of occurrence of MPS is close to 0.5, and the probability of MPS occurrence is closer to 62. Means close to 1.0.

Hereinafter, a method of determining the first estimated bit rate corresponding to the estimated bit rate for the MPS for each context group will be described in detail.

When a bit string consisting only of MPS comes in, the bits input by the CABAC algorithm are always processed as MPS, and the value of the probability status indicator is increased by 1 for each bit processed. No longer increasing, staying at 62). Using the above feature, the value σ _M of one representative probability state indicator is determined for a bit string consisting only of MPS, the estimated bit rate for σ _M is read from the first bit rate estimation table, and the number of MPSs is added to this value. By multiplying, it is possible to estimate the total bit rate of a bit string consisting only of MPS. Here, the first bit rate estimation table is the same as the bit rate estimation table described in the first embodiment. Since the configuration of the bit rate estimation table and the method of obtaining the bit rate estimation table have been described above in connection with the first embodiment, detailed descriptions of the configuration of the first bit rate estimation table and the method of obtaining the first bit rate estimation table will be omitted. When the above process is expressed as an equation, Equation 7 and Equation 8 below.

[Equation 7]

[Equation 8]

Here, B (S _MPS ) represents the first estimated bit rate, and B (σ _M , MPS) represents an estimated bit rate read from the first bit rate estimation table for the case where the probability state indicator is σ _M and MPS is input. N _MPS denotes the first number, σ _old denotes the value of the probability state indicator before the first estimated bit rate calculation, and [] denotes a Gaussian symbol.

Through the above process, it is possible to determine the first estimated bit rate by reading the first bit rate estimation table without directly encoding a bit string composed only of the MPS by the CABAC algorithm to determine the first estimated bit rate for the MPS.

Hereinafter, a method of determining the second estimated bit rate corresponding to the estimated bit rate for the LPS for each context group will be described in detail.

As described above, in the case where the MPS-only bit string is input, the input bit is always treated as the MPS and the probability state indicator simply needs to be incremented by 1. However, when the LPS-only bit string is input, the probability state indicator is It is not simply decremented by 1, but every bit processing, it is necessary to find out what the next probability status indicator is by referring to the table provided inside CABAC, and when LPS is input even after the probability status indicator becomes 0, the input bit thereafter. Since the is processed in MPS, the method of determining the second bit rate is different from the method of determining the first bit rate.

Figure 7 shows the change in the probability state indicator value when LPS input according to the CABAC standard algorithm. Referring to FIG. 7, values between 0 and 62 of a probability state indicator are represented by two values, group number (G _num (σ)) and group offset (G _off (σ)). In Fig. 7, numbers 0 to 13 at the top represent group numbers, and numbers in subscripts of respective probability state indicators (σ) indicate group offsets. The group number refers to the number of consecutive LPS needed to zero a particular probability state indicator. When the group number and the group offset are defined as described above, the change in the probability state indicator value when n _LPS (meaning the second number) LPSs are continuously input can be estimated by Equation 9 below. have.

[Equation 9]

Here, sigma _LPS represents a changed probability state indicator when n _LPS LPSs are continuously input, and G- ¹ represents a function that maps a group number and a group offset to a probability state indicator. 8 illustrates a change in the probability state indicator value when LPS is input according to Equation 9 above. On the other hand, if the LPS is input again when the probability state indicator is 0, the MPS value is changed, so all unprocessed bits are treated as MPS. In this case, that is, when G _num (σ) <n _LPS is established. The probability state indicator changed after the processing of all the bits is terminated by Equation 10 below.

[Equation 10]

In the case of the probability state indicator having the same group number, assuming that the LPS is the same estimated bit rate, the estimated bit rate for each probability state indicator group may be determined by Equation 11 below.

[Equation 11]

)는 확률상태 지시자 그룹

의 추정 비트율을 나타내고, B(s,LPS)는 확률상태 지시자가 s이고 LPS가 입력되었을 경우에 대해 상기 제 1 비트율 추정 테이블로부터 판독한 추정 비트율을 나타내고, p_σ(s)는 CABAC 부호화 과정 중에 확률상태 지시자가 특정 값 s를 가질 확률을 나타낸다. p_σ(s)는 아래의 [수학식 12]로 정의될 수 있다.Where B _G (

) Is the probability status indicator group

A represents the estimated bit-rate, B (s, LPS) are the show the estimated bit rate a probability state indicator is s and the LPS is read out from the estimated table of the first bit rate for the case is entered, p _σ (s) is CABAC encoding process The probability state indicator indicates the probability of having a specific value s. p _σ (s) may be defined by Equation 12 below.

[Equation 12]

Where q represents a real number greater than 0 and less than 1, and p _σ (s) is assumed to follow a geometric distribution. By adjusting the q value, the estimated bit rate for each probability state indicator group may be adjusted.

The second bit rate estimation table stores the estimated bit rate for each probability state indicator group. In more detail, the second bit rate estimation table classifies values of probability state indicators of 0 to 62 into 14 probability state indicator groups having group numbers of 0 to 13, for each probability state indicator group. The estimated bit rate for the case where the bits included in the bit string coincides with the LPS is stored. In addition, as described above, the probability state indicator groups indicate that the CABAC algorithm groups probability state indicators having the same number of successive LPSs required to make the probability state indicator zero for each value of the probability state indicator. The status indicator group number indicates the number of the contiguous LPS needed to zero the probability status indicator.

는 확률상태 지시자 그룹번호를 나타내고, B_G(

)는 확률상태 지시자 그룹

의 추정 비트율을 나타내고, W_G(

)는 1.00을 8에 대응시키고 0.50을 4에 대응시키는 것과 같이 B_G(

)를 양자화하여 정수값의 가중치로 표현한 것을 나타낸다. W_G(

)는 하드웨어 구현을 용이하게 하기 위해 추정 비트율을 정수화한 것이고, 추후에 상기 정수화된 추정 비트율을 오른쪽으로 3비트 쉬프트(shift) 하여 추정 비트율을 생성할 수 있다. 도 9에서는 1.00이 8에 대응되도록 양자화 하였으나, 다른 방식의 양자화를 적용할 수 도 있다. 예를 들면, 1.00이 16에 대응되도록 양자화 할 수도 있으며, 이 경우에는 정수화된 추정 비트율을 오른쪽으로 4비트 쉬프트(shift) 하여 추정 비트율을 생성할 수 있다.9 shows an example of the second bit rate estimation table, which is created using a value of q = 0.08. In Figure 9

Indicates the probability status indicator group number, and B _G (

) Is the probability status indicator group

Denotes the estimated bit rate of W _G (

) Is equivalent to B _G (for mapping 1.00 to 8 and 0.50 to 4).

) Is expressed by the weight of an integer value. W _G (

) Is an integer of the estimated bit rate to facilitate hardware implementation, and the estimated bit rate may be later generated by shifting the integerized estimated bit rate to the right by three bits. In FIG. 9, 1.00 is quantized to correspond to 8. However, another quantization method may be applied. For example, 1.00 may be quantized to correspond to 16. In this case, the estimated bit rate may be shifted to the right by 4 bits to generate the estimated bit rate.

Through the above-described process, a second bit rate estimation table can be obtained. When the second bit rate estimation table is provided, the second estimated bit rate is obtained by reading the second bit rate estimation table without directly encoding a bit string consisting of only lPS by the CABAC standard algorithm to determine the second estimated bit rate for the LPS. You can decide. Specifically, by determining one representative group number G _M for a bit string composed only of LPS, and reading the estimated bit rate for G _M from the second bit rate estimation table and multiplying this value by the number of LPS, the total number of bit strings composed only of LPS The bit rate can be estimated. When the above process is expressed as an equation, Equation 13 to Equation 15 below are given.

[Equation 13]

[Equation 14]

[Equation 15]

_, if G_num(σ_MPS) ≥ n_LPS

_, if G _num (σ _MPS ) ≥ n _LPS

_, if G_num(σ_MPS) < n_LPS

_, if G _num (σ _MPS ) <n _LPS

Σ _MPS denotes the value of the probability state indicator after the first estimated bit rate calculation, σ _old denotes the value of the probability state indicator before the first estimated bit rate calculation, n _MPS denotes the first number, n _LPS represents the second number, B (S _LPS ) represents the second estimated bit rate, and B _G (G _M ) represents the second bit rate for a case where the group number of the probability state indicator group is G _M The estimated bit rate read from the estimation table is shown. As shown in Equation 15, when G _num (σ _MPS ) < n _LPS is established to process the MPS, the bit rate corresponding to each MPS is assumed to be 1. That is, it is assumed that the number of bits to be processed by the MPS is equal to the number of additionally output bits.

When the first estimated bit rate for the MPS and the second estimated bit rate for the LPS are determined through the above-described process, the estimated bit rate for each context group may be determined by summing the first estimated bit rate and the second estimated bit rate. Equation 16 used to determine the estimated bit rate for each context group is expressed by Equation 16 below.

[Equation 16]

Here, x denotes specific coding unit data, R _{x, j} denotes an estimated bit rate of the context group for the j-th context group in coding unit data x, and B _{x, j} (S _MPS ) denotes in coding unit data x The first estimated bit rate for the j-th context group is represented, and B _{x, j} (S _LPS ) represents the second estimated bit rate for the j-th context group in coding unit data x.

After the estimated bit rate for each context group is determined, the bit rate output can be generated by summing the estimated bit rates for each context group for all context groups. The equation used to determine the bit rate output is as shown in Equation 17 below.

[Equation 17]

Where R _x denotes the bit rate output for coding unit data x, m _x denotes the number of contexts used by coding unit data x, and R _{x, j} denotes the jth context group for coding unit data x. Indicates an estimated bit rate for each context group.

10 is a block diagram showing a bit rate estimation apparatus according to a second embodiment of the present invention.

Referring to FIG. 10, a bit rate estimating apparatus 700 includes a binarizer 710, a context grouping unit 720, a bit calculator 730, and a switch 740. It includes a context loader 750, a plurality of group rate estimator 760, and an accumulator 770.

The binarization unit 710 receives an input signal and performs binarization to output a bit stream that is a sequence of bits having a value of 0 or 1.

In the CABAC standard algorithm, input data may be encoded in bypass mode or regular mode according to the logic level of the bypass signal. Therefore, the present embodiment includes a switch 740 for selectively connecting the output of the binarizer 710 to the context grouper 720 or the bit number calculator 730 according to the logic level of the bypass signal. .

The context grouping unit 720 receives the bit string and groups data using the same context to generate a context group. The bit number calculator 730 receives the bit string and provides the number of bits belonging to the bit string.

The context loading unit 750 receives the context group, grasps the context used by the context group, and indicates a probability state indicator indicating an occurrence of MPS and MPS representing a symbol having a higher probability of occurrence among 0 and 1 for the context. gives (σ). A plurality of group bit rate estimator 760 receives the MPS and the probability state indicator for the context group and the context group, respectively, and determines the estimated bit rate for each context group. The accumulator 770 accumulates and sums outputs of the plurality of group bit rate estimation units and outputs of the bit number calculator to provide a bit rate output.

Since the bit rate estimator 700 has the above structure, the bit rate estimator 700 can perform the bit rate estimation for each context group in parallel using the plurality of group bit rate estimator 760, thereby increasing the bit rate estimation speed. You can.

FIG. 10 illustrates a case in which two bit rate estimators 760 are used for each group. However, those skilled in the art may use bit rate estimator 760 for three or more groups. It is also possible to implement 700.

As described above, according to the second embodiment, the bit rate can be estimated by summing values obtained by reading the estimated bit rates from the first and second bit rate estimation tables provided in advance without actually performing encoding on the coding units. In addition, the encoder can not only reduce the complexity of the encoder but also increase the speed of the encoder because it can perform bit rate estimation in parallel for each context group by context grouping the data using the same context without sequentially processing the bit strings. have.

The bit rate estimation method and apparatus according to the embodiment of the present invention can reduce the complexity of the encoder and perform the bit rate estimation quickly, and thus can be efficiently used for the high performance encoder based on CABAC.

As described above, the present invention has been described with reference to a preferred embodiment of the present invention, but those skilled in the art may vary the present invention without departing from the spirit and scope of the present invention as set forth in the claims below. It will be understood that modifications and changes can be made.

1 is a block diagram illustrating a bit rate estimation apparatus according to a first embodiment of the present invention.

2 shows an example of a bit rate estimation table.

3 is a flowchart illustrating a bit rate estimation method according to a first embodiment of the present invention.

FIG. 4 is a flowchart showing the table reading step of FIG. 3 in detail.

5 is a flowchart illustrating a bit rate estimation method according to a second embodiment of the present invention.

6 is a flowchart illustrating the bit rate determination step of FIG. 5 in detail.

Figure 7 shows the change in the probability state indicator value when LPS input according to the CABAC standard algorithm.

8 illustrates a change in the probability state indicator value during LPS input according to the second embodiment.

9 shows an example of a second bit rate estimation table.

10 is a block diagram showing a bit rate estimation apparatus according to a second embodiment of the present invention.

Claims (28)

A binarizer configured to receive an input signal, perform binarization, and provide a bit stream; Receive the bit strings to determine the context used by the bit strings; Most Probable Symbol (MPS) indicating a symbol having a higher probability of occurrence between 0 and 1 in the data using the context; A context loader for providing context data including a probability state indicator indicating a probability of MPS occurrence; And Occurs when the number of probable bits and the lowest probability symbol (LPS) are input when the MPS is input for each value of the probability state indicator by receiving the bit string and the context data. Providing a bit rate output by reading the estimated bit rate for each bit included in the bit string from a bit-rate estimation table that stores an estimated bit rate representing the number of probabilistic bits and summing the read estimated bit rates. A bit rate estimating apparatus comprising a table reading section. The method of claim 1, And a memory unit which stores the context data and the bit rate estimation table. delete The apparatus of claim 1, wherein the probability state indicator has an integer value of 0 to 62, and the probability of occurrence of the MPS increases as the value of the probability state indicator increases. The method of claim 4, wherein the bit rate estimation table, For each value of the probability state indicator, an estimated bit rate for each of the cases where the bits included in the bit string coincide with the MPS and the bits included in the bit string do not coincide with the MPS. Bit rate estimation device. The method of claim 5, wherein the estimated bit rate, Bit rate estimation, characterized in that it is determined probabilistically for each value of the probability state indicator by modeling a range that is an internal variable of CABAC (Context-Adaptive Binary Arithmetic Coding) algorithm as a random variable having an integer value of 256 or more and 510 or less Device. The method of claim 6, wherein the estimated bit rate, A bit rate estimating apparatus, which is generated using Equation 1 below (where σ represents the probability state indicator, α represents each bit included in the bit string, and B (σ, α) Denotes the estimated bit rate produced probabilistic for a given σ and α,

Represents a changed range after interval subdivision when the range is A,

(k) | _{σ, α} is for the given σ and α

Denotes the probability of being a specific value k, and b (k) is

Is a specific value k, it represents the number of bits generated through the renormalization process). [Equation 1]

8. The method of claim 7, wherein

(k) | _{σ, α} is generated by Equation 2 below, where p _A represents the probability distribution of the range, δ is 1 when the two factors are equal to each other, and two factors If is different, it represents a function with 0,

| _{σ, α} represents the result of interval segmentation for σ and α given when the range is a). [Equation 2]

9. The apparatus of claim 8, wherein the bit rate output is generated by Equation 3 below (where x represents specific coding unit data and R _x is estimation of the specific coding unit data). Represents a bit rate, α _k represents a k-th bit included in the bit string, σ _k represents a value of the probability state indicator loaded during k-th bit processing included in the bit string, and B (σ _k , α _k ) Represents the estimated bit rate read from the bit rate estimation table for a given σ _k and α _k ). &Quot; (3) &quot;

A binarization step of receiving an input signal to perform binarization and providing a bit stream; Receive the bit strings to determine the context used by the bit strings; Most Probable Symbol (MPS) indicating a symbol having a higher probability of occurrence between 0 and 1 in the data using the context; A context loading step of providing context data comprising a probability state indicator indicative of the probability that an MPS will occur; And Occurs when the number of probable bits and the lowest probability symbol (LPS) are input when the MPS is input for each value of the probability state indicator by receiving the bit string and the context data. Providing a bit rate output by reading the estimated bit rate for each bit included in the bit string from a bit-rate estimation table that stores an estimated bit rate representing the number of probabilistic bits and summing the read estimated bit rates. A bit rate estimation method comprising a table reading step. delete The method of claim 10, wherein the bit rate estimation table, For each value of the probability state indicator, an estimated bit rate for each of the cases where the bits included in the bit string coincide with the MPS and the bits included in the bit string do not coincide with the MPS. Bit rate estimation method. The method of claim 12, wherein the table reading step, Determining whether a bit included in the bit string matches the MPS; If the bit included in the bit string matches the MPS, the estimated bit rate for the case where the bit included in the bit string matches the MPS with respect to the current value of the probability state indicator is read from the bit rate estimation table. If the bit included in the bit string does not match the MPS, the estimated bit rate for the case where the bit included in the bit string does not match the MPS is read with respect to a current value of the probability state indicator from the bit rate estimation table. Making; And And summing the read estimated bit rates for each bit included in the bit string to provide a bit rate output. A binarization step of receiving an input signal to perform binarization and providing a bit stream; A context grouping step of generating the context group by receiving the bit string and grouping data using the same context; Loading a maximum probability symbol (MPS) representing a symbol having a higher probability of occurrence among 0 and 1 and a probability state indicator indicating a probability of occurrence of the MPS for each of the context groups; For each of the context groups, the MPS and the least probability symbol (LPS) included in the context group are separated to determine a first number, the number of MPSs, and a second number, the number of LPSs. Separation step; For each of the context groups, for each value of the probability state indicator, store an estimated bit rate indicating the number of probable bits that will occur when the MPS is input and the number of probable bits that will occur when the LPS is input. Determine a first estimated bit rate for the MPS based on the estimated bit rate and the first number read from the first bit-rate estimation table, group each of the values of the probability state indicator, and determine the probability state. A second estimated bit rate for the LPS based on the second number and an estimated bit rate read from a second bit rate estimation table that stores an estimated bit rate for each bit included in the bit string corresponding to the LPS for each indicator group; A table reading step of determining; A bit rate determining step for each group determining the estimated bit rate for each context group by summing up the first estimated bit rate and the second estimated bit rate for each of the context groups; And And summing the estimated bit rates for each context group to provide a bit rate output. delete 15. The method of claim 14, wherein the probability state indicator has an integer value between 0 and 62, and the probability of occurrence of the MPS increases as the value of the probability state indicator increases. The method of claim 16, wherein the first bit rate estimation table, For each value of the probability state indicator, an estimated bit rate for each of the cases where the bits included in the bit string coincide with the MPS and the bits included in the bit string do not coincide with the MPS. Bit rate estimation method. The method of claim 17, wherein the estimated bit rate stored in the first bit rate estimation table, Bit rate estimation, which is determined probabilistically for each value of the probability state indicator by modeling an internal variable range of the CABAC algorithm as a random variable having an integer value of 256 or more and 510 or less. Chung way. The method according to claim 18, wherein the estimated bit rate stored in the first bit rate estimation table, A bit rate estimation method characterized by using Equation 4 below (where σ represents the probability state indicator, α represents each bit included in the bit string, and B (σ, α) Denotes the estimated bit rate produced probabilistic for a given σ and α,

Represents a changed range after interval subdivision when the range is A,

(k) | _{σ, α} is for the given σ and α

Denotes the probability of being a specific value k, and b (k) is

Is a specific value k, it represents the number of bits generated through the renormalization process). &Quot; (4) &quot;

20. The method of claim 19, wherein

(k) | _{σ, α} is a bit rate estimation method characterized by the following Equation 5 (where p _A represents the probability distribution of the range, δ is 1 when the two factors are equal to each other and two factors If is different, it represents a function with 0,

| _{σ, α} represents the result of interval segmentation for σ and α given when the range is a). [Equation 5]

21. The method of claim 20, wherein the first estimated bit rate of each context group is determined by Equation 6 and Equation 7 below, wherein B (S _MPS ) is B (σ _M , MPS) denotes the first estimated bit rate, and B (σ _M , MPS) denotes an estimated bit rate read from the first bit rate estimation table for the case where the probability state indicator is σ _M and MPS is input, and n _MPS denotes the first 1 represents a number, σ _old represents a value of a probability state indicator before the first estimated bit rate calculation, and [] represents a Gaussian symbol). &Quot; (6) &quot;

[Equation 7]

The method of claim 21, wherein the second bit rate estimation table, Each value of the probability state indicator of 0 or more and 62 or less is classified into 14 groups of probability state indicators having group numbers of 0 to 13, and the bits included in the bit string for each probability state indicator group match the LPS. And a bit rate estimation method for storing the estimated bit rate. The method of claim 22, wherein the probability state indicator groups, For each value of the probability state indicator in a context-adaptive binary arithmetic coding (CABAC) algorithm, the number of consecutive LPS needed to make the probability state indicator 0 is grouped with the same probability state indicators, The probability state indicator group number is, And indicating the number of the contiguous LPS needed to zero the probability state indicator. The method of claim 23, wherein the estimated bit rate stored in the second bit rate estimation table, Bit rate estimation method, characterized in that generated using [Equation 8] below, where B _G (

) Group

B (s, LPS) represents an estimated bit rate of the estimated bit rate read from the first bit rate estimation table for the case where the probability state indicator is s and LPS is input, and p _σ (s) represents the CABAC encoding process. The probability state indicator has a specific value s). [Equation 8]

25. The method of claim 24, wherein p _σ (s) is generated by Equation (9) below, wherein q is a real number greater than zero and less than one. [Equation 9]

26. The method of claim 25, wherein the second estimated bit rate of each of the context groups is determined by Equation 10, Equation 11, and Equation 12 below. σ _MPS denotes the value of the probability state indicator after the first estimated bit rate calculation, σ _old denotes the value of the probability state indicator before the first estimated bit rate calculation, n _MPS denotes the first number, n _LPS represents the second number, G _num (σ) represents the probability state indicator group number to which the probability state indicator σ belongs, B (S _LPS ) represents the second estimated bit rate, and B _G (G _M ) Represents an estimated bit rate read from the second bit rate estimation table for the case where the group number of the probability state indicator group is G _M ). [Equation 10]

[Equation 11]

[Equation 12]

_, if G _num (σ _MPS ) ≥ n _LPS

_, if G _num (σ _MPS ) <n _LPS 27. The method of claim 26, wherein the estimated bit rate for each context group is generated by Equation 13 below, and the bit rate output is generated by Equation 14 below. x represents specific coding unit data, R _{x, j} represents the estimated bit rate for each context group for the j-th context group in coding unit data x, and B _{x, j} (S _MPS ) represents the j in coding unit data x The first estimated bit rate for the first context group, B _{x, j} (S _LPS ) represents the second estimated bit rate for the jth context group in coding unit data x, and R _x is used for coding unit data x For the bit rate output, m _x denotes the number of contexts used by the coding unit data x). [Equation 13]

[Equation 14]

A binarizer configured to receive an input signal, perform binarization, and provide a bit stream; A context grouper for generating a context group by receiving the bit string and grouping data using the same context; A bit number calculator configured to receive the bit string and provide the number of bits belonging to the bit string; A switch for selectively coupling the output of the binarizer to the context grouper or the bit number calculator according to a logic level of a bypass signal; Receives the context group to identify the context used by the context group, and indicates a maximum probability symbol (MPS) representing a symbol having a higher probability of occurrence among 0 and 1 and a probability of generating the MPS for the context. A context loading unit providing a probability state indicator; A plurality of group bit rate estimators for determining the estimated bit rate for each group for the context group by receiving the MPS and the probability state indicator for the context group and the context group, respectively; And And a cumulative unit for accumulating and summing outputs of the plurality of group-specific bit rate estimation units and outputs of the bit number calculator to provide a bit rate output.

Images