WO2019208190A1 - Dispositif de codage d'image, procédé de codage d'image, et programme - Google Patents

Dispositif de codage d'image, procédé de codage d'image, et programme Download PDF

Info

Publication number
WO2019208190A1
WO2019208190A1 PCT/JP2019/015380 JP2019015380W WO2019208190A1 WO 2019208190 A1 WO2019208190 A1 WO 2019208190A1 JP 2019015380 W JP2019015380 W JP 2019015380W WO 2019208190 A1 WO2019208190 A1 WO 2019208190A1
Authority
WO
WIPO (PCT)
Prior art keywords
unit
block
coefficient
intra prediction
prediction
Prior art date
Application number
PCT/JP2019/015380
Other languages
English (en)
Japanese (ja)
Inventor
淳真 大澤
Original Assignee
ソニー株式会社
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by ソニー株式会社 filed Critical ソニー株式会社
Publication of WO2019208190A1 publication Critical patent/WO2019208190A1/fr

Links

Images

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
    • H04N19/124—Quantisation
    • H04N19/126—Details of normalisation or weighting functions, e.g. normalisation matrices or variable uniform quantisers
    • 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
    • H04N19/136—Incoming video signal characteristics or properties
    • H04N19/14—Coding unit complexity, e.g. amount of activity or edge presence estimation
    • 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
    • H04N19/157—Assigned coding mode, i.e. the coding mode being predefined or preselected to be further used for selection of another element or parameter
    • H04N19/159—Prediction type, e.g. intra-frame, inter-frame or bidirectional frame prediction
    • 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
    • H04N19/169—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding
    • H04N19/18—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding the unit being a set of transform coefficients
    • 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
    • H04N19/50—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using predictive coding
    • H04N19/593—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using predictive coding involving spatial prediction techniques

Definitions

  • the present disclosure relates to an image encoding device, an image encoding method, and a program. More specifically, an image encoding device that performs an encoding process that enables reduction of banding, which is a striped pseudo contour that tends to appear in a gentle gradation area of an image when the encoded image is decoded, and an image code And a program.
  • AVC Advanced Video Coding
  • HEVC High Efficiency Video Coding
  • inter prediction coding interframe prediction
  • intra prediction coding intraframe prediction coding
  • a shaded pseudo contour may occur in the decoded image.
  • This banding composed of a striped pattern is likely to appear in a flat image area with little change in pixel value such as “sky” in the image, and becomes conspicuous as image quality degradation of the decoded image.
  • Such a pseudo contour can be suppressed by reviving the high frequency component lost by the quantization at the time of coding on the encoder side.
  • the coefficient component is revived indefinitely, the coding efficiency is improved. There is a problem of getting worse.
  • the present disclosure has been made in view of the above-described problems, for example, and performs an encoding process that enables reduction of striped pseudo contours appearing in an image when the encoded image is decoded, that is, so-called banding.
  • An object is to provide an encoding device, an image encoding method, and a program.
  • the first aspect of the present disclosure is: A block determination unit that determines the complexity of the constituent pixels of the block to be encoded; For a block that is determined to be low in complexity, it has a coefficient correction unit that performs correction processing of a quantized frequency transform coefficient, The coefficient correction unit is According to the image encoding apparatus, the correction range of the quantization frequency conversion coefficient is determined according to the gradation direction of the block, and the quantization frequency conversion coefficient in the determined correction range is corrected.
  • the second aspect of the present disclosure is: An image encoding method executed in an image encoding device, A block determination step in which the block determination unit determines the complexity of the constituent pixels of the block to be encoded; and A coefficient correction unit executes a coefficient correction step for performing a correction process of a quantization frequency transform coefficient on a block determined to have low complexity in the block determination step,
  • the coefficient correction step includes An image encoding method includes a process of determining a correction range of a quantization frequency transform coefficient according to a gradation direction of a block and correcting the quantization frequency conversion coefficient of the determined correction range.
  • the third aspect of the present disclosure is: A program for executing an image encoding process in an image encoding device; A block determination step for causing the block determination unit to determine the complexity of the constituent pixels of the block to be encoded; A coefficient correction unit is caused to execute a coefficient correction step for executing a correction process of a quantization frequency transform coefficient for a block determined to have low complexity in the block determination step, In the coefficient correction step, A program for determining a correction range of a quantization frequency conversion coefficient according to a gradation direction of a block and executing a process of correcting the quantization frequency conversion coefficient of the determined correction range.
  • the program of the present disclosure is a program that can be provided by, for example, a storage medium or a communication medium provided in a computer-readable format to an information processing apparatus or a computer system that can execute various program codes.
  • a program in a computer-readable format, processing corresponding to the program is realized on the information processing apparatus or the computer system.
  • system is a logical set configuration of a plurality of devices, and is not limited to one in which the devices of each configuration are in the same casing.
  • an encoding process that can reduce the occurrence of pseudo contour (banding) is realized.
  • a block determination unit that determines the complexity of the constituent pixels of the block to be encoded, and a process of correcting the quantization frequency transform coefficient for the block determined to have low complexity
  • a coefficient correction unit is included.
  • the coefficient correction unit determines a correction range of the coefficient according to the gradation direction of the block, and corrects the coefficient in the determined correction range.
  • the in-screen prediction unit determines whether the encoding process to which DC prediction, plane prediction, or direction prediction is applied is the lowest cost, and outputs the determination result to the coefficient correction unit as optimal intra prediction mode information To do.
  • the coefficient correction unit performs a process of correcting the quantized frequency transform coefficient when the optimal intra prediction mode information is directional prediction. With this configuration, an encoding process that can reduce the occurrence of pseudo contour (banding) is realized. Note that the effects described in the present specification are merely examples and are not limited, and may have additional effects.
  • the CTU is divided into CUs (Coding Units) that are units of coding units.
  • the encoding process and the decoding process are executed for each CU.
  • a prediction mode is determined by a CU or a PU (Prediction Unit) set as a division unit thereof.
  • a CU is composed of one or more PUs.
  • intra-screen prediction intra prediction
  • inter-screen prediction inter prediction
  • which of these two prediction processes is applied is selected in units of CUs. Which one to select is determined based on the cost. For example, a method (low cost method) capable of reducing more data is selected by comparing the amount of data that can be reduced when encoding is performed by applying each prediction in each CU, which is each encoding unit. .
  • intra prediction When performing intra prediction (intra prediction), prediction is performed by predicting a block to be encoded from pixel values of a block (a block on the left or on the top of the picture) in the vicinity of the block to be encoded (CU or the like). As an image, an encoding process is performed for calculating a difference from a predicted image.
  • AVC and HEVC the following three prediction methods are defined.
  • c Directional prediction
  • DC prediction and planar prediction are both methods using the pixels on the upper side and the left side adjacent to the encoding target block as reference pixels.
  • DC prediction the average value of the reference pixels on the upper side and the left side adjacent to the encoding target block is used as a prediction value.
  • planar prediction the plane approximate value of the reference pixels on the upper side and the left side adjacent to the encoding target block. Is used as a predicted value.
  • Directional (Angular) prediction is a method in which a reference pixel is a pixel in one specific direction with respect to an encoding target block.
  • the encoding target block has one specific direction, and a reference pixel is generated by performing interpolation prediction based on the directionality.
  • FIG. 1 is a diagram illustrating a prediction mode of (Angular) 35 defined in HEVC (H.265 / HEVC).
  • 33 types of directional predictions of prediction mode numbers 0 (planar prediction) and 1 (DC prediction) 2 to 34 are defined for the constituent pixels of the encoding target block.
  • reference pixels are generated by interpolating and predicting 45 ° pixels in the lower left direction of the constituent pixels of the encoding target block.
  • reference pixels are generated by interpolating and predicting 45 ° pixels in the upper left direction of the constituent pixels of the encoding target block.
  • a reference pixel is generated by interpolating and predicting 45 ° pixels in the upper right direction of the constituent pixels of the encoding target block.
  • both AVC and HEVC use directional prediction for encoding processing, basically select a mode that has a predicted image with a pixel value that has a smaller pixel value than the pixel value of the encoding target block. Processing is performed.
  • a prediction mode number is defined. This number corresponds to the identifier of the prediction mode of HEVC intra-screen prediction (intra prediction), and the identifiers of the following prediction modes are defined.
  • Prediction mode number 0 is planar (Planar) prediction
  • Prediction mode number 1 is DC prediction
  • Prediction mode numbers 2 to 34 are directional predictions (Angular), It is. In the case of directional prediction, prediction mode numbers 2 to 34 are associated with each other as shown in FIG.
  • A DC prediction, (b) Planar prediction, (c) Directional prediction (c) Direction of prediction (c) Directional prediction Which prediction is used, that is, which prediction of the prediction mode numbers 0 to 34 is to be executed, is determined based on the cost, that is, the degree of code amount reduction, for each coding block. A method that can maximize the data reduction amount in units of blocks is selected and executed. Note that the cost is calculated based on, for example, the amount of encoded data when the difference between the encoded block and the reference block is encoded.
  • the lowest cost prediction mode is selected for each coding block, the difference between the coding block and the reference block is calculated, and coding processing such as frequency conversion of the difference data, quantization processing, etc. is executed. Details of the specific encoding process will be described later.
  • inter prediction motion compensation is performed on the preceding and succeeding frames (pictures) in units of encoding target blocks (such as CU), and encoding targets and reference positions between different frames Is calculated, and encoding processing is performed by frequency conversion of the difference data, quantization processing, or the like.
  • the striped pseudo contour ((b) decoded image shown in FIG. (Banding) may occur.
  • This pseudo contour is likely to appear mainly in a flat portion of an image or a gradation region where the gradation changes gradually. This is due to the loss of data due to the quantization processing at the time of image encoding.
  • a pseudo contour as shown in FIG. 2B may appear.
  • the image encoding apparatus according to the present disclosure realizes an encoding process that reduces the occurrence of such pseudo contours.
  • FIG. 3 is a diagram illustrating a configuration example of the image encoding device 100 according to the present disclosure.
  • the image encoding device 100 includes an image encoding unit 110 and a control unit 150.
  • the image encoding unit 110 includes a block determination unit 111, a subtraction unit 112, a frequency conversion unit 113, a quantization unit 114, a coefficient correction unit 115, an entropy encoding unit 116, an inverse quantization unit 121, an inverse frequency conversion unit 122, and an addition.
  • the image encoding unit 110 receives the input image 10 to be encoded, generates an encoded image (stream) 20, and outputs it. The processing of the image encoding unit 110 is performed under the control of the control unit 150.
  • the input image 10 to be encoded is a moving image, and the number of pixels of one image frame of the image to be encoded is, for example, 4 ⁇ 4 pixels, 8 ⁇ 8 pixels, 16 ⁇ 16 pixels, or the like. Are divided into rectangular blocks, and an encoding process is performed for each block.
  • the image encoding unit 110 is sequentially input with constituent blocks of image frames constituting a moving image, and an encoding process in units of blocks is executed.
  • 4 ⁇ 4 pixels, 8 ⁇ 8 pixels, 16 ⁇ 16 pixels, and 32 ⁇ 32 pixels can be encoded in block units as block sizes.
  • the block determination unit 111 determines the complexity of the block constituent pixels for each block input as an encoding target. For example, an index value indicating the complexity of the pixel values of the constituent pixels of the input block, such as the variance value of the constituent pixels of the rectangular block that is the encoding unit, the difference between the maximum luminance and the minimum luminance, the total variation (TV), The complexity of the constituent pixels of the encoding target block is determined.
  • the block determination unit 111 compares the complexity index value calculated for each block with a predetermined threshold value, and sets one of the following two determination values for each block.
  • A) High complexity block complexity determination value 1
  • B) Low complexity block complexity determination value 0
  • a predetermined fixed value is used as the threshold value.
  • a variable value set in a predetermined processing unit such as every image sequence or every image frame in the control unit 150 may be used.
  • the block unit complexity determination information (complexity determination values 0 and 1), which is a block unit complexity determination result by the block determination unit 111, is notified to the coefficient correction unit 115.
  • the subtraction unit 112 inputs a block for which the block unit complexity determination has been completed from the block determination unit 111, and receives a prediction signal (predicted image data: intra predicted image data or inter predicted image data) input from the prediction method switching unit 134. ) Is calculated, and the calculated residual signal (prediction error data) is output to the frequency converter 113.
  • a prediction signal predicted image data: intra predicted image data or inter predicted image data
  • the frequency conversion unit 113 performs frequency conversion on the residual signal (prediction error data) input from the subtraction unit 112 and converts the residual signal into frequency domain coefficients. For example, discrete cosine transform (DCT transform) is performed to convert the residual signal (prediction error data) input from the subtracting unit 112 from a spatial domain image signal to frequency domain transform coefficient data.
  • DCT transform discrete cosine transform
  • the conversion coefficient data calculated by the frequency conversion unit 113 is output to the quantization unit 114.
  • the quantization unit 114 performs a quantization process on the frequency conversion coefficient input from the frequency conversion unit 113 to generate a quantization coefficient.
  • the quantization unit 114 quantizes the frequency conversion coefficient input from the frequency conversion unit 113 in a quantization step determined based on, for example, information related to the code amount target value supplied from the control unit 150.
  • the quantization unit 115 inputs the quantized transform coefficient to the coefficient correction unit 115.
  • the coefficient correction unit 115 receives the quantized frequency conversion coefficient from the quantization unit 114 and executes a correction process for the quantized frequency conversion coefficient after quantization.
  • the coefficient correction unit 115 executes the correction process of the frequency conversion coefficient calculated corresponding to the block only for the block that is highly likely to generate a pseudo contour (banding).
  • This correction process is a process executed to suppress the occurrence of pseudo contour (banding).
  • a specific example of the coefficient correction process executed by the coefficient correction unit 115 will be described in detail later.
  • the coefficient corrected by the coefficient correction unit 115 is output to the entropy encoding unit 116 and the inverse quantization unit 121.
  • the coefficient generated by the quantization unit 114 without correction in the coefficient correction unit 115 is directly used as the entropy encoding unit 116 and the inverse quantum. Is output to the conversion unit 121.
  • the entropy encoding unit 116 generates an encoded image stream 20 by performing entropy encoding processing (variable length encoding processing) on the quantized frequency conversion coefficient after quantization input from the coefficient correction unit 115. Output.
  • the inverse quantization unit 121 inversely quantizes the quantized data in the same quantization step as executed in the quantization unit 114, and restores the frequency transform coefficient data.
  • the inverse quantization unit 121 outputs the restored frequency transform coefficient data to the inverse frequency transform unit 122.
  • the inverse frequency transform unit 122 performs an inverse frequency transform process on the frequency transform coefficient data input from the inverse quantization unit 121 to restore a residual signal (prediction error data) set as a quantization target.
  • the restored prediction error data is output to the adding unit 123.
  • the addition unit 123 adds the decoded prediction error data input from the inverse frequency conversion unit 122 and the predicted image data input from the prediction method switching unit 134 to generate decoded image data (reconstructed image).
  • the adding unit 123 outputs the generated decoded image data to the intra-loop filter 124 and the intra-screen prediction unit (intra prediction unit) 131.
  • the in-loop filter 124 applies distortion correction to the reconstructed image supplied from the adding unit 123 by applying a deblocking filter (DBF: De-Blocking Filter), a sample adaptive offset (SAO), or the like. Image correction processing such as value correction is performed.
  • the deblocking filter (DBF) application process is a process for performing filtering processing on block boundaries of motion prediction and conversion processing in order to reduce block-like distortion due to motion compensation and conversion processing.
  • Sample adaptive offset (SAO) is a process of correcting a pixel value by adding a predetermined offset value to a pixel value in a block.
  • the filter processing result image (decoded image) that has been subjected to distortion removal, pixel value correction, and the like by the in-loop filter 124 is stored in the frame memory 125.
  • the filter processing result image (decoded image) stored in the frame memory 125 is supplied to the inter-screen prediction unit (inter prediction unit) 132.
  • the intra-screen prediction unit (intra prediction unit) 131 receives the decoded image data (reconstructed image) generated by the addition unit 123, that is, the restored prediction error data input from the inverse frequency transform unit 122 and the prediction method switching unit 134.
  • the decoded image data (reconstructed image) which is the addition data with the predicted image data to be input is input, and the intra-screen prediction (intra prediction) process using the decoded image data (reconstructed image) is executed.
  • the intra prediction unit (intra prediction unit) 131 has the following three predictions described above with reference to FIG. (A) DC prediction (b) Planar prediction (c) Directionality (angular) prediction It is determined which of these methods has the lowest cost (the amount of data reduction is large). Regarding the directional prediction, the reference pixel setting direction is also determined based on the cost. The cost is calculated based on, for example, the amount of encoded data when the difference between the encoded block and the predicted image block acquired from the decoded image data (reconstructed image) is encoded.
  • Intra prediction unit 131 is, for example, H.264. H.264 / AVC and H.264.
  • Intra prediction images corresponding to all intra prediction modes including directional prediction defined by H.265 / HEVC and the like are generated, and cost calculation is performed when the difference from the encoding target block is encoded.
  • the in-screen prediction unit (intra prediction unit) 131 executes this cost calculation process to optimize the intra prediction mode that is most similar to the block to be encoded and can reduce the amount of code most when the difference is encoded. Selected as the intra prediction mode.
  • the intra prediction mode determination information determined by the in-screen prediction unit (intra prediction unit) 131 and the cost information corresponding to the mode are output to the prediction method determination unit 133, and the intra prediction image data generated according to the selected optimal intra prediction mode is And output to the prediction method switching unit 134. Furthermore, the optimal intra prediction mode information determined by the intra prediction unit (intra prediction unit) 131 is also output to the coefficient correction unit 115.
  • the inter-screen prediction unit (inter prediction unit) 132 frames the input image 10 to be encoded and the filter processing result image (decoded image) subjected to distortion removal, pixel value correction, and the like by the in-loop filter 124. Input from the memory 125, and perform inter prediction processing (motion detection and motion compensation) using each image.
  • the inter-screen prediction unit (inter prediction unit) 132 evaluates the cost based on the prediction error and the generated code amount for each prediction candidate included in the search range defined in AVC or HEVC, and the prediction candidate that minimizes the cost That is, the prediction candidate with the highest compression rate is selected as the optimal prediction in the inter prediction.
  • the inter-screen prediction unit (inter prediction unit) 132 generates inter prediction image data according to the selected optimal prediction, and the determined inter prediction information and mode-corresponding cost information are output to the prediction method determination unit 133 and selected optimal
  • the inter prediction image data generated according to the inter prediction is output to the prediction method switching unit 134.
  • the prediction method determination unit 133 inputs the optimal intra prediction mode determination information determined by the in-screen prediction unit (intra prediction unit) 131 based on the cost and the cost information corresponding to the mode. Further, optimal inter prediction determination information and cost information are input from the inter-screen prediction unit (inter prediction unit) 132. The prediction method determination unit 133 compares the costs of these two pieces of information, and determines which is the low-cost (large data reduction amount) encoding mode. Further, the switch of the prediction method switching unit 134 is controlled according to the determination result.
  • the switch control by the prediction method determination unit 133 is executed as follows.
  • the intra prediction unit (intra prediction unit) 131 sets the switch setting of the prediction method switching unit 134.
  • the intra prediction image data generated based on the prediction is set to be output to the subtraction unit 112.
  • the switch setting of the prediction method switching unit 134 is changed to the inter-screen prediction unit (inter prediction unit) 132. Is set to output the inter prediction image data generated based on the inter prediction to the subtraction unit 112.
  • the subtraction unit 112 is a residual signal (prediction) that is a difference from the prediction signal (prediction image data: intra prediction image data or inter prediction image data) input from the prediction method switching unit 134. Error data), and outputs the calculated residual signal (prediction error data) to the frequency converter 113.
  • prediction residual signal
  • the transform coefficient data is transform coefficient data calculated by the frequency converter 113 and quantized by the quantizer 114.
  • FIG. 4 to FIG. 6 show an example of setting a coding block having a gradual gradation that is a block that is highly likely to generate a pseudo contour (banding) and quantization frequency transform coefficient data corresponding to each block.
  • the blocks 201 to 203 shown in the upper left part of FIGS. 4 to 6 are 4 ⁇ 4 pixel blocks, all of which have a gentle gradation.
  • a block 201 illustrated in FIG. 4 is an example of an encoding target block, and is a block having a gentle gradation in the horizontal direction. This block 201 can be expressed as a product sum of 16 blocks shown on the right side of the block 201 in FIG.
  • Each of the 16 blocks shown on the right side of the block 201 in FIG. 4 is an example of a frequency conversion base, and each represents a different frequency component.
  • a block having a gentle gradient in the horizontal direction similar to the block 201 shown in FIG. 4 can be expressed. it can.
  • the number of frequency conversion bases (16 in FIG. 4) differs depending on the number of pixels to be converted. .
  • the coefficient data is conversion coefficient data to be corrected in the coefficient correction unit 115.
  • the transform coefficients to be corrected are not all of the coefficients a00 to a33 but some coefficients selected according to the gradation setting of the encoding target block, that is, the block 201. This correction target selection process will be described later.
  • a block 202 shown in FIG. 5 is a block having a gentle gradation in the vertical direction.
  • This block 202 can also be expressed as a product-sum result of 16 blocks shown on the right side of the block 202 in FIG.
  • the 16 blocks shown on the right side of the block 202 in FIG. 5 are an example of a frequency conversion base having the same pattern as the 16 blocks shown on the right side of the block 201 in FIG.
  • Each of the 16 pattern data is multiplied and multiplied by predetermined multiplication coefficients b00 to b33 different from the multiplication coefficients a00 to a33 shown in FIG. 4, so that they are summed in the vertical direction as in the block 202 shown in FIG. It is possible to express a block with a gentle gradient.
  • a block 203 shown in FIG. 6 is a block having a gentle gradation in the upper left diagonal 45 ° direction.
  • This block 203 can also be expressed as a product sum of 16 blocks shown on the right side of the block 203 in FIG.
  • the 16 blocks shown on the right side of the block 202 in FIG. 6 are also examples of frequency conversion bases having the same pattern as the 16 blocks shown on the right side of the block 201 in FIG.
  • Each of the 16 pieces of pattern data is multiplied and multiplied by predetermined multiplication coefficients c00 to c33 different from the multiplication coefficients a00 to a33 shown in FIG. 4, so that the diagonal 45 ° as in the block 203 shown in FIG. It is possible to express a block with a gentle gradient in the direction.
  • FIG. 7 is a diagram showing 32 directions of the directional prediction defined in HEVC (H.265 / HEVC) described above with reference to FIG. As described above, in HEVC, 33 types of prediction mode numbers 2 to 34 are defined for the directionality prediction.
  • the intra prediction unit (intra prediction unit) 131 of the image encoding unit 110 illustrated in FIG. 3 determines the intra prediction method in units of blocks based on the cost.
  • the direction selection in the directional (angular) prediction is also a cost. It is executed based on.
  • the direction of the reference target pixel or block is a direction in which a pixel similar to the constituent pixel of the encoding target block exists. That is, a direction in which a pixel having a small difference exists is selected.
  • FIG. 7 shows the directional prediction that is estimated to be selected corresponding to each of the encoding target blocks 201 to 203 having the three different gradient directions described with reference to FIGS.
  • An example of the direction at, that is, the prediction mode number is shown.
  • the block is estimated to be a block having a horizontal gradation. That is, when the encoding target block is the block 201 having the horizontal gradation direction described with reference to FIG. 4, the intra prediction unit (intra prediction unit) 131 performs prediction as illustrated in (Example 1) in FIG. 7. It is estimated that the directionality prediction of mode number 26 is selected.
  • the block is estimated to be a block having a vertical gradation. That is, when the encoding target block is the block 202 having the vertical gradation direction described with reference to FIG. 5, the intra prediction unit (intra prediction unit) 131 performs prediction as shown in (Example 2) of FIG. It is estimated that the directionality prediction of mode number 10 is selected.
  • the block is estimated to be a block having a gradation in the upper left 45 ° direction. That is, when the encoding target block is the block 203 whose gradation direction is the upper left 45 ° direction described with reference to FIG. 6, the intra-screen prediction unit (intra prediction unit) 131 is illustrated in (Example 3) of FIG. 7. Thus, it is estimated that the directionality prediction of the prediction mode number 2 or 34 is selected.
  • the coefficient correcting unit 115 refers to the coefficients, that is, referring to FIGS. 4 to 6 in advance, according to the intra prediction mode determination information such as the encoding mode numbers 0 to 34 input from the intra prediction unit (intra prediction unit) 131.
  • the correction target of the coefficient (a00 to a33 etc.) to be multiplied with each of the pattern data (DCT base) described above is determined, and correction of the coefficient to be corrected is executed.
  • the intra prediction unit (intra prediction unit) 131 has the following three predictions: (A) DC prediction (b) Planar prediction (c) Directionality (angular) prediction It is determined which of these methods has the lowest cost (the amount of data reduction is large). Regarding the directional prediction, the reference pixel setting direction is also determined based on the cost.
  • Intra prediction unit 131 is, for example, H.264. H.264 / AVC and H.264.
  • Intra prediction images corresponding to all intra prediction modes including directional prediction defined by H.265 / HEVC and the like are generated, and cost calculation is performed when the difference from the encoding target block is encoded. By executing this cost calculation process, the intra prediction mode that is most similar to the encoding target block and that can reduce the amount of code most when the difference is encoded is selected as the optimal intra prediction mode.
  • the intra prediction unit (intra prediction unit) 131 outputs the predicted image data generated according to the selected optimal intra prediction mode to the prediction method switching unit 134. Further, the intra prediction mode determination information determined by the intra prediction unit (intra prediction unit) 131 and the cost information corresponding to the mode are output to the prediction method determination unit 133. Further, the intra prediction mode determination information determined by the intra prediction unit (intra prediction unit) 131 is also output to the coefficient correction unit 115.
  • the determination value is “0”, it can be estimated that the block has gradation in a direction corresponding to the direction of the directionality prediction mode.
  • directionality prediction in 33 directions is defined regardless of the prediction block size.
  • direction prediction in a specific direction for example, HEVC stipulated as an intra prediction mode optimal for the encoding target block.
  • the encoding target block can be estimated to be a block having a gradation having a gradient in a direction perpendicular to the selected prediction mode direction.
  • the optimal intra prediction mode selected by the intra prediction unit (intra prediction unit) 131 is the directionality of the prediction mode number 22 shown in FIG.
  • this encoding target input block can be estimated as a block having gradation in the upper right 30 ° direction as shown in FIG. 8A.
  • the optimal intra prediction mode selected by the intra prediction unit (intra prediction unit) 131 is the direction prediction of the prediction mode number 26 shown in FIG. If it is determined that the input block to be encoded is a block having gradation in the horizontal 0 ° direction as shown in FIG.
  • the block is a gradation block having an inclination in the direction perpendicular to each directional prediction.
  • FIG. 9 shows an example of the gradation pattern of the encoding target block estimated from the optimum intra prediction mode (prediction mode number of HEVC direction prediction) selected by the intra prediction unit (intra prediction unit) 131.
  • FIG. 9 shows the following eight examples.
  • it is estimated that the gradation direction of the encoding target block is the upper left 60 ° direction. Is done.
  • the gradation direction of the encoding target block is estimated to be the vertical direction. .
  • the gradation direction of the encoding target block is the upper right 60 ° direction. Is done.
  • the optimal intra prediction mode (for example, prediction mode number of HEVC direction prediction) information selected by the intra prediction unit (intra prediction unit) 131 is notified to the coefficient correction unit 115.
  • the prediction method determination unit 133 determines whether to apply intra prediction (intra prediction) or inter prediction (inter prediction) to each block. In any case, the in-screen prediction unit (intra prediction unit) 131 executes the optimal intra prediction mode selection process for the encoding target block in parallel.
  • the gradation direction of the block can be estimated according to the result of the optimum prediction mode, and the coefficient correction unit 115 performs coefficient correction according to the estimated gradation direction. That is, the intra prediction unit (intra prediction unit) 131 is the optimal intra intra prediction unit regardless of whether the prediction method applied to the block encoding process is intra prediction (intra prediction) or inter prediction (inter prediction).
  • the prediction mode (prediction mode number of HEVC direction prediction) information is notified to the coefficient correction unit 115, and the coefficient correction unit 115 performs coefficient correction according to the estimated gradation direction.
  • the coefficient correction unit 115 is based on the block-corresponding optimal intra prediction mode (for example, the prediction mode number of HEVC direction prediction) input from the intra-screen prediction unit (intra prediction unit) 131.
  • a final coefficient correction target block and its range are selected from the candidate blocks.
  • the optimum intra prediction mode corresponding to the block input from the intra prediction unit (intra prediction unit) 131 is one of the following three prediction modes.
  • the optimal intra prediction mode corresponding to the block input from the intra prediction unit (intra prediction unit) 131 is (a) DC prediction, (b) planar prediction,
  • the block is estimated not to be a block having gradation in a specific direction, and is excluded from the coefficient correction target block.
  • the block is in a specific direction. Since it is estimated that the block has gradation, it is determined as the final coefficient correction target block.
  • the coefficient correcting unit 115 receives the quantized quantized frequency transform coefficient from the quantizing unit 114 and executes a correction process for the quantized quantized frequency transform coefficient.
  • the coefficient corrected by the coefficient correction unit 115 is a coefficient (a00 to a33 etc.) to be multiplied with each of the frequency conversion bases described above with reference to the examples of FIGS. 4 to 6, that is, a quantized frequency conversion coefficient.
  • the optimal intra prediction mode information corresponding to the encoding target block the input information from the intra prediction unit (intra prediction unit) 131 is “directional prediction” and the prediction mode number defined by HEVC is 26. To do.
  • the prediction mode number is 26, it corresponds to FIG. 9 (7), and the gradation direction of the block is estimated to be the horizontal direction.
  • the number of pattern data (DCT base) is determined in accordance with the block size to be encoded, and various numbers can correspond to the block size.
  • the example shown in FIG. 10 is an example using 64 pattern data (DCT base) (D00 to D77) corresponding to an 8 ⁇ 8 pixel block.
  • the pseudo contour is generated when high-frequency components necessary for smooth gradation expression disappear due to quantization processing.
  • the high frequency component necessary for this smooth gradation expression depends on the direction of gradation.
  • the coefficient correction unit 115 selects pattern data (DCT base) of high frequency components necessary for smoothly expressing the gradation of the block, and a coefficient corresponding to the selected pattern data (DCT base), that is, quantization frequency conversion Correct the coefficient.
  • These 64 pattern data (DCT base) (D00 to D77) all have different patterns.
  • the pattern whose shading changes only in the horizontal direction is mainly set to one line at the top.
  • These pattern data D00 to D07 are pattern data (DCT base) that has a great influence on the gradation in the horizontal direction.
  • the pattern data (DCT base) (D00 to D07) indicated by the white frame in FIG.
  • the gradation is expressed by outputting the data (DCT basis) stronger than D10 to D77.
  • the coefficient correction unit 115 performs pattern data (DCT base) (D00 to DCT) of a portion surrounded by a white frame shown in FIG. D07) is determined to be an important element for representing a gradation having a change in the horizontal direction, and the coefficient of this portion is corrected mainly. By performing this correction, the pseudo contour (banding) appearing in the decoded image is suppressed.
  • 8 ⁇ 8 64 pieces of pattern data (DCT base) shown in FIGS. 10 and 11 increase as the horizontal frequency progresses from left to right, and as the vertical frequency progresses from top to bottom. To increase. From this property, it is possible to classify into pattern data (DCT base) having a great influence on gradations in various different directions and other pattern data.
  • the coefficient correction unit 115 estimates the gradation direction of an encoding target block having gradations in various different directions from the block-corresponding optimal intra prediction mode information from the intra-screen prediction unit (intra prediction unit) 131.
  • the coefficient correction unit 115 selects and selects pattern data (DCT basis) that is important for expressing gradation in the direction based on the gradation direction of the block estimated from the block-corresponding optimal intra prediction mode information.
  • a coefficient (quantized frequency conversion coefficient) corresponding to pattern data (DCT base) is set as a correction target.
  • coefficients (quantized frequency conversion coefficients) of pattern data (DCT base) selected by the coefficient correction unit 115 as correction targets for blocks having various different gradation directions will be described.
  • FIG. 12A shows a coefficient range of quantization frequency transform coefficients selected as a correction target when the gradation direction of the encoding target block is the horizontal direction.
  • the case where the gradation direction of the encoding target block is the horizontal direction corresponds to the example shown in FIG. 9 (7) described above with reference to FIG.
  • the coefficient correction unit 115 estimates that the gradation direction of the encoding target block is the horizontal direction. .
  • the coefficient correction unit 115 determines the correction range of the quantization frequency transform coefficient set corresponding to each piece of pattern data (DCT basis) for the encoding target block input from the quantization unit 114.
  • the pattern data (DCT base) D00 to D07
  • the quantization frequency transform coefficients associated with the pattern data (DCT base) D00 to D07
  • a00 to a07 are selected as correction targets.
  • FIG. 12B shows a coefficient range of quantization frequency transform coefficients selected as a correction target when the gradation direction of the encoding target block is a vertical direction.
  • the case where the gradation direction of the encoding target block is the vertical direction corresponds to the example shown in FIG.
  • the coefficient correction unit 115 estimates that the gradation direction of the encoding target block is the vertical direction. .
  • the coefficient correction unit 115 determines the correction range of the quantization frequency transform coefficient set corresponding to each piece of pattern data (DCT basis) for the encoding target block input from the quantization unit 114.
  • the pattern data (DCT basis) (D00 to D70) of the portion surrounded by the white frame shown in FIG. 12B, which is an important element for expressing gradation having a change in the vertical direction, is selected.
  • the quantization frequency transform coefficients associated with the pattern data (DCT base) (D00 to D70) for example, b00 to b70 are selected as correction targets.
  • FIG. 12C shows the coefficient range of the quantized frequency transform coefficient selected as the correction target when the gradation direction of the encoding target block is the upper left 45 ° direction.
  • the case where the gradation direction of the encoding target block is the upper left 45 ° direction corresponds to the example shown in FIG.
  • the coefficient correction unit 115 determines the correction range of the quantization frequency transform coefficient set corresponding to each piece of pattern data (DCT basis) for the encoding target block input from the quantization unit 114.
  • pattern data (DCT basis) ((D00 to D04) of a portion surrounded by a white frame shown in FIG. 12C, which is an important element for expressing a gradation having a change in the upper left 45 ° direction. )
  • quantized frequency transform coefficients associated with the pattern data (DCT base) ((D00 to D04) to (D40 to D44)
  • (c00 to c04) to (C40 to c44) are selected as correction targets.
  • FIG. 12D shows a coefficient range of quantization frequency transform coefficients selected as a correction target when the gradation direction of the encoding target block is the upper left 30 ° direction.
  • the case where the gradation direction of the encoding target block is the upper left 30 ° direction corresponds to the example shown in FIG.
  • the coefficient correction unit 115 indicates that the gradation direction of the encoding target block is the upper left 30 ° direction. presume.
  • the coefficient correction unit 115 determines the correction range of the quantization frequency transform coefficient set corresponding to each piece of pattern data (DCT basis) for the encoding target block input from the quantization unit 114.
  • pattern data (DCT basis) ((D00 to D05) of a portion surrounded by a white frame shown in FIG. 12D, which is an important element for expressing a gradation having a change in the upper left 30 ° direction. )
  • quantized frequency transform coefficients associated with the pattern data (DCT base) ((D00 to D05) to (D30 to D35)
  • (d00 to d05) to (D30 to c35) are selected as correction targets.
  • FIG. 12E shows the coefficient range of the quantized frequency transform coefficient selected as the correction target when the gradation direction of the encoding target block is the upper left 60 ° direction.
  • the case where the gradation direction of the encoding target block is the upper left 60 ° direction corresponds to the example shown in FIG.
  • the coefficient correction unit 115 indicates that the gradation direction of the encoding target block is the upper left 60 ° direction. presume.
  • the coefficient correction unit 115 determines the correction range of the quantization frequency transform coefficient set corresponding to each piece of pattern data (DCT basis) for the encoding target block input from the quantization unit 114.
  • pattern data (DCT basis) ((D00 to D03) of a portion surrounded by a white frame shown in FIG. 12E, which is an important element for expressing a gradation having a change in the upper left 60 ° direction. )
  • quantization frequency transform coefficients associated with the pattern data (DCT base) ((D00 to D03) to (D40 to D43)
  • E40 to e43 quantization frequency transform coefficients associated with the pattern data (E40 to e43) are selected as correction targets.
  • the coefficient correction unit 115 selects pattern data (DCT base) that is an important element for output of each gradation according to the gradation direction of the encoded block, and corrects the coefficient component. By this correction processing, it is possible to suppress pseudo contour (banding) that tends to appear in the decoded image when the encoded image data is decoded.
  • DCT base pattern data
  • pseudo contour banding
  • the coefficient correction unit 115 acquires the frequency transform coefficient before the quantization process, and further performs the frequency transform coefficient after the quantization when the quantization process is performed with the quantization value corresponding to the preset target coding rate. To get. Furthermore, the bit values truncated by the quantization process are calculated by comparing these frequency conversion coefficients. Finally, at least a part of the truncated value is added to the frequency conversion coefficient after quantization, and a correction coefficient is calculated by restoring the value truncated by the quantization.
  • FIG. 13 is a processing example for a block in which the gradation direction of the encoding target block is horizontal, as described with reference to FIG. 11 and FIG.
  • FIG. 13A shows a discrete cosine transform (DCT) for the output of the frequency transform unit 113, that is, the residual signal (prediction error data) input from the subtractor 112 when a block whose gradation direction is horizontal is input. It is an example of frequency conversion coefficient data (DCT coefficient) as a result of (conversion).
  • DCT coefficient frequency conversion coefficient data
  • the data shown in FIG. 13A is generated by the following process.
  • a block whose gradation direction is horizontal is input to the subtraction unit 112 illustrated in FIG. 3, and the subtraction unit 112 receives a prediction signal (prediction image data: intra prediction image data or inter prediction image data) input from the prediction method switching unit 134. ) Is calculated as a residual signal (prediction error data), and the calculated residual signal is input to the frequency converter 113.
  • the frequency conversion unit 113 performs frequency conversion on the residual signal (prediction error data) input from the subtraction unit 112 and converts the residual signal into frequency domain coefficients. That is, discrete cosine transform (DCT transform) is executed to convert the residual signal (prediction error data) input from the subtracting unit 112 from a spatial domain image signal to frequency domain transform coefficient data.
  • DCT transform discrete cosine transform
  • the coefficients shown in FIG. 13A are frequency conversion coefficients calculated by the frequency conversion unit 113 by the above processing.
  • the frequency conversion coefficient calculated by the frequency conversion unit 113 is input to the quantization unit 114 shown in FIG. 3 and subjected to quantization processing.
  • An example of the quantization frequency conversion coefficient after this quantization processing is the data shown in FIG.
  • the coefficient values shown in FIG. 13B are also expressed in decimal numbers.
  • the quantization unit 114 performs a quantization process on the frequency conversion coefficient input from the frequency conversion unit 113 to generate a quantization frequency conversion coefficient.
  • the quantization unit 114 quantizes the frequency conversion coefficient input from the frequency conversion unit 113 in a quantization step determined based on, for example, information on the target value of the code amount supplied from the control unit 150, and obtains a quantization frequency. Generate conversion coefficients.
  • the quantization unit 115 inputs the calculated quantization frequency conversion coefficient to the coefficient correction unit 115.
  • the coefficient correction unit 115 first determines a coefficient correction range.
  • the coefficient correction range determination process is the process described above with reference to FIG. That is, the coefficient correction range is determined according to the gradation direction. For example, for a block having a gradation in the horizontal direction, as shown in FIG. 12A, an important element for representing a gradation having a change in the horizontal direction is determined as a coefficient correction range. That is, in the example shown in FIG. 13, the quantized frequency transform coefficient in the portion surrounded by the dotted line frame shown in FIG. 13B is selected as the correction target.
  • the value truncated by the quantization process (Binary number) is a value (binary number) shown in FIG.
  • the coefficient correction unit 115 calculates a value (binary number) rounded down by the quantization process, that is, a value (binary number) shown in FIG. 13C, and evaluates the calculated value.
  • the value to be truncated in the quantization process is 3 bits, the value to be truncated takes a value from 0 to 7.
  • QD rdd > QD th (conditional expression 1)
  • a coefficient that adds a previously expected correction amount, for example, (+1) to the quantized transform coefficient is performed on the coefficient that satisfies the above conditional expression. As a result of this correction, the corrected quantized frequency conversion coefficient shown in FIG. 13D is obtained.
  • Each of these four coefficients is QD rdd > QD th (conditional expression 1)
  • the coefficient satisfying the above conditional expression, and the correction unit 115 performs coefficient correction on the coefficients by adding a correction amount expected in advance, for example, (+1) to these four quantization frequency conversion coefficients. Specifically, the following correction is executed.
  • the coefficient correction unit 115 calculates a value lost by the quantization processing in the quantization unit 114, and performs coefficient correction processing for correcting the quantized frequency transform coefficient when the calculated value is larger than the specified threshold value.
  • the correction range is a coefficient element that is important for expressing a coefficient selected according to the gradation direction, that is, a change in the gradient direction, as described above. By performing this correction, it becomes possible to reduce the pseudo contour (banding) when decoding the encoded image.
  • the corrected quantized frequency transform coefficient generated by the correction process executed in the coefficient correction unit 115 is output to the entropy encoding unit 116 and the inverse quantization unit 121.
  • the entropy encoding unit 116 performs entropy encoding processing (variable length encoding processing) on the corrected quantized frequency transform coefficient input from the coefficient correction unit 115 to generate the encoded image stream 20 and output it. To do.
  • the inverse quantization unit 121 inversely quantizes the quantized data in the same quantization step as executed in the quantization unit 114, and restores transform coefficient data.
  • the inverse quantization unit 121 outputs the restored transform coefficient data to the inverse frequency transform unit 122.
  • the coefficient correction unit 115 determines which direction the gradation of the block has based on the block-compatible optimal intra prediction mode information input from the intra-screen prediction unit (intra prediction unit) 131. Then, a process for determining the range of coefficients to be corrected is executed.
  • FIGS. 8A and 8B described above show a plurality (three types) of examples of blocks each having the same gradation direction.
  • all of the three blocks shown in FIG. 8A are blocks having a gradation in the upper right 30 ° direction, but the gradations are different from each other.
  • Each of the three blocks shown in FIG. 8B is a block having a gradation in the horizontal direction, but the form of each gradation is different.
  • the coefficient correction unit 115 can determine from the information input from the in-screen prediction unit (intra prediction unit) 131 whether the block belongs to one of the groups shown in FIGS. 8A and 8B. It is impossible to determine which of the three types of gradation in FIG.
  • the coefficient correction unit 115 is configured to acquire not only gradation direction information but also information about a gradation mode, for example, a gradient inclination (pixel value inclination). It is possible to determine the correction range of the corresponding coefficient and perform coefficient correction.
  • FIG. 14 shows an example of gradation mode information that can be used when the coefficient correction unit 115 performs processing for determining a coefficient correction range according to the gradation mode and coefficient correction.
  • the example shown in FIG. 14 is gradation mode information including correspondence data between pixel positions and luminance. If the coefficient correction unit 115 can acquire such correspondence data between the pixel position and the luminance, for example, the coefficient correction unit 115 can perform coefficient correction range determination processing and coefficient correction according to the gradation mode.
  • the 14 is generated in, for example, the intra prediction unit (intra prediction unit) 131 and provided to the coefficient correction unit 115.
  • the gradation mode information including correspondence data between the pixel position and the luminance shown in FIG.
  • a configuration in which the control unit 150 generates and provides the coefficient correction unit 115 is possible.
  • an independent data analysis unit (not shown in FIG. 3) may be provided, generated in the data analysis unit, and provided to the coefficient correction unit 115.
  • the coefficient correction unit 114 may generate and use the coefficient correction unit 114 itself.
  • the coefficient operation range corresponding to the gradation can be fixed for each gradation, or can be set in the encoding control unit.
  • the encoded stream generated by the encoding apparatus of the present disclosure is H.264. H.264 / AVC and H.264
  • decoding is possible with decoders that comply with various encoding standards that have the definition of directional prediction in intra-frame prediction encoding (intra prediction encoding), such as encoding standards that will be standardized in the future. It is.
  • the analysis of gradation is not limited to the analysis using luminance data, and other analysis using, for example, a color difference is also possible.
  • FIG. 15 a sequence of processing executed by the image encoding device 100 according to the present disclosure will be described with reference to a flowchart illustrated in FIG.
  • the processing according to the flowchart shown in FIG. 15 can be executed according to a program stored in the storage unit of the image encoding device 100, for example, and a program execution process by a processor such as a CPU having a program execution function, for example. As can be done.
  • a processor such as a CPU having a program execution function
  • Step S101 block determination processing for the block to be encoded is performed.
  • This process is a process executed by the block determination unit 111 of the image encoding unit 110 shown in FIG.
  • the block determination unit 111 determines the complexity of the block constituent pixels for each block input as an encoding target. For example, at least one of the variance value of the constituent pixels of the rectangular block as the encoding unit, the difference between the maximum luminance and the minimum luminance, or the total variation (TV) is calculated, and the constituent pixels of the input block are calculated based on the calculated value An index value indicating the complexity of the pixel value is calculated, and the complexity of the constituent pixels of the encoding target block is determined.
  • the block determination unit 111 compares the complexity index value calculated for each block with a predetermined threshold value, and sets the following two determination values for the block.
  • A) High complexity block complexity determination value 1
  • B) Low complexity block complexity determination value 0
  • Block unit complexity determination information (complexity determination values 0 and 1), which is a block unit complexity determination result by the block determination unit 111, is notified to the coefficient correction unit 115.
  • Step S102 The process of step S102 is an optimal intra prediction (intra prediction) mode determination process in the intra prediction unit (intra prediction unit) 131.
  • the process of step S102 is executed in parallel with the block determination process of step S101.
  • the intra prediction unit (intra prediction unit) 131 has the following three predictions described above with reference to FIG. (A) DC prediction (b) Planar prediction (c) Directionality (angular) prediction It is determined which of these methods has the lowest cost (the amount of data reduction is large).
  • the reference pixel setting direction is also determined based on the cost. The cost is calculated based on, for example, the amount of encoded data when the difference between the encoded block and the reference block acquired from the decoded image data (reconstructed image) is encoded.
  • the intra prediction mode determination information determined by the in-screen prediction unit (intra prediction unit) 131 and the cost information corresponding to the mode are output to the prediction method determination unit 133, and the intra prediction image data generated according to the selected optimal intra prediction mode is And output to the prediction method switching unit 134. Furthermore, the optimal intra prediction mode information determined by the intra prediction unit (intra prediction unit) 131 is also output to the coefficient correction unit 115.
  • Step S103 Depending on the result of the block determination process in step S101, the process branches in step S103. That is, the processing branches depending on whether the complexity of the processing target block is low.
  • step S103 Yes
  • step S104 determines with the complexity of a process target block not being low
  • step S105 determines with the complexity of a process target block not being low
  • Step S104 If it is determined in step S103 that the complexity of the processing target block is low, the process is further branched in step S104. In step S104, the process branches depending on whether or not the optimal intra-screen prediction (intra prediction) mode in step S102 is directional prediction.
  • step S104 Yes
  • step S106 the process proceeds to step S105.
  • the coefficient correction unit 115 performs the quantization frequency only when the optimal in-screen prediction (intra prediction) mode determined by the in-plane prediction unit (intra prediction unit) 131 is directional prediction.
  • a conversion coefficient correction process is executed. In step S106 and subsequent steps, this coefficient correction process is performed.
  • Step S105 The process of step S105 is executed in either of the following two cases.
  • Step S106 The processes in steps S106 to S107 are processes related to the correction of the quantization frequency transform coefficient executed in the coefficient correction unit 115.
  • the processes in steps S106 to S107 are executed in the following cases.
  • the coefficient correction unit 115 first determines a coefficient correction range in step S106.
  • This coefficient correction range determination process is the process described above with reference to FIGS. That is, the coefficient correction unit 115 acquires the gradation direction of the encoding target block having gradations in various different directions from the block-corresponding optimal intra prediction mode information from the in-screen prediction unit (intra prediction unit) 131. To do.
  • the coefficient correction unit 115 selects and selects pattern data (DCT basis) that is important for expressing the gradation in the direction based on the gradation direction of the block acquired from the block-corresponding optimal intra prediction mode information.
  • a coefficient corresponding to pattern data (DCT base) is set as a correction target.
  • FIG. 12A which is an important element for expressing the gradation in the horizontal direction.
  • the pattern data (DCT base) (D00 to D07) surrounded by the white frame shown in FIG. 6 is selected, and the quantized frequency transform coefficients associated with the pattern data (DCT base) (D00 to D07), for example, a00 to a07 is selected as a correction target.
  • Step S107 When the determination of the correction range according to the gradation direction is completed in step S106, next, in step S107, a quantization frequency conversion coefficient correction process is executed.
  • This process is, for example, the process described above with reference to FIG.
  • the coefficient correction unit 115 compares a threshold value (QD th ) applied to coefficient correction with a value (QD rdd ) that is rounded down in the quantization process.
  • QD rdd > QD th (conditional expression 1)
  • Correction is performed by adding a correction amount expected in advance, for example, (+1) to the quantized transform coefficient with respect to the coefficient satisfying the conditional expression.
  • the coefficient correction unit 115 performs the correction process of the quantization frequency transform coefficient that brings the value lost by the quantization process in the quantization unit 114 close to the value before the quantization.
  • the correction range is a coefficient within the range selected according to the gradation direction in step S106. That is, only coefficients that are important for expressing the change in the gradient direction, and by performing these corrections, it becomes possible to reduce pseudo contours (banding) when decoding an encoded image.
  • step S107 the corrected quantized frequency transform coefficient generated by the coefficient correction unit 115 is output to the entropy encoding unit 116 and the inverse quantization unit 121.
  • the entropy encoding unit 116 performs entropy encoding processing (variable length encoding processing) on the corrected quantized frequency transform coefficient input from the coefficient correction unit 115 to generate the encoded image stream 20 and output it.
  • the inverse quantization unit 121 inversely quantizes the quantized data in the same quantization step as executed in the quantization unit 114, and restores transform coefficient data.
  • the inverse quantization unit 121 outputs the restored transform coefficient data to the inverse frequency transform unit 122.
  • a CPU (Central Processing Unit) 501 functions as a control unit or a data processing unit that executes various processes according to a program stored in a ROM (Read Only Memory) 502 or a storage unit 508. For example, processing according to the sequence described in the above-described embodiment is executed.
  • a RAM (Random Access Memory) 503 stores programs executed by the CPU 501 and data.
  • the CPU 501, ROM 502, and RAM 503 are connected to each other by a bus 504.
  • the CPU 501 is connected to an input / output interface 505 via a bus 504.
  • An input unit 506 including various switches, a keyboard, a mouse, a microphone, and a sensor, and an output unit 507 including a display and a speaker are connected to the input / output interface 505.
  • the CPU 501 executes various processes in response to a command input from the input unit 506 and outputs a processing result to the output unit 507, for example.
  • the storage unit 508 connected to the input / output interface 505 includes, for example, a hard disk and stores programs executed by the CPU 501 and various data.
  • a communication unit 509 functions as a transmission / reception unit for Wi-Fi communication, Bluetooth (BT) communication, and other data communication via a network such as the Internet or a local area network, and communicates with an external device.
  • BT Bluetooth
  • the drive 510 connected to the input / output interface 505 drives a removable medium 511 such as a magnetic disk, an optical disk, a magneto-optical disk, or a semiconductor memory such as a memory card, and executes data recording or reading.
  • a removable medium 511 such as a magnetic disk, an optical disk, a magneto-optical disk, or a semiconductor memory such as a memory card
  • the technology disclosed in this specification can take the following configurations.
  • a block determination unit that determines the complexity of the constituent pixels of a block to be encoded; For a block that is determined to be low in complexity, it has a coefficient correction unit that performs correction processing of a quantized frequency transform coefficient,
  • the coefficient correction unit is An image coding apparatus that determines a correction range of a quantization frequency conversion coefficient according to a gradation direction of a block and corrects the quantization frequency conversion coefficient of the determined correction range.
  • the image encoding device includes: It has an intra-screen prediction unit that performs intra-screen prediction (intra prediction) processing, The in-screen prediction unit determines whether the encoding process using DC prediction, planar prediction, or directionality (angular) prediction is the lowest cost, and the determination result is an optimal intra prediction mode. Output to the coefficient correction unit as information, The coefficient correction unit is The image coding apparatus according to (1), wherein the quantization frequency transform coefficient correction process is performed when the optimal intra prediction mode information input from the intra-screen prediction unit is directional prediction.
  • the coefficient correction unit When the optimal intra prediction mode information input from the intra prediction unit is directional prediction, the gradation direction of the block is analyzed based on the optimal intra prediction mode information, and based on the analyzed gradation direction.
  • the in-screen prediction unit It is determined which encoding process to which the directional prediction in 33 directions specified in HEVC (High Efficiency Video Coding) is the lowest cost, and the coefficient correction is performed using the determination result as the optimum intra prediction mode information.
  • Output to The coefficient correction unit is The image encoding device according to (2) or (3), wherein a correction range of the quantization frequency transform coefficient is determined based on direction information included in optimal intra prediction mode information input from the intra prediction unit.
  • the coefficient correction unit The image encoding device according to any one of (1) to (4), wherein a quantization frequency transform coefficient corresponding to pattern data that is an important element for representing block gradation is selected as a correction target.
  • the coefficient correction unit The frequency conversion coefficient generated by the frequency conversion unit; Compare the quantized frequency transform coefficient generated by the quantization process for the frequency transform coefficient in the quantization unit, A quantization frequency transform coefficient whose bit value lost by the quantization process is larger than a specified threshold value is selected, and correction of the selected quantization frequency transform coefficient is executed (1) to (5) The image encoding device described.
  • the image encoding device includes: Frequency at which a frequency conversion coefficient is calculated by performing frequency conversion of a residual signal that is a difference between a block to be encoded and intra prediction (intra prediction) image data or inter prediction (inter prediction) image data.
  • the coefficient correction unit is The image encoding device according to any one of (1) to (6), wherein correction processing of a quantization frequency transform coefficient calculated by the quantization unit is executed.
  • the coefficient correction unit Select pattern data (DCT base) including high-frequency components necessary for smoothly expressing the gradation of the block, and execute the correction processing of the quantized frequency transform coefficient corresponding to the selected pattern data (DCT base) ( The image encoding device according to any one of 1) to (8).
  • the block determination unit Whether at least one of the dispersion value of the block constituent pixels, the difference between the maximum luminance and the minimum luminance, or the total variation (TV) is calculated, and the input block is a block having constituent pixels with high complexity based on the calculated value
  • the image encoding device according to any one of (1) to (9), which determines whether or not.
  • An image encoding method executed in the image encoding device A block determination step in which the block determination unit determines the complexity of the constituent pixels of the block to be encoded; and A coefficient correction unit executes a coefficient correction step for performing a correction process of a quantization frequency transform coefficient on a block determined to have low complexity in the block determination step,
  • the coefficient correction step includes An image encoding method including processing for determining a correction range of a quantization frequency transform coefficient in accordance with a gradation direction of a block and correcting the quantization frequency transform coefficient of the determined correction range.
  • a program for executing an image encoding process in an image encoding device A block determination step for causing the block determination unit to determine the complexity of the constituent pixels of the block to be encoded;
  • a coefficient correction unit is caused to execute a coefficient correction step for executing a correction process of a quantization frequency transform coefficient for a block determined to have low complexity in the block determination step,
  • the series of processes described in the specification can be executed by hardware, software, or a combined configuration of both.
  • the program recording the processing sequence is installed in a memory in a computer incorporated in dedicated hardware and executed, or the program is executed on a general-purpose computer capable of executing various processing. It can be installed and run.
  • the program can be recorded in advance on a recording medium.
  • the program can be received via a network such as a LAN (Local Area Network) or the Internet and installed on a recording medium such as a built-in hard disk.
  • the various processes described in the specification are not only executed in time series according to the description, but may be executed in parallel or individually according to the processing capability of the apparatus that executes the processes or as necessary.
  • the system is a logical set configuration of a plurality of devices, and the devices of each configuration are not limited to being in the same casing.
  • an encoding process that can reduce the occurrence of pseudo contour (banding) is realized.
  • a block determination unit that determines the complexity of the constituent pixels of the block to be encoded, and a process of correcting the quantization frequency transform coefficient for the block determined to have low complexity
  • a coefficient correction unit is included.
  • the coefficient correction unit determines a correction range of the coefficient according to the gradation direction of the block, and corrects the coefficient in the determined correction range.
  • the in-screen prediction unit determines whether the encoding process to which DC prediction, plane prediction, or direction prediction is applied is the lowest cost, and outputs the determination result to the coefficient correction unit as optimal intra prediction mode information To do.
  • the coefficient correction unit performs a process of correcting the quantized frequency transform coefficient when the optimal intra prediction mode information is directional prediction.

Landscapes

  • Engineering & Computer Science (AREA)
  • Multimedia (AREA)
  • Signal Processing (AREA)
  • Compression Or Coding Systems Of Tv Signals (AREA)

Abstract

Un but de la présente invention est d'obtenir un processus de codage qui peut réduire l'apparition d'un pseudo-contour (cerclage). Un dispositif de codage d'image comprend : une unité de détermination de bloc qui détermine la complexité des pixels constitutifs d'un bloc à coder ; et une unité de correction de coefficient qui réalise un processus de correction d'un coefficient de conversion de fréquence de quantification pour un bloc déterminé pour avoir une faible complexité. L'unité de correction de coefficient détermine une plage de correction du coefficient selon la direction de gradation du bloc et corrige le coefficient dans la plage de correction déterminée. Une unité de prédiction dans l'écran détermine ce qui présente le coût le plus bas, un processus de codage auquel une prédiction de courant continu est appliquée, un processus de codage auquel une prédiction de plan est appliquée, ou un processus de codage auquel une prédiction directionnelle de laquelle une direction est appliquée, et délivre le résultat de détermination à l'unité de correction de coefficient en tant qu'informations de mode de prédiction intra optimal. L'unité de correction de coefficient réalise le processus de correction du coefficient de conversion de fréquence de quantification lorsque les informations de mode de prédiction intra optimal sont la prédiction directionnelle.
PCT/JP2019/015380 2018-04-26 2019-04-09 Dispositif de codage d'image, procédé de codage d'image, et programme WO2019208190A1 (fr)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
JP2018-085515 2018-04-26
JP2018085515 2018-04-26

Publications (1)

Publication Number Publication Date
WO2019208190A1 true WO2019208190A1 (fr) 2019-10-31

Family

ID=68295344

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/JP2019/015380 WO2019208190A1 (fr) 2018-04-26 2019-04-09 Dispositif de codage d'image, procédé de codage d'image, et programme

Country Status (1)

Country Link
WO (1) WO2019208190A1 (fr)

Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2015103970A (ja) * 2013-11-25 2015-06-04 富士通株式会社 動画像符号化装置、動画像符号化方法、動画像符号化プログラム、及び動画像撮影装置
WO2017033560A1 (fr) * 2015-08-25 2017-03-02 Kddi株式会社 Appareil de codage d'image animée, appareil de décodage d'image animée, procédé de codage d'image animée, procédé de décodage d'image animée et programme

Patent Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2015103970A (ja) * 2013-11-25 2015-06-04 富士通株式会社 動画像符号化装置、動画像符号化方法、動画像符号化プログラム、及び動画像撮影装置
WO2017033560A1 (fr) * 2015-08-25 2017-03-02 Kddi株式会社 Appareil de codage d'image animée, appareil de décodage d'image animée, procédé de codage d'image animée, procédé de décodage d'image animée et programme

Similar Documents

Publication Publication Date Title
KR102426721B1 (ko) 색차 성분 양자화 매개 변수 결정 방법 및 이러한 방법을 사용하는 장치
JP6606794B2 (ja) データ符号化及び復号化
JP6060229B2 (ja) ビデオ符号化における符号化ユニット量子化パラメータ
RU2696311C1 (ru) Устройство и способ для компенсации движения видео с выбираемым интерполяционным фильтром
JP5750579B2 (ja) イントラ予測モード復号化方法及び装置
JP5711098B2 (ja) 画像符号化方法，画像復号方法，画像符号化装置，画像復号装置およびそれらのプログラム
US11778179B2 (en) Image encoding method/device, image decoding method/device and recording medium having bitstream stored thereon
US20150063452A1 (en) High efficiency video coding (hevc) intra prediction encoding apparatus and method
WO2017030198A1 (fr) Dispositif de codage d'image, dispositif de décodage d'image et programmes correspondants
US10432961B2 (en) Video encoding optimization of extended spaces including last stage processes
TW202315408A (zh) 以區塊為基礎之預測技術
US20130235938A1 (en) Rate-distortion optimized transform and quantization system
US10277907B2 (en) Rate-distortion optimizers and optimization techniques including joint optimization of multiple color components
AU2011316747A1 (en) Internal bit depth increase in deblocking filters and ordered dither
US20210051320A1 (en) Adaptive Bilateral Filtering Using Look-Up Tables
JP2023101812A (ja) 画像処理装置、及び画像処理方法
US8792549B2 (en) Decoder-derived geometric transformations for motion compensated inter prediction
JP2021515494A (ja) 適応乗算係数を用いた画像フィルタリングのための方法および装置
JP2011239365A (ja) 動画像符号化装置及びその制御方法、コンピュータプログラム
WO2019208190A1 (fr) Dispositif de codage d'image, procédé de codage d'image, et programme
JP5302256B2 (ja) 符号化装置、復号装置及びプログラム
CN112313950A (zh) 视频图像分量的预测方法、装置及计算机存储介质
RU2800683C2 (ru) Способ и устройство предсказывания компонента видеоизображения и компьютерный носитель данных
WO2021045059A1 (fr) Dispositif de codage, dispositif de décodage et programme
JP2022070176A (ja) 符号化装置及びプログラム

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 19792145

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 19792145

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: JP