WO2005081540A1 - 画像符号化方法、その装置及びその制御プログラム - Google Patents
画像符号化方法、その装置及びその制御プログラム Download PDFInfo
- Publication number
- WO2005081540A1 WO2005081540A1 PCT/JP2005/002243 JP2005002243W WO2005081540A1 WO 2005081540 A1 WO2005081540 A1 WO 2005081540A1 JP 2005002243 W JP2005002243 W JP 2005002243W WO 2005081540 A1 WO2005081540 A1 WO 2005081540A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- dead zone
- image
- quantization
- width
- block
- Prior art date
Links
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
- H04N19/134—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
- H04N19/152—Data rate or code amount at the encoder output by measuring the fullness of the transmission buffer
-
- 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
-
- 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
- H04N19/105—Selection of the reference unit for prediction within a chosen coding or prediction mode, e.g. adaptive choice of position and number of pixels used for 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
- H04N19/102—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
- H04N19/102—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
- H04N19/132—Sampling, masking or truncation of coding units, e.g. adaptive resampling, frame skipping, frame interpolation or high-frequency transform coefficient masking
-
- 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
-
- 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
- H04N19/149—Data rate or code amount at the encoder output by estimating the code amount by means of a model, e.g. mathematical model or statistical model
-
- 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
- H04N19/17—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 an image region, e.g. an object
- H04N19/176—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 an image region, e.g. an object the region being a block, e.g. a macroblock
-
- 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
Definitions
- the present invention relates to an image encoding method, an apparatus therefor, and a control program therefor, and more particularly, to an image encoding method for performing adaptive quantization for improving subjective image quality, an apparatus therefor, and a control program therefor.
- an image frame forming a moving image is divided into a plurality of regions called macroblocks (MB), and encoding is performed on blocks obtained by finely dividing the MB.
- AVC ISO / IEC 14496-10
- Figure 2 shows the image frame configuration of Advanced Video Coding
- the MB is supplied by an intra-frame prediction device 5108 for predicting from within the same image frame reconstructed in the past, or an inter-frame prediction device 5109 for predicting from a previously reconstructed image frame.
- the predicted value is reduced.
- the MB signal whose predicted value has been reduced is called a prediction error signal.
- the prediction error signal is further divided into smaller blocks (hereinafter simply referred to as blocks), and a spatial domain force is also transformed into a frequency domain by an orthogonal transformer 5101.
- the orthogonal transform coefficient of the block transformed into the frequency domain by the quantization device 5102 is quantized by the quantization step size corresponding to the quantization parameter supplied in MB units from the quantization control device 5103. .
- the quantization control device 5103 monitors the generated code amount and increases the quantization parameter if the generated code amount is larger than the target code amount. Conversely, the quantization control device 5103 increases the quantization parameter. If it is less than the amount, reduce the quantization parameter. Thus, the moving image can be encoded with the target code amount.
- the quantized orthogonal transform coefficient is called a quantized transform coefficient, and is a variable-length coding device.
- Entropy coding is performed by 5104 and output.
- the quantized transform coefficient is inversely quantized by an inverse quantizer 5105, and inversely orthogonally transformed by an inverse orthogonal transformer 5106 to return to the original spatial domain.
- the predicted value is added to the block returned to the spatial area and stored in the frame memory 5107.
- An image frame reconstructed from the stored blocks is called a reference frame.
- the intra-frame prediction device 5108 is a prediction direction that minimizes the prediction error signal of the current MB from the reference frame
- the inter-frame prediction device 5109 is that the prediction error signal of the current MB is minimized from the reference frame. Detect a motion vector.
- the prediction determination switch 5110 compares the prediction error due to the intra-frame prediction with the prediction error due to the inter-frame prediction, and selects a prediction with a small prediction error.
- the quantization control device 5103 monitors an input image signal and a prediction error signal that are not limited to the generated code amount, and If the visual sensitivity of the MB is high, the quantization parameter is made small (fine quantization), and if the visual sensitivity is low, the quantization parameter is made large (coarse quantization) (the finer the quantization, the better the image quality).
- the conventional technology has the following three problems.
- the first problem is that the pictures of each block constituting the MB are not always the same. In such a case, the conventional technology cannot perform quantization suitable for the picture of each block constituting the MB.
- the second problem is that the individual blocks that make up the MB can be independently intra-frame predicted, or the individual blocks that make up the MB can be inter-frame predicted using independent motion vectors.
- the performance of minimizing the prediction error differs for each block constituting the MB (hereinafter, referred to as prediction performance).
- the conventional technology cannot perform quantization suitable for the prediction performance of each block constituting the MB.
- the third problem is that, for the first and second reasons, the distribution of the orthogonal transform coefficients corresponding to the coordinates inside the block (hereinafter, referred to as the spatial frequency) is different, and the distribution of each block constituting the MB is different. Is not uniform. In such a case, the conventional technology cannot perform quantization suitable for the distribution of orthogonal transform coefficients of each block.
- the quantization coefficient of the MB is adjusted according to the conversion coefficient having the highest visual sensitivity in the frequency domain within the MB or the block having the highest visual sensitivity in the spatial domain within the MB. Have no power to decide. As a result, other transform coefficients with low visual sensitivity in the frequency domain and blocks with low visual sensitivity in the spatial domain are quantized more than necessary. That is, an unnecessary information amount is assigned to a conversion coefficient having low visual sensitivity.
- the high-frequency transform coefficients are cut off from the transform coefficients in all the blocks constituting the MB more than the low-frequency transform coefficients, and in the inter-frame prediction, the cutoff of the coefficients is performed.
- Japanese Patent Application Laid-Open No. 2003-230142 (Document 1) describes a technique for improving the average subjective image quality of the entire image frame by turning it off without transmitting the quantization characteristic additional information.
- the pattern of the block, the block cannot perform quantization suitable for the prediction performance and the distribution of block transform coefficients.
- the code amount of the quantization parameter hereinafter referred to as quantization characteristic additional information
- the present invention has been made in view of the above problems, and has as its object to provide each transform coefficient without using the quantization characteristic additional information, and each block having a plurality of transform coefficients as constituent elements.
- the aim is to provide high-quality image coding technology by enabling quantization with flexible intensity.
- an object of the present invention is to provide a transform coding technique that performs quantization on a plurality of transform coefficients with the same quantization width, and visually recognizes transform coefficients in a frequency domain without adding additional information to a bit stream.
- An object of the present invention is to provide a higher quality image by enabling quantization according to sensitivity.
- an object of the present invention is to provide a technique of transform coding of an image in which a set of blocks having a plurality of transform coefficients as components is quantized with the same quantization width, and additional information is added to a bit stream.
- An object of the present invention is to provide a higher-quality image by enabling quantization according to the visual sensitivity of a block in a spatial domain without adding an image.
- an image encoding method uses a step of generating a transform coefficient by transforming an image from a spatial domain to a frequency domain, and using the same quantization width as at the time of decoding. And a step of quantizing the transform coefficient with a quantization characteristic different from the quantization characteristic at the time of decoding.
- the image coding apparatus of the present invention uses a transform unit that generates a transform coefficient by transforming an image from a spatial domain to a frequency domain, and uses the same quantization width as that used for decoding, And a quantization means for quantizing the transform coefficient with a quantization characteristic different from the quantization characteristic.
- control program for image encoding uses a computer that converts an image into a spatial domain and a frequency domain to generate transform coefficients, and uses the same quantization width as used in decoding.
- present invention is characterized in that it functions as a quantization means for quantizing a transform coefficient with a quantization characteristic different from the quantization characteristic at the time of decoding. The invention's effect
- a means for setting a visual sensitivity in a frequency domain of a transform coefficient and a dead zone width according to a visual sensitivity in a spatial domain of a block having a plurality of transform coefficients as components This provides a quantization function according to the visual sensitivity of the transform coefficient in the frequency domain and the visual sensitivity in the spatial domain of a block having a plurality of transform coefficients as components.
- the transform coefficient having low visual sensitivity in the frequency domain and the code amount wastedly consumed in blocks having low visual sensitivity in the spatial domain Can be reduced. Due to the reduction in the code amount, the quantization of the entire image frame becomes finer than in the conventional method, and a transform coefficient having high visual sensitivity in the frequency domain and a block having high visual sensitivity in the spatial domain are encoded with high image quality. .
- FIG. 1 is a diagram showing a configuration of a conventional technique.
- FIG. 2 is a diagram showing an image frame (only a luminance signal when the resolution is QCIF).
- FIG. 3 is a diagram showing an example of the configuration of the first embodiment.
- FIG. 4 is a flowchart of dead zone generation.
- FIG. 5 is a flowchart for generating a block dead zone scale.
- FIG. 6 is a quantization flow chart for one orthogonal transform coefficient.
- FIG. 7 is a diagram showing quantization characteristics (quantization step size q) of the conventional method.
- FIG. 10 is a view for explaining effects of the present invention.
- a indicates the complexity of each block (the smaller the value, the flatter)
- b indicates the quantization intensity according to the prior art
- c indicates the quantization intensity according to the present invention.
- the quantization intensity of MB is 20.
- FIG. 11 is a diagram illustrating an example of a configuration of a second embodiment.
- Replacement form (Rule 26)
- FIG. 12 is a diagram showing an example of a configuration of a spatial frequency dead zone scale generation device.
- FIG. 13 is an operation flowchart of the spatial frequency device characteristic type setting device.
- FIG. 14 is a view for explaining effects of the present invention.
- d indicates a block spatial frequency characteristic type inside MB
- e indicates a quantization type according to the related art
- f indicates a quantization type according to the present invention.
- “1” is a bidirectional prediction block
- “2” is a non-isolated motion block
- “3” is a normal motion block.
- FIG. 15 is a diagram showing quantization intensity characteristics for each type (only in the horizontal direction in a block).
- FIG. 16 is a diagram showing an example of the configuration of the third embodiment.
- FIG. 17 is a diagram showing an example of a configuration of a hybrid dead zone scale generation device.
- FIG. 18 is a diagram showing an example of the configuration of the fourth embodiment.
- FIG. 19 is an operation flowchart of the gap correction dead zone scale generation device.
- FIG. 20 is a diagram showing a configuration of an information processing apparatus using the present invention.
- a dead zone is generated using the same quantization width as at the time of decoding by using a dead zone generation device 201 and a block dead zone scale generation device 202 as shown in FIG.
- Each transform coefficient is quantized with a quantization characteristic different from the quantization characteristic at the time of decoding by quantizing each transform coefficient using the quantization coefficient.
- the visual sensitivity of the transform coefficient in the frequency domain and the multiple transform coefficients are obtained. It is possible to provide a quantization function according to the visual sensitivity in the spatial domain of a block including a coefficient as a component, and further reduce the code amount.
- a conversion coefficient having a high visual sensitivity in the frequency domain, or a high visual sensitivity in the spatial domain, and a low visual sensitivity in a frequency domain in which the dead zone width becomes narrower in the block, the conversion coefficient, or the spatial domain The lower the visual sensitivity, the wider the dead zone width is set for a block.
- the width of the dead zone is adaptively changed according to the flatness of the image.
- the flatness of the image is determined by the prediction mode of the image, the direction of intra-frame prediction of the image, the motion of the image, the direction of inter-frame prediction of the image, the average absolute value error of the image, the variance of the image, and the maximum value of the image. It is calculated from at least one of the difference between the minimum values, the average absolute value error of the image prediction error signal, and the variance of the image prediction error signal.
- FIG. 3 is an example showing the configuration of the first embodiment.
- an image frame constituting a moving image is divided into a plurality of regions called macroblocks (MB), and coding is performed on blocks obtained by finely dividing the MB. .
- MB macroblocks
- the MB is an intra-frame prediction device 108 that predicts from within the same image frame reconstructed in the past or an inter-frame prediction device 109 that predicts from an image frame reconstructed in the past. The value is reduced.
- the MB signal whose predicted value has been reduced is called a prediction error signal.
- the prediction error signal is further divided into smaller blocks (hereinafter simply referred to as blocks), and is transformed by the orthogonal transformer 101 into a spatial domain force frequency domain.
- the orthogonal transform coefficients of the block transformed into the frequency domain are quantized by the quantization device 102 at a quantization step size corresponding to the quantization parameter.
- the quantization parameter is supplied from the quantization control device 103 to the quantization device 102 in MB units.
- the quantization control device 103 monitors the generated code amount, and if the generated code amount is larger than the target code amount, increases the quantization parameter if the generated code amount is larger than the target code amount. If less, the quantization parameter is reduced.
- the moving image can be encoded with the target code amount.
- the quantized orthogonal transform coefficient is called a quantized transform coefficient, and is a variable-length coding device.
- the quantized transform coefficients are inversely quantized by the inverse quantizer 105, subjected to inverse orthogonal transform by the inverse orthogonal transformer 106, and returned to the original spatial domain.
- the predicted value is added to the block returned to the spatial area and stored in the frame memory 107.
- the image frame reconstructed from the stored blocks is called a reference frame.
- the intra-frame prediction device 108 minimizes the prediction error signal of the current MB from the reference frame
- the inter-frame prediction device 109 minimizes the prediction error signal of the current MB from the reference frame. Detect a motion vector.
- the prediction determination switch 110 compares the prediction error due to the intra-frame prediction with the prediction error due to the inter-frame prediction, and selects a prediction with a small prediction error.
- the quantization device 102 uses a dead zone when quantizing the orthogonal transform coefficient supplied from the orthogonal transform device 101. Dead zone means that the output corresponding to an input near 0 (zero) is set to 0 (zero). The range of input for performing such operations is called the dead zone width.
- the quantization device 102 sets the output obtained by quantizing the orthogonal transform coefficient, that is, the quantized transform coefficient to 0 (zero).
- the dead zone width is generated by dead zone generating device 201 and block dead zone scale generating device 202.
- the block dead zone scale generation device 202 receives the image signal and the prediction error, analyzes the picture or prediction performance of the target block, and generates a dead zone scale suitable for the picture and prediction performance of the block. Output to the device 201.
- the dead zone generation device 201 receives the dead zone scale from the block dead zone scale generation device 202 and the MB quantization parameter from the quantization control device 103 as inputs, and obtains a dead zone from the dead zone scale and the MB quantization parameter.
- the zone width is calculated, and the dead zone width is output to the quantization device 102.
- the dead zone width is obtained by multiplying the dead zone scale by the MB quantization parameter. Therefore, the dead zone scale is a coefficient of the MB quantization parameter used to determine the dead zone width.
- the size of the image frame is set to QCIF (176 x 144) size
- the size of MB is set to 16 x 16 size
- the size of the block constituting MB is set to 4 x 4 size.
- a dead zone generation device 201 a block dead zone scale generation device 202, and a quantization device 102 accompanied by a change in internal operation by the dead zone generation device 201, which are features of the present embodiment, will be described. I do.
- the input / output and operation of the dead zone generation device 201 will be described below.
- the input of the dead zone generation device 201 is a dead zone scale dz_scale (b, i, j) (0 ⁇ b ⁇ ) corresponding to the b-th block in the raster scan order of the MB currently targeted by the quantization device 102. 15, 0 ⁇ 3, 0 ⁇ j ⁇ 3), with the quantization parameter mb--q supplied from the quantization controller 103. is there.
- the output of the dead zone generation device 201 is the orthogonal transform coefficient co b, i, j of the b-th block in the raster scan order in the MB currently targeted by the quantization device 102 (0 ⁇ b ⁇ 15, The dead zone width dz (b, i, j) (0 ⁇ b ⁇ 15, 0 ⁇ i ⁇ 3, 0 ⁇ j ⁇ 3) corresponding to 0 ⁇ i ⁇ 3, 0 ⁇ j ⁇ 3).
- the operation of the dead zone generation device 201 will be described below with reference to FIG.
- a reference dead zone base_dz (i, j) (0 ⁇ i ⁇ 3, 0 ⁇ j ⁇ 3) is calculated from the quantization parameter mb_q.
- the calculation method of the reference dead zone is such that the encoding device to which the present invention is connected (hereinafter referred to as a base encoder) uses a quantization matrix WM (i, j) (0 ⁇ i ⁇ 3, 0 ⁇ j ⁇ 3). ) Depends on whether or not to use.
- the quantization matrix is a weighting parameter (quantization additional information) according to a spatial frequency for division in quantization and multiplication in inverse quantization.
- the quantization width for each spatial frequency can be made variable.
- the quantization step size q_step_table [q] is a quantization step size corresponding to the quantization parameter q defined by the base encoder (Q_MIN ⁇ p ⁇ Q_MAX, Q_MIN and Q_MAX also depend on the base encoder) .
- step S101A a reference dead zone base_dz (i, j) is calculated by equation (1).
- step S101B a reference dead zone base_dz (i, j) is calculated by equation (2).
- base— dz (i, j) mb—q—step (2)
- step S102 the dead zone width dz (i, j) is calculated from the reference dead zone base_dz (i, j) and the dead zone scale dz_scale (b, i, j) according to equation (3).
- dead zone width dz (b, i, j) can be set arbitrarily by the value of the dead zone scale dz_scale (b, i, j).
- the dead zone scale generation device 202 generates a dead zone scale suitable for the picture or prediction performance of each block having a plurality of transform coefficients as constituent elements.
- the input to the block dead zone scale generating device 202 is an input image signal org (b, i, j) (b, i, j) corresponding to the b-th block in the raster scan order of the MB currently targeted by the quantizing device 102. 0 ⁇ b ⁇ 15, 0 ⁇ i ⁇ 3, 0 ⁇ j ⁇ 3) and prediction error signal pd (b, i, j) (0 ⁇ b ⁇ 15, 0 ⁇ i ⁇ 3, 0 ⁇ j ⁇ 3) is there.
- bit precision of the input signal is n bits without sign.
- the output of the block dead zone scale generator 202 is a dead zone scale dz-scale (b, i, j) corresponding to the b-th block in the raster scan order within the MB currently targeted by the quantizer 102. (0 ⁇ b ⁇ 15, 0 ⁇ i ⁇ 3, 0 ⁇ j ⁇ 3).
- the block dead zone scale generator 202 uses the signal used to generate the dead zone scale as an image feature signal im b, i, j) (0 ⁇ b ⁇ 15, 0 ⁇ 3, 0 ⁇ j ⁇ 3)
- block dead zone scale generating apparatus 202 Referring to FIG. 5, the operation of block dead zone scale generating apparatus 202 will be described.
- step 301 an image feature signal is selected. There are three choices below.
- the quantization control device 102 of the base encoder determines a quantization parameter using an input image signal other than the generated code amount
- the quantization control device 102 inputs the image feature amount signal im b, i, j). Connect the force image signal org (b, i, j).
- the quantization controller 102 of the base encoder determines the quantization parameter using the prediction error signal pd in addition to the generated code amount and the input image signal, the image feature amount signal im b, i , j) is connected to the prediction error signal pd (b, i, j).
- step 302 the average absolute value error LlAC (b) corresponding to each block number b (0 ⁇ b ⁇ 15)
- abs (x) is a function that returns the absolute value of the input x.
- the average absolute value error L1AC (b) (l ⁇ LlAC (b) ⁇ n) indicates the distribution of the image feature signal within the block b.
- step 303 the block complexity bcm (b) (0 ⁇ b ⁇ 15) corresponding to each block number b (0 ⁇ b ⁇ 15) is calculated using equation (6).
- max (x, y) is a function that returns the magnitude of the value of the input x, y! /.
- the quantization intensity can be set according to the visual sensitivity of the block in the spatial domain (picture / predictability).
- step 304 the block dead zone scale bdz_scale (b) (0 ⁇ b ⁇ 15) corresponding to each block number b (0 ⁇ b ⁇ 15) is calculated using equation (7).
- bdz_scale (b) clip (bdz_limit, (bcm (b) / min— bcm)) ⁇ ⁇ )
- Parameter, clip (x, y) is the function that returns the value of input x, y which is smaller! /
- Min (bcm (b) is bcm (b) This function returns the minimum value of (l ⁇ bcm (b) ⁇ n) .
- the block dead zone scale is calculated in consideration of the complexity around the block. Then, the following equation (7A) may be used instead of equation (7)!
- locaLbcm (b) is a function that returns the minimum bcm value of the target block b and its surrounding blocks
- min (local_bcm (b) is a function that returns the minimum value of local_bcm (b) (l ⁇ bcm (b) ⁇ n) It is.
- step 305 the dead zone scale dz_scale (b, i, j) (0 ⁇ b ⁇ 15, 0 ⁇ i ⁇ ) corresponding to each block number b (0 ⁇ b ⁇ 15) is calculated using equation (9). 3. Set the block dead zone scale bdz_scale (b) to 0 ⁇ j ⁇ 3).
- the dead zone scale dz_scale of the block with low visual sensitivity in the spatial region where the visual sensitivity is high and the block dead zone scale dz_scale is small in the spatial region is large.
- the pixel range of the block (the maximum pixel value and the minimum pixel Value difference) may be used. That is, any information may be used as long as the complexity of the block can be obtained.
- the input of the quantization device 102 is a dead zone width dz (b, i, j) (0 ⁇ b ⁇ 15, 0 ⁇ i ⁇ 3, 0 ⁇ j ⁇ 3) supplied from the dead zone generation device 201.
- dz dead zone width
- Orthogonal transform coefficients co b, i, j) supplied from the orthogonal transformer 101 (0 ⁇ b ⁇ 15, 0 ⁇ i ⁇ 3, 0 ⁇ j ⁇ 3)
- the quantizer supplied from the quantization controller 103 Parameter mb_q.
- the output of the quantization device 102 is a quantized transform coefficient q_cof (b, i, j) (0 ⁇ b ⁇ 15, 0 ⁇ i ⁇ 3, 0 ⁇ j ⁇ 3).
- the input / output added to the conventional configuration is only the input dead zone width dz (b, i, j) (0 ⁇ b ⁇ 15, 0 ⁇ i ⁇ 3, 0 ⁇ j ⁇ 3).
- the output value of the quantization conversion coefficient q_co b, i, j) differs from the conventional method due to the influence of the operation described below.
- step S201 it is compared whether the absolute value abs_col of the orthogonal transform coefficient co b, i, j) is smaller than the dead zone width dz (b, i, j). If it is smaller, step S202 is executed. Otherwise, step S203 is executed.
- step S202 the quantized transform coefficient q_cof (b, i, j) is set to 0.
- step S203 a quantized transform coefficient q_cof (b, i, j) is obtained by the following calculation method.
- the method of calculating the quantized transform coefficients differs depending on whether or not the base encoder uses the quantization matrix WM (i, j) (0 ⁇ i ⁇ 3, 0 ⁇ j ⁇ 3). Each step will be described below with step S203A if the quantization matrix is used, and S / B if so!
- the quantization step size q_step_table [q] is a quantization step size corresponding to the quantization parameter q defined by the base encoder (Q_MIN ⁇ p ⁇ Q_MAX, Q_MIN and Q_MAX also depend on the base encoder) .
- step S203A the quantized transform coefficient q_cof (b, i, j) is calculated by equation (10A).
- abs (x) is a function that returns the absolute value of the input x
- f is a parameter that depends on the base encoder and is less than 1, which is 0.5 if rounded off and 0 if truncated.
- step S203B the quantized transform coefficient q_cof (b, i, j) is calculated by equation (10B).
- abs (x) is a function that returns the absolute value of the input x
- f is a parameter that depends on the base encoder and is less than 1, which is 0.5 for rounding and 0 for truncation.
- Quantization for one MB is completed by applying ⁇ 3).
- the quantization characteristic means the relationship between the input cof to the quantization device 102 and the output icof of the inverse quantization device 105.
- the output i_coi3 ⁇ 4S0 of the input cof smaller than 2q is obtained by quantizing the quantization step size four times.
- the visual sensitivity of a block is controlled by controlling the dead zone width dz in consideration of the pattern of the block, the prediction performance of the block, or the distribution of orthogonal transform coefficients in the block, which is simply determined by the prediction mode of the block.
- quantization optimal for the visual sensitivity of the transform coefficients in the block can be realized.
- the dead zone scale dz_scale supplied by the block dead zone scale generation device 202 of Embodiment 1 of the present invention can be controlled in consideration of the prediction performance of the picture / block of the block without adding the quantization additional information. That is, the quantization strength can be set according to the visual sensitivity of the block in the spatial domain as shown in FIG. According to the present invention, it is possible to set a quantization strength suitable for the visual sensitivity of a block in a spatial domain, and it is possible to reduce an unnecessary code amount generated in the block having a low visual sensitivity. As a result, the generated code amount of the entire image frame is also reduced, and the quantization parameter of the entire image frame is reduced. As a result, quantization of a block having high visual sensitivity in the spatial domain becomes finer than in the conventional method, and coding is performed with higher image quality.
- FIG. 11 shows the configuration of Embodiment 2 of the present invention.
- the configuration of the second embodiment includes a spatial frequency dead zone scale generator 203 instead of the block dead zone scale generator 202 in the configuration of the first embodiment.
- the spatial frequency dead zone scale generator 203 generates a dead zone scale dz_scale (b, i, j) (0 ⁇ b ⁇ 15, 0 ⁇ i ⁇ 3, 0 ⁇ j ⁇ 3) to the dead zone generator 201
- the present invention can be applied to other sizes.
- the spatial frequency dead zone scale generating device 203 generates a dead zone scale suitable for the distribution of the orthogonal transform coefficients of each block constituting the MB.
- the input to the spatial frequency dead zone scale generator 203 is an input image signal org (b, i, j) corresponding to the b-th block in the raster scan order of the MB currently targeted by the quantizer 102. (0 ⁇ b ⁇ 15, 0 ⁇ i ⁇ 3, 0 ⁇ j ⁇ 3), prediction mode mode (b) (0) corresponding to the b-th block in the raster scan order of the MB currently targeted by quantizer 102 ⁇ b ⁇ 15) Mv (b, dir) (0 ⁇ b ⁇ 15, 0 ⁇ dir ⁇ l).
- dir indicates the direction of the motion vector, where 0 is the horizontal direction and 1 is the vertical direction.
- the prediction mode includes an intra-frame prediction mode (0 motion vectors) for prediction from within the same image frame, and an inter-frame prediction mode for prediction of one past or future image frame.
- the output of the spatial frequency dead zone scale generator 203 is a dead zone scale dz_scale (b, i, j) corresponding to the b-th block in the raster scan order within the MB currently targeted by the quantizer 102. (0 ⁇ b ⁇ 15, 0 ⁇ i ⁇ 3, 0 ⁇ j ⁇ 3).
- the input / output of the spatial frequency dead zone scale generation device 203 has been described above.
- FIG. 12 shows the internal configuration of the spatial frequency dead zone scale generation device 203, and its operation will be described.
- the spatial frequency dead zone scale generating device 203 includes a spatial frequency characteristic setting device 2031 and a dead zone scale device 2032 for each characteristic type.
- the spatial frequency characteristic setting device 2031 uses the input image, prediction mode, and motion vector to calculate the distribution of the orthogonal transform coefficient of the b-th block in the raster scan order of the MB currently targeted by the quantization device 102. Outputs the characteristic type type (b) (0 ⁇ b ⁇ 15, 0 ⁇ type (b) ⁇ 3) according to.
- spatial frequency characteristic setting device 2031 With reference to FIG. 13, the operation of spatial frequency characteristic setting device 2031 will be described.
- step S4101 it is determined whether the prediction mode of block b is intra-frame prediction.
- step S41011 is executed.
- Range max_v (o, i, j) —mm—v (b, i, j) (12)
- max_v (b, i, j) is a function that returns the largest pixel value org (b, i, j) (0 ⁇ i ⁇ 3, 0 ⁇ j ⁇ 3) of block b
- min_v (b, i, j) is a function that returns the minimum pixel value org (b, i, j) (0 ⁇ 3, 0 ⁇ j ⁇ 3) of block b.
- the block is flat or textured, it is desirable to quantize the transform coefficient of the low-frequency component in the block finely and coarsely quantize the transform coefficient of the high-frequency component.
- step S4102 it is determined whether the prediction mode of block b is the bidirectional prediction mode.
- the inside of the bidirectional prediction block is a pan area or a still area, and has high visual sensitivity.
- the prediction error signal having a small power is noise generated by compressing a future or past frame, it is better to set the quantization strength at which the power is small and the prediction error signal is reduced!
- abs (x) is a function that returns the absolute value of the input x
- u_mv (b, dir) is a function that returns the motion vector mv in the dir direction of the block adjacent above block b
- l_mv (b, dir) is This function returns the motion vector mv in the dir direction of the block adjacent to the left side of block b.
- the non-isolated motion block is a pan area or a still area and has high visual sensitivity.
- the prediction error signal having a small power is noise generated by compressing a frame in the future or the past, the quantization power for reducing the power and reducing the prediction error signal is good.
- the characteristic-type dead zone scale device 2032 is a characteristic type type (b) corresponding to the b-th block in the raster scan order of the MB currently targeted by the quantization device 102 supplied by the spatial frequency characteristic setting device 2031.
- the dead zone scale dz_scale (b, i, j) (0 ⁇ b ⁇ 15, 0 ⁇ i ⁇ 3, 0 ⁇ j ⁇ 3) of the b-th block is calculated.
- the calculation method of the dead zone scale of block b according to the characteristic type (from typeO to 3) is shown below.
- B0, Bl, B2, B3, and B4 are predetermined parameters.
- B0 (i, j) ⁇ 0, 1.1, 1.3, 1.6 ⁇ , ⁇ 1.1, 1.3, 1.6, 1.8 ⁇ , ⁇ 1.3, 1.6, 1.8, 2.0 ⁇ , ⁇ 1.6, 1.8, 2.0,2.8 ⁇
- B3 (x) ⁇ 0, 1.1, 1.3, 1.4 ⁇ , ⁇ 1.1, 1.3, 1.4, 1.6 ⁇ , ⁇ 1.3, 1.4, 1.6, 1.8 ⁇ , ⁇ 1.4, 1.6, 1.8, 2.0 ⁇ , with the relationship B4>B1>B2> 1.
- the block width bw is a numerical value other than 4 in the present embodiment, it is shown that the values of B0 and B3 can be calculated by Expression (19).
- K (i, j) is also a large value depending on the spatial frequency (i, j) .
- the prediction mode in the intra-frame prediction mode and the prediction direction pred_dir can be supplied from the prediction determination 110, the prediction The inclination of type (0) should be changed according to the direction (vertical, horizontal, diagonal, etc.) of the direction pred_dir. For example, if the prediction direction is horizontal, the pattern inside the block is flat in the horizontal direction, and the quantization coefficient of the conversion coefficient corresponding to the frequency in the horizontal direction i is calculated from the conversion coefficient of the frequency in the vertical direction j. It is better to generate a dead zone scale dz_scale that quantizes finely
- the dead zone width according to the distribution of the transform coefficient of each block can be set by the dead zone scale dz_scale supplied by the spatial frequency dead zone scale generation device 203. That is, as shown in FIGS. 14 and 15, it is possible to perform quantization in consideration of visual sensitivity in the frequency domain of each transform coefficient without adding quantization additional information.
- a dead zone is set according to the distribution of transform coefficients of each block, and as a result, it is possible to reduce the generated code amount of transform coefficients with low visual sensitivity in the frequency domain. As a result, the generated code amount of the entire image frame is also reduced, and the quantization parameter of the entire image frame is reduced. As a result, a transform coefficient having high visual sensitivity in the frequency domain is quantized more finely than in the conventional method, and it is possible to perform encoding with higher image quality.
- Example 3 of the present invention will be described.
- FIG. 16 shows the configuration of Embodiment 3 of the present invention.
- the configuration of the third embodiment includes a hybrid dead zone scale generation device 204 instead of the block dead zone scale generation device 202 in the configuration of the first embodiment.
- Spatial frequency dead zone scale generator 203 Dead zone scale dz_scale (b, i, j) (0 ⁇ b ⁇ 15, 0 ⁇ i ⁇ 3, 0 ⁇ j ⁇ 3) corresponding to the b-th block in the raster scan order of the image frame Supply to
- the size of the image frame is set to QCIF (176 x 144) size
- the size of MB is set to 16 x 16 size
- the size of block constituting MB is set to 4 x 4 size.
- the hybrid dead zone scale generation device 204 generates a dead zone scale suitable for the picture of each block, the prediction performance of each block, and the distribution of the orthogonal transform coefficient of each block.
- the input to the hybrid dead zone scale generator 204 is a prediction mode mode (b) (0 ⁇ b ⁇ 15) corresponding to the b-th block in the raster scan order of the MB currently targeted by the quantizer 102.
- Motion vector mv (b, dir) (0 ⁇ b ⁇ 15, 0 ⁇ dir ⁇ l)
- input image signal org (b, i, j) (0 ⁇ b ⁇ 15, 0 ⁇ i ⁇ 3, 0 ⁇ j ⁇ 3)
- the prediction error signal pd (b, i, j) (0 ⁇ b ⁇ 15, 0 ⁇ i ⁇ 3, 0 ⁇ j ⁇ 3).
- the output of the hybrid dead zone scale generator 204 is a dead zone scale dz—scale (b, i, j) corresponding to the b-th block in the raster scan order within the MB currently targeted by the quantizer 102. ) (0 ⁇ b ⁇ 15, 0 ⁇ i ⁇ 3, 0 ⁇ j ⁇ 3).
- FIG. 17 shows the internal configuration of the hybrid dead zone scale generator 204, and its operation will be described below.
- the hybrid dead zone generation device 204 as shown in FIG. 17 includes a block dead zone scale generation device 202, a spatial frequency dead zone scale generation device 203, and a mixer 2041.
- the block dead zone scale generation device 202 is a spatial frequency dead zone according to the first embodiment.
- the scale generation device 203 is as described in the second embodiment.
- the input to the mixer 2041 is a dead zone scale ldz_scale l (b, b) corresponding to the b-th block in the raster scan order within the MB currently targeted by the quantization device 102 supplied by the block dead zone device 202.
- the output of the mixer 2041 is a dead zone scale dz_scale (b, i, j) (0 ⁇ b ⁇ 15) corresponding to the b-th block in the raster scan order within the MB currently targeted by the quantizer 102. , 0 ⁇ i ⁇ 3, 0 ⁇ j ⁇ 3).
- the mixer 2041 calculates the dead zone scale dz.scale (b, i, j) according to the characteristic type type (b) of the block b.
- dz_scale (b, i, j) dz_scalel (b, i, j) X dz_scale2 (b, i, j) (20)
- dz_scale (b, i, j) dz_scalel (b, i, j) X dz_scale2 (b, i, j) (20)
- the visual sensitivity (pattern, prediction performance) in the spatial region of the block, and the It enables quantization suitable for visual sensitivity (distribution) in the frequency domain of transmutation coefficients.
- the generated code amount of the entire image frame is also reduced, and the quantization parameter of the entire image frame is reduced.
- a block having a high visual sensitivity in the spatial domain and a transform coefficient having a high visual sensitivity in the frequency domain are quantized more finely than in the conventional method, and encoding can be performed with higher image quality.
- Example 4 of the present invention will be described.
- FIG. 18 shows the configuration of Example 4 of the present invention.
- the configuration of the fourth embodiment includes a gap correction dead zone scale generation device 205 instead of the block dead zone scale generation device 202 in the configuration of the first embodiment.
- the gap correction dead zone scale generation device 205 is a dead zone scale dz_scale (b, i, j) corresponding to the b-th block in the raster scan order of the image frame (0 ⁇ b ⁇ 15, 0 ⁇ i ⁇ 3, 0 ⁇ j ⁇ 3) is supplied to the dead zone generator 201.
- the size of the image frame is
- the present invention can be applied to other sizes.
- the input to the gap correction dead zone scale generation device 205 is supplied from the quantization control device 103, the actual quantization parameter mb_q of the MB currently targeted by the quantization device 102, and the quantization control device 103. This is the ideal quantization parameter ideaLq of the MB currently targeted by the quantizer 102 to be implemented.
- the output of the gap correction dead zone scale generation device 205 is the dead zone scale dz_scale (b, i, j) corresponding to the b-th block in the raster scan order within the MB currently targeted by the quantization device 102. (0 ⁇ b ⁇ 15, 0 ⁇ i ⁇ 3, 0 ⁇ j ⁇ 3).
- step S501 the gap quantization width qstep_gap of the real quantization parameter mb_q and the ideal quantization parameter ideaLq is calculated using Expression (22).
- the quantization step size q_st mark _table [q] is a quantization step size corresponding to the quantization parameter q defined by the base encoder (Q_MIN ⁇ p ⁇ Q_MAX, Q_MIN and Q--MAX are also included). Base encoder dependent).
- step S502 the dead zone scale dz_scale (b, i, j) is calculated from the gap quantization width qstep_gap using Expression (23).
- the first is that the prediction mode of the MB selected by the prediction decision of the base encoder 110 cannot transmit the difference between the quantization parameter of the current MB or the quantization parameter of the previous MB, and This is the case where the ideal MB quantization parameter of the quantization control device 103 of the encoder is larger than the actual MB quantization parameter.
- the base encoder has a limitation on the difference delta_mb_Q from the quantization parameter of the previous MB that can be transmitted for each MB (for example, -2 ⁇ delta_mb_Q ⁇ 2), and This is a case where the ideal MB quantization parameter of the quantization control device 103 of the encoder is larger than the actual MB quantization parameter.
- Example 5 of the present invention will be described.
- the image encoding apparatus can also be realized by a computer program that can be configured by hardware.
- FIG. 20 is a general block configuration diagram of an information processing system that implements the moving picture coding apparatus according to the present invention.
- the information processing system (computer) shown in FIG. 20 includes a processor A1001, a program memory A1002, and storage media A1003 and A1004.
- the storage media A1003 and A1004 may be separate storage media or storage areas having the same storage medium power.
- a magnetic storage medium such as a hard disk can be used.
- the present invention relates to an image transform coding technique, in which the visual sensitivity in the frequency domain of transform coefficients and the visual sensitivity in the spatial domain of a block having a plurality of transform coefficients as components are improved.
- Means for setting a corresponding dead zone width, whereby quantization according to the visual sensitivity in the frequency domain of the transform coefficient and the visual sensitivity in the spatial domain of a block having a plurality of transform coefficients as components is provided. It is possible to provide functions.
- the present invention wastefully consumes conversion coefficients having low visual sensitivity in the frequency domain and blocks having low visual sensitivity in the spatial domain without depending on the quantization width determined by the quantization parameter.
- the amount of code that has been used can be reduced, and by reducing the amount of code, the quantization of the entire image frame becomes more intense than in the conventional method, and the conversion coefficient with high visual sensitivity in the frequency domain and the high visual sensitivity in the spatial domain
- the blocks are coded with high quality.
Landscapes
- Engineering & Computer Science (AREA)
- Multimedia (AREA)
- Signal Processing (AREA)
- Physics & Mathematics (AREA)
- Algebra (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Analysis (AREA)
- Mathematical Optimization (AREA)
- Pure & Applied Mathematics (AREA)
- Compression Or Coding Systems Of Tv Signals (AREA)
- Compression, Expansion, Code Conversion, And Decoders (AREA)
Abstract
Description
Claims
Priority Applications (4)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
JP2006510197A JP4613909B2 (ja) | 2004-02-20 | 2005-02-15 | 画像符号化方法、その装置及びその制御プログラム |
US10/598,154 US8743956B2 (en) | 2004-02-20 | 2005-02-15 | Image encoding method, device thereof, and control program thereof |
EP05719145A EP1718080A4 (en) | 2004-02-20 | 2005-02-15 | BILDCODE PROCESS, DEVICE AND CONTROL PROGRAM THEREFOR |
CN2005800054937A CN1922886B (zh) | 2004-02-20 | 2005-02-15 | 图像编码方法及其设备 |
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
JP2004-044011 | 2004-02-20 | ||
JP2004044011 | 2004-02-20 |
Publications (1)
Publication Number | Publication Date |
---|---|
WO2005081540A1 true WO2005081540A1 (ja) | 2005-09-01 |
Family
ID=34879326
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/JP2005/002243 WO2005081540A1 (ja) | 2004-02-20 | 2005-02-15 | 画像符号化方法、その装置及びその制御プログラム |
Country Status (7)
Country | Link |
---|---|
US (1) | US8743956B2 (ja) |
EP (3) | EP2611157A1 (ja) |
JP (2) | JP4613909B2 (ja) |
KR (1) | KR100860147B1 (ja) |
CN (1) | CN1922886B (ja) |
TW (1) | TWI255651B (ja) |
WO (1) | WO2005081540A1 (ja) |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2010268508A (ja) * | 2004-02-20 | 2010-11-25 | Nec Corp | 画像符号化方法、その装置及びその制御プログラム |
JP2010541454A (ja) * | 2007-10-04 | 2010-12-24 | サムスン エレクトロニクス カンパニー リミテッド | 復号化器での量子化係数調整方法及び装置 |
WO2015045301A1 (ja) * | 2013-09-27 | 2015-04-02 | 日本電気株式会社 | 映像符号化装置、映像符号化方法および映像符号化プログラム |
Families Citing this family (31)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US8422546B2 (en) * | 2005-05-25 | 2013-04-16 | Microsoft Corporation | Adaptive video encoding using a perceptual model |
JP2007151075A (ja) * | 2005-10-31 | 2007-06-14 | Seiko Epson Corp | 印刷装置、印刷方法、画像処理装置、画像処理方法、印刷プログラム、画像処理プログラム、及び記録媒体 |
US8059721B2 (en) | 2006-04-07 | 2011-11-15 | Microsoft Corporation | Estimating sample-domain distortion in the transform domain with rounding compensation |
US7995649B2 (en) | 2006-04-07 | 2011-08-09 | Microsoft Corporation | Quantization adjustment based on texture level |
US20070237237A1 (en) * | 2006-04-07 | 2007-10-11 | Microsoft Corporation | Gradient slope detection for video compression |
US8503536B2 (en) | 2006-04-07 | 2013-08-06 | Microsoft Corporation | Quantization adjustments for DC shift artifacts |
US8711925B2 (en) | 2006-05-05 | 2014-04-29 | Microsoft Corporation | Flexible quantization |
US8238424B2 (en) | 2007-02-09 | 2012-08-07 | Microsoft Corporation | Complexity-based adaptive preprocessing for multiple-pass video compression |
US8498335B2 (en) * | 2007-03-26 | 2013-07-30 | Microsoft Corporation | Adaptive deadzone size adjustment in quantization |
US20080240257A1 (en) * | 2007-03-26 | 2008-10-02 | Microsoft Corporation | Using quantization bias that accounts for relations between transform bins and quantization bins |
US8243797B2 (en) | 2007-03-30 | 2012-08-14 | Microsoft Corporation | Regions of interest for quality adjustments |
US8442337B2 (en) | 2007-04-18 | 2013-05-14 | Microsoft Corporation | Encoding adjustments for animation content |
US8331438B2 (en) * | 2007-06-05 | 2012-12-11 | Microsoft Corporation | Adaptive selection of picture-level quantization parameters for predicted video pictures |
US20090060027A1 (en) * | 2007-08-30 | 2009-03-05 | Tektronix, Inc. | Compressed Signal Subjective Quality Ratings Prediction |
JP5055078B2 (ja) * | 2007-10-01 | 2012-10-24 | キヤノン株式会社 | 画像処理装置及びその方法 |
EP2046053A1 (en) * | 2007-10-05 | 2009-04-08 | Thomson Licensing | Method and device for adaptively quantizing parameters for image coding |
KR100939917B1 (ko) | 2008-03-07 | 2010-02-03 | 에스케이 텔레콤주식회사 | 움직임 예측을 통한 부호화 시스템 및 움직임 예측을 통한부호화 방법 |
US8189933B2 (en) | 2008-03-31 | 2012-05-29 | Microsoft Corporation | Classifying and controlling encoding quality for textured, dark smooth and smooth video content |
US20090262801A1 (en) * | 2008-04-17 | 2009-10-22 | Qualcomm Incorporated | Dead zone parameter selections for rate control in video coding |
US8897359B2 (en) | 2008-06-03 | 2014-11-25 | Microsoft Corporation | Adaptive quantization for enhancement layer video coding |
CN101742323B (zh) * | 2008-11-05 | 2013-05-01 | 上海天荷电子信息有限公司 | 无再损视频编码和解码的方法和装置 |
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 |
JP5484083B2 (ja) * | 2010-01-14 | 2014-05-07 | 株式会社メガチップス | 画像処理装置 |
JP5536071B2 (ja) | 2010-10-05 | 2014-07-02 | エンパイア テクノロジー ディベロップメント エルエルシー | 空間光パターンに基づく深さデータの生成 |
KR20130108457A (ko) | 2011-03-09 | 2013-10-02 | 닛본 덴끼 가부시끼가이샤 | 영상 부호화 장치, 영상 복호 장치, 영상 부호화 방법 및 영상 복호 방법 |
CN103548350B (zh) | 2011-06-28 | 2017-03-01 | 日本电气株式会社 | 图像编码设备和图像解码设备 |
US20130188691A1 (en) * | 2012-01-20 | 2013-07-25 | Sony Corporation | Quantization matrix design for hevc standard |
US10356428B2 (en) * | 2015-04-13 | 2019-07-16 | Qualcomm Incorporated | Quantization parameter (QP) update classification for display stream compression (DSC) |
WO2017082698A1 (ko) * | 2015-11-11 | 2017-05-18 | 삼성전자 주식회사 | 비디오 복호화 방법 및 그 장치 및 비디오 부호화 방법 및 그 장치 |
US20180041870A1 (en) * | 2016-08-05 | 2018-02-08 | Wymond Choy | Detection Using NFC Open Circuit |
KR102636100B1 (ko) | 2016-12-16 | 2024-02-13 | 삼성전자주식회사 | 데드존에 기초하여 양자화를 수행하는 인코더 및 이를 포함하는 비디오 처리 시스템 |
Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JPS6146685A (ja) * | 1984-08-13 | 1986-03-06 | Nec Corp | 動画像信号の予測符号化装置 |
JPS62209984A (ja) * | 1986-03-10 | 1987-09-16 | Nec Corp | 画像信号の符号化方法およびその装置 |
JPH02105792A (ja) * | 1988-10-14 | 1990-04-18 | Nippon Telegr & Teleph Corp <Ntt> | 直交変換係数量子化回路 |
JPH0440118A (ja) * | 1990-06-05 | 1992-02-10 | Matsushita Electric Ind Co Ltd | 量子化器 |
JPH05308629A (ja) * | 1992-05-01 | 1993-11-19 | Olympus Optical Co Ltd | 動画像符号化方式 |
JPH07236142A (ja) * | 1993-10-18 | 1995-09-05 | Mitsubishi Electric Corp | 高能率符号化装置及び復号化装置 |
JP2003230142A (ja) * | 2001-11-30 | 2003-08-15 | Sony Corp | 画像情報符号化方法及び装置、並びにプログラム及び記録媒体 |
Family Cites Families (15)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US1066079A (en) * | 1906-06-30 | 1913-07-01 | Ashton Valve Company | Signal for trains. |
CA1277416C (en) * | 1984-08-13 | 1990-12-04 | Akihiro Furukawa | Inter-frame predictive coding apparatus for video signal |
JPH06146685A (ja) * | 1992-10-30 | 1994-05-27 | Oi Seisakusho Co Ltd | 自動車用ドアロックの操作装置 |
US5724097A (en) * | 1993-10-18 | 1998-03-03 | Mitsubishi Denki Kabushiki Kaisha | Adaptive quantization of video based on edge detection |
JP3788823B2 (ja) * | 1995-10-27 | 2006-06-21 | 株式会社東芝 | 動画像符号化装置および動画像復号化装置 |
JP2830855B2 (ja) | 1996-08-22 | 1998-12-02 | 日本電気株式会社 | 適応量子化制御装置 |
US6434196B1 (en) * | 1998-04-03 | 2002-08-13 | Sarnoff Corporation | Method and apparatus for encoding video information |
US6408026B1 (en) * | 1999-08-06 | 2002-06-18 | Sony Corporation | Deadzone quantization method and apparatus for image compression |
US7003038B2 (en) | 1999-09-27 | 2006-02-21 | Mitsubishi Electric Research Labs., Inc. | Activity descriptor for video sequences |
CN1199468C (zh) * | 2000-01-12 | 2005-04-27 | 皇家菲利浦电子有限公司 | 图像数据压缩 |
US7280700B2 (en) * | 2002-07-05 | 2007-10-09 | Microsoft Corporation | Optimization techniques for data compression |
TWI323613B (en) * | 2002-07-29 | 2010-04-11 | Qualcomm Inc | Digital image encoding |
US6853318B1 (en) * | 2003-12-30 | 2005-02-08 | Eastman Kodak Company | Digital image compression utilizing shrinkage of subband coefficients |
EP1564997A1 (en) * | 2004-02-12 | 2005-08-17 | Matsushita Electric Industrial Co., Ltd. | Encoding and decoding of video images based on a quantization with an adaptive dead-zone size |
KR100860147B1 (ko) | 2004-02-20 | 2008-09-24 | 닛본 덴끼 가부시끼가이샤 | 화상 부호화 방법, 그 장치 및 제어 프로그램을 기록한 컴퓨터 판독가능한 기록 매체 |
-
2005
- 2005-02-15 KR KR20067016716A patent/KR100860147B1/ko not_active IP Right Cessation
- 2005-02-15 US US10/598,154 patent/US8743956B2/en not_active Expired - Fee Related
- 2005-02-15 EP EP20130160117 patent/EP2611157A1/en not_active Ceased
- 2005-02-15 EP EP20100191897 patent/EP2290986A1/en not_active Ceased
- 2005-02-15 CN CN2005800054937A patent/CN1922886B/zh not_active Expired - Fee Related
- 2005-02-15 WO PCT/JP2005/002243 patent/WO2005081540A1/ja not_active Application Discontinuation
- 2005-02-15 JP JP2006510197A patent/JP4613909B2/ja not_active Expired - Fee Related
- 2005-02-15 EP EP05719145A patent/EP1718080A4/en not_active Withdrawn
- 2005-02-18 TW TW94104821A patent/TWI255651B/zh not_active IP Right Cessation
-
2010
- 2010-07-22 JP JP2010164660A patent/JP4952828B2/ja not_active Expired - Fee Related
Patent Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JPS6146685A (ja) * | 1984-08-13 | 1986-03-06 | Nec Corp | 動画像信号の予測符号化装置 |
JPS62209984A (ja) * | 1986-03-10 | 1987-09-16 | Nec Corp | 画像信号の符号化方法およびその装置 |
JPH02105792A (ja) * | 1988-10-14 | 1990-04-18 | Nippon Telegr & Teleph Corp <Ntt> | 直交変換係数量子化回路 |
JPH0440118A (ja) * | 1990-06-05 | 1992-02-10 | Matsushita Electric Ind Co Ltd | 量子化器 |
JPH05308629A (ja) * | 1992-05-01 | 1993-11-19 | Olympus Optical Co Ltd | 動画像符号化方式 |
JPH07236142A (ja) * | 1993-10-18 | 1995-09-05 | Mitsubishi Electric Corp | 高能率符号化装置及び復号化装置 |
JP2003230142A (ja) * | 2001-11-30 | 2003-08-15 | Sony Corp | 画像情報符号化方法及び装置、並びにプログラム及び記録媒体 |
Non-Patent Citations (1)
Title |
---|
See also references of EP1718080A4 * |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2010268508A (ja) * | 2004-02-20 | 2010-11-25 | Nec Corp | 画像符号化方法、その装置及びその制御プログラム |
US8743956B2 (en) | 2004-02-20 | 2014-06-03 | Nec Corporation | Image encoding method, device thereof, and control program thereof |
JP2010541454A (ja) * | 2007-10-04 | 2010-12-24 | サムスン エレクトロニクス カンパニー リミテッド | 復号化器での量子化係数調整方法及び装置 |
WO2015045301A1 (ja) * | 2013-09-27 | 2015-04-02 | 日本電気株式会社 | 映像符号化装置、映像符号化方法および映像符号化プログラム |
Also Published As
Publication number | Publication date |
---|---|
JP2010268508A (ja) | 2010-11-25 |
EP2611157A1 (en) | 2013-07-03 |
TW200529672A (en) | 2005-09-01 |
EP1718080A1 (en) | 2006-11-02 |
CN1922886A (zh) | 2007-02-28 |
EP2290986A1 (en) | 2011-03-02 |
JP4952828B2 (ja) | 2012-06-13 |
EP1718080A4 (en) | 2011-01-12 |
KR20060118593A (ko) | 2006-11-23 |
TWI255651B (en) | 2006-05-21 |
JP4613909B2 (ja) | 2011-01-19 |
CN1922886B (zh) | 2012-08-01 |
KR100860147B1 (ko) | 2008-09-24 |
US20070140333A1 (en) | 2007-06-21 |
JPWO2005081540A1 (ja) | 2008-03-06 |
US8743956B2 (en) | 2014-06-03 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
WO2005081540A1 (ja) | 画像符号化方法、その装置及びその制御プログラム | |
US8498335B2 (en) | Adaptive deadzone size adjustment in quantization | |
JP4747975B2 (ja) | 画像処理装置および方法、プログラム、並びに、記録媒体 | |
JP4187405B2 (ja) | 符号化方式におけるオブジェクトベースのレート制御装置及びその方法 | |
KR0184905B1 (ko) | 부호량제어장치 및 이것을 사용한 부호화장치 | |
KR20060072070A (ko) | 이미지 또는 픽쳐 시퀀스를 인코딩하는데 사용될 수 있는양자화 매트릭스를 발생하는 방법 및 장치 | |
US7430329B1 (en) | Human visual system (HVS)-based pre-filtering of video data | |
CN105379269A (zh) | 兴趣区域感知的视频编码 | |
KR19990076563A (ko) | 동화상 부호화 방식 | |
JP2004527960A (ja) | メディアプロセッサにおけるmpeg2デコード処理の動的複雑性予測及び調整 | |
GB2306846A (en) | Determining quantization interval in video signal encoder | |
WO2002054772A2 (en) | Method of performing video encoding rate control using bit budget | |
KR20080108538A (ko) | 비디오 코딩 방법 | |
KR100708182B1 (ko) | 동영상 부호화기의 비트율 제어 장치 및 방법 | |
JP6946979B2 (ja) | 動画像符号化装置、動画像符号化方法、及び動画像符号化プログラム | |
KR20100012149A (ko) | 양자화/역 양자화 장치 및 방법과 그를 이용한 영상부호화/복호화 장치 | |
JP3069144B2 (ja) | 動画像符号化装置 | |
KR20020087810A (ko) | 미세입자 스케일러블 코딩의 적응적 선택 강화 적용 방법 | |
JP2005005862A (ja) | 画像符号化装置 | |
Tun et al. | An efficient rate control algorithm for a wavelet video codec | |
Mahmood et al. | A content-aware quantisation mechanism for transform domain distributed video coding | |
JPH05219496A (ja) | 画像符号化装置及び動画像符号化装置 | |
US20060182175A1 (en) | Image encoding apparatus, image encoding method, and computer program product | |
Tun et al. | Rate control algorithm based on quality factor optimization for Dirac video codec | |
KR0141297B1 (ko) | 비선형 양자화방법 |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
AK | Designated states |
Kind code of ref document: A1 Designated state(s): AE AG AL AM AT AU AZ BA BB BG BR BW BY BZ CA CH CN CO CR CU CZ DE DK DM DZ EC EE EG ES FI GB GD GE GH GM HR HU ID IL IN IS JP KE KG KP KR KZ LC LK LR LS LT LU LV MA MD MG MK MN MW MX MZ NA NI NO NZ OM PG PH PL PT RO RU SC SD SE SG SK SL SY TJ TM TN TR TT TZ UA UG US UZ VC VN YU ZA ZM ZW |
|
AL | Designated countries for regional patents |
Kind code of ref document: A1 Designated state(s): GM KE LS MW MZ NA SD SL SZ TZ UG ZM ZW AM AZ BY KG KZ MD RU TJ TM AT BE BG CH CY CZ DE DK EE ES FI FR GB GR HU IE IS IT LT LU MC NL PL PT RO SE SI SK TR BF BJ CF CG CI CM GA GN GQ GW ML MR NE SN TD TG |
|
121 | Ep: the epo has been informed by wipo that ep was designated in this application | ||
DPEN | Request for preliminary examination filed prior to expiration of 19th month from priority date (pct application filed from 20040101) | ||
WWE | Wipo information: entry into national phase |
Ref document number: 2006510197 Country of ref document: JP |
|
WWE | Wipo information: entry into national phase |
Ref document number: 2005719145 Country of ref document: EP |
|
WWE | Wipo information: entry into national phase |
Ref document number: 2007140333 Country of ref document: US Ref document number: 10598154 Country of ref document: US |
|
WWE | Wipo information: entry into national phase |
Ref document number: 200580005493.7 Country of ref document: CN Ref document number: 1020067016716 Country of ref document: KR |
|
NENP | Non-entry into the national phase |
Ref country code: DE |
|
WWW | Wipo information: withdrawn in national office |
Country of ref document: DE |
|
WWP | Wipo information: published in national office |
Ref document number: 2005719145 Country of ref document: EP |
|
WWP | Wipo information: published in national office |
Ref document number: 1020067016716 Country of ref document: KR |
|
WWP | Wipo information: published in national office |
Ref document number: 10598154 Country of ref document: US |