WO2012117955A1 - 画像符号化装置 - Google Patents
画像符号化装置 Download PDFInfo
- Publication number
- WO2012117955A1 WO2012117955A1 PCT/JP2012/054526 JP2012054526W WO2012117955A1 WO 2012117955 A1 WO2012117955 A1 WO 2012117955A1 JP 2012054526 W JP2012054526 W JP 2012054526W WO 2012117955 A1 WO2012117955 A1 WO 2012117955A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- picture
- code amount
- quantization parameter
- value
- image
- Prior art date
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
- H04N19/134—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
- H04N19/142—Detection of scene cut or scene change
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T9/00—Image coding
- G06T9/007—Transform coding, e.g. discrete cosine transform
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
- H04N19/102—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the element, parameter or selection affected or controlled by the adaptive coding
- H04N19/124—Quantisation
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
- H04N19/102—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
- H04N19/146—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
- H04N19/134—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
- H04N19/15—Data rate or code amount at the encoder output by monitoring actual compressed data size at the memory before deciding storage at 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
- H04N19/10—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
- H04N19/172—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 picture, frame or field
-
- 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
- H04N19/177—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 group of pictures [GOP]
-
- 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
- H04N19/60—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 apparatus, and more particularly to an image encoding apparatus that encodes image data using a characteristic value indicating the complexity of input image data.
- Image encoding devices are MPEG2 and H.264.
- Image data broadcast by digital broadcasting or the like is recorded on a recording medium such as a DVD using an image encoding technology such as H.264.
- the image encoding device executes a code amount control process based on recording conditions such as the capacity and recording time of the recording medium.
- Non-Patent Document 1 describes TM5 (Test Model 5), which is one of the code amount control methods.
- TM5 is a technique proposed in the process of standardization of the MPEG2 encoding system.
- the TM5 executes a code amount control process using a characteristic value of image data called an activity.
- the activity is a characteristic value indicating the complexity of the image.
- the activity of a macroblock is calculated according to the following procedure. An absolute difference value between a pixel value of a certain pixel in the macro block and a pixel average value of the pixels in the macro block is calculated. The total sum of absolute differences of each pixel in the macroblock is calculated as the activity of the macroblock.
- Patent Document 1 describes a technique for detecting a scene change based on an activity of image data.
- JP 2009-232148 A “Test Model 5”, ISO / IEC-JTC1 / SC29 / WG11, April 1993
- Activity is a parameter used to determine image data encoding conditions such as code amount control and scene change detection. However, by using an activity as the complexity of image data, an appropriate encoding condition may not be selected. If an appropriate encoding condition is not selected, the bit rate of the encoded image data may be greatly deviated from a preset target bit rate, or the image quality of the encoded image data may be deteriorated.
- An image encoding device is an image encoding device that encodes uncompressed image data in units of pictures, generates frequency component data by Hadamard transform of the first picture, and includes AC included in the frequency component data.
- a Hadamard transform unit that calculates the characteristic value of the first picture by summing up the absolute values of the component values, and an encoding unit that encodes the first picture using the characteristic value as a parameter indicating the complexity of the first picture And comprising.
- an appropriate encoding condition for the image data can be selected by using the characteristic value when encoding the first picture.
- the image coding apparatus further includes a code amount calculation unit that calculates a target code amount of a picture that is a target value of a code amount generated by encoding the first picture, a characteristic value, and a target code amount of the picture
- a code amount calculation unit that calculates a target code amount of a picture that is a target value of a code amount generated by encoding the first picture, a characteristic value, and a target code amount of the picture
- a first quantization parameter determination unit that determines a quantization parameter used for encoding the first picture based on the first picture, and the encoding unit encodes the first picture using the quantization parameter.
- the accuracy of code amount control can be improved by determining the quantization parameter of the first picture based on the characteristic value.
- the image coding apparatus further includes a characteristic value of a first picture and a characteristic value of a coded first picture closest to the first picture among the first pictures of an image group composed of a plurality of pictures. And a scene change determination unit that determines that a scene change has occurred in the first picture when the absolute difference between and is larger than the first threshold.
- the characteristic value includes the frequency component of the picture
- the presence / absence of a scene change can be determined in consideration of the fluctuation of the frequency component between pictures.
- the image coding apparatus further includes the first picture when the coded first picture closest to the first picture among the first pictures of the picture group composed of a plurality of pictures is the first picture.
- a first difference calculation unit for calculating a first difference absolute value of the quantization parameter of the first head picture and a quantization parameter of the first head picture, and a predetermined number of encoded head pictures calculated from the first picture to the second head In the case of a picture, a second difference calculation unit that calculates a second difference absolute value between the quantization parameter of each second leading picture and the quantization parameter of the coded leading picture located immediately before each second leading picture And a correction unit that corrects the quantization parameter of the first picture so that the sum of the first difference absolute value and all the second difference absolute values is equal to or less than a predetermined value.
- an object of the present invention is to provide a technique capable of appropriately selecting an encoding condition of image data.
- FIG. 1 It is a functional block diagram which shows the structure of the image coding apparatus by the 1st Embodiment of this invention. It is a flowchart of the encoding process which the image coding apparatus shown in FIG. 1 performs. It is a figure which shows the calculation procedure of the Hadamard value which the Hadamard transformation part shown in FIG. 1 calculates. As shown in FIG. It is a figure which shows the arrangement
- FIG. 1 determines a quantization parameter. It is a figure which shows the correlation with the code amount of an intra picture, and an activity when the quantization parameter is determined based on the activity of a picture. It is a flowchart of the quantization parameter determination process by the 2nd Embodiment of this invention. 10 is a flowchart of a quantization parameter correction process shown in FIGS. 6 and 9. As shown in FIG. 2 is a diagram illustrating a configuration of H.264 data. FIG. It is a figure which shows the correction direction set by the correction process shown in FIG. As shown in FIG. It is a figure which shows the change of the quantization parameter of H.264 data.
- FIG. 1 is a block diagram showing a functional configuration of an image encoding device 1 according to the first embodiment of the present invention.
- the image encoding device 1 converts the uncompressed image data 21 into the H.264 format.
- H.264 data 29 is output.
- the image encoding device 1 includes a Hadamard transform unit 11, a scene change determination unit 12, a quantization parameter determination unit 13, an encoding unit 14, and a QP correspondence table 15.
- the Hadamard transform unit 11 performs Hadamard transform on the uncompressed image data 21 which is moving image data, and generates frequency component data 22 (see FIG. 3).
- the Hadamard transform unit 11 generates an Hadamard value 23 by adding the AC component values included in the frequency component data 22. Since the Hadamard value 23 is calculated for each picture of the uncompressed image data 21, H.264 data 29 corresponds to each picture.
- the scene change determination unit 12 determines whether or not a scene change has occurred in the current picture using the Hadamard value 23 of each picture and the generated code amount of GOP (Group Of Picture).
- the current picture is a picture to be encoded.
- the quantization parameter determination unit 13 determines the quantization parameter 24 of the current picture based on the Hadamard value 23 of the current picture, the picture target code amount, and the QP correspondence table 15.
- the picture target code amount is a target value of the code amount generated when the current picture is encoded.
- the QP correspondence table 15 is a table in which quantization parameters 24 corresponding to both the Hadamard value 23 and the picture target code amount are set.
- the quantization parameter determination unit 13 includes a code amount calculation unit 131, an error calculation unit 132, and a determination method selection unit 133.
- the code amount calculation unit 131 calculates a GOP ideal code amount, a GOP target code amount, and a picture target code amount.
- the GOP ideal code amount is the H.264 per GOP unit. It is an ideal value of the code amount of the H.264 data 29, and is calculated based on the target bit rate set before encoding.
- the GOP target code amount is a value obtained by adjusting the GOP ideal code amount based on the GOP generated code amount.
- the GOP generation code amount is H.264 in GOP units. This is a code amount of H.264 data 29.
- the error calculation unit 132 calculates an overall error and a period error based on the GOP ideal code amount and the GOP generated code amount.
- the overall error and the period error are used for calculating the picture target code amount. Details of the overall error and the period error will be described later.
- the determination method selection unit 133 selects a method for determining the quantization parameter 24 of the current picture from one of the following two methods.
- the first method is a method of determining a quantization parameter using the Hadamard value 23 of the current picture.
- the second method is a method of determining the quantization parameter 24 of the immediately previous encoded I (Intra) picture as the quantization parameter 24 of the current picture.
- the encoding unit 14 inputs the uncompressed image data 21.
- the encoding unit 14 encodes the current picture using the quantization parameter 24 of the current picture, H.264 data 29 is generated.
- the image encoding device 1 generates frequency component data 22 by Hadamard transform of the current picture.
- the sum of the AC component values of the frequency component data 22 is calculated as the Hadamard value 23.
- the image encoding device 1 uses the Hadamard value 23 as a characteristic value indicating the complexity of the image in the picture.
- the complexity indicates the degree of variation in the pixel value of each pixel included in the picture.
- the Hadamard value 23 includes a frequency component of a picture. For this reason, when the Hadamard value 23 is used as the characteristic value indicating the complexity of the image, the picture can be encoded in consideration of variations in the frequency components of the picture. Therefore, it is possible to determine the coding condition of the picture with high accuracy.
- the image encoding device 1 uses the Hadamard value 23 to determine whether or not a scene change has occurred in the current picture.
- the image encoding device 1 can determine the presence / absence of a scene change in consideration of changes in frequency components between pictures. Accordingly, it is possible to improve the scene change detection accuracy.
- the image encoding device 1 determines the quantization parameter 24 based on the Hadamard value 23 of the current picture. Since the correlation between the Hadamard value 23 and the generated code amount of the picture is higher than the activity, the accuracy of the code amount control can be improved.
- FIG. 2 is a flowchart of the encoding process executed by the image encoding device 1.
- the code amount calculation unit 131 calculates the GOP ideal code amount.
- the GOP ideal code amount is H.264. H.264 data 29 frame rate; It is calculated based on the target bit rate of H.264 data 29 and the number of pictures per GOP.
- the Hadamard transform unit 11 starts calculating the Hadamard value 23 of each picture of the uncompressed image data 21 (step S1).
- the Hadamard transform unit 11 calculates the Hadamard value 23 of each picture in parallel with the processing of steps S2 to S6 described later.
- the image encoding device 1 determines a picture to be encoded (current picture) (step S2).
- the scene change determination unit 12 determines whether or not a scene change has occurred in the current picture based on the Hadamard value 23 of the current picture (step S3).
- the quantization parameter determination unit 13 determines the quantization parameter 24 of the current picture based on the scene change determination result (step S4). When the scene change occurs in the current picture, the quantization parameter determination unit 13 determines the quantization parameter 24 of the current picture based on the Hadamard value 23 of the current picture.
- the encoding unit 14 encodes the current picture using the quantization parameter determined by the quantization parameter determining unit 13 (step S5). After encoding the current picture, the image encoding device 1 determines whether or not to end the encoding process of the uncompressed image data 21 (step S6). When the encoding process ends (Yes in step S6), the image encoding device 1 ends the process shown in FIG. When the encoding process is not finished (No in step S6), the image encoding apparatus 1 repeatedly executes the processes in steps S2 to S5.
- the Hadamard transform unit 11 calculates the Hadamard value 23 of each picture in parallel with the determination of the quantization parameter (Step S4) and the coding of the picture (Step S5).
- FIG. 3 is a schematic diagram showing a flow of calculating the Hadamard value 23.
- the picture 21P is a picture of the uncompressed image data 21, and is original image data that has not been subjected to preprocessing such as prediction processing. In FIG. 3, the sizes of the pixels 21a to 21h are exaggerated.
- the Hadamard transform unit 11 performs Hadamard transform on the pixel values of the eight pixels 21a to 21h arranged in the horizontal direction, and generates frequency component data 22 including a DC component H0 and AC components H1 to H7. In this way, the Hadamard transform unit 11 performs Hadamard transform on each pixel of the picture 21P in units of 8 pixels in the horizontal direction. Note that the encoding unit 14 does not use the frequency component data 22 when encoding the current picture.
- the encoding unit 14 performs Hadamard transform separately from the Hadamard transform unit 11 when encoding the current picture.
- the absolute sum of all AC components obtained by the horizontal Hadamard transform is calculated as the Hadamard value 23. That is, the Hadamard value 23 is a total value of absolute values of all AC components obtained by Hadamard transforming all pixels of a picture in units of 8 pixels, and is calculated in units of pictures.
- the Hadamard transform unit 11 outputs a Hadamard value 23 to the scene change determination unit 12 and the quantization parameter determination unit 13. Since the Hadamard value 23 can be calculated without performing the Hadamard transform in the vertical direction, the amount of calculation when calculating the Hadamard value 23 can be reduced.
- FIG. 2 is a diagram illustrating an arrangement of pictures of H.264 data 29.
- FIG. 4 “I” indicates an I picture.
- “B” indicates a B (Bi-Directional Predictive) picture.
- “P” indicates a P (Predictive) picture.
- the I picture, the P picture, and the B picture are collectively referred to as “picture”.
- GOPs 30, 40, and 50 have one I picture.
- the I picture is arranged at the head of each GOP.
- FIG. 5 is a flowchart of the scene change determination process (step S3).
- the scene change determination unit 12 determines whether or not a scene change has occurred, based on two criteria: a change in Hadamard value 23 and a generated code amount of GOP.
- the flow of the scene change determination process will be described by taking as an example the case where the GOP 40 including the pictures 41 to 49 is the GOP to be encoded (current GOP).
- the scene change determination unit 12 determines a picture (comparison target picture) to be compared with the current picture (step S31).
- the comparison target picture is the first picture (I picture 41) of the GOP 40.
- the comparison target picture is the first picture (I picture 31) of the GOP 30 encoded immediately before the GOP 40. That is, the scene change determination unit 12 determines an encoded I picture closest to the current picture as a comparison target picture.
- the scene change determination unit 12 determines whether or not a scene change has occurred based on a change in the Hadamard value 23.
- the scene change determination unit 12 calculates a Hadamard difference value, which is an absolute difference between the Hadamard value 23 of the current picture and the Hadamard value 23 of the comparison target picture (step S32).
- the scene change determination unit 12 compares the Hadamard difference value with a first SC (Scene Change) threshold value (step S33).
- the first SC threshold is calculated by multiplying the Hadamard value 23 of the comparison target picture by a predetermined first SC coefficient. Since the comparison target picture is the first picture (I picture) of the GOP, the first SC threshold value changes in GOP units.
- the first SC threshold value may be a fixed value set before the start of encoding of the uncompressed image data 21.
- the scene change determination unit 12 determines that a scene change has occurred in the current picture (step S37). That is, when the Hadamard value 23 of the current picture changes beyond a threshold value obtained from the Hadamard value 23 of the comparison target picture, the scene change determination unit 12 determines that a scene change has occurred.
- the scene change determination unit 12 checks whether or not the current picture is the I picture 41 (step S34). If the current picture is not the I picture 41 (No in step S34), the scene change determination unit 12 ends the process of FIG.
- the scene change determination unit 12 uses the generated code amount of the GOP encoded immediately before to determine whether or not a scene change has occurred. judge. Specifically, the scene change determination unit 12 calculates a code amount difference value (step S35).
- the code amount difference value is an absolute difference between the GOP ideal code amount and the generated code amount of the GOP 30 encoded immediately before the GOP 40.
- the scene change determination unit 12 determines that a scene change has occurred in the current picture (I picture 41) (Step S37).
- the second SC threshold is calculated by multiplying the GOP ideal code amount by a predetermined second SC coefficient indicating a scene change determination criterion. That is, when the ratio of the code amount difference value to the GOP ideal code amount exceeds the threshold value obtained from the GOP ideal code amount, it is determined that a scene change has occurred in the current picture (I picture 41).
- the scene change determination unit 12 determines that no scene change has occurred in the current picture, and ends the process of FIG.
- the scene change determination unit 12 determines whether or not a scene change has occurred in the current picture using the Hadamard difference value. Since the Hadamard value 23 is calculated by Hadamard transforming the picture, the frequency component of the picture is taken into consideration. That is, since a scene change can be detected based on a change in frequency components between the current picture and the comparison target picture, the accuracy of scene change determination can be improved.
- the scene change determination unit 12 determines the presence / absence of a scene change based on the GOP ideal code amount and the generated code amount of the GOP 30 encoded immediately before the current GOP (GOP 40). To do. As described above, the scene change determination accuracy can be improved by determining the presence / absence of a scene change by using the two parameters of the Hadamard value 23 and the generated code amount of the GOP 30 encoded immediately before the current picture. .
- step S4 the quantization parameter determination process (step S4, see FIG. 2) will be described.
- the quantization parameter 24 of the encoded I picture closest to the current picture is used as the quantization parameter 24 of the current picture.
- the quantization parameter 24 of the current picture is set to It is determined based on the Hadamard value 23.
- FIG. 6 is a flowchart of the quantization parameter determination process (step S4).
- the quantization parameter determination unit 13 checks whether or not a scene change has occurred in the current picture (step S401).
- the quantization parameter determination unit 13 determines the quantization parameter 24 using the Hadamard value 23 of the current picture regardless of the picture type of the current picture. To do.
- the code amount calculation unit 131 calculates the picture target code amount of the current picture based on the GOP ideal code amount (step S402).
- the code amount calculation unit 131 regards the current picture as an I picture and calculates a picture target code amount. Adjustment of the quantization parameter 24 according to the picture type is performed in step S411 described later. Specifically, the picture target code amount is calculated by multiplying the GOP ideal code amount by the I picture ratio regardless of the picture type of the current picture.
- the I picture ratio is calculated as a ratio of the generated code amount of the I picture 31 to the generated code amount of the GOP 30 located immediately before the GOP 40.
- the quantization parameter determination unit 13 determines the quantization parameter 24 using the Hadamard value 23 of the current picture, the picture target code amount, and the QP correspondence table 15 (step S403).
- the QP correspondence table 15 is a two-dimensional table in which quantization parameters corresponding to both the Hadamard value 23 and the picture target code amount are set.
- the quantization parameter determination unit 13 determines the quantization parameter 24 with reference to the QP correspondence table 15 using the Hadamard value 23 of the current picture and the picture target code amount as input parameters.
- step S403 it is desirable to determine the quantization parameter 24 by converting the Hadamard value 23 and the picture target code amount of the current picture into an average value per macroblock.
- the Hadamard value 23 and the picture target code amount per macroblock are set as input parameters. Thereby, it is not necessary to prepare the QP correspondence table 15 for each picture size.
- the quantization parameter determination unit 13 determines a quantization parameter based on the Hadamard value 23 and the picture target code amount regardless of the type of the picture. To do. This is because when a scene change occurs, the image quality may be deteriorated by setting the quantization parameter 24 of the encoded I picture to the quantization parameter 24 of the current picture.
- the quantization parameter determination unit 13 checks whether the current picture is an I picture (step S404). When the current picture is a P picture or a B picture (No in step S404), the quantization parameter 24 of the current picture is determined to be the quantization parameter 24 of the encoded I picture closest to the current picture (step S405). ). When the current picture is any of the pictures 42 to 49, the quantization parameter 24 of the I picture 41 is determined as the quantization parameter 24 of the current picture.
- the quantization parameter determination unit 13 executes the process of step S406.
- the determination method selection unit 133 selects the determination method of the quantization parameter 24 from either the first method or the second method based on whether or not the code amount difference value exceeds the selection reference value (Step S1). 406).
- the first method is a method of determining based on the Hadamard value 23 of the current picture.
- the second method uses the quantization parameter 24 of the immediately preceding encoded I picture.
- the code amount difference value is calculated as an absolute difference value between the GOP ideal code amount and the generated code amount of the GOP encoded immediately before the current GOP.
- the selection reference value will be described later.
- steps S406 to S410 the case where the I picture 41 is the current picture and the GOP 40 is the current GOP will be described as an example unless otherwise specified.
- the quantization parameter 24 of the first picture (I picture 31) of the GOP 30 encoded immediately before is set as the quantization parameter 24 of the I picture 41.
- the difference between the generated code amount of GOP 30 and the GOP ideal code amount is not so large as to determine that a scene change has occurred in I picture 41, but is relatively large.
- the quantization parameter 24 of the first picture of the GOP 30 is set to the quantization parameter 24 of the picture 41, the difference between the generated code amount of the GOP 40 and the GOP ideal code amount is relatively large as in the GOP 30. Is likely to be.
- the quantization parameter determination unit 16 selects a method for determining the quantization parameter 24 of the I picture 41 based on the code amount difference value in step S406.
- the quantization parameter determination unit 13 obtains a code amount difference value, which is an absolute difference between the GOP ideal code amount and the generated code amount of GOP30, as in step S35 (see FIG. 5). calculate.
- the quantization parameter determination unit 13 confirms whether the code amount difference value exceeds the selection reference value.
- the selection reference value is a value serving as a reference for determining whether or not the Hadamard value 23 is used to determine the quantization parameter 24, and is smaller than the second SC threshold value.
- the selection reference value is calculated by multiplying the GOP ideal code amount by a predetermined selection coefficient.
- the selection coefficient is smaller than the second SC coefficient used for scene change determination. This is because, as described above, if a scene change has occurred (Yes in step S401), the quantization parameter 24 of the I picture 41 is calculated based on the Hadamard value 23. (Step S403).
- the determination method selection unit 133 selects a method using the quantization parameter 24 of the encoded I picture.
- the quantization parameter 24 of the I picture 41 is determined to be the quantization parameter 24 of the encoded I picture 31 closest to the I picture 41 (step S405). This is because even if the quantization parameter 24 of the I picture 31 is used for encoding the GOP 40, it is considered that the difference between the generated code amount of the GOP 40 and the GOP ideal code amount does not increase.
- the determination method selection unit 133 determines that the difference between the generated code amount of GOP30 and the GOP ideal code amount is relatively large. For this reason, a method for determining the quantization parameter 24 based on the Hadamard value 23 is selected.
- the code amount calculation unit 131 calculates the target code amount (GOP target code amount) of the GOP 40 (step S407). In order to converge the generated code amount in GOP units to the GOP ideal code amount, the GOP target code amount is calculated based on the GOP ideal code amount and the generated code amount of the encoded GOP.
- the error calculation unit 132 uses (Equation 1) to calculate an overall error that occurs with the encoding of the uncompressed image data 21.
- ET represents the overall error.
- Qd represents the GOP ideal code amount.
- Qg indicates a generated code amount of the encoded GOP. That is, the error calculation unit 132 calculates a value (individual error) obtained by subtracting the GOP ideal code amount from the generated code amount of the encoded GOP, and sums the individual errors of each encoded GOP, thereby reducing the total error. calculate.
- the error calculation unit 132 calculates a period error using (Expression 2).
- Ep indicates a period error.
- range indicates the number of encoded GOPs for which a period error is to be calculated. That is, the error calculation unit 132 specifies a predetermined number of encoded GOPs based on the current GOP among the encoded GOPs, and sums the individual errors of the specified encoded GOPs to reduce the period error. calculate.
- the error determination unit 132 calculates the GOP target code amount using (Equation 3).
- Qa indicates the GOP target code amount.
- Ce is a coefficient of 1 or less that is multiplied by the overall error and the period error, and is set before the uncompressed image data 21 is encoded.
- the GOP target code amount is calculated based on the GOP ideal code amount, the overall error, and the period error.
- a lower limit may be set for the GOP target code amount.
- the quantization parameter 24 of the I picture 41 is determined based on the Hadamard value 23 and the picture target code amount calculated from the GOP target code amount.
- the GOP target code amount is significantly lower than the GOP ideal code amount, it is assumed that the quantization parameter 24 is determined to be an extremely high value.
- the image quality of the GOP 40 to which the I picture 41 belongs greatly deteriorates.
- the H.P. The image quality of the H.264 data 29 can be maintained at a certain level or higher.
- the code amount calculation unit 131 calculates the picture target code amount (step S408).
- the picture target code amount is calculated by multiplying the GOP target code amount by the I picture ratio.
- the quantization parameter determination unit 13 determines the quantization parameter 24 of the I picture 41 based on the Hadamard value 23 of the I picture 41, the picture target code amount, and the QP correspondence table 15 (step S409). ).
- the quantization parameter of the I picture 41 is changed. 24 is determined based on the Hadamard value 23.
- the bit rate of the H.264 data 29 can be converged to the target bit rate.
- the quantization parameter determination unit 13 corrects the quantization parameter 24 of the I picture 41 based on the quantization parameter 24 of the encoded I picture (step S410). As a result, H.C. The image quality of the H.264 data 29 is prevented from changing rapidly.
- the quantization parameter determination unit 13 specifies the quantization parameter 24 of a predetermined number (for example, three) of encoded I pictures with reference to the GOP 40.
- the quantization parameter determination unit 13 calculates the sum of absolute differences between the quantization parameter 24 of the I picture 41 and the quantization parameter 24 of the identified encoded I picture.
- the quantization parameter 24 of the I picture 41 is corrected so that the difference absolute value sum is equal to or less than the change upper limit value. Details of step S410 will be described in the second embodiment.
- step S411 will be described.
- the process of step S411 is executed after the quantization parameter 24 of the current picture is determined in steps S403 and S405, and after the quantization parameter 24 of the current picture is corrected in step S410.
- the quantization parameter 24 of the current picture determined in any of steps S403, S405, and S410 is adjusted according to the picture type of the current picture.
- the quantization parameter determination unit 13 adds an offset value corresponding to the picture type of the current picture to the quantization parameter 24 (step S411).
- the offset value is set to a value larger than 0.
- the offset value of the I picture may be a value larger than 0.
- the quantization parameter determination unit 13 confirms whether the quantization parameter 24 of the current picture exceeds a preset upper limit value and does not fall below a lower limit value (step S412). If the quantization parameter 24 exceeds the upper limit value, the quantization parameter 24 is set to the upper limit value. If the quantization parameter 24 is below the lower limit value, the quantization parameter 24 is set to the lower limit value. In this way, the quantization parameter 24 of the current picture is determined.
- the encoding unit 14 encodes the current picture using the quantization parameter 24 determined by the quantization parameter determination unit 13 (see step S5, FIG. 2).
- the image encoding device 1 sets the quantization parameter 24 of the current picture based on the Hadamard value 23 of the current picture. Thereby, the accuracy of the code amount control when encoding the uncompressed image data 21 can be improved. Hereinafter, this reason will be described.
- FIG. 7 is a diagram illustrating a correlation between the code amount of the encoded I picture and the Hadamard value of the encoded I picture when the quantization parameter of each picture is determined based on the above procedure.
- FIG. 8 is a diagram illustrating a correlation between the code amount of the encoded I picture and the activity of the encoded I picture when the quantization parameter of the I picture is determined based on the activity. 7 and 8, the vertical axis indicates the code amount per macroblock.
- the variation in the generated code amount of the I picture is greater when the quantization parameter is determined using the Hadamard value than when the quantization parameter is determined using the activity. small. Therefore, when the quantization parameter of the I picture is determined using the Hadamard value, variation in the generated code amount of the picture can be suppressed, so that the accuracy of code amount control can be improved.
- FIG. 9 is a flowchart of the quantization parameter determination process (step S4) according to the second embodiment of the present invention.
- the second embodiment is different from the first embodiment in that the quantization parameter 24 correction process (step S410) is executed even after the quantization parameter 24 of the current picture is determined in step S403. It is.
- FIG. 10 is a flowchart of the quantization parameter correction process (step S410).
- FIG. 2 is a diagram illustrating an array of GOPs that form H.264 data 29.
- step S410 will be described in detail with reference to FIG. 10 and FIG. 11, taking as an example the case where the first picture (I picture 61) of GOP 60 is the current picture.
- the quantization parameter determination unit 13 determines the quantization parameter 24 of the I picture 61 when a scene change occurs (Yes in step S451), the quantization parameter determination unit 13 sets the correction direction using the Hadamard value 23 of the I picture 61 ( Step S452).
- the correction direction is a parameter indicating whether the quantization parameter 24 of the I picture 61 determined in step S403 or S409 is increased or decreased with reference to the quantization parameter 24 of the encoded I picture 51.
- FIG. 12 is a diagram illustrating the correction direction of the quantization parameter 24 of the I picture 61. In FIG. 12, the horizontal axis is the number of each picture, and the sign of the picture is used as the value on the horizontal axis for convenience.
- the quantization parameter determination unit 13 specifies the encoded I picture 51 closest to the I picture 61. Specifically, the quantization parameter determination unit 13 specifies the GOP 50 encoded immediately before the GOP 60 to which the I picture 61 belongs, and specifies the I picture 51 as the first picture of the GOP 50.
- the quantization parameter determination unit 13 determines that the complexity increases from the I picture 51 to the I picture 61, and the correction direction is upward. (Direction of arrow 65).
- the quantization parameter 24 of the I picture 61 is corrected so as to be equal to or greater than the quantization parameter 24 of the I picture 51. It should be noted that the quantization parameter 24 of the I picture 61 is not corrected at the time of step S452, but is corrected in step S456 described later.
- the quantization parameter determination unit 13 determines that the complexity has decreased, and sets the correction direction downward (in the direction of the arrow 66). decide.
- the quantization parameter 24 of the I picture 61 is corrected so as to be equal to or less than the quantization parameter 24 of the I picture 51.
- step S453 determines the quantization parameter 24 of the I picture 61 when the code amount difference value is larger than the selection reference value (No in step S451), the correction direction is determined using the GOP generation code amount. Is determined (step S453). That is, step S453 is executed when the quantization parameter of the current picture is determined by the processing of steps S404 to SS09 shown in FIG. 6 or FIG.
- the quantization parameter determination unit 13 specifies the encoded GOP (GOP50) closest to the I picture 61.
- the quantization parameter determination unit 13 determines the correction direction to be upward (the direction of the arrow 65) in order to increase the code amount.
- the quantization parameter determination unit 13 determines the correction direction downward (in the direction of the arrow 66) in order to suppress the code amount.
- FIG. 13 is a diagram illustrating changes in the quantization parameter 24.
- the horizontal axis indicates the picture number, and the code of the I picture is used for convenience.
- the quantization parameters 24 of the I pictures 31, 41 and 51 are “25”, “24” and “26”, respectively, and the quantization parameter 24 of the I picture 61 (current picture) is determined to be “23”. Steps S454 and S455 will be described by taking the case as an example.
- the quantization parameter determination unit 13 calculates the amount of change in the quantization parameter 24 of the I picture 61 (step S454). That is, the absolute difference between the quantization parameter 24 of the I picture 61 and the quantization parameter 24 of the first picture (I picture 51) of the encoded GOP 50 closest to the I picture 61 is calculated.
- the quantization parameter determination unit 13 calculates the absolute value of the change amount of the quantization parameter 24 of the encoded I picture (step S455). Specifically, the quantization parameter determination unit 13 specifies the two leading pictures 51 and 41 of the encoded GOP with reference to the I picture 61. The absolute difference between the quantization parameter 24 of the I picture 51 and the quantization parameter 24 of the I picture 41 is calculated. An absolute difference value between the quantization parameter 24 of the I picture 41 and the quantization parameter 24 of the I picture 31 is calculated.
- the quantization parameter determination unit 13 sets the correction range so that the total value of the difference absolute value calculated in step S454 and all the difference absolute values calculated in step S455 is equal to or less than a predetermined upper limit value (Ste S456).
- the total value is expressed by the following (formula 4).
- “S” indicates a total value.
- “QP” indicates the quantization parameter 24 of the I picture 61 (current picture).
- PrevQP1”, “PrevQP2”, and “PrevQP3” indicate the quantization parameters 24 of the I pictures 51, 41, and 31, respectively.
- the correction direction is downward, and the quantization parameter (PrevQP1) of the I picture 51 is “26”. Therefore, the quantization parameter determination unit 13 sets the correction range of the quantization parameter 24 of the I picture 61 to “25” to “26” as shown in FIG. 13 (step S456).
- the quantization parameter determination unit 13 corrects the quantization parameter 24 of the I picture 61 determined in step S403 or S409 (see FIG. 9) so as to fall within the set correction range (step S457). As illustrated in FIG. 13, when the quantization parameter 24 of the I picture 61 is “23”, the quantization parameter determination unit 13 corrects the quantization parameter 24 of the I picture 61 to “25”. When the quantization parameter 24 of the I picture 61 is determined to be larger than “26”, the quantization parameter determination unit 13 corrects the quantization parameter 24 of the I picture 61 to “26”.
- the quantization parameter determination unit 13 corrects the quantization parameter of the current picture so that the sum of the absolute differences calculated in steps S454 and S455 is equal to or less than the predetermined upper limit value. As a result, the amplitude of the oscillation of the quantization parameter (repetition of the increase and decrease of the quantization parameter) can be reduced. Since it is possible to prevent repeated improvements in image quality and image quality degradation, It is possible to prevent the overall image quality of the H.264 data 29 from being lowered.
- the quantization parameter determination unit 13 may omit step S455.
- the quantization parameter determination unit 13 sets the correction range so that
- the change upper limit value may be different from the change upper limit value used when step S455 is executed.
- the Hadamard transform unit 11 may calculate the Hadamard value by performing the Hadamard transform in the vertical direction in addition to the Hadamard transform in the horizontal direction. Thereby, since a Hadamard value including frequency components in the horizontal direction and the vertical direction can be obtained, it is possible to improve scene change detection accuracy and code amount control accuracy.
- the Hadamard transform unit 11 may perform Hadamard transform only in the vertical direction.
- the code amount calculation unit 131 uses the GOP ideal code amount, the overall error, and the period error for calculating the GOP target code amount.
- the code amount calculation unit 131 may calculate the GOP target code amount using only the GOP ideal code amount and the overall error.
- the code amount calculation unit 131 may calculate the GOP target code amount using only the GOP ideal code amount and the period error. Thereby, the calculation amount at the time of encoding of the uncompressed image data 21 can be further suppressed.
- the code amount is controlled in GOP units by determining the quantization parameter of the current picture based on the Hadamard value 23 and the picture target code amount.
- the image coding apparatus 1 may control the code amount in units of groups of a plurality of pictures different from the GOP, instead of controlling the code amount in GOP units.
- the image coding apparatus 1 may use two or more consecutive GOPs as a code amount control unit, or may use a group of pictures smaller than the GOP as a code amount control unit.
- the scene change determination unit 12 determines the first picture of the group closest to the current picture as a comparison target picture.
- step S402 the code amount calculation unit 131 may calculate the picture target code amount by multiplying the ideal code amount of the group by the head picture ratio.
- the leading picture ratio can be obtained by calculating the ratio of the generated code amount of the leading picture to the generated code amount of the immediately preceding encoded group.
- step S404 the quantization parameter determination unit 13 may determine whether or not the current picture is the first picture of the code amount control unit.
- the quantization parameter determination unit 13 calculates the absolute difference value using the current picture (I picture 61) and the first picture of the group instead of the I pictures 51, 41, and 61. do it.
Landscapes
- Engineering & Computer Science (AREA)
- Multimedia (AREA)
- Signal Processing (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Discrete Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Algebra (AREA)
- Mathematical Analysis (AREA)
- Mathematical Optimization (AREA)
- Pure & Applied Mathematics (AREA)
- Compression Or Coding Systems Of Tv Signals (AREA)
Abstract
Description
{1.全体構成}
図1は、本発明の第1の実施の形態に係る画像符号化装置1の機能的構成を示すブロック図である。画像符号化装置1は、非圧縮画像データ21をH.264方式で符号化して、H.264データ29を出力する。画像符号化装置1は、アダマール変換部11と、シーンチェンジ判定部12と、量子化パラメータ決定部13と、符号化部14と、QP対応テーブル15とを備える。
画像符号化装置1は、カレントピクチャをアダマール変換して周波数成分データ22を生成する。周波数成分データ22の交流成分値の総和が、アダマール値23として算出される。画像符号化装置1は、カレントピクチャの符号化条件を決定する際に、ピクチャにおける画像の複雑度を示す特性値としてアダマール値23を利用する。複雑度は、ピクチャが有する各画素の画素値のばらつきの度合いを示す。アダマール値23は、ピクチャの周波数成分を含む。このため、画像の複雑度を示す特性値としてアダマール値23を利用した場合、ピクチャの周波数成分のばらつきを考慮した上で、ピクチャを符号化することができる。したがって、ピクチャの符号化条件を精度よく決定することができる。
以下、画像符号化装置1の動作について詳しく説明する。図2は、画像符号化装置1が実行する符号化処理のフローチャートである。
アダマール値の算出について詳しく説明する。アダマール変換部11は、量子化パラメータの決定(ステップS4)及びピクチャの符号化(ステップS5)と並行して、各ピクチャのアダマール値23を算出する。
以下、シーンチェンジ判定処理(ステップS3、図2参照)について詳しく説明する。
以下、量子化パラメータ決定処理(ステップS4、図2参照)について説明する。基本的には、カレントピクチャに最も近い符号化済みIピクチャの量子化パラメータ24が、カレントピクチャの量子化パラメータ24として用いられる。しかし、シーンチェンジが発生した場合、あるいは、GOP理想符号量と符号化済みGOPの発生符号量との差分が後述する選択基準値よりも大きい場合、カレントピクチャの量子化パラメータ24は、カレントピクチャのアダマール値23に基づいて決定される。
以下、図9を参照しながら、本発明の第2の実施の形態を説明する。図9は、本発明の第2の実施の形態に係る量子化パラメータ決定処理(ステップS4)のフローチャートである。第2の実施の形態が第1の実施の形態と異なる点は、ステップS403でカレントピクチャの量子化パラメータ24が決定された後にも量子化パラメータ24の補正処理(ステップS410)が実行される点である。
Claims (13)
- 非圧縮画像データをピクチャ単位で符号化する画像符号化装置であって、
第1ピクチャをアダマール変換して周波数成分データを生成し、前記周波数成分データに含まれる交流成分値の絶対値を合計することにより、前記第1ピクチャの特性値を算出するアダマール変換部と、
前記第1ピクチャの複雑さを示すパラメータとして前記特性値を用いて前記第1ピクチャを符号化する符号化部と、
を備える。 - 請求項1に記載の画像符号化装置であって、
前記アダマール変換部は、水平方向に配列された所定数の画素単位で前記第1ピクチャをアダマール変換する。 - 請求項1に記載の画像符号化装置であって、さらに、
前記第1ピクチャの符号化により発生する符号量の目標値であるピクチャ目標符号量を算出する符号量算出部と、
前記特性値と前記ピクチャ目標符号量とに基づいて、前記第1ピクチャの符号化に用いられる量子化パラメータを決定する第1量子化パラメータ決定部と、
を備え、
前記符号化部は、前記量子化パラメータを用いて前記第1ピクチャを符号化する。 - 請求項3に記載の画像符号化装置であって、
前記符号量算出部は、予め設定された目標ビットレートに基づいて、複数のピクチャにより構成される画像グループの符号量の理想値であるグループ理想符号量を算出し、
前記画像符号化装置は、さらに、
前記第1ピクチャの量子化パラメータとして、画像グループの先頭に位置するとともに前記第1ピクチャに最も近い符号化済み先頭ピクチャの量子化パラメータの使用を決定する第2量子化パラメータ決定部と、
前記グループ理想符号量と、前記第1ピクチャが属する画像グループの直前に符号化された画像グループの発生符号量とに基づいて、前記第1量子化パラメータ決定部及び前記第2量子化パラメータ決定部のいずれかを選択する決定方法選択部と、
を備える。 - 請求項4に記載の画像符号化装置であって、さらに、
前記特性値に基づいて、前記第1ピクチャでシーンチェンジが発生したか否かを判定するシーンチェンジ判定部、
を備え、
前記決定方法選択部は、前記第1ピクチャでシーンチェンジが発生したと判定された場合、前記第1量子化パラメータ決定部を選択する。 - 請求項4に記載の画像符号化装置であって、さらに、
前記決定方法選択部は、前記第1ピクチャが画像グループの先頭ピクチャである場合、前記第1量子化パラメータ決定部及び前記第2量子化パラメータ決定部のいずれかを選択する。 - 請求項4に記載の画像符号化装置であって、さらに、
符号化済み画像グループの発生符号量から前記グループ理想符号量を差し引いた値を個別誤差とした場合、全ての符号化済み画像グループの個別誤差の総和を全体誤差として算出する全体誤差算出部、
を備え、
前記符号量算出部は、前記グループ理想符号量と前記全体誤差とに基づいて、前記ピクチャ目標符号量を算出する。 - 請求項4に記載の画像符号化装置であって、さらに、
符号化済み画像グループの発生符号量から前記グループ理想符号量を差し引いた値を個別誤差とした場合、前記ピクチャが属する画像グループから起算して所定数の符号化済み画像グループの個別誤差の総和を期間誤差として算出する期間誤差算出部、
を備え、
前記符号量算出部は、前記グループ理想符号量と前記期間誤差とに基づいて、前記ピクチャ目標符号量を算出する。 - 請求項1に記載の画像符号化装置であって、さらに、
前記第1ピクチャの特性値と、複数のピクチャにより構成される画像グループの先頭ピクチャのうち前記第1ピクチャに最も近い符号化済みの先頭ピクチャの特性値との差分絶対値が第1しきい値よりも大きい場合、前記第1ピクチャでシーンチェンジが発生したと判定するシーンチェンジ判定部、
を備える。 - 請求項9に記載の画像符号化装置であって、さらに、
予め設定された目標ビットレートに基づいて、複数のピクチャにより構成される画像グループの符号量の理想値であるグループ理想符号量を設定する符号量算出部、
を備え、
前記シーンチェンジ判定部は、前記第1ピクチャが画像グループの先頭ピクチャであり、かつ、前記グループ理想符号量と、前記第1ピクチャが属する画像グループの直前に符号化された画像グループの発生符号量との差分絶対値が第2しきい値よりも大きい場合、前記第1ピクチャでシーンチェンジが発生したと判定する。 - 請求項3に記載の画像符号化装置であって、さらに、
複数のピクチャにより構成される画像グループの先頭ピクチャのうち前記第1ピクチャに最も近い符号化済み先頭ピクチャを第1先頭ピクチャとした場合、前記第1ピクチャの量子化パラメータと、前記第1先頭ピクチャの量子化パラメータとの第1差分絶対値を計算する第1差分計算部と、
前記第1ピクチャから起算して所定数の符号化済み先頭ピクチャを第2先頭ピクチャとした場合、各第2先頭ピクチャの量子化パラメータと、各第2先頭ピクチャの直前に位置する符号化済み先頭ピクチャの量子化パラメータとの第2差分絶対値を計算する第2差分計算部と、
前記第1差分絶対値と全ての第2差分絶対値との合計値が所定値以下となるように、前記第1ピクチャの量子化パラメータを補正する補正部と、
を備える。 - 請求項11に記載の画像符号化装置であって、
前記補正部は、画像グループの符号量の理想値であるグループ理想符号量と前記第1ピクチャが属する画像グループの直前に符号化された画像グループの発生符号量とに基づいて、前記第1ピクチャの量子化パラメータを前記第1先頭ピクチャの量子化パラメータよりも小さい値に補正するか大きい値に補正するかを決定する。 - 請求項11に記載の画像符号化装置であって、さらに、
前記特性値に基づいて、前記第1ピクチャでシーンチェンジが発生したか否かを判定するシーンチェンジ判定部、
を備え、
前記補正部は、シーンチェンジが前記第1ピクチャで発生したと判定された場合、前記第1ピクチャの特性量と前記第1先頭ピクチャの特性量とに基づいて、前記第1ピクチャの量子化パラメータを前記第1先頭ピクチャの量子化パラメータよりも小さい値に補正するか大きい値に補正するかを決定する。
Priority Applications (3)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
JP2013502282A JP6062356B2 (ja) | 2011-02-28 | 2012-02-24 | 画像符号化装置 |
US14/001,980 US9031341B2 (en) | 2011-02-28 | 2012-02-24 | Image coding apparatus |
US14/624,598 US9407919B2 (en) | 2011-02-28 | 2015-02-18 | Image coding apparatus |
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
JP2011041798 | 2011-02-28 | ||
JP2011-041798 | 2011-02-28 |
Related Child Applications (2)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
US14/001,980 A-371-Of-International US9031341B2 (en) | 2011-02-28 | 2012-02-24 | Image coding apparatus |
US14/624,598 Continuation US9407919B2 (en) | 2011-02-28 | 2015-02-18 | Image coding apparatus |
Publications (1)
Publication Number | Publication Date |
---|---|
WO2012117955A1 true WO2012117955A1 (ja) | 2012-09-07 |
Family
ID=46757892
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/JP2012/054526 WO2012117955A1 (ja) | 2011-02-28 | 2012-02-24 | 画像符号化装置 |
Country Status (3)
Country | Link |
---|---|
US (2) | US9031341B2 (ja) |
JP (1) | JP6062356B2 (ja) |
WO (1) | WO2012117955A1 (ja) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US9350989B2 (en) | 2012-12-11 | 2016-05-24 | Megachips Corporation | Moving image coding apparatus, code amount control method, and storage medium |
JP2021177369A (ja) * | 2019-06-12 | 2021-11-11 | シャンハイ カンブリコン インフォメーション テクノロジー カンパニー リミテッドShanghai Cambricon Information Technology Co., Ltd. | ニューラルネットワークにおける量子化パラメータの確定方法および関連製品 |
Families Citing this family (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US9661329B2 (en) * | 2014-04-30 | 2017-05-23 | Intel Corporation | Constant quality video coding |
US20180027256A1 (en) * | 2015-01-19 | 2018-01-25 | Nec Corporation | Video encoding device, video encoding method, and video encoding program |
US9872026B2 (en) * | 2015-06-12 | 2018-01-16 | Intel Corporation | Sample adaptive offset coding |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JPH04177991A (ja) * | 1990-11-13 | 1992-06-25 | Nec Home Electron Ltd | 動画データの圧縮符号化方式 |
JPH08317387A (ja) * | 1995-05-17 | 1996-11-29 | Nippon Columbia Co Ltd | 動画像圧縮符号化方式 |
JP2007318617A (ja) * | 2006-05-29 | 2007-12-06 | Renesas Technology Corp | 画像符号化装置、及び画像符号化プログラム |
JP2009044483A (ja) * | 2007-08-09 | 2009-02-26 | Panasonic Corp | 画像符号化装置 |
Family Cites Families (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP3560217B2 (ja) * | 1998-04-30 | 2004-09-02 | ソニー株式会社 | データ符号化装置、データ符号化方法及びデータ伝送方法 |
JP4747975B2 (ja) * | 2006-07-14 | 2011-08-17 | ソニー株式会社 | 画像処理装置および方法、プログラム、並びに、記録媒体 |
JP2007318817A (ja) | 2007-09-05 | 2007-12-06 | Fujitsu Component Ltd | アンテナ装置 |
JP5039976B2 (ja) | 2008-03-24 | 2012-10-03 | 株式会社メガチップス | トランスコーダ |
US8363727B2 (en) * | 2008-09-30 | 2013-01-29 | Microsoft Corporation | Techniques to perform fast motion estimation |
-
2012
- 2012-02-24 JP JP2013502282A patent/JP6062356B2/ja active Active
- 2012-02-24 WO PCT/JP2012/054526 patent/WO2012117955A1/ja active Application Filing
- 2012-02-24 US US14/001,980 patent/US9031341B2/en active Active
-
2015
- 2015-02-18 US US14/624,598 patent/US9407919B2/en active Active
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JPH04177991A (ja) * | 1990-11-13 | 1992-06-25 | Nec Home Electron Ltd | 動画データの圧縮符号化方式 |
JPH08317387A (ja) * | 1995-05-17 | 1996-11-29 | Nippon Columbia Co Ltd | 動画像圧縮符号化方式 |
JP2007318617A (ja) * | 2006-05-29 | 2007-12-06 | Renesas Technology Corp | 画像符号化装置、及び画像符号化プログラム |
JP2009044483A (ja) * | 2007-08-09 | 2009-02-26 | Panasonic Corp | 画像符号化装置 |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US9350989B2 (en) | 2012-12-11 | 2016-05-24 | Megachips Corporation | Moving image coding apparatus, code amount control method, and storage medium |
US9538191B2 (en) | 2012-12-11 | 2017-01-03 | Megachips Corporation | Moving image coding apparatus, code amount control method, and storage medium |
JP2021177369A (ja) * | 2019-06-12 | 2021-11-11 | シャンハイ カンブリコン インフォメーション テクノロジー カンパニー リミテッドShanghai Cambricon Information Technology Co., Ltd. | ニューラルネットワークにおける量子化パラメータの確定方法および関連製品 |
JP7166704B2 (ja) | 2019-06-12 | 2022-11-08 | 寒武紀(西安)集成電路有限公司 | ニューラルネットワークにおける量子化パラメータの確定方法および関連製品 |
Also Published As
Publication number | Publication date |
---|---|
US9407919B2 (en) | 2016-08-02 |
JPWO2012117955A1 (ja) | 2014-07-07 |
US20130336593A1 (en) | 2013-12-19 |
JP6062356B2 (ja) | 2017-01-18 |
US20150163495A1 (en) | 2015-06-11 |
US9031341B2 (en) | 2015-05-12 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
EP1372113B1 (en) | Variable bit rate video encoding method and device | |
JP5212372B2 (ja) | 画像処理装置及び画像処理方法 | |
JP4529919B2 (ja) | 適応量子化装置及び適応量子化プログラム | |
US8559501B2 (en) | Method and apparatus for adaptively determining a bit budget for encoding video pictures | |
US8107529B2 (en) | Coding device, coding method, program of coding method, and recording medium recorded with program of coding method | |
JP6062356B2 (ja) | 画像符号化装置 | |
JPWO2009035143A1 (ja) | 画像処理装置及び方法 | |
US8891892B2 (en) | Image encoding method using adaptive preprocessing scheme | |
KR20140042845A (ko) | 지각적 비디오 코딩을 위한 구조적 유사성 기반의 레이트-왜곡 최적화 방법 및 시스템 | |
JP2002010259A (ja) | 画像符号化装置及び画像符号化方法及び画像符号化プログラムを記録した記録媒体 | |
EP1978745B1 (en) | Statistical adaptive video rate control | |
US8831089B1 (en) | Method and apparatus for selecting optimal video encoding parameter configurations | |
JP5618128B2 (ja) | 符号化装置、符号化方法、およびプログラム | |
JP5706759B2 (ja) | 画像符号化装置及び画像符号化方法 | |
US8559738B2 (en) | Predictive coding method for coding texture of image | |
US20090161758A1 (en) | Code amount control method, code amount control device, and camera system | |
JP5649296B2 (ja) | 画像符号化装置 | |
US8792562B2 (en) | Moving image encoding apparatus and method for controlling the same | |
JP6313614B2 (ja) | 動画像符号化装置及びその制御方法 | |
JP5871602B2 (ja) | 符号化装置 | |
US8971393B2 (en) | Encoder | |
JP6239838B2 (ja) | 動画像符号化装置、その制御方法、及び撮像装置 | |
Xie et al. | A novel parallel-friendly rate control scheme for HEVC | |
Sun et al. | A novel incremental rate control scheme for H. 264 video coding | |
Hsia et al. | High-performance adaptive group-of-picture rate control for H. 264/AVC |
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: 12752373 Country of ref document: EP Kind code of ref document: A1 |
|
ENP | Entry into the national phase |
Ref document number: 2013502282 Country of ref document: JP Kind code of ref document: A |
|
NENP | Non-entry into the national phase |
Ref country code: DE |
|
WWE | Wipo information: entry into national phase |
Ref document number: 14001980 Country of ref document: US |
|
32PN | Ep: public notification in the ep bulletin as address of the adressee cannot be established |
Free format text: NOTING OF LOSS OF RIGHTS PURSUANT TO RULE 112(1) EPC (EPO FORM 1205A DATED 05/12/13) |
|
122 | Ep: pct application non-entry in european phase |
Ref document number: 12752373 Country of ref document: EP Kind code of ref document: A1 |