JP5447239B2 - Moving picture coding apparatus and moving picture coding method - Google Patents

Description

  The present invention relates to a moving image encoding apparatus and a moving image encoding method.

  MPEG (Moving Picture Experts Group) -2, which is an international standard, H.264, and H.264 are code formats for handling moving image information as digital signals. H.264 / AVC (Advanced Video Coding) is used.

  In recent years, H.C. A technology for compressing and encoding full HD (High Definition) video data using H.264 / AVC has spread from a digital video camera to a digital still camera, and further to a movie shooting function of a mobile phone. In a stationary apparatus, H.264 Devices that record full HD video data on a high-density and large-capacity storage medium such as a Blu-ray (registered trademark) disk recorder using the H.264 / AVC compression method have also appeared.

  MPEG-2 and H.264 In a moving image compression process such as H.264 / AVC, in each frame image, a rectangular area having a similar pattern from a temporally close image called a reference image is referred to as a rectangular area (macroblock) serving as a compression processing unit. Explored. Then, the spatial position difference between the two rectangular areas is compression-coded as motion vector data, and the residual data of these images is compression-coded as coefficient data.

  In this compression coding process, the moving image compression coding device exchanges image data such as reference images with the memory. Therefore, when the amount of data such as a full HD image and a high-definition frame image increases, the moving image coding device The amount of data transferred between the memory and the memory increases.

  Therefore, conventionally, the original image input is supplied to the moving image encoding processing unit while buffering the original image input, and the reduction process is performed when the reference image is written to the memory, and the reduced image and the error data associated therewith are stored in the memory. Technology has been used.

  According to this technique, by performing a reduction process before writing a reference image generated for use in motion detection into a memory, the amount of data transferred between the moving picture coding apparatus and the memory at the time of writing and reading is reduced. . The reference image read from the memory is enlarged, and error data read from the memory is added to restore the original reference image, which is used for motion detection.

JP 2007-221697 A JP 2007-036738 A

  However, in the technique of reducing the reference image and storing it in the memory, the error data increases as the reduction rate of the reference image is increased in order to reduce the data transfer amount. Since the error data is also written in the memory to restore the reference image, there is a problem that if the error data becomes large, the reduction of the data transfer amount is limited, and the data transfer amount cannot be reduced sufficiently.

  According to one aspect of the invention, a first encoding unit that encodes original image data and generates first compressed encoded data, and generates reference image data by decoding the first compressed encoded data A first decoding unit that performs intra-frame encoding of the original image data and outputs second compressed encoded data to a memory; and decodes the second compressed encoded data A moving image code comprising: a second decoding unit that generates pseudo reference image data; and an error calculation unit that generates error data between the reference image data and the pseudo reference image data and outputs the error data to the memory A device is provided.

  According to the disclosed moving image encoding device and moving image encoding method, the data transfer amount between the moving image encoding device and the memory can be reduced.

It is a figure which shows an example of the moving image encoder of 1st Embodiment. It is a flowchart which shows the flow of a process of a moving image encoder. It is a figure which shows an example of the moving image encoder of 2nd Embodiment. It is a figure which shows the example which adjusts the quantization parameter value of a reference image compression encoding part. It is a figure which shows an example of an encoding process part. It is a figure which shows an example of the updated conversion table. It is a figure which shows the example which adjusts the quantization parameter value of a reference image compression encoding part using an evaluation formula.

Embodiments of a moving image encoding apparatus and a moving image encoding method of the present invention will be described below with reference to the drawings.
(First embodiment)
FIG. 1 is a diagram illustrating an example of a moving image encoding apparatus according to the first embodiment.

The moving image encoding apparatus 10 according to the first embodiment includes an encoding unit 11, a decoding unit 12, an encoding unit 13, a decoding unit 14, an error calculation unit 15, and a reference image restoration unit 16. .
The encoding unit 11 encodes input original image data to generate compressed encoded data. The encoding unit 11 performs an intra-frame encoding process on a reference frame image that is not subjected to the inter-frame encoding process in the original image data, and performs inter-frame encoding on the other frame images. Process. The encoding unit 11 inputs the generated compressed encoded data to the decoding unit 12 in order to generate reference image data. The encoding unit 11 outputs the generated compressed encoded data from the moving image encoding device 10.

The compressed encoded data input to the decoding unit 12 is, for example, a quantized coefficient before entropy encoding, but may be entropy encoded data.
The decoding unit 12 decodes the compression encoded data to generate reference image data.

  The encoding unit 13 performs intra-frame encoding processing on the original image data to generate compressed encoded data, and outputs the compressed encoded data to the memory 20. Further, the encoding unit 13 inputs the compression encoded data to the decoding unit 14.

  Note that the compression encoded data input to the decoding unit 14 is, for example, a quantized coefficient before entropy encoding, but may be entropy encoded data. The same applies to data output to the memory 20.

The decoding unit 14 decodes the compressed encoded data input from the encoding unit 13 to generate pseudo reference image data.
The error calculation unit 15 generates error data between the reference image data generated by the decoding unit 12 and the pseudo reference image data generated by the decoding unit 14 and outputs the error data to the memory 20.

  The reference image restoration unit 16 reads out the compressed encoded data and the error data from the memory 20. Then, the reference image restoration unit 16 decodes the compressed encoded data, performs error data addition processing on the decoding result, restores the reference image data, and inputs the restored reference image data to the encoding unit 11 To do.

Note that the memory 20 may be in the moving image encoding apparatus 10.
Hereinafter, the operation of the moving image encoding apparatus 10 will be described.
FIG. 2 is a flowchart showing a flow of processing of the moving image encoding apparatus.

Step S1: The encoding units 11 and 13 input original image data.
Step S2: The encoding unit 11 performs an intra-frame encoding process or an inter-frame encoding process on the input original image data, and generates compressed encoded data. The generated compressed encoded data is input to the decoding unit 12. In the inter-frame encoding process, the reference image restoration unit 16 reads out the compressed encoded data and the error data from the memory 20, and performs the decoding of the compressed encoded data and the addition process of the error data, so that the reference image Restore the data.

Step S3: The decoding unit 12 decodes the compressed encoded data generated by the encoding unit 11 to generate reference image data.
Step S4: On the other hand, the encoding unit 13 performs intra-frame encoding processing on the input original image data to generate compressed encoded data. The generated compressed encoded data is written into the memory 20. The compressed encoded data is input to the decoding unit 14.

Step S5: The decoding unit 14 decodes the compressed encoded data input from the encoding unit 13 to generate pseudo reference image data.
Step S6: The error calculation unit 15 calculates an error between the reference image data generated by the decoding unit 12 and the pseudo reference image data generated by the decoding unit 14. Error data obtained by the calculation is written in the memory 20.

  Since the encoding unit 11 performs irreversible compression processing by quantization, the image quality of the decoded reference image data is deteriorated compared to the original image data. In the moving image encoding apparatus 10 according to the present embodiment, intra-frame encoding processing is performed on original image data in a path different from the path for generating reference image data, and the obtained compressed encoded data is stored in the memory 20. And pseudo reference image data is generated by decoding.

  The pseudo reference image data generated here is generated by performing encoding and decoding on the original image data in the same manner as the reference image data, so that the degradation quality is the same as that of the reference image data, and the error calculation unit The error data calculated in 15 can be reduced.

Thereby, the data transfer amount between the moving image encoder 10 and the memory 20 can be reduced.
(Second Embodiment)
Hereinafter, as a second embodiment, a moving picture coding apparatus capable of reducing the amount of data transferred to and from the memory will be described in more detail.

FIG. 3 is a diagram illustrating an example of a moving image encoding apparatus according to the second embodiment.
The moving image coding apparatus 100 includes an input image buffer 110, an interframe coding unit 120, an intraframe coding unit 130, a reference image generation unit 140, and a reference image restoration unit 150.

The input image buffer 110 holds the original image data, and supplies the held original image data to the inter-frame encoding unit 120, the intra-frame encoding unit 130, and the reference image generation unit 140.
In the original image data, for example, a 4: 2: 0 YUV format is used as a frame image format. In the case of full HD video data (1920 × 1080 pixels, 30 fps), the data transfer amount per second is 1920 × 1080 [pixels] × 1.5 [bytes / pixel] × 30 [fps] = about 90 MB / s It is.

  The frame image includes an intra image in which compression encoding processing is performed without referring to other images, and an inter image in which encoding is performed with reference to other images. Further, in each macroblock in the frame image, there are an intra macroblock that can be compressed only by information in the frame and an inter macroblock that compresses a difference between the blocks of other frame images.

The inter-frame coding unit 120 and the intra-frame coding unit 130 have the function of the coding unit 11 in the moving picture coding apparatus 10 according to the first embodiment shown in FIG.
The interframe coding unit 120 includes a motion vector search memory 121, a motion vector search unit 122, and a difference image generation unit 123.

The motion vector search memory 121 temporarily and locally holds a portion to be subjected to motion vector search among the reference image data restored by the reference image restoration unit 150.
The motion vector search unit 122 detects a motion vector for the macroblock to be encoded in the original image data held in the input image buffer 110 using the reference image data held in the motion vector search memory 121. . Then, the motion vector search unit 122 reads appropriate reference image data from the motion vector search memory 121 using the detected motion vector, generates a prediction image block, and outputs it.

  The difference image generation unit 123 obtains a difference between pixel data between the macro block to be processed and the predicted image block generated by the motion vector search unit 122 and outputs the difference as block image block data.

The intraframe coding unit 130 includes a frequency conversion unit 131, a quantization unit 132, and an entropy coding unit 133.
When the macro block to be processed is the above-described inter macro block, the frequency conversion unit 131 performs frequency conversion on the difference image block data output from the difference image generation unit 123 of the inter-frame encoding unit 120. . In addition, when the processing target macroblock is an intra macroblock, the frequency conversion unit 131 performs frequency conversion on the processing target macroblock in the original image data held in the input image buffer 110. The quantization unit 132 performs quantization on the output of the frequency conversion unit 131. The entropy encoding unit 133 performs entropy encoding on the compressed encoded data from the quantization unit 132 and outputs the result.

  It should be noted that parameters such as quantization parameter values used for processing in the intra-frame coding unit 130 differ depending on the compression standard when processing an intra macroblock and when processing an inter macroblock. Things are set.

The reference image generation unit 140 includes a local decoding unit 141, a reference image compression encoding unit 142, a reference image local decoding unit 143, and an error calculation unit 144.
The local decoding unit 141 has the function of the decoding unit 12 in the video encoding device 10 of the first embodiment shown in FIG. The local decoding unit 141 performs inverse quantization corresponding to the compression coding standards in the frequency conversion unit 131 and the quantization unit 132 on the compression coded data output from the quantization unit 132 during the inter-macroblock processing. The difference image block data is reconstructed by performing conversion and inverse frequency conversion. Then, the local decoding unit 141 generates local decoded image data by adding the pixel data of the predicted image block output from the motion vector search unit 122 and the difference image block data.

  When processing an intra macroblock, the local decoding unit 141 performs inverse quantization and inverse frequency conversion on the compressed encoded data from the quantization unit 132 to generate local decoded image data.

Note that the locally decoded image data corresponds to the reference image data described in the first embodiment.
The reference image compression encoding unit 142 has the function of the encoding unit 13 in the moving image encoding apparatus 10 according to the first embodiment shown in FIG. The reference image compression encoding unit 142 performs an intra-frame encoding process on the original image data held in the input image buffer 110. The reference image compression encoding unit 142 inputs the compression encoded data, which is the quantization result in the intraframe encoding process, to the reference image local decoding unit 143. Further, the reference image compression encoding unit 142 writes the compression encoded data obtained by further entropy encoding the quantization result in the reference image memory area 201 of the memory 200.

  The reference image local decoding unit 143 has the function of the decoding unit 14 in the video encoding device 10 of the first embodiment shown in FIG. The reference image local decoding unit 143 decodes the compressed encoded data generated by the reference image compression encoding unit 142 according to the compression standard applied by the reference image compression encoding unit 142, and performs pseudo reference Generate image data.

  The error calculation unit 144 has the function of the error calculation unit 15 in the moving picture coding apparatus 10 according to the first embodiment shown in FIG. The error calculation unit 144 calculates an error between the locally decoded image data output from the local decoding unit 141 and the pseudo reference image data. Then, the error calculation unit 144 writes the error data obtained by the calculation into the error memory area 202 of the memory 200.

  The reference image restoration unit 150 has the function of the reference image restoration unit 16 in the moving picture coding apparatus 10 according to the first embodiment illustrated in FIG. 1, and includes a reference image decoding unit 151 and an error addition unit 152. ing.

The reference image decoding unit 151 reads and decodes the compressed encoded data stored in the reference image memory area 201 of the memory 200.
The error adding unit 152 reads error data stored in the error memory area 202 of the memory 200, adds the error data to the output of the reference image decoding unit 151, and restores the reference image data. The restored reference image data is written into the motion vector search memory 121.

Hereinafter, an example of the operation of the moving image encoding apparatus 100 will be described.
When the original image data is taken into the input image buffer 110, a control unit (not shown) determines whether the macro block to be processed is an inter macro block or an intra macro block.

  In the case of an intra macroblock, the interframe coding unit 120 and the reference image restoration unit 150 do not operate, and the intraframe coding unit 130 performs an intraframe coding process. At this time, the macroblock to be processed is frequency-converted by the frequency converter 131, quantized by the quantizer 132, entropy-encoded by the entropy encoder 133, and compressed as a bit stream. It is output from the moving picture encoding apparatus 100.

  At this time, the reference image compression coding unit 142 performs intra-frame coding processing on the macroblock to be processed of the original image data, and the generated compression coded data is stored in the reference image memory area of the memory 200. 201 is written.

  Also, the compressed encoded data from the quantization unit 132 of the intra-frame encoding unit 130 is decoded by the local decoding unit 141 and input to the error calculation unit 144 as local decoded image data. The compressed encoded data generated by the reference image compression encoding unit 142 is decoded by the reference image local decoding unit 143 and input to the error calculation unit 144 as pseudo reference image data. The error calculation unit 144 calculates an error between the locally decoded image data that is reference image data and the pseudo reference image data, and stores the error data in the error memory area 202 of the memory 200.

  In such intra macroblock processing, the reference image compression encoding unit 142 uses the same intra prediction mode and quantization parameter values as the intra-frame encoding unit 130, thereby reducing the error data in the error calculation unit 144. can do. In particular, error data can be set to 0 in an intra frame in which all macro blocks in the frame are intra macro blocks. Note that intra prediction means that the processing result of the surrounding macroblock in the same frame image processed before the macroblock being processed is used as prediction data. The compression standard defines an intra prediction mode as to which peripheral macroblock data is processed and used.

  On the other hand, when processing an inter macroblock, the reference image decoding unit 151 reads and decodes the compressed encoded data stored in the reference image memory area 201 of the memory 200. Further, the error adding unit 152 reads out the error data stored in the error memory area 202 of the memory 200, adds it to the output of the reference image decoding unit 151, and restores the reference image data.

  The restored reference image data is held in the motion vector search memory and read by the motion vector search unit 122. The motion vector search unit 122 detects a motion vector for the macroblock to be processed using the reference image data held in the motion vector search memory 121.

  Then, the motion vector search unit 122 reads appropriate reference image data from the motion vector search memory 121 using the detected motion vector, generates a prediction image block, and outputs it. The difference image generation unit 123 obtains a difference between pixel data between the macro block to be processed and the predicted image block generated by the motion vector search unit 122 and outputs the difference as block image block data. The difference image block data is frequency-converted by the frequency converter 131, quantized by the quantizer 132, entropy-encoded by the entropy encoder 133, and encoded as a compressed bit stream. Output from the device 100.

  At this time, also in the reference image compression encoding unit 142, intra-frame encoding processing is performed on the inter macroblock to be processed, and the generated compressed encoded data is written in the reference image memory area 201 of the memory 200. .

  In addition, the compression encoded data from the quantization unit 132 of the intraframe encoding unit 130 is decoded by the local decoding unit 141 and reconstructed as difference image block data. Furthermore, the local decoding unit 141 generates local decoded image data by adding the pixel data of the predicted image block output from the motion vector search unit 122 and the difference image block data, and sends the decoded data to the error calculation unit 144. input. The compression encoded data generated by the reference image compression encoding unit 142 is decoded by the reference image local decoding unit 143 and input to the error calculation unit 144 as pseudo reference image data. The error calculation unit 144 calculates error data between the locally decoded image data that is reference image data and the pseudo reference image data, and stores the error data in the error memory area 202 of the memory 200.

  As described above, the moving image encoding apparatus 100 performs the intraframe encoding process on the original image data in a path different from the path for generating the reference image data, and stores the obtained compressed encoded data in the memory 200. Write to. At this time, H.C. By applying intra-frame coding processing (also referred to as intra compression) defined in H.264 / AVC, it is possible to compress the data amount to about 5 MB / s with respect to 90 MB / s original image data input. is there.

  In addition, the moving image encoding apparatus 100 generates pseudo reference image data by decoding the compression encoded data output from the reference image compression encoding unit 142. The pseudo reference image data generated here is generated by performing encoding and decoding on the original image data in the same manner as the reference image data, so that the degradation quality is the same as that of the reference image data, and the error calculation unit The error data calculated at 144 can be reduced. Thereby, the data transfer amount between the moving image coding apparatus 100 and the memory 200 can be reduced, and the memory cost can be reduced.

  By the way, when processing an inter macroblock, the inter-frame encoding process is performed as described above. Even at this time, since the reference image compression encoding unit 142 performs the intra-frame encoding process, a different encoding process is performed between the local decoding unit 141 and the local decoding result of the reference image local decoding unit 143. An error caused by.

Hereinafter, a method for reducing the error data by reducing the error data written from the error calculation unit 144 to the memory 200 will be described.
FIG. 4 is a diagram illustrating an example of adjusting the quantization parameter value of the reference image compression encoding unit.

  In FIG. 4, the local decoding unit 141, the reference image compression encoding unit 142, the reference image local decoding unit 143, the error calculation unit 144, and the interface provided in the reference image compression encoding unit 142 shown in FIG. A conversion table 300 for intra Q value conversion is shown.

  In the conversion table 300, an inter quantization parameter value QP1 used when the quantization unit 132 of the intra-frame encoding unit 130 quantizes the inter macroblock is set as an index. An intra quantization parameter value QP2 used in the reference image compression encoding unit 142 is set corresponding to the inter quantization parameter value QP1. The conversion table 300 is initialized by a control unit (not shown) before the moving image coding process according to the image size and target bit rate handled by the moving image coding apparatus 100.

  FIG. 4 illustrates an encoding processing unit 142a that performs intra-frame encoding processing on a macroblock to be processed of original image data in the reference image compression encoding unit 142.

FIG. 5 is a diagram illustrating an example of the encoding processing unit.
The encoding processing unit 142a includes a reference image frequency converting unit 142-1, a reference image quantizing unit 142-2, a selector 142-3, and a reference image entropy encoding unit 142-4.

The reference image frequency conversion unit 142-1 receives the macroblock to be processed from the input image buffer 110 and performs frequency conversion.
The reference image quantization unit 142-2 performs quantization on the frequency conversion result using the three quantization parameter values QL, QM, and QH having different sizes supplied from the conversion table 300.

  The selector 142-3 selects one of the compression encoded data (quantized coefficients CL, CM, and CH) obtained by quantization using the quantization parameter values QL, QM, and QH in the error calculation unit 144. Select and output according to the selection result.

  The reference image entropy encoding unit 142-4 performs entropy encoding on the quantized coefficient selected by the selector 142-3, generates compressed encoded data, and outputs the compressed encoded data to the reference image memory area 201 of the memory 200. .

Hereinafter, a method for adjusting the quantization parameter value by each unit shown in FIGS. 4 and 5 will be described.
When the inter macroblock is encoded by the intraframe encoding unit 130 illustrated in FIG. 3, the reference image compression encoding unit 142 acquires the inter quantization parameter value QP <b> 1 used by the quantization unit 132. Then, the encoding processing unit 142a inputs a value corresponding to the acquired inter quantization parameter value QP1 from the intra quantization parameter value QP2 of the conversion table 300 as a quantization parameter value QM to be used.

In the example of FIG. 4, QM = 10 is used for QP1 (Index) = 12.
In addition to the QM, the encoding processing unit 142a performs quantization using a quantization parameter value QH larger than QM and a quantization parameter value QL smaller than QM. As shown in FIG. Quantization coefficients CL, CM, and CH are generated.

  The reference image local decoding unit 143 decodes the compression encoded data given by the three types of quantization coefficients CL, CM, and CH, and generates the three types of decoded image data L, M, and H as pseudo reference image data. .

  The error calculation unit 144 compares the decoded image data L, M, and H with the local decoded image data output from the local decoding unit 141 by the comparison unit 144a. Then, the error calculation unit 144 selects the decoded image data L, M, and H that have the smallest difference as the optimum pseudo reference image, and outputs the difference to the memory 200 as error data. The comparison unit 144a notifies the encoding processing unit 142a of the selection result.

  The selector 142-3 of the encoding processing unit 142a selects and outputs any of the three types of quantization coefficients CL, CM, and CH according to the selection result. The reference image entropy encoding unit 142-4 performs entropy encoding on the selected quantized coefficient, generates compressed encoded data, and outputs the compressed encoded data to the memory 20.

  For example, when the difference between the decoded image data H and the local decoded image data is minimum, the encoding processing unit 142a applies the entropy code to the quantization coefficient CM generated by encoding using the quantization parameter value QH. Process and output.

  The comparison unit 144a also notifies the reference image compression encoding unit 142 of a table update instruction signal for updating the conversion table 300 based on the selection result. The reference image compression encoding unit 142 updates the conversion table 300 based on the table update instruction signal.

FIG. 6 is a diagram illustrating an example of the updated conversion table.
In the example shown in FIG. 6, the conversion table 300a updated when QH = 11 is optimal with respect to the inter quantization parameter value QP1 = 12. An intra quantization parameter value QP2 = 11 is assigned to Index = 12.

In the above example, three types of quantization parameter values having different sizes are used as candidates, but two types, or four or more types of quantization parameter values may be used as candidates.
As described above, the reference image compression encoding unit 142 calculates the error by determining the quantization parameter value using the inter quantization parameter value QP1 used by the quantization unit 132 of the intra-frame encoding unit 130. The error data in the unit 144 can be reduced.

  In addition, encoding and decoding are performed using a plurality of quantization parameter values having different sizes, a plurality of decoded image data are generated, and an optimum one is selected from among the decoded image data. Error data can be further reduced.

  In addition, the conversion table 300 can be optimized by updating the conversion table 300 according to the selection result in the error calculation unit 144, and the reference image compression encoding unit 142 performs the inter quantization parameter value GP1. An optimum quantization parameter value to be used can be obtained.

  Further, the reference image compression encoding unit 142 performs entropy encoding processing on the optimal quantization coefficient selected from the plurality of quantization coefficients, thereby reducing the hardware load in encoding. be able to.

FIG. 7 is a diagram illustrating an example of adjusting the quantization parameter value of the reference image compression encoding unit using the evaluation formula.
The same elements as those shown in FIG.

  Unlike the reference image entropy encoding unit 142-4 illustrated in FIG. 5, the reference image entropy encoding unit 142-5 of the encoding processing unit 142b performs entropy coding on a plurality of quantization coefficients CL, CM, and CH. To generate encoded streams SL, SM, and SH.

The error calculation unit 144b generates and outputs error data YL, YM, and YH between the plurality of decoded image data L, M, and H and the local decoded image data.
Further, in the example of FIG. 7, an evaluation unit 310 and a selector 311 are provided.

  The evaluation unit 310 receives stream amounts (encoded data amounts) XL, XM, and XH and error data YL, YM, and YH, and inputs (XL, YL), (XM, YM), and (XH, YH). Data transfer costs (= stream amount × weighting coefficient + error data) are obtained for the three sets. The evaluation unit 310 outputs a set that minimizes the data transfer cost as an evaluation result. Moreover, the evaluation unit 310 outputs a table update instruction signal for optimizing the conversion table 300 with the quantization parameter value used when the encoded stream indicated by the set is generated.

The selector 311 selects and outputs a set of error data designated by the evaluation result.
According to the method as described above, for example, when the combination of the stream amount XH and the error data YH for the encoded stream SH minimizes the data transfer cost, the selector 142-6 allows the encoded stream SH to Are output to the memory 200 as compressed encoded data. The selector 311 outputs the error data YH to the memory 200 as error data. Further, the conversion table 300a is updated as shown in FIG. 6 by assuming that QH = 11 is optimal for the inter quantization parameter value QP1 = 12, according to the table update instruction signal.

  Also by the method as described above, the error data output to the memory 200 can be reduced, and the total data transfer cost including the compressed encoded data written in the reference image memory area 201 of the memory 200 can be reduced.

  As mentioned above, although one viewpoint of the moving image encoding device and the moving image encoding method of the present invention has been described based on the embodiments, these are merely examples, and are not limited to the above description.

The following additional notes are further disclosed with respect to the plurality of embodiments described above.
(Supplementary note 1) a first encoding unit that encodes original image data and generates first compressed encoded data;
A first decoding unit that generates reference image data by decoding the first compression-encoded data;
A second encoding unit for intra-encoding the original image data and outputting second compressed encoded data to a memory;
A second decoding unit that decodes the second compressed encoded data to generate pseudo reference image data;
Generating error data between the reference image data and the pseudo reference image data, and outputting the error data to the memory;
A moving picture encoding apparatus comprising:

  (Supplementary Note 2) The second compression-encoded data and the error data are read from the memory, the second compression-encoded data is decoded, and the error data is added to a decoding result, whereby the reference image The moving picture coding apparatus according to supplementary note 1, further comprising a reference picture restoration unit for restoring data.

  (Supplementary Note 3) The second encoding unit uses the first quantization parameter value used in the quantization in the first encoding unit, and the second quantization used in the intra-frame encoding The moving image encoding apparatus according to appendix 1 or 2, wherein a parameter value is determined.

(Supplementary Note 4) The second encoding unit generates a plurality of the second compressed encoded data based on the plurality of second quantization parameter values having different sizes,
The second decoding unit decodes a plurality of the second compression-encoded data to generate a plurality of the pseudo reference image data,
The moving image according to claim 3, wherein the error calculation unit selects the pseudo reference image data that minimizes the error data with respect to the reference image data, and outputs the minimum error data to the memory. Encoding device.

  (Supplementary Note 5) When the second encoding unit generates the pseudo reference image data that minimizes the error data among a plurality of quantization results using the plurality of second quantization parameter values. The entropy encoding is performed on the quantization result based on the second quantization parameter value used in the above, and the second compressed encoded data to be output to the memory is generated. Video encoding device.

  (Supplementary Note 6) The second encoding unit includes a conversion table for obtaining the second quantization parameter from the first quantization parameter, and includes a selection result of the pseudo reference image in the error calculation unit. The moving image encoding apparatus according to appendix 4 or 5, wherein the conversion table is updated accordingly.

  (Supplementary Note 7) Data transfer cost based on encoded data amounts and error data of a plurality of compressed encoded data generated based on a plurality of second quantization parameter values having different sizes The moving picture coding apparatus according to claim 3, further comprising: an evaluation unit that obtains the compression-coded data that minimizes the output and outputs the compressed coding data to the memory from the second coding unit.

(Supplementary Note 8) A first encoding unit generates original compressed encoded data by encoding original image data,
A first decoding unit that decodes the first compressed encoded data to generate reference image data;
A second encoding unit performs intra-frame encoding on the original image data and outputs second compressed encoded data to a memory;
A second decoding unit that generates pseudo reference image data by decoding the second compressed encoded data;
An error calculation unit generates error data between the reference image data and the pseudo reference image data, and outputs the error data to the memory.

DESCRIPTION OF SYMBOLS 10 Moving image encoder 11, 13 Encoding part 12, 14 Decoding part 15 Error calculation part 16 Reference image decompression | restoration part 20 Memory

Claims (6)

A first encoding unit that encodes original image data and generates first compressed encoded data;
A first decoding unit that generates reference image data by decoding the first compression-encoded data;
A second encoding unit for intra-encoding the original image data and outputting second compressed encoded data to a memory;
A second decoding unit that decodes the second compressed encoded data to generate pseudo reference image data;
Generating error data between the reference image data and the pseudo reference image data, and outputting the error data to the memory;
A moving picture encoding apparatus comprising:   The second encoding unit determines a second quantization parameter value to be used at the time of the intra-frame encoding based on a first quantization parameter value used in quantization in the first encoding unit. The moving picture coding apparatus according to claim 1, wherein: The second encoding unit generates a plurality of second compressed encoded data based on the plurality of second quantization parameter values having different sizes,
The second decoding unit decodes a plurality of the second compression-encoded data to generate a plurality of the pseudo reference image data,
The moving image according to claim 2, wherein the error calculation unit selects the pseudo reference image data that minimizes the error data with respect to the reference image data, and outputs the minimum error data to the memory. Image encoding device.   The second encoding unit is used when generating the pseudo reference image data that minimizes the error data among a plurality of quantization results using the plurality of second quantization parameter values. 4. The moving image according to claim 3, wherein entropy encoding is performed on the quantization result based on the second quantization parameter value to generate the second compressed encoded data to be output to the memory. 5. Encoding device.   The second encoding unit includes a conversion table for obtaining the second quantization parameter from the first quantization parameter, and according to a selection result of the pseudo reference image in the error calculation unit, 5. The moving picture coding apparatus according to claim 3, wherein the conversion table is updated. A first encoding unit that encodes the original image data to generate first compressed encoded data;
A first decoding unit that decodes the first compressed encoded data to generate reference image data;
A second encoding unit performs intra-frame encoding on the original image data and outputs second compressed encoded data to a memory;
A second decoding unit that generates pseudo reference image data by decoding the second compressed encoded data;
An error calculation unit generates error data between the reference image data and the pseudo reference image data, and outputs the error data to the memory.

Images