KR101761131B1 - Apparatus and method for estimating distortion of video encoding device with high speed - Google Patents

Abstract

According to an embodiment of the present invention, a device for estimating a restoration error of a video encoding device at a high speed comprises: a transform/quantization unit for outputting a conversion coefficient (Vs) obtained by performing discrete cosine transform (DCT) and scaling with respect to an inputted prediction error image (U) and a quantization coefficient (Vg) obtained by performing inverse scaling and quantization with respect to the conversion coefficient (Vs); and a new distortion calculator unit for calculating an estimated restoration error (D_(mode)) in a transform region on the basis of the conversion coefficient (Vs) and the quantization coefficient (Vg), and transferring the estimated restoration error (D_(mode)) to a rate distortion optimization (RDO) mode decision unit.

Description

[0001] APPARATUS AND METHOD FOR ESTIMATING DISTORTION OF VIDEO ENCODING DEVICE WITH HIGH SPEED [0002]

The present invention relates to an apparatus and a method for estimating a fast reconstruction error of a high efficiency video coding (HEVC) video encoder.

Recently, with the development of DTV, mobile devices, various online video streaming services, and the like, the accessibility of multimedia contents has become easier and the demand for high quality video is increasing.

In particular, there is an increasing demand for 4K to 8K class ultra high resolution video. Also, portable display devices are required to reproduce full HD images, and a refresh rate of 60 frames or more, which is twice as large as that of the conventional 30 frames, is required.

Such ultra high resolution and high frame video content requires as much as ten times as much data throughput as conventional video content. In order to process these video contents more efficiently, a new video standard with higher compression efficiency than the existing H.264 / AVC has become necessary. Accordingly, Jointly collaborative team on video coding (JCTVC) Standardization of HEVC (High Efficiency Video Coding) was completed and the standardization of HEVC Main Profile was completed in January 2013.

On the other hand, the HEVC compression standard adopts various coding techniques in order to increase the coding efficiency. In particular, the previous standard, AVC, uses prediction modes for various block sizes within a maximum of 16x16 macroroblocks, while HEVC supports prediction modes for various block sizes within a maximum of 64x64 Coding Tree Block (CTU).

Among the prediction modes for the various block sizes, the complexity of the HEVC encoder is increased by about 300% in comparison with the AVC in order to determine the optimum mode combination in terms of compression efficiency. In order to commercialize the real-time UHD video service, the HEVC encoder To optimize the complexity of the system.

It is an object of the present invention to provide a fast reconstruction error estimation apparatus for a video encoder including a New Distortion calculator capable of calculating an estimated reconstruction error without inputting an original image and a reconstructed image.

It is another object of the present invention to provide an apparatus and method for estimating a fast reconstruction error of a video encoder that shortens the execution time required for encoding by omitting the image reconstruction process for the N modes which has been required in the past.

It is another object of the present invention to provide an apparatus and method for estimating a fast reconstruction error of a video encoder that can save memory access and power consumption, which are required for image reconstruction in N modes.

)를 계산하고, 상기 추정 복원 오차(

)를 RDO(rate distortion optimization) Mode Decision 부에 전달하는 New Distortion Calculator부를 포함할 수 있다.According to an aspect of the present invention, there is provided an apparatus for estimating a fast reconstruction error of a video encoder, the apparatus comprising: a discrete cosine transform (DCT) and a scaling a transform coefficient Vs for performing a scaling operation and a quantization coefficient Vq for performing inverse scaling and quantization on the transform coefficient Vs, And an estimated reconstruction error (Vq) in the transform region based on the quantization coefficient (Vq)

), And the estimated restoration error (

) To a rate distortion optimization (RDO) mode decision unit.

)를 계산할 수 있다.In an exemplary embodiment, the New Distortion Calculator may use a constant value defined in HEVC (High Efficiency Video Coding) for quantization and inverse quantization of integer operations, The estimated restoration error ("

) Can be calculated.

) 및 비트율(

)에 기반하여, 최소의 율-왜곡 비용을 가지는 율-왜곡 최적화(RDO)된 모드를 결정하는 상기 RDO Mode Decision부를 더 포함할 수 있다In an embodiment, the estimated restoration error (

) And bit rate (

), The RDO Mode Decision section may determine a rate-distortion optimized (RDO) mode having a minimum rate-distortion cost

In an embodiment, the apparatus may further include a reconstruction unit that performs image reconstruction only once per coding tree block (CTU) only for the determined mode.

In an embodiment, transform, quantization, entropy coding, and image reconstruction of the prediction error image U are performed only for the determined mode, Can be generated.

In an exemplary embodiment, the final reconstructed image may be stored in a decoded picture buffer unit through a deblock / SAO filter unit and used for predicting a next image.

)를 계산하는 단계를 포함할 수 있다.According to another aspect of the present invention, there is provided a method for estimating a fast reconstruction error of a video encoder, comprising the steps of: performing discrete cosine transform (DCT) and scaling generating a transform coefficient by performing inverse scaling and quantization on the transform coefficient Vs to generate a quantization coefficient Vq; and performing the inverse scaling and quantization on the transform coefficient Vs, Based on the quantization coefficient (Vq), an estimated reconstruction error ("

). ≪ / RTI >

) 및 비트율(

)에 기반하여, 최소의 율-왜곡 비용을 가지는 율-왜곡 최적화(RDO, rate distortion optimization)된 모드를 결정하는 단계, 상기 결정된 모드에 대해서만 영상 복원(Reconstruction)을 CTU(coding tree block)당 한번만 수행하여 최종 비트 스트림과 최종 복원 영상을 생성하는 단계, 및 상기 최종 복원 영상을 저장하여 다음 영상의 예측에 사용하는 단계를 더 포함할 수 있다.In an embodiment, the estimated restoration error (

) And bit rate (

Determining a rate distortion optimization (RDO) mode having a minimum rate-distortion cost based on the first and second prediction modes, and reconstructing the reconstructed mode for the determined mode only once per CTU (coding tree block) Generating a final bitstream and a final reconstructed image, and storing the final reconstructed image to be used for predicting a next image.

The effects of the fast reconstruction error estimation apparatus and method of the video encoder according to the present invention will be described as follows.

According to at least one of the embodiments of the present invention, a new distortion calculator unit that can calculate the estimated reconstruction error without inputting the original image and the reconstructed image may be included.

In addition, according to at least one of the embodiments of the present invention, it is possible to shorten the execution time required for encoding by omitting the image reconstruction process for the N modes, which has been conventionally required.

In addition, according to at least one embodiment of the present invention, memory access and power consumption, which are conventionally required for image reconstruction for N modes, can be saved.

1 is a diagram showing a general HEVC encoder.
FIG. 2 is a diagram specifically illustrating a reconstruction unit of FIG. 1. FIG.
FIG. 3 is a diagram specifically illustrating the distortion calculator of FIG. 1. FIG.
FIG. 4 is a diagram specifically illustrating the RDO mode decision unit of FIG.
5 is a block diagram illustrating a fast reconstruction error estimation apparatus of a video encoder according to an embodiment of the present invention.
6 is a diagram specifically illustrating a transform / quantization unit included in the fast reconstruction error estimation apparatus of a video encoder according to an embodiment of the present invention.
7 is a block diagram illustrating a New Distortion Calculator included in a fast reconstruction error estimation apparatus of a video encoder according to an embodiment of the present invention.

Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings, wherein like reference numerals are used to designate identical or similar elements, and redundant description thereof will be omitted. The suffix "module" and " part "for the components used in the following description are given or mixed in consideration of ease of specification, and do not have their own meaning or role. In the following description of the embodiments of the present invention, a detailed description of related arts will be omitted when it is determined that the gist of the embodiments disclosed herein may be blurred. It is to be understood that both the foregoing general description and the following detailed description are exemplary and explanatory and are intended to provide further explanation of the invention as claimed. , ≪ / RTI > equivalents, and alternatives.

Terms including ordinals, such as first, second, etc., may be used to describe various elements, but the elements are not limited to these terms. The terms are used only for the purpose of distinguishing one component from another.

It is to be understood that when an element is referred to as being "connected" or "connected" to another element, it may be directly connected or connected to the other element, . On the other hand, when an element is referred to as being "directly connected" or "directly connected" to another element, it should be understood that there are no other elements in between.

The singular expressions include plural expressions unless the context clearly dictates otherwise.

In the present application, the terms "comprises", "having", and the like are used to specify that a feature, a number, a step, an operation, an element, a component, But do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, or combinations thereof.

1 is a view showing a general HEVC encoding apparatus.

1, a general HEVC encoding apparatus includes a transform / quantization unit 110, a reconstruction unit 120, an entropy coding unit 130, a distortion calculator 140, a rate distortion optimization (RDO) A mode decision unit 150, an intra / inter prediction unit 160, a deblock / SAO filter unit 170, and a decoded picture buffer unit 180.

First, the transform / quantization unit 110 can transform and quantize the prediction error image U, which is a difference value between the original image S and the prediction image P from the intra / inter prediction unit 160, for encoding .

)을 생성하기 위해 Entropy Coding 부(130)에 전달될 수 있고, 이와 함께 복원 영상(

) 생성을 위해 Reconstruction 부(120)로 전달될 수 있다.Here, the quantized coefficients (Vq) are the bit rate

To the Entropy Coding unit 130 to generate the reconstructed image

To the reconstruction unit 120 in order to generate the reconstructed image.

)를 계산할 수 있다.The distortion calculator 140 receives the reconstructed image through the original image S and the reconstruction unit 120 and calculates a reconstruction error

) Can be calculated.

)와 Entropy Coding 부(130)에서 생성된 비트율(

)은 RDO Mode Decision 부(150)의 입력으로 전달될 수 있다.Here, the restoration error calculated by the distortion calculator 140

) And the bit rate generated by the entropy coding unit 130

May be passed to the input of the RDO Mode Decision Unit 150. [

)와 비트율(

)을 전달받아 최소의 율-왜곡 비용을 가지는 율-왜곡 최적화(RDO)된 모드(

)를 결정할 수 있다.The RDO mode decision unit 150 compares the reconstruction error (

) And the bit rate (

Rate-distortion optimized (RDO) mode with minimal rate-distortion cost

Can be determined.

)가 결정되면 최종적으로 결정된 모드에 대한 예측 오차의 변환/양자화, 엔트로피코딩, 영상 복원을 거쳐 최종 비트 스트림과 최종 복원 영상을 생성할 수 있다. 이렇게 생성된 최종 복원 영상은 Deblock/SAO Filters 부(170)를 거쳐 Decoded Picture Buffer부(180)에 저장되어 다음 영상의 예측에 사용될 수 있다.And, the rate-distortion optimized (RDO) mode

), The final bitstream and the final reconstructed image can be generated through transformation / quantization, entropy coding, and image reconstruction of the prediction error for the finally determined mode. The final reconstructed image thus generated is stored in the decoded picture buffer unit 180 through the Deblock / SAO Filters unit 170 and can be used for predicting the next image.

FIG. 2 is a diagram specifically illustrating a reconstruction unit of FIG. 1. FIG.

Referring to FIG. 2, the reconstruction unit 120 may include an inverse quantization unit 121 and an inverse transform unit 122.

)을 생성하게 된다.Specifically, the reconstruction unit 120 dequantizes the quantized coefficient Vq input from the transform / quantization unit 110 through the inverse quantization unit 121, inversely transforms the quantized coefficient Vq through the inverse transform unit 122, (SUM) from the inter prediction unit 160 to the reconstructed image (P)

).

FIG. 3 is a diagram specifically illustrating the distortion calculator of FIG. 1. FIG.

)을 입력받을 수 있다. 이 때 복원 영상(

)을 얻기 위해서는 도 2에서 설명한 바와 같이 Reconstruction 부(120)는 많은 연산 과정들을 수행하게 된다.Referring to FIG. 3, the distortion calculator 140 calculates an original image S and a reconstructed image S

Can be input. At this time,

2, the reconstruction unit 120 performs a number of calculation processes.

FIG. 4 is a diagram specifically illustrating the RDO mode decision unit of FIG.

)와 비트율(

)로부터 율-왜곡 비용(

)를 계산하고, 율-왜곡 비용(

)이 최소가 되는 모드(

)를 최종 예측 모드로 결정할 수 있다.Referring to FIG. 4, the RDO Mode Decision Unit 150 calculates a restoration error

) And the bit rate (

) From the rate-distortion cost (

), And the rate-distortion cost (

) Is the minimum mode

) As the final prediction mode.

On the other hand, the HEVC enables selection of a coding block (CB) ranging from a maximum size of 64x64 to a minimum size of 8x8, unlike the conventional moving picture compression standard method. Here, CB may be composed of different prediction blocks (PB) and transform blocks (TB).

)이 최소가 되는 모드(

)의 결정을 수행하여야 한다.In order to determine the optimum combination of the prediction modes of the various sizes from the viewpoint of the compression performance, the rate-distortion cost

) Is the minimum mode

) Shall be carried out.

)를 구하기 위해서는 각각의 코딩 블록 크기(64x64, 32x32, 16x16, 8x8)마다 12개의 예측 블록 모드(INTER_SKIP, INTER_MERGE, INTER_2Nx2N, INTER_2NxN, INTER_Nx2N, INTER_2NxnU, INTER_2NxnD, INTER_nLx2N, INTER_nRx2N, INTER_NXN, INTRA_2Nx2N, INTRA_NxN) X 변환 블록 깊이만큼의 영상 복원(Reconstruction) 연산량이 필요하게 된다. 예를 들어, 변환 블록 깊이가 3인 경우, 64x64 코딩 블록 크기의 모드 결정에 필요한 영상 복원(Reconstruction)의 수는 (12) x (3) = (36)과 같다.In particular,

12 prediction block modes (INTER_SKIP, INTER_MERGE, INTER_2Nx2N, INTER_2NxN, INTER_2NxNU, INTER_2NxnD, INTER_nLx2N, INTER_nRx2N, INTER_NXN, INTRA_2Nx2N, INTRA_NxN) of each coding block size (64x64, 32x32, 16x16, 8x8) A reconstruction operation amount corresponding to the depth of the conversion block is required. For example, when the transform block depth is 3, the number of image reconstructions required for mode determination of a 64x64 coding block size is equal to (12) x (3) = (36).

)이 최소가 되는 모드(

)의 결정시에 Reconstruction 부(120)에서 수행되는 연산을 생략하는 과정이 절실히 요구된다.Finally, to optimize the computational complexity of the HEVC encoder, the rate-distortion cost

) Is the minimum mode

A process of omitting the operation performed in the reconstruction unit 120 is desperately required.

)이 최소가 되는 모드(

)의 결정시에 원 영상(S)과 Reconstruction 부(120)에서 생성된 복원 영상(

)을 이용하지 않고, 추정 복원 오차(

)를 계산할 수 있도록 한다.The present invention addresses the need for rate-distortion costs

) Is the minimum mode

The original image S and the reconstructed image generated in the reconstruction unit 120

), The estimated restoration error (

) Can be calculated.

)의 입력 없이도 추정 복원 오차(

)를 계산할 수 있는 New Distortion calculator 부(도 5의 540)를 포함한다. 그 결과, 종래에 요구되었던 N개 모드에 대한 영상 복원(Reconstruction) 과정이 생략될 수 있고, 그에 따라 부호화에 필요한 수행 시간을 단축시킬 수 있게 된다.More specifically, the present invention relates to an image reconstruction method,

) Without input of the estimated restoration error (

(540 in Fig. 5) capable of calculating a new distortion calculator. As a result, the image reconstruction process for the N modes, which has been required in the past, can be omitted, thereby shortening the execution time required for encoding.

Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings. It will be apparent to those skilled in the art that the present invention may be embodied in other specific forms without departing from the spirit or essential characteristics thereof.

5 to 7, only the necessary components are shown in explaining the characteristic functions according to the present invention, and various other components are provided in the video encoder It is obvious to those skilled in the art that the present invention can be embodied in a high-speed reconstruction error estimation apparatus of the present invention.

5 is a block diagram illustrating a fast reconstruction error estimation apparatus of a video encoder according to an embodiment of the present invention.

5, a fast reconstruction error estimation apparatus of a video encoder includes a transform / quantization unit 510, a reconstruction unit 520, an entropy coding unit 530, a new distortion calculator unit 540, an RDO Mode Decision unit 550 An Intra / Inter Prediction unit 560, a Deblock / SAO Filters unit 570, and a Decoded Picture Buffer unit 580.

The reconstruction unit 520, the entropy coding unit 530, the RDO mode decision unit 550, and the intra / inter prediction unit 560, which are the remaining components except for the transform / quantization unit 510 and the new distortion calculator unit 540, ), The Deblock / SAO Filters unit 570 and the Decoded Picture Buffer unit 580 are the same as those described above with reference to FIG. 1 to FIG. 4, and thus a repeated description will be omitted.

First, the transform / quantization unit 510 can receive the prediction error image U, which is a difference value between the original image S and the prediction image P from the intra / inter prediction unit 560.

The transform / quantization unit 510 generates a transform coefficient Vs by performing discrete cosine transform (DCT) and scaling on the input predictive error image U, To the calculator unit 540.

In addition, the transform / quantization unit 510 generates a quantization coefficient Vq by performing inverse scaling and quantization on the generated transform coefficient Vs, and outputs the quantization coefficient Vq to the New Distortion Calculator 540 and Reconstruction To the unit 520.

Meanwhile, a more detailed description of the Transform / Quantization unit 510 will be continued with reference to FIG.

)를 계산할 수 있다. The New Distortion Calculator unit 540 calculates an estimated reconstruction error in the transform region based on the transform coefficient Vs and the quantization coefficient Vq transmitted from the transform /

) Can be calculated.

)를 RDO Mode Decision 부(550)에 전달할 수 있다.Then, the New Distortion Calculator unit 540 calculates an estimated restoration error (

) To the RDO Mode Decision unit 550. [

)을 입력으로 하여 복원 오차(

)를 계산한 기존의 Distortion calculator 부(도 1의 140)와 달리 원 영상(S)과 복원 영상(

)의 입력 없이, Transform/Quantization 부(510)에서 생성된 변환 계수(Vs)와 양자화 계수(Vq)를 입력으로 하여 추정 복원 오차(

)를 계산할 수 있게 된다.That is, an apparatus for estimating a fast reconstruction error of a video encoder according to the present invention includes an original image S and a reconstructed image generated by a reconstruction unit 120

) As the input,

Unlike the conventional distortion calculator (140 in FIG. 1) which calculates the original image (S) and the restored image

(Vs) and the quantization coefficient (Vq) generated by the Transform / Quantization unit 510 without inputting the estimated reconstruction error

Can be calculated.

As a result, the fast reconstruction error estimation apparatus of the video encoder according to the present invention can omit the image reconstruction process for the N modes, which was required in the conventional method, The memory access and the power consumption required for the memory access and the power consumption can be saved. As a result, the coding time can be significantly reduced.

Meanwhile, the New Distortion Calculator unit 540 will be described in more detail with reference to FIG.

6 is a diagram specifically illustrating a transform / quantization unit included in the fast reconstruction error estimation apparatus of a video encoder according to an embodiment of the present invention.

Referring to FIG. 6, the transform / quantization unit 510 may include a transform unit 511 and a quantization unit 512 for an integer unit transform and a quantization operation in the HEVC. have.

는 DCT matrix를 나타내고,

는 transpose된 DCT kernel matrix를 나타낸다. Specifically, the transform unit 511 may perform an integer-based discrete cosine transform (DCT) on the input predictive error image (511a). 511a

Represents a DCT matrix,

Represents the transposed DCT kernel matrix.

만큼의 스케일링(크기 변환, Scaling)을 수행(511b)하여 변환계수(Vs)를 출력할 수 있다. 인용부호 511b에서의

는

를 만족하고,

은

를 만족한다. 여기서,

는 화소 bit 크기(8, 10, 12 등)를 나타내고,

은

Transform 크기(L=4, 8, 16, 32 등)를 나타낸다.Then, the converting unit 511 converts the DCT result (V)

(511b) to output the conversion coefficient (Vs). 511b

The

Lt; / RTI >

silver

. here,

Represents the pixel bit size (8, 10, 12, etc.)

silver

Transform size (L = 4, 8, 16, 32, etc.).

는 양자화 파라미터(QP, Quantization Parameter)에 따른 양자화 크기(Quantization Step Size)를 나타낸다.The quantization unit 512 may perform inverse scaling (512a) on the transform coefficient Vs and perform quantization on the inverse scaled result (V) (512b) to output the quantization coefficient Vq. At point 512b

Represents a quantization step size according to a quantization parameter (QP).

Therefore, the prediction error image U in the transform / quantization unit 510 included in the fast reconstruction error estimation apparatus of the video encoder according to the present invention is transformed through the DCT 511a and the scaling 511b in the transform 511, The transform coefficient Vs can be outputted as the quantization coefficient Vq via the inverse scaling 512a and the quantization 512b in the quantization 512. [

)를 계산할 수 있게 된다.5). As described above, the transform coefficient Vs and the quantization coefficient Vq outputted as described above are input to the New Distortion Calculator (540 in Fig. 5), and the New Distortion Calculator (540 in Fig. 5) Estimated restoration error (

Can be calculated.

7 is a block diagram illustrating a New Distortion Calculator included in a fast reconstruction error estimation apparatus of a video encoder according to an embodiment of the present invention.

)에 대한 차이는 앞서 도 2를 통해 설명된 영상 복원(Reconstruction) 과정을 이용하여 아래의 수학식 1과 같이 예측 오차(U)와 역양자화(Inverse Quantization) 및 역변환(Inverse Transform)된 예측 오차(

)의 차이로 나타낼 수 있다.In the HEVC encoder, the original image (S) and the reconstructed image (

The difference between the prediction error (U) and the inverse quantization and inverse transformed prediction error (U) as shown in the following Equation 1 using the image reconstruction process described above with reference to FIG. 2

).

In general, the DCT 511a is a transform in which energy is conserved by Parseval's Theorem, and the transform of the HEVC maintains the property of the DCT 511a. Thus, the distortion of Equation 1 can be expressed as Equation 2 below .

은 아래의 수학식 3과 같이 양자화와 역양자화를 거친 계수를 나타낸다.V in Equation (2) represents a coefficient after the DCT 511a is performed as shown in FIG. 6,

Represents a coefficient obtained through quantization and inverse quantization as shown in Equation 3 below.

은 HEVC에서 아래의 수학식 4와 같이 나타낼 수 있다.In Equation 3,

Can be expressed by the following Equation (4) in the HEVC.

)를 계산할 수 있다.The fast reconstruction error estimation apparatus of the video encoder according to the present invention uses the coefficients of the transform region from Equations 1 to 4 and the transform unit 511 and quantization unit 512 shown in FIG. 6, The estimated restoration error (

) Can be calculated.

)를 계산하기 위해, 수학식 2와 같이 변환부(511)의 출력인 변환계수(Vs)와 양자화부(512)의 출력인 양자화 계수(Vq)를 이용한다. 즉, 수학식 3 및 4를 이용하여 수학식 2를 아래의 수학식 5와 같이 표현할 수 있다.As described above, in the present invention, the estimated reconstruction error (

The transform coefficient Vs output from the transforming unit 511 and the quantization coefficient Vq output from the quantizing unit 512 are used as shown in Equation 2 below. That is, Equation (2) can be expressed as Equation (5) below using equations (3) and (4).

Equation (6) and Equation (7) are substituted into Equation (5) and expressed as Equation (8) below.

과

는 정수 연산의 양자화와 역양자화를 위해 HEVC에 정의된 상수값이다.here,

and

Is a constant value defined in HEVC for quantization and dequantization of integer operations.

는

를 만족하고,

는

를 만족한다. 그리고, 수학식 8로부터 추정 복원 오차(

)는 아래의 수학식 9와 같이 표현될 수 있다.On the other hand, in Equation (8)

The

Lt; / RTI >

The

. From Equation (8), the estimated restoration error (

) Can be expressed by Equation (9) below.

Since Equation (9) is scaled in integer units using Equations (6) and (7) for integer operation in Equation (5), it can be calculated as Equation (10) by integer unit multiplication and shift operation.

는

를 나타내고,

는

를 나타낸다.In Equation (10)

The

Lt; / RTI >

The

.

과

는 정수 연산의 양자화와 역양자화를 위해 HEVC에 정의된 상수값이므로, 본 발명에 따른 비디오 부호화기의 고속 복원 오차 추정 장치는 추가적인 상수 테이블을 필요로 하지 않는다.Meanwhile,

and

Is a constant value defined in the HEVC for quantization and dequantization of integer arithmetic operations. Therefore, the fast reconstruction error estimation apparatus of the video encoder according to the present invention does not require an additional constant table.

)이 최소가 되는 모드(

)의 결정을 위한 복원 오차의 계산시에 원 영상(S) 과 복원 영상(

)을 필요로 하지 않는다. 5 and 7, the New Distortion Calculator 540 included in the fast reconstruction error estimation apparatus of the video encoder according to the present invention calculates the rate-

) Is the minimum mode

(S) and reconstructed image (S) in the calculation of the reconstruction error

).

)를 계산한다. Instead, an estimated restoration error (Vs) is obtained by using a transform coefficient (Vs) and a quantization coefficient (Vq) obtained from the transform unit (511 in Fig. 6) of the HEVC and the quantization unit

).

)를 계산할 수 있다. 본 발명에 따른 비디오 부호화기의 고속 복원 오차 추정 장치는 32x32부터 4x4까지 모든 크기의 Transform 연산과 8-bit/10-bit 등의 화소 크기에 모두 적용 가능하고, HEVC에서 지원하는 4x4 블록에서의 Transform Skip 모드에도 적용 가능하다.In this process. The New Distortion Calculator unit 540 effectively estimates the restoration error ("

) Can be calculated. The apparatus for fast reconstruction error estimation of a video encoder according to the present invention can be applied to all sizes of 32x32 to 4x4 transform operations and pixel sizes such as 8-bit and 10-bit, and Transform Skip in a 4x4 block supported by HEVC Mode.

As a result, the fast reconstruction error estimation apparatus of the video encoder according to the present invention includes the New Distortion Calculator unit 540, thereby omitting the image reconstruction process for the N modes required in the conventional system, , Memory access and power consumption required for image reconstruction in the conventional N modes can be saved, and coding time can be significantly reduced.

Accordingly, the foregoing detailed description should not be construed in any way as limiting and should be considered illustrative. The scope of the present invention should be determined by rational interpretation of the appended claims, and all changes within the scope of equivalents of the present invention are included in the scope of the present invention.

Claims (8)

A transform coefficient Vs obtained by performing discrete cosine transform (DCT) and scaling on an input predictive error image U and inverse scaling and quantization A Transform / Quantization unit for outputting the quantization coefficient Vq performed; And
Based on the transform coefficient (Vs) and the quantization coefficient (Vq), an estimated reconstruction error (

), And the estimated restoration error (

) To a rate distortion optimization (RDO) mode decision unit,
Wherein the Transform / Quantization unit scales the result of performing the DCT by 2 ^ST , and outputs the transform coefficient Vs, wherein ST satisfies ST = 15-BM, L = L x L Transform size (L = 2n, n is a natural number of 2 or more), where M = log ₂ L, B denotes a pixel bit size, and L = L x L.
The method according to claim 1,
The New Distortion Calculator unit,
A constant value defined in HEVC (High Efficiency Video Coding) is used for quantization and inverse quantization of an integer operation, and the estimated restoration error

) Of the video encoder.
The method according to claim 1,
The estimated restoration error (

) And bit rate (

(RDO) mode determining unit that determines a rate-distortion optimized (RDO) mode having a minimum rate-distortion cost, based on the RDO mode decision unit.
The method of claim 3,
And a reconstruction unit which performs reconstruction only once per coding tree block (CTU) only for the determined mode.
The method of claim 3,
Quantization, entropy coding, and image reconstruction of the predictive error image U only for the determined mode to generate a final bitstream and a final reconstructed image, Speed reconstruction error estimation apparatus.
6. The method of claim 5,
And the final reconstructed image is stored in a decoded picture buffer unit through a deblock / SAO filter unit and used for prediction of a next image.
Generating transform coefficients by performing discrete cosine transform (DCT) and scaling on the input predictive error image (U);
Performing inverse scaling and quantization on the transform coefficient (Vs) to generate a quantization coefficient (Vq); And
Based on the transform coefficient (Vs) and the quantization coefficient (Vq), an estimated reconstruction error (

), ≪ / RTI >
Wherein the step of generating the transform coefficients performs the scaling by 2 ^ST on the result of performing the DCT to output the transform coefficient Vs, wherein ST satisfies ST = 15-BM , M denotes a fast reconstruction error estimate of a video encoder that satisfies M = log ₂ L, B denotes a pixel bit size, and L = L x L Transform size (L = 2n and n is a natural number of 2 or more) Way.
8. The method of claim 7,
The estimated restoration error (

) And bit rate (

Determining a rate distortion optimized (RDO) mode with a minimum rate-distortion cost,
Performing reconstruction only once per coding tree block (CTU) only for the determined mode to generate a final bitstream and a final reconstructed image; And
And storing the final reconstructed image to be used for prediction of a next image.

Images