WO2009035144A1 - 画像処理装置及び画像処理方法 - Google Patents
画像処理装置及び画像処理方法 Download PDFInfo
- Publication number
- WO2009035144A1 WO2009035144A1 PCT/JP2008/066911 JP2008066911W WO2009035144A1 WO 2009035144 A1 WO2009035144 A1 WO 2009035144A1 JP 2008066911 W JP2008066911 W JP 2008066911W WO 2009035144 A1 WO2009035144 A1 WO 2009035144A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- prediction mode
- intra
- quantization
- offset
- difference
- Prior art date
Links
- 238000012545 processing Methods 0.000 title claims description 119
- 238000003672 processing method Methods 0.000 title claims description 7
- 230000015556 catabolic process Effects 0.000 claims abstract description 20
- 238000006731 degradation reaction Methods 0.000 claims abstract description 20
- 230000002093 peripheral effect Effects 0.000 claims abstract description 15
- 238000013139 quantization Methods 0.000 claims description 315
- 238000004364 calculation method Methods 0.000 claims description 17
- 230000006866 deterioration Effects 0.000 claims description 11
- 238000001514 detection method Methods 0.000 description 109
- 239000011159 matrix material Substances 0.000 description 89
- 238000000034 method Methods 0.000 description 79
- 238000011156 evaluation Methods 0.000 description 72
- 230000008569 process Effects 0.000 description 52
- 230000009466 transformation Effects 0.000 description 11
- 238000010586 diagram Methods 0.000 description 10
- 230000000694 effects Effects 0.000 description 9
- 230000007423 decrease Effects 0.000 description 8
- 238000010606 normalization Methods 0.000 description 8
- 230000008859 change Effects 0.000 description 4
- 230000000007 visual effect Effects 0.000 description 4
- 238000006243 chemical reaction Methods 0.000 description 3
- 230000006870 function Effects 0.000 description 3
- 238000012937 correction Methods 0.000 description 2
- 230000006872 improvement Effects 0.000 description 2
- 230000000630 rising effect Effects 0.000 description 2
- 238000013341 scale-up Methods 0.000 description 2
- LFVLUOAHQIVABZ-UHFFFAOYSA-N Iodofenphos Chemical compound COP(=S)(OC)OC1=CC(Cl)=C(I)C=C1Cl LFVLUOAHQIVABZ-UHFFFAOYSA-N 0.000 description 1
- 230000003044 adaptive effect Effects 0.000 description 1
- 230000001174 ascending effect Effects 0.000 description 1
- 238000012790 confirmation Methods 0.000 description 1
- 238000007796 conventional method Methods 0.000 description 1
- 238000013144 data compression Methods 0.000 description 1
- 230000003247 decreasing effect Effects 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 230000002441 reversible effect Effects 0.000 description 1
- 238000000926 separation method Methods 0.000 description 1
- 230000008685 targeting Effects 0.000 description 1
- 238000012546 transfer Methods 0.000 description 1
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
- H04N19/102—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the element, parameter or selection affected or controlled by the adaptive coding
- H04N19/103—Selection of coding mode or of prediction mode
- H04N19/11—Selection of coding mode or of prediction mode among a plurality of spatial predictive coding modes
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
- H04N19/102—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the element, parameter or selection affected or controlled by the adaptive coding
- H04N19/124—Quantisation
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
- H04N19/134—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the element, parameter or criterion affecting or controlling the adaptive coding
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
- H04N19/134—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the element, parameter or criterion affecting or controlling the adaptive coding
- H04N19/146—Data rate or code amount at the encoder output
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/60—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using transform coding
- H04N19/61—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using transform coding in combination with predictive coding
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/60—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using transform coding
- H04N19/625—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using transform coding using discrete cosine transform [DCT]
Definitions
- the present invention relates to an encoding apparatus and method for detecting, for example, a quantization matrix, a quantization parameter and the like used in the previous encoding in a predetermined unit.
- the MPEG Moving picture experts group
- back search refers to the quantization step used in the previous coding, or the quantization step that is in the multiple relation thereof, as the remainder of the discrete cosine transfer (DCT) coefficient. It is a method to find the quantization step related to the minimum value as the optimal quantization step by using the property that the total sum becomes minimum.
- the quantization step in the previous encoding is reproduced by the back search method, and again with the same quantization step and GOP (Group of pictures) phase as in the previous time.
- a video data compression device is disclosed that prevents degradation of video quality when encoding and decoding input video data by tandeming an encoder and a decoder to repeat video data encoding and decoding. It is done.
- AVC Advanced Video Coding
- AVC Intra In this AVC intraframe coding, in order to improve the coding efficiency, a pixel value in the macro block is predicted from peripheral pixels in the screen, and a screen in which the differential image data from the prediction is coded A forecast has been introduced.
- the AVC stream encoded using this intraframe coding only is hereinafter referred to as AVC Intra. That is, in this AVC Intra, an input image is separated into a predicted image from peripheral pixels and a difference at the time of encoding, and the difference is encoded. :.
- the prediction image is prepared in nine modes, such as prediction from the upper pixel and prediction from the horizontal pixel, and usually the sum of differences from the prediction image per unit block (eg 4 ⁇ 4 samples etc.) Has adopted the smallest prediction mode.
- Patent Document 1 Japanese Patent Application Laid-Open No. 10-17040.
- ⁇ ' However, although the generated code amount at the time of encoding tends to decrease in the above-mentioned conventional method, since the part of AVC encoding is a non-reversible transform, the encoding and AVC do not Decoding ⁇ : causes quantization distortion to the original input image, so the quantization distortion degrades the image quality each time encoding and decoding are repeated during dubbing.
- a difference image data consisting of difference values from peripheral pixels of the processing target pixel to be processed is generated for each of a plurality of predetermined in-screen prediction modes.
- the difference image data generation unit On the basis of the difference image data generation unit, the difference sum value based on the sum of the absolute values of the respective processing target pixels in the difference image data table, and the priority set in advance in the in-screen prediction mode.
- a ranking unit has been added to prioritize internal prediction modes.
- the image processing apparatus can set the priority of a particular in-screen prediction mode to a high level each time, which makes it easy to select a particular in-screen prediction mode, and the in-screen used in the previous encoding.
- the prediction mode can be easily selected to suppress quantization distortion during dubbing.
- a difference image data that is different from the difference value from the peripheral pixel of the processing target pixel to be processed is generated for each of a plurality of predetermined in-screen prediction modes.
- the intra prediction mode is selected according to the difference image data generation step, the difference total value which is the sum of the absolute values of the processing target pixels in the difference image data, and the priority set in advance in the intra prediction mode.
- a ranking step has been established.
- the image processing apparatus can set the priority of a particular in-screen prediction mode to a high level each time, it becomes easy to select a particular in-screen prediction mode, and the screen used in the previous encoding is selected. By making it easy to select the inner prediction mode, it is possible to suppress quantization distortion during dubbing.
- this enables the image processing apparatus to set the priority of a specific intra prediction mode to a high level each time, which makes it easy to select a specific intra prediction mode, and the previous encoding It is possible to easily select the intra-frame prediction mode that is sometimes used, to suppress quantization distortion during dubbing, and to realize an image processing apparatus and an image processing method that can suppress deterioration in image quality.
- FIG. 1 is a conceptual diagram of an encoding apparatus according to the first and second embodiments of the present invention.
- FIG. 2 is a schematic diagram for explaining transformation and quantization in the AVC standard.
- FIG. 3 is a diagram showing the relationship between the quantization parameter QP and the sum of remainders ( ⁇ r) when an image that has never been encoded is used as the input image.
- FIG. 4 is a diagram showing the relationship between the quantization parameter QP and the sum of remainders ( ⁇ ⁇ ⁇ ⁇ r) for an input image that has been encoded and decoded. '
- Figure 5 shows the quantization parameters after normalization with the rescaling factor RF.
- FIG. 6 is a diagram showing the relationship between the normalization with rescheduling factor 1 RF and the quantization parameter QP after correction of a predetermined region and the sum ⁇ of evaluation values.
- FIG. 7 is a block diagram of a coding apparatus according to the first embodiment of the present invention.
- FIG. 9 shows the relationship between macro projects and intra coding units.
- Figure 10 (a). ⁇ (:)) Is an illustration of nine different 4 x 4 Intra prediction modes.
- FIG. 11 shows the flow of processing according to the first embodiment of the present invention.
- FIG. 11 shows the flow of processing according to the first embodiment of the present invention.
- FIG. According to the processing of Q matrix ⁇ outpouring in slice units by Ronitija. 'FIG. 13 explains the processing of QP ⁇ ⁇ out in macro unit by the QP. Detection unit of the encoder according to the first embodiment of the present invention.
- FIG. 14 is a detailed flowchart of offset priority application processing according to the first embodiment of the present invention.
- FIG. 15 is a block diagram of a coding apparatus according to a second embodiment of the present invention.
- FIG. 16 is a flowchart describing processing of prediction mode detection by the encoding device according to the second embodiment of the present invention.
- FIG. 17 is a detailed flowchart of the offset-first application process according to another embodiment of the present invention. Low chart Best mode for carrying out the invention
- coding according to the Advanced Video Coding (AVC) standard is lossy conversion, and thus coding and decoding cause distortion to the original baseband image. Therefore, for example, each time encoding and decoding are repeated by dubbing or the like by tandem connection, the image quality is degraded due to the distortion.
- AVC Advanced Video Coding
- the intraframe coding of the AVC standard when re-coding an image that has been once coded and decoded, it is used in the previous coding.
- One of the features is that the distortion of the image can be suppressed only to the operation error of the encoder and the decoder by reusing the following three parameters.
- FIG. 1 is a conceptual diagram of the coding apparatus and method according to the first embodiment of the present invention, and will be described.
- the image processing apparatus 1 includes a first encoding unit 2, a second encoding unit 3, and a back search unit 3.
- the back search unit 3 further includes a prediction mode detection unit 5, quantum A quantization matrix (Q matrix) detection unit 6 and a quantization parameter (QP) detection unit 7 are provided.
- the first encoding unit 2 performs the first encoding on the input image information.
- the back search unit 3 calculates integer precision DCT coefficients obtained in the process of the first encoding with a resizing factor (RF) by combining various quantization matrices Qmatrix and quantization parameter QP.
- RF resizing factor
- the quantization matrix Qmatrix, the quantization parameter QP, and the in-screen prediction mode, which were used in the first encoding, are evaluated by evaluating the size of the remainder r obtained when the division is performed. Then, the second encoding unit 4 performs the second encoding using the quantization matrix Qmatrix, the quantization parameter QP and the intra prediction mode detected by the back search unit 3 in the first encoding. .
- the Qmatrix detection unit 6 detects the integer precision DCT in the first coding based on the integer precision DCT coefficients obtained in the first coding process.
- the quantization matrix Qmatrix is detected for each slice using the remainder obtained by dividing the coefficients by the combination of various quantization matrices Qmatrix and quantization parameter QP and RF by one.
- the QP detection unit 7 rescales the integer precision DCT coefficients based on the integer precision DCT coefficients obtained in the first coding process by combining the various quantization matrices Qmatrix and the quantization parameter QP. Divided by: The remainder is detected, and the quantization parameter QP is detected for each macro block.
- the prediction mode detection unit 5 detects the intra prediction mode used in the previous encoding.
- the detected prediction mode is adopted in the second encoding unit 4.
- DCT processing is performed on the difference image data, and the obtained DCT coefficients are quantized by dividing by the resizing factor RF by combining Qmatrix and the quantization parameter QP.
- This quantization data is further coded by a predetermined coding method to form a coded stream.
- the intra-frame prediction mode used for coding, the quantization matrix Qmatrix, and the quantization parameter QP are set. : '
- the quantization data is decoded using the set in-frame prediction mode, quantization, matrix Qmatrix and quantization parameter QP.
- the remainder portion rounded off during encoding Can not be restored. Therefore, in the AV.C standard, the image quality is degraded by the remainder of this quantization.
- this image quality is called quantization distortion. ,.
- the quantization matrix Qmatrix used in the previous encoding By detecting the parameter QP or the intra prediction mode and performing coding using them, the image quality degradation of the video data is suppressed.
- quantization matrix Qmatrix used in the previous encoding
- the back search divides the DCT coefficients by each quantization scale Q in Moving Picture Experts Group (MPEG) 12 and has the smallest remainder r.
- the quantization scale Q can be detected as the quantization scale Q used at the previous encoding.
- the A V C standard differs from M P E G-2 in the following points, and if the M P E G E-2 method is applied as it is, the detection accuracy will decrease.
- the present invention adopts AVC encoding, detects quantization matrix Qmatrix and quantization parameter QP used in the previous encoding with a high probability with a predetermined unit, and reuses them. It is therefore an issue to reduce rounding errors due to quantization during dubbing and improve the dubbing characteristics.
- the quantization level Z is multiplied by the quantization matrix Qmatrix and the rescaling factor one RF which is a function of the quantization parameter QP, and the precision is improved.
- the shifted one is decoded.
- the remainder r is considered to be 0 if (W ⁇ 6) is the same. From these characteristics, the combination of various quantization matrices Qmatrix and quantization parameter QP can be used to divide integer precision DCT coefficients (W ⁇ 6) that are shifted using resizing factor-one RF. By evaluating the size of the remainder r, it is possible to detect the quantization matrix Qmatrix and the quantization parameter QP 'used in the previous coding. In addition to the above, in the image processing apparatus 1 according to the first embodiment, in order to improve detection accuracy, the following (i) to (vi Consider the point of). The following is a group statement for each point.
- the DCT part is divided into an integer precision part and a non-integer precision part, the integer precision part is called integer precision DCT, and the non-integer precision part and quantization are collectively called quantization.
- integer precision DCT used for encoding hereinafter referred to simply as “DCT” and integer used for decoding
- the precision inverse DCT hereinafter simply referred to as “inverse DCT” is not an inverse transform, therefore, the DCT coefficient W used in encoding and the inverse DCT coefficient W used in decoding (hereinafter: “W ′” Not to be equal).
- a quantization error ER expressed by the following equation occurs during quantization. If Eq. (7) is expressed as an actual numerical value, it can be represented as Eq. (8). That is, in AVC Intra, when quantization processing and inverse quantization processing are performed on the DCT coefficients by encoding and decoding, the values of the elements of the inverse DCT coefficients are scaled up by the quantization error ER.
- the value of the decoded differential image data is made the same level as the differential image data by dividing the decoded differential image data by “64”. There is.
- the rescaling factor at the time of decoding is given by the following equation
- decoding re-scaling factor one RF r the re-scaling factor RF used as a value to be divided in the back search process is calculated.
- the effect of scale-up accompanying DCT processing can be removed from the remainder r, and the remainder r can be compared on the same scale.
- the image processing apparatus 1 calculates the rescaling factor-one RF by multiplying the value of the transformation matrix D with respect to the decoding rescheduling factor RF r.
- the image processing apparatus 1 by dividing the difference image data by the re-scaling factor-one RF, it is possible to simultaneously divide the portion that is uniformly scaled up by the DCT processing, so that the scale-up It is designed to reduce the variation of remainder r and improve the accuracy of back search detection.
- the sum of absolute differences (SAD: Sum of Absolute Differences) with the predicted image from surrounding pixels is encoded.
- SAD Sum of Absolute Differences
- the quantization level Z is multiplied by the decoding rescheduling factor-one RF r at the time of decoding
- the decoding rescheduling factor-one RF r is carried in advance in the standard by 6 bits in order to prevent calculation rounding at the time of decoding. This is why the DCT coefficient W 'is shifted 6 bits to the left at the time of.
- inverse quantization and inverse DCT processing are performed with a 6-bit carry condition, After adding up the image with a 6-bit carry, the sum is reduced by 6 bits to obtain a baseband image. Since the data contained in the lower 6 bits is rounded off by this 6-bit digit reduction, an operation error E is generated, and the shifted DCT coefficient (W ⁇ 6) generated in the next encoding is re-scaled. It can not be divided by factor one RF. Therefore, the present invention detects the quantization parameter QP whose remainder is the minimum value, not the quantization parameter QP whose remainder r becomes 0.
- the operation error E at this time of decoding may be a negative value
- the actual remainder r is compared with the value obtained by subtracting the remainder r from the rescaling factor one RF, and the smaller one is evaluated value Y I assume.
- the evaluation value Y is calculated according to the equation (12).
- the quantization parameter QP that minimizes the evaluation value Y is detected.
- the image processing apparatus 1 can appropriately detect the quantization parameter QP using the appropriate evaluation value ⁇ .
- the residual r is calculated by dividing the 6-bit shifted DCT coefficient (W ⁇ 6) by the various quantization parameter respinning factor one RF by using an image that has never been encoded as an input image.
- the horizontal axis represents the quantization parameter QP, and the vertical axis represents the sum of remainders ( ⁇ r), resulting in the curve rising to the right as shown in FIG.
- the remainder r is calculated by dividing the 6-bit shifted DCT coefficient (W ⁇ 6) by the various quantization parameter QPs and the rescaling factor one RF by the QP
- the horizontal axis represents the quantization parameter
- the vertical axis the sum of remainders ( ⁇ r), as shown in Fig.4.
- a local minimum value of the sum ( ⁇ . ⁇ ) of the remainder r is generated, there is also a tendency of rising to the right, regardless of whether encoding and decoding have already been performed. It can be seen that the sum ( ⁇ r) of small small residues r also decreases.
- the rescaling factor RF Standardize.
- the DCT coefficient W that can take a large residue has a small residue, it can be considered as random and can be weighted (usually, the low frequency component is weighted because the low frequency component has a larger DCT coefficient W).
- Figure 6 shows the relationship between the sum of the evaluation value Y normalized with resizing factor 1 RF and the quantization parameter QP after scaling the remainder r by the absolute value IWI of the DCT coefficient. ing. From the figure, the sum of the evaluation values Y in the quantization parameter QP used in the previous encoding ( ⁇ ⁇ ) and the sum of the evaluation values of the quantization parameter QP with a shift of 6 n: ( ⁇ ⁇ ) It is clear that there is almost no change compared to Figure 5. :,
- normalization with rescaling factor one RF is performed only in a region where the sum ( ⁇ ) of the evaluation values Y has a slope (
- the value of the DCT coefficient absolute value I WI can also be used as the evaluation value Y.
- the sum of the evaluation values Y ( ⁇ ⁇ ) is minimized in the region where (IWI 7 7) ⁇ RF, and the wrong quantization parameter ⁇ QP is the last used quantization. It is designed to be able to reliably prevent detection as a parameter or QP.
- the image processing apparatus 1 stores the values of approximately 5 quantization parameter values QP. In an array in order from the one with the smallest sum ( ⁇ ⁇ r) of the remainder r, and If the difference between the QP with the smallest r) and the QP with the second smallest sum ( ⁇ r) of the remainder r is small and the difference is 6n, the larger of the QP is adopted. Furthermore, the image processing apparatus .1 compares the adopted quantization parameter QP with the quantization parameter QP having the third smallest sum r of residual r ( ⁇ r), and the difference between the two is 6n. Then, the larger quantization parameter QP is adopted, and the adopted quantization parameter QP is replaced.
- the image processing apparatus 1 when the image processing apparatus 1 detects a plurality of local minimum values of the sum ( ⁇ r) of the remainder r, the quantization parameter QP having a large value is prioritized and used in the previous encoding. Adopted as quantization parameter QP. As a result, the image processing apparatus 1 can calculate the quantization parameter in which the quantization parameter 1Q shifted by 6n is used in the previous encoding.
- the image processing apparatus ⁇ which is designed to be able to suppress false detection as a QP, further checks whether the sum ( ⁇ r) of the detected plurality of remainders r is 6 ⁇ cycles or not. In order to be able to suppress false detection of accidentally existing local minimum values as quantization parameter QP.
- the resizing factor 1 RF is calculated as described above for various quantization parameter QPs, and is used in the previous encoding using the evaluation value Y calculated from the remainder r. Detect the quantised quantization parameter QP. Therefore, the larger the number of quantization parameter QPs that can be taken, the larger the amount of computation and evaluation. In order to avoid this problem, if the approximate value of the quantization parameter QP used for the previous encoding is known, it is sufficient to evaluate only with the quantization parameter QP around it, and the amount of operation To reduce Note that this quantization parameter QP can be calculated, for example, in the original image determination processing using the quantization coefficient in the third embodiment. It can also be inferred from the activity of the input image data.
- quantization matrix Qmatrix several patterns are defined in advance (for example, it can be identified by an ID number).
- the combination of the quantization matrix Qmatrix and the quantization parameter QP is changed in macro block units, and the rescheduling factor one RF is calculated for each quantization matrix Qmatrix and the quantization parameter QP.
- the minimum value of the remainder r has a period of 6 n for quantization parameter Q.
- QP even if quantization parameter Q P deviates by 6 n. It only needs to be able to detect the quantization matrix Qmatrix. For this reason, in the image processing apparatus 1, if the approximate value of the quantization parameter used in the previous encoding is known, the continuous 6 quantization parameters including the value of the QP are known. It is assumed that it is sufficient to evaluate by QP.
- the quantization matrix execute quantization parameter detection processing to detect quantization parameter QP using Qmatrix.
- the configuration of the coding apparatus 10 according to the first embodiment of the present invention will be shown and described below with reference to FIG.
- the encoding device 10 corresponds to the image processing device 1 in FIG.
- the back search unit 30 is a back search unit 3
- the parameter one time encoder 50 is a second encoding unit 4
- the Qmatrix detection unit 31 is a Qmatrix detection unit 6
- the QP detection unit 32 is a QP detection unit It corresponds to 7 respectively.
- the encoder 10 has a pre-encoder 20 for the first pass (pass) and a first-order section 30 for the first pass, a parameter encoder for the second pass (pass) 50, the code amount A control unit 40, a Qmatrix / QP decision unit 41, and an in-screen prediction mode decision unit 45 are provided.
- the first-pass pre-encoder 20 further comprises an Intra prediction mode determining unit 21 with offset, integer precision 0 2 2, a quantizing unit 2 3 and an entropy calculation unit 2 4.
- the back search unit 30 further comprises a Qmatrix detection unit 31, a QP detection unit 32, an in-screen prediction mode detection unit 33.
- the second-pass parameter encoder 50 includes an intra-frame prediction processing unit 51, an integer precision DCT unit .52, a quantization unit 5.3 and an entropy coding unit 54.
- 2-pass coding is adopted, but the basic idea is the same even if it is multi-pass coding more than that.
- FIG. 8 (a) a detailed configuration example of the Qmatrix detection unit 31 is shown in FIG. 8 (a).
- a detailed configuration example of the QP detection unit 3.2 is shown in Fig. 8 (b). ⁇ ⁇ ⁇ :.
- the Qmatrix detection unit 31 is composed of a remainder calculation unit 31a, an evaluation. Value determination unit 31b, and a Qmatrix determination unit 31c.
- the Q P detection unit 32 includes a remainder calculation unit 3 2 a, an evaluation value determination unit 3. 2 b, and a QP determination unit 3 2 c.
- the input image is input to the first pass pre-encoder 20.
- the input image data is input to the Intra prediction mode determination unit 21 with offset.
- the Intra prediction mode determining unit 21 with offsets 1 gives priority to offsets described later.
- the application processing determines the optimal in-screen prediction mode from the nine in-screen prediction modes, and generates the difference image data and the difference absolute value sum SAD from the prediction image in which the input image uses peripheral pixels, Input to the integer precision DCT unit 2 2.
- the integer precision DCT unit 22 generates a DCT coefficient W from the difference image data by orthogonal transform using discrete cosine transform, and supplies the DCT coefficient W to the quantization unit 23 and the backtracking unit 3.
- the quantization unit 23 quantizes the DCT coefficient to generate quantized data.
- the quantization level Z output from the quantization unit 2 3 by this quantization is sent to the entropy calculation unit 2 4.
- the entropy calculation unit 24 binarizes and quantizes the input quantization level Z to generate a coded stream, and supplies the coded stream to the code amount control unit 40.
- the code amount control unit 40 estimates the quantization matrix Qmatri and the quantization parameter 1 QP. When performing code amount control for the parameter encoder 50 based on the generated code amount of the encoded stream. , This is sent to the Qmatrix / QP determination unit 41. .
- the DCT coefficient W sent from the integer precision DCT unit 22 to the back search unit 30 in slice units is input to the Q matrix detection unit 31.
- the Qmatrix detection unit 3.1 divides the DCT coefficient W by the remainder calculation unit 3 1 a by the various quantization matrices Q matrix by the rescaling factor 1 RF and supplies the result to the evaluation value determination unit 3 1 b.
- the evaluation value determination unit 31b converts the remainder r. Into the evaluation value Y according to the concept described in (iii) and sends it to the -Q matrix determination unit 31c.
- the Qmatrix determination unit 31 c compares the evaluation values Y of various quantization matrices Qmatrix, and calculates a quantization matrix having the smallest evaluation value Y.
- the Qmatrix is used as the previously used quantization matrix Qmatrix. Output. The details of the quantization matrix detection processing procedure by the Qmatrix detection unit 31 will be described later.
- the quantization matrix Qmatrix detected at this time is sent to the QP detection unit 32.
- the remainder calculating unit 32 a divides the DCT coefficient W by the resizing factor 1 RF based on various quantization parameter QPs, and supplies the result to the evaluation value judging unit 32 b.
- the evaluation value determination unit 32b converts the remainder r into an evaluation value Y according to the concept described in (iii), using the quantization matrix Qmatrix detected by the evaluation value Qmatrix detection unit 31. Send to QP judgment unit 3 2 c.
- the QP deciding unit 32c compares the evaluation values Y of various quantization parameter QPs, and determines the quantization parameter QP having the smallest evaluation value Y as the quantization parameter QP used last time. Output.
- the details of the quantization parameter detection processing procedure by the 0 to 3 detection unit 3 2 will be described later.
- the detected previous quantization parameter QP and quantization matrix Qmatrix are sent to the Qmatrix / QP determination unit 41.
- the back search unit 30 detects whether the one estimated in the first pass pre-encoder 20 is adopted for the quantization matrix Qmatrix and the quantization parameter QP. Decide what to use. Normally, if the input image data is an original image that has never been encoded, the back search will fail because there is no previous encoding, and it is estimated by the encoding process of the first pass Preen coder 20. Adopt the Conversely, if the input image is a non-original image that has already been encoded, the quantization distortion may be repeated to avoid the occurrence of the: detected by the back search unit 30.
- the quantization matrix Qmatrix, the quantization parameter QP, and the intra prediction mode decision unit with offset 2 1 determined by the Qmatrix / QP determination unit 4 1. Encode the input image using the determined intra prediction mode: and output it as an encoded stream (Stream). .
- the in-screen prediction processing unit 51 it is selected as the optimal in-screen prediction mode: and the in-screen prediction mode determined in the Intra prediction mode determining unit with offset 21:
- the prediction image is scanned to generate a difference image data from the input image, and the difference image data is input to the integer precision DCT unit 5: 2.
- the difference image data is output as DCT coefficient W by orthogonal transform using discrete cosine transform:
- the quantization unit 5 3 receives the DCT coefficient W as input and performs quantization using the quantization matrix Qma trix X determined by the Qmatrix / QP determination unit 4 1 and the quantization parameter QP, and the quantization level Z is entropic.
- the input quantization level Z is binarized and arithmetically coded and output as a coded stream.
- the quantization level Q is a function of the quantization matrix Qmatrix and the quantization parameter QP.
- the scaling factor RF is multiplied and the DCT coefficient W 6 bits left shifted is decoded. If the intra prediction mode changes, the distribution of DCT coefficients changes, so the probability that the remainder r is 0 is low.
- the above-mentioned quantization matrix detection processing and quantization parameter detection processing are executed in units of macroblocks MB (16 ⁇ 16 pixels).
- the intra prediction mode in which the code amount is minimized is detected in advance, and the above-mentioned quantization matrix detection / processing and quantization parameter processing is performed in the detected intra prediction mode. Execute detection processing.
- the detected in-screen prediction mode is not limited to the previously coded in-screen prediction mode, since the in-screen prediction mode 'in which the generated code amount is minimized is used, Among the Indra coding units IA, it is assumed that many of them were previously coded and they are intra prediction modes.
- the encoder 10 0 detects the local minimum value using the sum of remainders ( ⁇ r) for each macro block. For this reason, even if the intra-frame prediction mode different from the one used last in the intra-frame coding unit IA of 16 is mixed, the image processing apparatus 1 generates many previously-coded intra-frame prediction modes. It is possible to appropriately detect the quantization matrix Qmatrix and the quantization parameter QP based on.
- the encoding apparatus 10 of the present invention performs nine intra-screen (Intra (intra)) predictions defined by the AVC standard when encoding is performed by the AVC Intra. Of the modes, a specific in-screen prediction mode is selected preferentially. As a result, the encoding device 10 prevents the change in the intra prediction mode during dubbing. As a result, it is possible to suppress the occurrence of quantization distortion again and improve the dubbing characteristics.
- this embodiment is characterized in that the dubbing characteristic is improved such that a predetermined in-screen prediction mode is preferentially selected.
- Figures 10 (a) to (;)) show the prediction directions for the nine intra prediction modes (mode numbers 0 to 8). '
- the prediction direction is one prediction direction for each 4 ⁇ 4 pixel block (ie, in-screen prediction). ⁇ ⁇ ⁇ ⁇ )))) will be encoded in blocks of 4 ⁇ 4 pixels using this selected prediction direction.
- the screen with mode number 2 If in the inner prediction mode, the 4 pixels to the left (horizontally adjacent 'of the processing target pixel of 4 to .4 to be processed and the upper (adjacent to the vertical direction) Based on the average value of 8 pixels in total (for example, mean (A, B, C ⁇ D, ⁇ , J ⁇ , L)) of 4 pixels, and all 4 ⁇ 4 pixels of the pixel to be processed are predicted. Ru .. -.
- DC Direct Current prediction mode
- intra-screen prediction modes mode numbers 0 to I, 3 to 8
- prediction is performed according to the directions of arrows shown in the figure.
- -Fig. 11 shows the flow of processing performed when the input image data is a non-original image, and shows a part of the configuration in the encoding device 10 of Fig. 7 extracted.
- the encoder 10 serves as an intra prediction mode determining means.
- Intra offset mode with offset (offset) 2 1, integer precision DCT 2 as discrete cosine transform means 2 2, back search portion 30 as quantization parameter calculation means 3 0, as quantization means
- the process is performed by the quantization unit 53, the entropy encoding unit 54 consisting of CABAC (Context-based Adaptive Binary Arithmetic Code, etc.) as encoding means. It uses binary arithmetic coding that adaptively codes according to the context:
- the input image data is input to the Intra prediction mode determination unit 21 with offset, and the Intra prediction mode determination unit 21 with offset is one of nine types of in-screen prediction modes.
- the intra prediction mode of is adopted.
- the predicted image for each intra coding unit IA that is, difference image data
- the smallest difference absolute value sum SAD which is the sum of the absolute values of the processing target pixel in, is adopted. In the present embodiment, processing is performed along the flow of the flowchart of FIG. 14 described later.
- Intra.prediction mode determination unit 21 with offset of encoding device 10 sets in advance the value of offset for each intra-frame prediction mode, and is encoded according to each intra-frame prediction mode. Calculate the sum of absolute differences SAD in the differential image data. Then, the Intra prediction mode determination unit with offset 2.1 adds an offset- corresponding to the intra-frame prediction mode to the difference absolute value sum SAD, and also adds an offset-added difference absolute value sum to which an offset is added. SAD f Comparison is performed, and the intra prediction mode that minimizes the sum of absolute differences SAD is adopted as the intra prediction mode to be used for actual coding.
- the Intra prediction mode determining unit 21 with offset supplies the mode number X of the intra prediction mode selected in this way and the difference image data to the integer precision DCT unit 22.
- the integer precision DCT unit 22 orthogonally transforms the difference image data by DCT, and sends it to the back search unit 30 as a DCT coefficient W.
- the quantization parameter QP and the quantization matrix are obtained by the above-mentioned back-slash processing.
- the Q matrix is detected, and these and the DCT coefficient W are supplied to the quantization unit 53.
- the quantization unit 53 quantizes this DCT coefficient W using the quantization parameter QP. And the quantization matrix Qmatrix, and outputs the result as the quantization level Z to the entropy coding unit 5 4.
- the entropy coding unit 54 binary-arithmetically codes the quantization level Z and outputs it as a coded stream. .
- the Intra prediction mode decision unit with offset 21 exerts the following characteristic actions.
- the correlation in a certain direction is extremely strong and fixed to DC prediction mode, the difference absolute value sum SAD from the predicted image becomes extremely large. If the amount of generated code increases, select the intra prediction mode in the direction of the larger correlation, and suppress the increase in the amount of generated code.
- the encoding apparatus realizes flexible control of the image quality and dubbing characteristics according to the characteristics of the input image.
- the coding apparatus 10 when only one single intra prediction mode is permitted among the nine intra prediction modes as described above, DC is used to obtain a predicted image from the average value of peripheral pixels.
- the prediction mode ie, the intra prediction mode number 2 in FIG. 10 (c)
- the coding apparatus 10 preferentially selects the DC prediction mode.
- the intra-frame prediction mode with the smallest difference absolute value sum SAD to be originally selected and the DC prediction mode are estimated from the predicted image. Compare the absolute difference sum SAD, and if the difference is less than offset, adopt DC prediction mode.
- the Intra prediction mode decision unit with offset 21 sets the value of offset [n] for each intra prediction mode.
- the value of offset [nI is predetermined for each in-screen prediction mode.
- offset The Intra prediction mode determination unit 21 sets the value of offset [n] by reading o lpfset [n] from ROM (Read Only Memory) or the like (not shown).
- the value of this offsetb] is determined in accordance with the degree of deterioration of the image quality due to coding, and is set so that the smaller the decrease in the image quality is, the smaller the in-screen prediction mode is. That is, the value of offset [n] is set to “high” (does not add offset) to the offset to the DC prediction mode with the smallest image quality degradation. Further, the value of offset [n] is set so as to be large relative to the in-screen prediction mode in which the degree of degradation of image quality is large, and to be small relative to the intra-screen prediction mode in which the degree of degradation of image quality is small.
- the Intra prediction mode determining unit with offset 21 generates difference image data from input image data for each intra prediction mode and calculates difference absolute value sum .SAD.
- the Intra prediction mode determination unit with offset 2 1 adds offset [n] corresponding to the intra prediction mode to the difference absolute value sum SAD to calculate an offset difference absolute value sum SAD f, Do. Note: For DC prediction mode, the sum of absolute differences SAD, two offsets, and the sum of absolute differences SAD f. .
- Intra Intra prediction mode decision unit with offset 21 A screen that uses the intra prediction mode with the smallest offset absolute difference sum SAD f for encoding, by comparing offset absolute difference sums SAD f with each other. Select as inner prediction mode.
- the Intra prediction mode determination unit 21 with offset is supplied to the integer precision DCT processing unit 22 with the mode number X of the selected intra prediction mode and the differential image data. It is done.
- the following effects can be obtained. That is, by preferentially selecting the DC prediction mode according to the value of the offset, it is possible to suppress the change in the prediction mode in the screen at the time of dubbing. Further, the purpose of using the image according to the value of the offset.
- the dubbing characteristics can be controlled flexibly, and the same intra prediction mode as the previous encoding can be selected easily during dubbing to prevent repeated quantization distortion from being applied, and image quality degradation during dubbing. Can be reduced. (2-3) Processing procedure
- the encoding device 10 rescales the residual r by one rescaling factor only when the absolute value IWI of each DCT coefficient is equal to or less than the standardized threshold (ie,
- the evaluation value Y is calculated by normalization with RF, and if the absolute value I D W of the DCT coefficient is larger than the normalization threshold, the remainder r is the evaluation value Y. Then, the encoding device 10 is configured to detect the quantization matrix Q matrix based on the rescaling fac- tor R F that minimizes the evaluation value Y.
- the Qmatrix detection unit .3 1 is used in the previous encoding with the absolute value IW.I of the DCT coefficient W determined by the integer precision DCT unit 22 as input.Possible quantization matrix Qmatrix and quantum An initial value is determined from among the conversion parameter QP (step S 1). Subsequently, the Qmatrix detection unit 31 changes the combination of the quantization matrix Qmatrix and the quantization parameter QP in units of macroblocks, and changes the rescaling factor one for each quantization matrix Qmatrix. And each quantization parameter QP. Calculate RF.
- the remainder calculation unit 3 1 a calculates a remainder r obtained by dividing IW.I «6 by the re-scaling factor one RF for each sample in the slice: (step S 2).
- the minimum value of the remainder r has a period of 6n with respect to the quantization parameter QP, and even if the quantization parameter QP is shifted by 6n, Since it is only necessary to detect the quantization matrix Qmatrix in the processing of FIG. 10, if the approximate value of the quantization parameter QP used in the previous encoding is known, the continuous 6 It is assumed that it is sufficient to evaluate with the number of quantization parameter QPs.
- the evaluation value determination unit 3 lb determines whether or not
- «7> RF, and if the relationship is satisfied, the remainder r is normalized by the rescheduling factor one RF (The schedule The value of one ring is taken as the evaluation value Y (step S4). On the other hand, when the relationship does not hold, the evaluation value determination unit 3 lb sets IW ⁇ ⁇ 6 as the evaluation value Y (step S5). In this way, the evaluation value determination unit 3 lb calculates the remainder r obtained by dividing the shifted DCT coefficient (W ⁇ 6) by the rescaling factor -RF with respect to 256 ( 16xl6) samples in the macroblock. Let iii) be normalized and corrected for evaluation value Y, and calculate the sum of evaluation values Y for each quantization matrix Qmatrix and quantization parameter QP.
- the Qmatrix determination unit 31 c is a slice.
- the sum of the quantization matrix Qmatrix and the evaluation value Y for each quantization parameter QP is compared on a unit basis (step S 9), the quantization matrix Qmatiix for which the sum ( ⁇ ) becomes the smallest is detected, and the previously used quantum It is assumed that the quantization matrix Qmatrix (step S 10), and the ID number of the quantization matrix Qmatrix is output.
- the encoding device 10 multiplies the remainder r by the absolute value I Dw of the DCT coefficient, for example, to calculate a weighted multiplication value before the step S 2, and calculates the multiplication value, and rescheduling the multiplication value to a factor of 1 RF It is good also as an evaluation value Y by normalizing by. In this case, even if the absolute value IWI of each DCT coefficient is large in the coding apparatus 1, the evaluation value Y of the area may be increased to prevent false detection.
- the standardized value can be made uniform as the evaluation value Y. '
- the encoding device 10 detects local minimum values, for example, using the remainder r as the evaluation value Y as it is, and when multiple local minimum values are detected, the period of the local minimum in the quantization parameter QP: is 6 It is also possible to detect a large quantization parameter Q.sub.Q as the quantization parameter Q.sub.P used in the previous encoding. .
- encoder 10 rescales residual r only when the absolute value IWI of each DCT coefficient is equal to or less than the normalization threshold (ie,
- the evaluation value Y is calculated by normalizing with factor 1 RF, and when the absolute value IWI of the DCT coefficient is larger than the normalization threshold, the remainder r is set as the evaluation value Y. Then, the encoding device 10 is configured to detect the quantization parameter QP based on the resizing factor RF with which the evaluation value Y is minimum.
- the QP detection unit 32 receives the absolute value IWI of the DCT coefficient W calculated by the integer precision DCT unit 22 and uses the quantization matrix Qmatrix detected by to calculate various quantization parameters in macro block units. Calculate the rescaling factor one RF for each QP. At this point, the QP detection unit 32 knows the approximate value of the quantization parameter QP used last time: If it is, the calculation is performed by reducing the amount of detection by targeting only the QPs around that. be able to. When the QP detection unit 32 obtains an initial value from the quantization parameter QP that may have been used in the previous encoding (step S 21), all the DCTs in the macro block are calculated. It is determined whether the absolute value I WI of the coefficient is 0 (step S22).
- the QP detection unit 32 makes the quantization parameter QP undetectable when the absolute value [W “is 0. of all the DCT coefficients (step S 23). Immediate QP QP If the absolute value IWI of all D. CT. Coefficients in the macro block is 0, the detection unit 32 determines the quantization parameter QP by dividing by any quantization parameter QP, so that r is 0 ⁇ .
- Step S22 when the absolute value IWI of all the DCT coefficients is not 0, the remainder calculation unit 32a of the QP detection unit 32 Calculate the remainder r divided by the rescaling factor RF calculated according to the equation (9) for the DCT coefficient (
- ⁇ 6) shifted for each sample of 256 ( 16xl6) in the macro block. Do (Step S 2 4) o
- the evaluation value determination unit 32 b determines whether (
- ⁇ 6) be the evaluation value Y (step S27). That is, the evaluation value determination unit 32b performs the normalization and correction described in (iii) on the remainder obtained by dividing the shifted DCT coefficient (W ⁇ 6) by RF for 256 ( 16xl6) samples in the macro block. The evaluation value Y is set as the evaluation value Y, and the sum ( ⁇ ) is calculated for each quantization parameter QP (step S 28).
- the evaluation value determination unit 32b determines whether or not the evaluation value Y is calculated with respect to all possible quantization parameters. QP (step S29), and all quantization parameters are determined. If the evaluation value Y is not calculated for the QP, the quantization parameter QP is changed (step S30), and the processes of steps S26, S28, and S29 are repeated ⁇ To r-'-
- the QP deciding unit 32 c determines, in macro block units, each QP.
- the sum ( ⁇ ⁇ ⁇ ) of evaluation values Y is compared (step S 31), and the quantization parameter QP for which the sum becomes the smallest is taken as the quantization parameter QP used last time (step S 32). It will output the value of QP.
- the region where the quantization parameter QP is less than 6 is excluded because it increases the possibility of false detection. That is because, for example, in the practical area of 200 Mbps or less, the quantum parameter less than 6 is rarely used, and the detection rate is higher if it is excluded.
- the coding apparatus 10 is used in the previous encoding of DCT coefficients obtained in the process of encoding by the Qmatrix detection unit 3.1.
- Qmatrix can be detected with high probability for each slice using remainder r.
- the encoding device 10 can detect the QP used in the previous encoding from the DCT coefficients obtained in the encoding process by the QP detection unit 32 with high probability using the remainder r for each macroblock. .
- the encoding device 10 By reusing the quantization matrix Qmatrix and the quantization parameter QP that have been detected, the rounding error during dubbing can be reduced, and the dubbing characteristics can be improved.
- the encoding device 10 can treat the DCT coefficients in encoding and decoding equally. Furthermore, the coding device 10 can improve detection accuracy by weighting the remainder r with the DCT coefficient W and normalizing it with the rescheduling “ring factor one RF”.
- the coding device 10 takes into account the periodicity of 6n of the quantization parameter QP introduced from AVC, and the minimum value of the remainder r is a minimum value of 6n greater than the quantization parameter QP. If there is a value, let the quantization parameter QP be the quantization parameter QP used in the previous encoding, even if it is not the minimum value. And the encoding device
- the intra prediction mode determining unit 21 with offset of the encoding device 10 starts offset offset application processing procedure RT3 when the input image data is supplied, and proceeds to step S51. :. ''
- step S 51 the Intra prediction mode deciding unit with offset 2: 1 sets the offset [n] according to the mode number n for each image prediction mode, and the next step S 52 Move to
- step S 52 the Intra prediction mode determination unit with offset 21 1 generates difference image data corresponding to each mode number n in each image prediction mode, and also generates a difference absolute value sum SAD from the difference image data.
- Each step is calculated, then the process goes to the next step S53.
- step S 53 the Intra prediction mode determination unit with offset 2 1 performs an addition process. If the mode number n of the target intra prediction mode is set to “0” and the minimum difference absolute value sum minSAD is set to the integer maximum value INT_MAX assumed in calculation, the process proceeds to the next step S 54.
- step S54 the Intra prediction mode determination unit with offset 21 determines whether or not the mode number n of the current addition processing target is "2". If a negative result is obtained here, this indicates that it is necessary to add an offset to the absolute difference sum SAD, and in this case, the Intra prediction mode determination unit with offset 21 1 Move on to next step S55.
- step S55 the intra prediction mode determination unit with offset 21 adds the offset [n] set in step S51 to the difference absolute value sum SAD, and the offset difference absolute value sum SAD f is added. Is calculated, the next step S 56 is performed.
- step S54 if an affirmative result is obtained in step S54, this means that the difference absolute value sum SAD should be offset as it is, such as with the offset absolute value sum SAD.
- the intra prediction mode determination unit 21 moves to the next step S 56.
- step S 56 the Intra prediction mode decision unit with offset 2 1 determines whether the mode number of the current addition processing target is n or not. ⁇ Here, the result is good. If this is the case, this means that there is still an intra-screen prediction mode to be processed, and at this time, the Intra prediction mode determination unit with offset-2 1 moves to the next step S 57 and adds processing If “1” is added to the mode number n of interest, the process returns to step S 54 to continue the process. .
- step S 56 this is the sum of offset absolute values with offset S.
- AD f corresponding to all intra prediction modes from n 2 0 to 8
- the Intra prediction mode determination unit 21 with offset is moved to the next step S 59.
- step S 59 when the intra prediction mode determination unit with offset 21 sets the mode number n to be compared to “0”, the process proceeds to the next step S 59.
- step S59 the intra prediction mode determination unit with offset 21 determines whether the sum of absolute differences with offsets SAD f to be subjected to comparison processing is less than the minimum sum of absolute differences with differences minSAD.
- the Intra prediction mode determination unit with offset 21 moves to the next step S 61 and the minimum absolute difference sum minSAD is compared with the offset absolute difference sum SAD to be compared. Update to the value of f and move on to the next step S 6 2 ,.
- step S60 the Intra prediction mode decision unit with offset 21 does not need to update the minimum difference absolute value sum minSAD, so it proceeds to the next step S62. .
- step S62 the Intra prediction mode determination unit with offset 21 determines whether or not the mode number n of the comparison processing target is "8". If a negative result is obtained here, this means that the intra prediction mode to be processed still remains. At this time, the Intra prediction mode determination unit with offset 21 1 performs the following: Step S 6 3 : If the mode number n to be compared is incremented by one, the process returns to step S6.0 to continue the process.
- the encoding device 10 has a 9 'screen previously determined by the AVC standard based on the differential image data consisting of the difference values from the peripheral pixels of the processing target pixel to be processed. Generate in inner prediction mode. Then, the encoding device 10 calculates the difference absolute value sum SAD as the difference sum value based on the sum of the absolute values of the respective processing target pixels in the difference image data, and the offset [n] preset in the in-screen prediction mode.
- the in-screen prediction mode is prioritized according to the priority according to.
- the encoder 10 has a low value of offset [n] and a high priority specific screen.
- the intra prediction modes can be prioritized to high priority. .
- the encoding apparatus 10 adds an offset (offset [n]) according to the intra prediction mode to the difference absolute value sum SAD for a plurality of difference image data, and adds the offset difference absolute value sum SAD. Generate f Then, the encoding device 10 ranks the intra prediction modes by comparing the offset sum SAD with offsets.
- the coding apparatus 10 can easily set the priority of the ranking according to the intra prediction mode. Further, the encoding apparatus 10 can suppress an increase in the generated code amount by adding offset [n] to the last, instead of using a specific intra prediction mode in a fixed manner.
- the encoding device 10 determines the intra prediction mode when the differential image data with offset absolute difference sum SAf takes a minimum value as the intra prediction mode for encoding the differential image data when the differential image data is generated. It was made to do.
- the encoding apparatus 10 can easily determine the particular intra prediction mode as the intra prediction mode for encoding the differential image data, so that the screen used in the previous encoding.
- the coding device can improve the probability of using the intra prediction mode same as that of the previous encoding, the image quality during dubbing can be improved. It is made to be able to
- the intra-frame encoding apparatus 10 encodes the differential image data by not adding an offset to one intra-frame prediction mode with little deterioration in image quality.
- the prediction mode it is easy to select the intra prediction mode with less degradation of image quality, and it is possible to suppress the degradation of image quality at the time of decoding.
- One intra prediction mode in which the image quality degradation is small is a DC prediction mode in which the difference value from the average value of peripheral pixels is taken, and the difference value from the average value of peripheral pixels in the vertical and parallel directions in the AVC Standard is calculated. It is mode 2 to take.
- the coding apparatus 10 can suppress the image quality deterioration at the time of coding by using many DC prediction modes with the smallest image quality deterioration.
- offset [n] according to the degree of deterioration of the image quality in the in-screen prediction mode
- the encoding apparatus 10 ranks in such a way that the priority of the in-screen prediction mode becomes higher as the deterioration of the image quality decreases. It can be attached.
- the encoding device 10 detects the quantization factor that is the basis of the quantization step used in the previous encoding from the difference image data in the intra-frame prediction mode, thereby reducing the image quality during dubbing. Can be suppressed.
- the encoding apparatus 10 weighs the amount of generated code and the degradation of the image quality, and offset [according to the application and purpose] n] can be set to meet the user's request.
- the encoding device 10 compares the absolute difference value sum SAD in each intra-frame prediction mode, and prioritizes a specific intra-frame prediction mode with a higher priority, thereby a specific screen
- the intra prediction mode can be ranked high each time.
- the encoding apparatus 10 can easily rank the intra-frame prediction mode used at the time of the previous encoding in a highly selective manner and select it as the intra-frame prediction mode of the differential image data used for encoding.
- the present invention can realize an image processing apparatus and an image processing method that can suppress the degradation of image quality. '.
- intra prediction mode detection process only the input image data is used, and the intra prediction mode used in the previous encoding is arithmetically calculated using the remainder r. It is characterized by detecting. .
- the coding apparatus 11 uses the quantization matrix Qmatrix and the quantization parameter QP to display in the screen for each input coding unit IA.
- the prediction mode is changed and the minimum value is detected using the sum ( ⁇ r) of the remainder r. This makes it possible for the coding device 11 to detect the intra-frame prediction mode encoded last time for all intra-coding units IA in the macroblock MB ′.
- the encoder '11 ranks the offset difference absolute value sum SAD by offset priority application processing, and the difference absolute value sum SAD with offset is three intra-frame prediction modes from the smaller side.
- the prediction mode detection process is executed.
- the encoding device 11 can easily select the intra-frame prediction mode encoded last time. Therefore, the intra-frame prediction mode encoded last time is selected as the three intra-frame prediction modes to be processed. It is made to be able to include surely.
- the quantization matrix and quantization parameter used in the previous encoding are based on the concept described above in the quantization matrix detection process and the quantization parameter detection process. If is known, by dividing the DCT coefficients of each of the nine intra prediction modes by 6 bit shifts (W ⁇ 6) by the resizing factorer RF, the remainder r is minimized. Detect the in-screen prediction mode. .
- the image processing apparatus 1 takes into consideration the following points due to the unique characteristics of AVC Intra, which were not considered in MPG. That is, in this in-screen prediction command detection processing, in addition to (i) and (ii) described above in the quantization matrix detection processing and the quantization parameter detection processing, the following (vii) Consider.
- the intra prediction mode changes, the difference value also changes, and as a result, the distribution of DCT coefficients also changes. Therefore, it is generally considered fair to divide the sum of residues r in a sub-macroblock by the number of non-zero DCT coefficients and compare with the residue per sample.
- the prediction mode used in the previous encoding is chosen, then the prediction is high.
- the number of non-zero DCT coefficients tends to decrease, and the number of non-zero DCT coefficients tends to increase as the mode is shifted in direction from the previous prediction.
- the sum of the remainder r is not divided by the number of non-zero DCT coefficients, but multiplied by the number of non-zero DCT coefficients.
- the sum of remainder r and the number of non-zero DCT coefficients are multiplied by the evaluation value Y, and this evaluation value Y is compared between each prediction mode, and the prediction mode with the smallest evaluation value Y is the previous encoding. Think of it as a prediction mode.
- an optimal prediction mode is determined based on the characteristic viewpoint (vii) above.
- the point having the intra prediction mode detection unit 33 and the intra prediction mode determination unit 45, the intra prediction processing unit 51 and the integer precision DCT unit 52 are different from the encoder 10 in that they have. The following description focuses on the differences.
- the previous quantization parameter QP detected by the Q matrix detection unit 32 and the quantization matrix Qmatrix are sent to the Qmatrix / QP determination unit 41.
- the Qmatrix / QP decision section 41 with regard to the quantization matrix Qmatrix and the quantization parameter QP, whether the one estimated in the first-pass pre-encoder 20 is adopted or not? 3 Decide whether to use the one detected in 0.
- the back search fails because there is no previous encoding, and the encoding process by the pre-encoder 20 of the first pass is performed. Adopt what was estimated.
- the one detected by the back search unit 3 0 is employed in order to avoid repeated quantization distortion.
- the quantization parameter selector QP and the quantization matrix Qmatrix used for the previous encoding detected by the back search unit 30 are also sent to the intra prediction mode detection unit 33.
- the in-screen prediction mode detection unit 33 detects the previous in-screen prediction mode by the above-described in-screen prediction mode detection process by using the quantization parameter QP and the quantization matrix Qmatrix. The details of the in-screen prediction mode detection process procedure by the in-screen prediction mode detection unit 33 will be described later.
- This previous prediction mode is sent to the in-screen prediction mode determination unit 45. In the in-screen prediction mode determination unit 45,
- the in-screen prediction mode detection unit 33 detects the previous prediction mode on the basis of the quantization parameter Q.Q. and the quantization matrix Qmatrix. This previous prediction mode is sent to the in-screen prediction mode determination unit 45. In the in-screen prediction mode determination unit 45,
- the intra prediction mode detection unit 33 determines whether the input image data is an original image that has never been encoded. That is, if the input image data is an original image that has never been encoded, detection of the intra prediction mode (backtracking) fails because there is no previous encoding, and the Intra prediction mode with offset The one estimated by the decision unit 2 1 is adopted. Conversely, if the input image is a non-original image that has already been encoded, in order to avoid repeated quantization distortion, the signal detected by the intra prediction mode detection unit 33 is used. Do.
- the quantization matrix Qmatrix, the quantization parameter QP, the in-screen prediction mode determination unit 45 determined by the Qmatrix / QP determination unit 41 as described above Encode the input image using the estimated intra prediction mode, Output as a coded stream (Stream).
- the in-screen prediction mode determined by the in-screen prediction mode determining unit 45 is selected as the optimal in-screen prediction mode, and the prediction image of peripheral pixels is used.
- Differential image data is generated from the input image, and the differential image data is input to the integer precision DCT unit 52.
- the integer precision DCT unit 52 outputs the difference image data as a DCT coefficient W by orthogonal transformation using discrete cosine transformation.
- the quantization unit 53 receives the DCT coefficient W and performs quantization using the quantization matrix Qmatrix s quantization parameter QP determined by the Qmatrix / QP determination unit 41, and the quantization level Z is entropic.
- the input quantization level Z is binarized and arithmetically coded and outputted as a coding stream.
- the intra-frame prediction mode according to the second embodiment of the present invention will be described with reference to the flowchart in FIG.
- the process of prediction mode detection by the detection unit 33 will be further described. This corresponds to the image information coding method according to the embodiment.
- the in-screen prediction mode detection unit 33 sets the in-screen prediction mode with the smallest difference absolute value sum of the predicted image (ie, the sum of absolute values of difference image data) as the initial value;
- Step S 1 5 the in-screen prediction mode detection unit 33 performs DCT processing on the difference image data from the predicted image (step S 15 2).
- Intra-frame prediction mode detector 33 shifts the absolute value IW I. of the DCT coefficient W by 6 bits for each sample in the 4 ⁇ 4 sub-mac mouth block and divides it by (IW
- the in-screen prediction mode detection unit 33 sets an evaluation value Y as the sum of the remainder r and the even number of the DCT coefficient (step S 1 56).
- the in-screen prediction mode detection unit 33 determines whether Y has been calculated for all the in-screen prediction modes that may have no screen (step S 1 5 7), and the evaluation is performed. If the value Y has not been calculated, the mode is switched to the prediction mode in which the sum of absolute differences SAD is smaller next (step S158), and the processing of steps S52 to S57 is repeated.
- step S 157 branches to Yes
- the evaluation value Y between the in-screen prediction modes is determined. Are compared (step S159), the prediction mode with the smallest evaluation value Y is set as the prediction mode used last time (step S160), and the process is ended.
- the encoding apparatus 10 detects AVC by arithmetically detecting the intra prediction mode used in the previous encoding from the input image. It is possible to suppress the occurrence of repeated quantization distortion during dubbing of an intra frame. As a result, the encoding device 10 can improve the deterioration of the SN ratio and the visual characteristics during dubbing.
- the encoding device 10 when comparing the difference absolute value sum SAD from the predicted image in the sub-track between the intra-frame prediction modes, the encoding device 10 sequentially starts from the small difference material sum S. SAD.
- This comparison is equivalent to comparing from the high-prediction intra-frame prediction mode used in the previous encoding.
- the intra-prediction prediction mode with a small sum of absolute differences SAD It is possible to reduce the amount of computation by comparing only a predetermined number (for example, three) of modes in order. It is possible to treat DCT coefficients in conversion and decoding equally.
- the encoding apparatus 10 can detect regardless of the algorithm of intra prediction of the previous encoding, the intra prediction method in consideration of the visual characteristics in the first encoding is used: An algorithm is available.
- the encoding device 1 1 uses the difference image data for which the absolute value sum with offset SAD is minimum as the quantization factor to be the basis of the quantization step used in the previous encoding.
- the encoding device 1 1 is a target number of three differences in the ascending order of the offset difference absolute value sum by the division factor based on the quantization parameter QP and the quantization matrix Qmatrix detected by the back search. Calculate the remainder value r when dividing the image data.
- the encoding device 11 determines the intra prediction mode in which the differential image data for which the remainder value r is minimum is generated as the intra prediction mode for the differential image data.
- the encoding device 11 can reliably detect the intra prediction mode used in the previous encoding, it is possible to further suppress the image quality degradation during dubbing.
- the encoding device 11 1 prioritizes a specific in-screen prediction mode by offset priority application processing to be a target of the in-screen prediction mode detection processing.
- the three intra prediction modes can almost certainly be included in the intra prediction mode used at the previous encoding. .
- the coding device 11 can almost certainly detect the intra prediction mode used at the time of the previous encoding by the intra prediction mode detection process.
- the present invention can realize an image processing apparatus and an image processing method capable of suppressing a decrease in image quality at the time of encoding. (4) Other embodiments
- the Intra prediction mode determination unit with offset determines the value of offset [n] for each prediction mode (step S 2 0 1).
- n means a prediction mode number (0 to 8 (except 2)) to be compared with the DC prediction.
- X means a prediction mode number adopted in coding.
- the Intra prediction mode determination unit with offset is an intra prediction mode to be compared. It is determined whether the sum of absolute difference sums SAD and offset [n] is smaller than the sum of absolute difference values SAD in the DC prediction mode (step S 2 0 4).
- the Intra prediction mode determination unit with offset is the sum of difference absolute value sum with offset SAD f that is the sum of difference absolute value sum SAD in the in-screen prediction mode to be compared and offset [n]. If it is determined that is smaller than the sum of offsets with absolute value SAD f of mode number X (branch to step S 2 0 4 Yes), the intra prediction mode within the screen to be compared against X is selected. Substitute the code number n, and move to step S 2 0 5.
- the offset difference absolute value sum SAD f which is the addition value of the difference absolute value sum SAD of the prediction mode to be compared and the offset] is more than the offset absolute value sum SAD f of the mode number X If it is judged not to be small (the step S2 04 is branched to No), the Intra prediction mode deciding unit with offset judges whether or not comparison has been made with respect to all possible prediction modes ( Step S 2 0 5), Intra offset added Intra prediction mode determination unit updates the value of n (step S 2 0 5 branches to No) if all the determinations have not been completed (step S 2 0 5) S 2 0 6) Perform the comparison of step S 5 for the next candidate mode.
- step S 2 05 If it is determined in step S 2 05 that comparison has been made with respect to all possible prediction modes to be adopted (step S 2 0 5 is branched to “Yes”), the Intra prediction mode determination unit with offset is Mode X The prediction mode to be adopted in encoding (step S 2 0 7), ends the series of processing.
- the encoding apparatus selects: an intra prediction mode in which the sum of absolute differences SAD is minimum, and the sum of absolute differences SAD in the selected intra prediction mode and one of the plurality of intra prediction modes. Comparison with absolute value sum SAD in prediction mode (DC prediction mode). Then, when the difference between the absolute value sum SAD is less than the threshold value, the encoding apparatus determines the differential image data corresponding to the differential absolute value sum priority prediction mode as the differential image data to be actually encoded. good.
- the difference absolute value sum SAD He described the case where it was decided to prioritize.
- the present invention is not limited to this, and for example, it may be used based on SATD (Sum of Absolute Transiormed Diiierence) or S SD,.. .Sum 01 Square Difference, sum of squared error), which has been Hadamard transformed to SAD. It may be possible to rank the SATD (Sum of Absolute Transiormed Diiierence) or S SD,.. .Sum 01 Square Difference, sum of squared error), which has been Hadamard transformed to SAD. It may be possible to rank the SATD (Sum of Absolute Transiormed Diiierence) or S SD,.. .Sum 01 Square Difference, sum of squared error), which has been Hadamard transformed to SAD. It may be possible to rank the SATD (Sum of Absolute Transiormed Diiierence) or S SD,.. .Sum 01 Square Difference, sum of squared error), which has been Hadamard transformed to SAD. It
- offsettn] is not added only to the DC prediction mode.
- the present invention is not limited to this, and the offset] may not be added to other in-screen prediction modes. Further, according to the present invention, offset [n] may not be added to two or more in-screen prediction modes. .
- offsettn is added according to the degree of deterioration of the image quality.
- the present invention is not limited to this.
- offset [n] may be added according to the generation degree of the generated code amount.
- an index for adding offset [n] is not necessarily required.
- the encoding device performs both the backtracking process and the offset priority application process.
- the present invention is not limited to this, and only the offset priority application process may be executed.
- the integer precision D C T unit 2 2 supplies the D C T coefficient to the quantization parameter calculation unit (not shown).
- the quantization parameter calculation unit determines an appropriate quantization parameter QP according to the size of the DCT coefficient W and the distribution state of the frequency range, supplies this to the quantization unit 53, and performs entropy coding Output as a coded stream via part 54.
- the case has been described in which the encoding device executes the in-screen prediction mode detection process for the three in-screen prediction modes as the number of processes.
- the present invention is not limited to this, and the number of processes is not particularly limited.
- the case has been described where the sum of absolute differences SAD is calculated for each intra-coding unit consisting of submacroblocks of 4 ⁇ 4 pixels.
- the present invention is not limited to this, and may be, for example, 8 ⁇ 8 pixels or 16 ⁇ 8 pixels, and there is no limitation on the intra coding unit.
- the present invention is not limited to this, and the present invention can be applied to all methods for selecting one intra prediction mode out of a plurality of intra prediction modes.
- the coding apparatus as an image processing apparatus by the differential image data generation unit and the Intra prediction mode determination unit with offset 21 as a ranking unit 10 And 11 described the case where 1 was to be configured.
- the present invention is not limited to this, and the image processing apparatus of the present invention may be configured by a differential image data generation unit and a ranking unit having other various configurations.
- the image information encoding apparatus and method according to the first and second embodiments described above can also be implemented as a program executed by a conduit and a recording medium storing the program. It is a matter of course that f is also the case, in which case the above-mentioned effects can be achieved. .
Landscapes
- Engineering & Computer Science (AREA)
- Multimedia (AREA)
- Signal Processing (AREA)
- Physics & Mathematics (AREA)
- Discrete Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Compression Or Coding Systems Of Tv Signals (AREA)
Abstract
Description
Claims
Priority Applications (4)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
EP08830296A EP2187648A4 (en) | 2007-09-12 | 2008-09-12 | IMAGE PROCESSING DEVICE AND IMAGE PROCESSING METHOD |
CN2008800081548A CN101641958B (zh) | 2007-09-12 | 2008-09-12 | 图像处理设备和图像处理方法 |
US12/530,872 US8369403B2 (en) | 2007-09-12 | 2008-09-12 | Image processing device and image processing method |
JP2009532267A JP5212372B2 (ja) | 2007-09-12 | 2008-09-12 | 画像処理装置及び画像処理方法 |
Applications Claiming Priority (4)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
JP2007-236945 | 2007-09-12 | ||
JP2007236945 | 2007-09-12 | ||
JP2007-236943 | 2007-09-12 | ||
JP2007236943 | 2007-09-12 |
Publications (1)
Publication Number | Publication Date |
---|---|
WO2009035144A1 true WO2009035144A1 (ja) | 2009-03-19 |
Family
ID=40452144
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/JP2008/066911 WO2009035144A1 (ja) | 2007-09-12 | 2008-09-12 | 画像処理装置及び画像処理方法 |
Country Status (7)
Country | Link |
---|---|
US (1) | US8369403B2 (ja) |
EP (1) | EP2187648A4 (ja) |
JP (1) | JP5212372B2 (ja) |
KR (1) | KR20100065240A (ja) |
CN (1) | CN101641958B (ja) |
TW (1) | TWI392372B (ja) |
WO (1) | WO2009035144A1 (ja) |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2010146772A1 (ja) * | 2009-06-19 | 2010-12-23 | 三菱電機株式会社 | 画像符号化装置、画像復号装置、画像符号化方法及び画像復号方法 |
EP2282545A1 (en) | 2009-07-27 | 2011-02-09 | Sony Corporation | Image encoding device and image encoding method |
WO2015033510A1 (ja) * | 2013-09-09 | 2015-03-12 | 日本電気株式会社 | 映像符号化装置、映像符号化方法及びプログラム |
JP7540082B2 (ja) | 2021-06-30 | 2024-08-26 | テンセント・アメリカ・エルエルシー | オフセット・ベースのリファインメント及び複数参照ライン選択の調和的設計 |
Families Citing this family (15)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP5158003B2 (ja) * | 2009-04-14 | 2013-03-06 | ソニー株式会社 | 画像符号化装置と画像符号化方法およびコンピュータ・プログラム |
TWI395490B (zh) * | 2010-05-10 | 2013-05-01 | Univ Nat Central | 實施於電子裝置之影像編碼方法 |
TWI403175B (zh) * | 2010-06-11 | 2013-07-21 | Ind Tech Res Inst | 畫面內預測模式最佳化方法與影像壓縮方法以及裝置 |
MY173604A (en) * | 2011-03-11 | 2020-02-10 | Sony Corp | Image processing apparatus and method |
MX2013013909A (es) | 2011-06-13 | 2014-01-24 | Panasonic Corp | Metodo de decodificacion de imagenes, metodo de codificacion de imagenes, aparato de decodificacion de imagenes, aparato de codificacion de imagenes y aparato de codificacion y decodificacion de imágenes. |
KR101953522B1 (ko) * | 2011-06-17 | 2019-02-28 | 가부시키가이샤 제이브이씨 켄우드 | 화상 부호화 장치, 화상 부호화 방법 및 화상 부호화 프로그램, 및 화상 복호 장치, 화상 복호 방법 및 화상 복호 프로그램 |
KR20130050406A (ko) * | 2011-11-07 | 2013-05-16 | 오수미 | 머지 모드에서의 움직임 정보 생성 방법 |
KR20130050407A (ko) * | 2011-11-07 | 2013-05-16 | 오수미 | 인터 모드에서의 움직임 정보 생성 방법 |
KR101621854B1 (ko) | 2014-02-21 | 2016-05-17 | 연세대학교 산학협력단 | Tsm 율-왜곡 최적화 방법, 그를 이용한 인코딩 방법 및 장치, 그리고 영상 처리 장치 |
JP2015216626A (ja) * | 2014-04-23 | 2015-12-03 | ソニー株式会社 | 画像処理装置及び画像処理方法 |
KR20170058837A (ko) * | 2015-11-19 | 2017-05-29 | 한국전자통신연구원 | 화면내 예측모드 부호화/복호화 방법 및 장치 |
WO2017086746A1 (ko) | 2015-11-19 | 2017-05-26 | 한국전자통신연구원 | 화면내 예측 모드 부호화/복호화 방법 및 장치 |
EP3324628B1 (en) * | 2016-11-18 | 2021-12-29 | Axis AB | Method and encoder system for encoding video |
CN108804476B (zh) * | 2017-05-05 | 2021-05-25 | 北京京东尚科信息技术有限公司 | 图像搜索结果的排序方法、装置、电子设备和存储介质 |
US11997275B2 (en) * | 2018-08-27 | 2024-05-28 | AT Technologies ULC | Benefit-based bitrate distribution for video encoding |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JPH09294263A (ja) * | 1996-04-24 | 1997-11-11 | Victor Co Of Japan Ltd | 画像情報圧縮装置 |
JP2002290914A (ja) * | 2001-03-26 | 2002-10-04 | Sony Corp | 記録再生装置及びそのバックサーチ方法 |
JP2005192232A (ja) * | 2002-05-28 | 2005-07-14 | Sharp Corp | 画像符号化システム、画像復号システム、及び画像を分割したブロックの画素値の予測方法 |
JP2006148419A (ja) * | 2004-11-18 | 2006-06-08 | Nippon Telegr & Teleph Corp <Ntt> | 画像符号化装置,画像符号化方法,画像符号化プログラムおよびコンピュータ読み取り可能な記録媒体 |
JP2006270437A (ja) * | 2005-03-23 | 2006-10-05 | Toshiba Corp | ビデオエンコーダ及びこれを用いた携帯無線端末装置 |
JP2007067469A (ja) * | 2005-08-29 | 2007-03-15 | Nippon Telegr & Teleph Corp <Ntt> | フレーム内予測符号化制御方法、フレーム内予測符号化制御装置、フレーム内予測符号化制御プログラムおよびそのプログラムを記録した記録媒体 |
Family Cites Families (48)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US4979498A (en) | 1989-10-30 | 1990-12-25 | Machida Incorporated | Video cervicoscope system |
JPH04114585A (ja) | 1990-09-04 | 1992-04-15 | Matsushita Electric Ind Co Ltd | 動き補償フレーム間符号化装置 |
US5428396A (en) | 1991-08-03 | 1995-06-27 | Sony Corporation | Variable length coding/decoding method for motion vectors |
US5565921A (en) | 1993-03-16 | 1996-10-15 | Olympus Optical Co., Ltd. | Motion-adaptive image signal processing system |
US5703646A (en) | 1993-04-09 | 1997-12-30 | Sony Corporation | Picture encoding method, picture encoding apparatus and picture recording medium |
JP3264043B2 (ja) | 1993-04-30 | 2002-03-11 | ソニー株式会社 | 量子化コントロール装置および方法 |
EP0671816B1 (en) | 1993-09-28 | 2000-03-29 | Sony Corporation | Encoding/decoding device with all odd or all even value rounding |
US5508942A (en) | 1993-11-24 | 1996-04-16 | Intel Corporation | Intra/inter decision rules for encoding and decoding video signals |
JP3561962B2 (ja) | 1994-07-25 | 2004-09-08 | ソニー株式会社 | 量子化コントロール装置および方法 |
JP2882287B2 (ja) | 1994-08-22 | 1999-04-12 | 日本電気株式会社 | 動画像符号化装置 |
JP3890606B2 (ja) | 1995-03-09 | 2007-03-07 | ソニー株式会社 | 画像信号符号化装置及び画像信号符号化方法 |
US6100931A (en) | 1996-03-19 | 2000-08-08 | Sony Corporation | Method and apparatus for controlling a target amount of code and for compressing video data |
US6101276A (en) | 1996-06-21 | 2000-08-08 | Compaq Computer Corporation | Method and apparatus for performing two pass quality video compression through pipelining and buffer management |
US5805228A (en) | 1996-08-09 | 1998-09-08 | U.S. Robotics Access Corp. | Video encoder/decoder system |
US6337879B1 (en) | 1996-08-14 | 2002-01-08 | Sony Corporation | Video data compression apparatus and method of same |
JP3864461B2 (ja) | 1996-08-30 | 2006-12-27 | ソニー株式会社 | 映像データ圧縮装置およびその方法 |
JPH1079948A (ja) | 1996-09-03 | 1998-03-24 | Mitsubishi Electric Corp | 画像符号化装置 |
KR100599017B1 (ko) | 1996-12-12 | 2006-12-13 | 소니 가부시끼 가이샤 | 영상 데이터 압축 장치 및 그 방법 |
JP4114210B2 (ja) | 1996-12-12 | 2008-07-09 | ソニー株式会社 | 映像データ圧縮装置およびその方法 |
CA2264834C (en) | 1997-07-08 | 2006-11-07 | Sony Corporation | Video data encoder, video data encoding method, video data transmitter, and video data recording medium |
KR100563756B1 (ko) | 1997-11-05 | 2006-03-24 | 소니 가부시끼 가이샤 | 디지탈 신호 변환 방법 및 디지탈 신호 변환 장치 |
US6393060B1 (en) | 1997-12-31 | 2002-05-21 | Lg Electronics Inc. | Video coding and decoding method and its apparatus |
JP4511649B2 (ja) | 1999-02-03 | 2010-07-28 | 三菱電機株式会社 | 画像符号化方式および画像符号化装置 |
JP2001169281A (ja) | 1999-12-13 | 2001-06-22 | Matsushita Electric Ind Co Ltd | 動画像符号化装置、および動画像符号化方法 |
WO2001058166A1 (en) * | 2000-02-01 | 2001-08-09 | Koninklijke Philips Electronics N.V. | Video encoding with a two step motion estimation for p-frames |
JP2002010259A (ja) | 2000-06-21 | 2002-01-11 | Mitsubishi Electric Corp | 画像符号化装置及び画像符号化方法及び画像符号化プログラムを記録した記録媒体 |
JP2002359853A (ja) | 2001-03-26 | 2002-12-13 | Sony Corp | 画像処理装置、画像処理方法、画像処理プログラムおよび記録媒体 |
EP1377070A4 (en) | 2001-03-28 | 2010-08-11 | Sony Corp | QUANTIZATION DEVICE, QUANTIZATION PROCESS, QUANTIZATION PROGRAM AND RECORDING MEDIUM |
JP3846424B2 (ja) | 2001-03-30 | 2006-11-15 | ソニー株式会社 | 画像信号量子化装置およびその方法 |
US7099951B2 (en) * | 2001-05-24 | 2006-08-29 | Vixs, Inc. | Method and apparatus for multimedia system |
US6934330B2 (en) * | 2001-08-01 | 2005-08-23 | Sony Corporation | Image processing apparatus |
JP4135427B2 (ja) * | 2001-08-01 | 2008-08-20 | ソニー株式会社 | 画像処理装置および方法、ならびに、画像処理プログラム |
EP1827026A1 (en) | 2002-01-18 | 2007-08-29 | Kabushiki Kaisha Toshiba | Video decoding method and apparatus |
US7236524B2 (en) | 2002-05-28 | 2007-06-26 | Sharp Laboratories Of America, Inc. | Methods and systems for image intra-prediction mode communication |
US7289672B2 (en) | 2002-05-28 | 2007-10-30 | Sharp Laboratories Of America, Inc. | Methods and systems for image intra-prediction mode estimation |
US7386048B2 (en) | 2002-05-28 | 2008-06-10 | Sharp Laboratories Of America, Inc. | Methods and systems for image intra-prediction mode organization |
US7715477B2 (en) | 2002-05-29 | 2010-05-11 | Diego Garrido | Classifying image areas of a video signal |
US7054497B2 (en) | 2002-06-14 | 2006-05-30 | Koninklijke Philips Electronics N.V. | Method and system for optimizing image sharpness during coding and image enhancement |
US6944224B2 (en) | 2002-08-14 | 2005-09-13 | Intervideo, Inc. | Systems and methods for selecting a macroblock mode in a video encoder |
JP4114585B2 (ja) | 2002-09-27 | 2008-07-09 | 東洋インキ製造株式会社 | ガスバリア性積層体の製造方法 |
CN1225127C (zh) * | 2003-09-12 | 2005-10-26 | 中国科学院计算技术研究所 | 一种用于视频编码的编码端/解码端双向预测方法 |
JP4127818B2 (ja) | 2003-12-24 | 2008-07-30 | 株式会社東芝 | 動画像符号化方法及びその装置 |
JP3846489B2 (ja) | 2004-05-10 | 2006-11-15 | セイコーエプソン株式会社 | 画像データ圧縮装置、エンコーダ、電子機器及び画像データ圧縮方法 |
JP2006222555A (ja) | 2005-02-08 | 2006-08-24 | Matsushita Electric Ind Co Ltd | 符号化装置及び符号化方法 |
KR101088375B1 (ko) | 2005-07-21 | 2011-12-01 | 삼성전자주식회사 | 가변 블록 변환 장치 및 방법 및 이를 이용한 영상부호화/복호화 장치 및 방법 |
AU2007244443A1 (en) | 2006-04-28 | 2007-11-08 | Ntt Docomo, Inc. | Image predictive coding device, image predictive coding method, image predictive coding program, image predictive decoding device, image predictive decoding method and image predictive decoding program |
US8184712B2 (en) | 2006-04-30 | 2012-05-22 | Hewlett-Packard Development Company, L.P. | Robust and efficient compression/decompression providing for adjustable division of computational complexity between encoding/compression and decoding/decompression |
US7756348B2 (en) | 2006-10-30 | 2010-07-13 | Hewlett-Packard Development Company, L.P. | Method for decomposing a video sequence frame |
-
2008
- 2008-09-12 EP EP08830296A patent/EP2187648A4/en not_active Withdrawn
- 2008-09-12 CN CN2008800081548A patent/CN101641958B/zh not_active Expired - Fee Related
- 2008-09-12 JP JP2009532267A patent/JP5212372B2/ja not_active Expired - Fee Related
- 2008-09-12 TW TW97135177A patent/TWI392372B/zh not_active IP Right Cessation
- 2008-09-12 KR KR20097019021A patent/KR20100065240A/ko not_active Application Discontinuation
- 2008-09-12 US US12/530,872 patent/US8369403B2/en not_active Expired - Fee Related
- 2008-09-12 WO PCT/JP2008/066911 patent/WO2009035144A1/ja active Application Filing
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JPH09294263A (ja) * | 1996-04-24 | 1997-11-11 | Victor Co Of Japan Ltd | 画像情報圧縮装置 |
JP2002290914A (ja) * | 2001-03-26 | 2002-10-04 | Sony Corp | 記録再生装置及びそのバックサーチ方法 |
JP2005192232A (ja) * | 2002-05-28 | 2005-07-14 | Sharp Corp | 画像符号化システム、画像復号システム、及び画像を分割したブロックの画素値の予測方法 |
JP2006148419A (ja) * | 2004-11-18 | 2006-06-08 | Nippon Telegr & Teleph Corp <Ntt> | 画像符号化装置,画像符号化方法,画像符号化プログラムおよびコンピュータ読み取り可能な記録媒体 |
JP2006270437A (ja) * | 2005-03-23 | 2006-10-05 | Toshiba Corp | ビデオエンコーダ及びこれを用いた携帯無線端末装置 |
JP2007067469A (ja) * | 2005-08-29 | 2007-03-15 | Nippon Telegr & Teleph Corp <Ntt> | フレーム内予測符号化制御方法、フレーム内予測符号化制御装置、フレーム内予測符号化制御プログラムおよびそのプログラムを記録した記録媒体 |
Non-Patent Citations (1)
Title |
---|
See also references of EP2187648A4 * |
Cited By (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2010146772A1 (ja) * | 2009-06-19 | 2010-12-23 | 三菱電機株式会社 | 画像符号化装置、画像復号装置、画像符号化方法及び画像復号方法 |
JPWO2010146772A1 (ja) * | 2009-06-19 | 2012-11-29 | 三菱電機株式会社 | 画像符号化装置、画像復号装置、画像符号化方法及び画像復号方法 |
EP2282545A1 (en) | 2009-07-27 | 2011-02-09 | Sony Corporation | Image encoding device and image encoding method |
JP2011029953A (ja) * | 2009-07-27 | 2011-02-10 | Sony Corp | 画像符号化装置と画像符号化方法 |
WO2015033510A1 (ja) * | 2013-09-09 | 2015-03-12 | 日本電気株式会社 | 映像符号化装置、映像符号化方法及びプログラム |
JPWO2015033510A1 (ja) * | 2013-09-09 | 2017-03-02 | 日本電気株式会社 | 映像符号化装置、映像符号化方法及びプログラム |
JP7540082B2 (ja) | 2021-06-30 | 2024-08-26 | テンセント・アメリカ・エルエルシー | オフセット・ベースのリファインメント及び複数参照ライン選択の調和的設計 |
Also Published As
Publication number | Publication date |
---|---|
US20100046621A1 (en) | 2010-02-25 |
JPWO2009035144A1 (ja) | 2010-12-24 |
TWI392372B (zh) | 2013-04-01 |
EP2187648A1 (en) | 2010-05-19 |
TW200920142A (en) | 2009-05-01 |
CN101641958A (zh) | 2010-02-03 |
KR20100065240A (ko) | 2010-06-16 |
EP2187648A4 (en) | 2012-07-11 |
CN101641958B (zh) | 2011-10-19 |
US8369403B2 (en) | 2013-02-05 |
JP5212372B2 (ja) | 2013-06-19 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
WO2009035144A1 (ja) | 画像処理装置及び画像処理方法 | |
US9420279B2 (en) | Rate control method for multi-layered video coding, and video encoding apparatus and video signal processing apparatus using the rate control method | |
US6876703B2 (en) | Method and apparatus for video coding | |
JP5350404B2 (ja) | 早いマクロブロック・デルタqpの決定 | |
US7620107B2 (en) | Video transcoding method and apparatus and motion vector interpolation method | |
US6590936B1 (en) | Coded data transform method, transcoding method, transcoding system, and data storage media | |
JP5472105B2 (ja) | 画像処理装置及び画像処理方法 | |
JP5893002B2 (ja) | ビデオ符号化におけるレート制御 | |
JP5590133B2 (ja) | 動画像符号化装置、動画像符号化方法、動画像符号化用コンピュータプログラム、動画像復号装置及び動画像復号方法ならびに動画像復号用コンピュータプログラム | |
US20050265447A1 (en) | Prediction encoder/decoder, prediction encoding/decoding method, and computer readable recording medium having recorded thereon program for implementing the prediction encoding/decoding method | |
US9071844B2 (en) | Motion estimation with motion vector penalty | |
US20100329333A1 (en) | Method and apparatus for control of rate-distortion tradeoff by mode selection in video encoders | |
US8891892B2 (en) | Image encoding method using adaptive preprocessing scheme | |
KR20080096760A (ko) | 부호화 장치, 부호화 방법 및 프로그램 | |
JP2001145113A (ja) | 画像情報変換装置及び方法 | |
WO2009157581A1 (ja) | 画像処理装置及び画像処理方法 | |
US8442113B2 (en) | Effective rate control for video encoding and transcoding | |
JP4407249B2 (ja) | データ処理装置およびその方法と符号化装置、その方法及びプログラム | |
JP4719854B2 (ja) | 動画像符号化装置、動画像符号化プログラム、動画像復号化装置及び動画像復号化プログラム | |
WO1997016031A1 (en) | Apparatus and method for selecting a coding mode in a block-based coding system | |
US9167263B2 (en) | Methods and devices for image encoding and decoding, and corresponding computer programs | |
JP2008245201A (ja) | 符号化装置 | |
JP4517963B2 (ja) | 符号化装置、符号化方法、プログラムおよび記録媒体 | |
JP2006121538A (ja) | 画像符号化装置、及び画像符号化方法 | |
JP4100067B2 (ja) | 画像情報変換方法及び画像情報変換装置 |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
WWE | Wipo information: entry into national phase |
Ref document number: 200880008154.8 Country of ref document: CN |
|
121 | Ep: the epo has been informed by wipo that ep was designated in this application |
Ref document number: 08830296 Country of ref document: EP Kind code of ref document: A1 |
|
ENP | Entry into the national phase |
Ref document number: 2009532267 Country of ref document: JP Kind code of ref document: A |
|
WWE | Wipo information: entry into national phase |
Ref document number: 2008830296 Country of ref document: EP |
|
WWE | Wipo information: entry into national phase |
Ref document number: 12530872 Country of ref document: US Ref document number: 1020097019021 Country of ref document: KR |
|
NENP | Non-entry into the national phase |
Ref country code: DE |