JP2010283739A - Moving image encoding device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a moving image encoding device which reduces degradation in image quality of an encoded image even if a difference between an original image and a local decoded image is increased. <P>SOLUTION: The moving image encoding device includes an intra-prediction mode correction part 3 to determine a difference between a neighboring original image adjoining to a rectangular area to be intra-predicted and a local decoded image in an area adjoining to the rectangular area, and to output a selection command for an intra-prediction mode determined by an intra-prediction mode determination part 2 to a selector 4 if the difference is less than a given threshold value, or to output a selection command for a DC mode to the selector 4 if it is larger than the given threshold value. <P>COPYRIGHT: (C)2011,JPO&INPIT

Description

  The present invention relates to a moving image encoding apparatus that encodes a moving image by performing intra prediction.

H. is a video compression standard. H.264 employs intra prediction that generates a predicted image from block adjacent pixels.
As the prediction modes in intra prediction, nine types are specified in the luminance 4 × 4 intra prediction mode, nine types in the luminance 8 × 8 intra prediction mode, four types in the luminance 16 × 16 intra prediction mode, and four types in the color difference intra prediction. ing.

H. In the H.264 standard encoder, in order to perform intra prediction with high accuracy, it is desirable to generate a prediction image from a locally decoded image in all prediction modes and compare the plurality of prediction images to determine the intra prediction mode.
However, when a prediction image is generated from a locally decoded image in all prediction modes, it is necessary to generate a locally decoded image in all prediction modes, which increases the amount of calculation.

In view of this, in the moving picture encoding device disclosed in Patent Document 1 below, an intra prediction mode is determined from an original image.
In this case, since it is not necessary to generate a local decoded image in all prediction modes, the amount of calculation can be reduced. However, since an intra predicted image used for encoding is generated from a local decoded image, When a certain degree of difference occurs in the locally decoded image, an appropriate intra prediction mode is not selected, and the image quality of the encoded image may be adversely affected.

Japanese Patent Laying-Open No. 2007-150913 (FIG. 1)

  Since the conventional moving image encoding apparatus is configured as described above, it is not necessary to generate a local decoded image in all prediction modes, and thus the amount of calculation can be reduced. However, since the intra prediction mode is determined from the original image and the intra prediction image is generated from the locally decoded image, if a certain amount of difference occurs between the original image and the locally decoded image, an appropriate intra prediction mode is not selected and encoding is performed. There were problems such as the image quality being adversely affected.

  The present invention has been made in order to solve the above-described problems, and provides a moving image encoding apparatus capable of suppressing deterioration in image quality of an encoded image even when a difference between an original image and a locally decoded image increases. The purpose is to obtain.

  The moving image encoding apparatus according to the present invention includes an intra prediction mode determination unit that determines an intra prediction mode from an original image of a rectangular area to be intra predicted and an adjacent original image that is an original image of an area adjacent to the rectangular area. If the difference is smaller than a predetermined threshold, the intra prediction mode determined by the intra prediction mode determination means is selected. When the difference is larger than a predetermined threshold, a mode selection unit that selects a DC mode for generating a prediction image using an average value of pixel values is provided, and the intra prediction image generation unit performs intra prediction by the mode selection unit. When a mode is selected, an intra-prediction image of the rectangular area is generated from the locally decoded image in the intra-prediction mode. If DC mode is selected by, in the DC mode, in which so as to generate an intra prediction image of the rectangular region from the local decoded image.

  According to the present invention, the intra prediction mode determining means for determining the intra prediction mode from the original image of the rectangular area to be intra predicted and the adjacent original image that is the original image of the area adjacent to the rectangular area, and the adjacent original When the difference between the image and the locally decoded image in the region adjacent to the rectangular region is obtained and the difference is smaller than a predetermined threshold, the intra prediction mode determined by the intra prediction mode determination unit is selected, and the difference is A mode selection unit that selects a DC mode for generating a prediction image using an average value of pixel values when the threshold value is larger than a predetermined threshold value, and the intra prediction image generation unit selects the intra prediction mode by the mode selection unit. In this case, the intra prediction image of the rectangular area is generated from the locally decoded image in the intra prediction mode, and the DC mode is selected by the mode selection means. In the DC mode, the intra-prediction image of the rectangular area is generated from the local decoded image, so even if the difference between the original image and the local decoded image becomes large, the quality of the encoded image There exists an effect which can suppress deterioration.

It is a block diagram which shows the moving image encoder by Embodiment 1 of this invention. It is explanatory drawing which shows a mode that the macroblock is divided | segmented into 16 rectangular areas for every 4x4 pixel. It is explanatory drawing which shows the adjacent pixel used for intra prediction in one 4x4 block. It is explanatory drawing which shows a mode that the macroblock is divided | segmented into four rectangular areas for every 8x8 pixel. It is explanatory drawing which shows the adjacent pixel used for intra prediction in one 8x8 block. It is explanatory drawing which shows the macroblock comprised from one 16x16 block. It is explanatory drawing which shows the adjacent pixel used for intra prediction in one 16x16 block. It is explanatory drawing which shows the process order of the block in intra 4x4 prediction. It is explanatory drawing which shows the process order of the block in intra 8x8 prediction. It is explanatory drawing which shows the process order of the block in intra 16x16 prediction. It is explanatory drawing which shows the calculation timing of intra 4x4 prediction when not considering operation | movement of the intra prediction mode correction | amendment part 3. FIG. It is explanatory drawing which shows the calculation timing of intra 8x8 prediction when not considering operation | movement of the intra prediction mode correction | amendment part 3. FIG. It is explanatory drawing which shows the calculation timing of the intra 16 * 16 prediction in case the operation | movement of the intra prediction mode correction | amendment part 3 is not considered. It is explanatory drawing which shows the calculation timing of intra 4x4 prediction when the operation | movement of the intra prediction mode correction | amendment part 3 is considered. It is explanatory drawing which shows the calculation timing of intra 8x8 prediction when the operation | movement of the intra prediction mode correction | amendment part 3 is considered. It is explanatory drawing which shows the calculation timing of intra 16x16 prediction when the operation | movement of the intra prediction mode correction | amendment part 3 is considered. It is a block diagram which shows the moving image encoder by Embodiment 2 of this invention. It is a block diagram which shows the moving image encoder by Embodiment 3 of this invention.

Embodiment 1 FIG.
1 is a block diagram showing a moving picture coding apparatus according to Embodiment 1 of the present invention.
In FIG. 1, an original image peripheral pixel memory 1 is a memory that temporarily stores an adjacent original image that is an original image of an area adjacent to a rectangular area that is an intra prediction target.
The intra prediction mode determination unit 2 performs a process of determining the prediction block size and the intra prediction mode from the original image of the rectangular area to be intra predicted and the adjacent original image stored in the original image peripheral pixel memory 1.
That is, the intra prediction mode determination unit 2 uses, for example, a plurality of intra prediction modes (for example, nine types of luminance 4 × 4 intra prediction modes, luminance 8 × 8 from the adjacent original images stored in the original image peripheral pixel memory 1. Nine types of intra prediction modes, four types of luminance 16 × 16 intra prediction modes, and four types of color difference intra predictions) are used to generate prediction images, and these prediction images and the original image of the rectangular region to be intra-predicted are obtained. By comparing and selecting a prediction image having the smallest difference from the original image among the plurality of prediction images, a process of outputting a prediction block size and an intra prediction mode related to the prediction image is performed.
The intra prediction mode determination unit 2 constitutes an intra prediction mode determination unit.

The intra prediction mode correction unit 3 is adjacent to the adjacent original image stored in the original image peripheral pixel memory 1 and the local decoded image stored in the local decoded image peripheral pixel memory 13 (adjacent to the rectangular region targeted for intra prediction). When the difference is smaller than a predetermined threshold value, an intra prediction mode selection command determined by the intra prediction mode determination unit 2 is output to the selector 4, and the difference is determined to be a predetermined value. When it is larger than the threshold value, a process of outputting a selection command of a DC mode for generating a predicted image using an average value of pixel values to the selector 4 is performed.
However, the intra prediction mode correction unit 3 inputs a correction function ON / OFF flag (a flag indicating whether or not the intra prediction mode is corrected), and the correction function ON / OFF flag indicates “correction function ON”. As long as the intra prediction mode or the DC mode selected by the intra prediction mode determination unit 2 is selected and the correction function ON / OFF flag indicates “correction function OFF”, the adjacent original image and local A process of outputting a DC mode selection command to the selector 4 is performed without performing a selection process of calculating the difference of the decoded image and comparing it with a predetermined threshold.

When the selector 4 receives the instruction for selecting the intra prediction mode from the intra prediction mode correction unit 3, the selector 4 selects the intra prediction mode determined by the intra prediction mode determination unit 2 and sends the intra prediction mode to the intra prediction image generation unit 5. When the output is received and a DC mode selection command is received from the intra prediction mode correction unit 3, the DC mode is output to the intra prediction image generation unit 5.
The intra prediction mode correction unit 3 and the selector 4 constitute mode selection means.

  When receiving the intra prediction mode determined by the intra prediction mode determination unit 2 from the selector 4, the intra prediction image generation unit 5 receives the local decoded image (intra prediction image) stored in the local decoded image peripheral pixel memory 13 in the intra prediction mode. When an intra prediction image of a rectangular region to be intra-predicted is generated from a region adjacent to the rectangular region to be predicted) and received from the selector 4 in the DC mode, the intra-prediction image from the local decoded image in the DC mode is generated. The process which produces | generates the intra estimated image of the rectangular area | region of intra prediction object is implemented. In addition, the intra estimated image production | generation part 5 comprises the intra estimated image production | generation means.

If the inter / intra prediction determination result input from the outside indicates that the inter prediction is adopted, the selector 6 selects the inter prediction image input from the outside, and the inter / intra prediction determination result indicates the intra prediction. If it has shown that it employ | adopts, the process which selects the intra estimated image produced | generated by the intra estimated image production | generation part 5 will be implemented.
The subtracter 7 subtracts the image (inter prediction image or intra prediction image) selected by the selector 6 from the original image of the rectangular area to be intra predicted, and outputs the difference image as the subtraction result to the integer conversion unit 8. Perform the process.

The integer conversion unit 8 performs an integer conversion on the difference image output from the subtractor 7 and outputs the integer conversion coefficient to the quantization unit 9.
The quantization unit 9 quantizes the integer transform coefficient output from the integer transform unit 8 using a predetermined quantization parameter, outputs the quantized transform coefficient to the inverse quantization unit 10, and the quantized transform coefficient Is output to an encoding unit (not shown) such as an entropy encoding unit. The encoding unit is a processing unit that generates encoded data by performing variable-length encoding processing or arithmetic encoding processing on the quantized transform coefficient output from the quantization unit 9.
The inverse quantization unit 10 performs a process of inversely quantizing the quantized transform coefficient output from the quantizing unit 9 and outputting the inverse quantized transform coefficient to the inverse integer transform unit 11.

The inverse integer transform unit 11 performs an inverse integer transform on the inverse quantization transform coefficient output from the inverse quantization unit 10 and outputs the inverse integer transform coefficient to the adder 12.
The adder 12 adds the inverse integer transform coefficient output from the inverse integer transform unit 11 and the image (inter-predicted image or intra-predicted image) selected by the selector 6, and the addition result is a rectangular region to be intra-predicted. The local decoded image is output to the local decoded image peripheral pixel memory 13 as a local decoded image.
The subtractor 7, integer transform unit 8, quantization unit 9, inverse quantization unit 10, inverse integer transform unit 11, and adder 12 constitute local decoded image generation means.
The local decoded image peripheral pixel memory 13 is a memory for storing the local decoded image output from the adder 12.

Next, the operation will be described.
H. In the H.264 standard, nine modes are defined for luminance in intra 4 × 4 prediction, nine modes in intra 8 × 8 prediction, four modes in intra 16 × 16 prediction, and four modes in color difference. ing.

In luminance intra 4 × 4 prediction, as shown in FIG. 2, a macroblock is divided into 16 rectangular areas for each 4 × 4 pixel, and intra prediction is performed for each rectangular area.
Here, FIG. 3 shows adjacent pixels used for intra prediction in one 4 × 4 block.
In the luminance intra 8 × 8 prediction, as shown in FIG. 4, for each 8 × 8 pixel, the macroblock is divided into four rectangular areas, and intra prediction is performed for each rectangular area.
Here, FIG. 5 shows adjacent pixels used for intra prediction in one 8 × 8 block.
In the luminance intra 16 × 16 prediction, as shown in FIG. 6, intra prediction is performed for a 16 × 16 block.
Here, FIG. 7 shows adjacent pixels used for intra prediction in a 16 × 16 block.

First, the original image peripheral pixel memory 1 temporarily stores an adjacent original image that is an original image of a region adjacent to a rectangular region that is an intra prediction target.
That is, the original image peripheral pixel memory 1 includes an adjacent original image that is an original image of an area adjacent to 16 rectangular areas in the 4 × 4 block shown in FIG. 3, and 4 in the 8 × 8 block shown in FIG. 5. An adjacent original image that is an original image of an area adjacent to the rectangular areas and an adjacent original image that is an original image of an area adjacent to the rectangular area in the 16 × 16 block shown in FIG. 7 are temporarily stored. The
Similarly, for the color difference block, an adjacent original image that is an original image of an adjacent region is temporarily stored.

The local decoded image peripheral pixel memory is also adjacent to the 16 pixels in the 4 × 4 block shown in FIG. 3 and the 4 pixels in the 8 × 8 block shown in FIG. The local decoded image of the area being stored, and the local decoded image of the area adjacent to the rectangular area in the 16 × 16 block shown in FIG. 7 are temporarily stored.
Similarly, local decoded images of adjacent areas are temporarily stored for the color difference blocks.

The intra prediction mode determination unit 2 determines the prediction block size and the intra prediction mode from the original image of the rectangular area to be intra predicted and the adjacent original image stored in the original image peripheral pixel memory 1.
Specifically, the prediction block size and the intra prediction mode are determined as follows.
For example, the intra prediction mode determination unit 2 determines a plurality of intra prediction modes (for example, nine types of luminance 4 × 4 intra prediction modes, luminance 8 from the adjacent original image stored in the original image peripheral pixel memory 1. Nine types of × 8 intra prediction modes and four types of luminance 16 × 16 intra prediction modes) are generated.

When the prediction image is generated in the plurality of intra prediction modes, the intra prediction mode determination unit 2 compares the plurality of prediction images with the original image of the rectangular area to be predicted, and among the plurality of prediction images, the original image The prediction image with the smallest difference is selected.
When the intra prediction mode determination unit 2 selects a prediction image having the smallest difference from the original image, the intra prediction mode determination unit 2 selects a prediction block size and an intra prediction mode related to the prediction image as an optimal prediction block size and an intra prediction mode.
Similarly, the optimum intra prediction mode is selected for the color difference.

Here, FIG. 8 is an explanatory diagram showing the processing order of blocks in intra 4 × 4 prediction, and FIG. 9 is an explanatory diagram showing the processing order of blocks in intra 8 × 8 prediction.
FIG. 10 is an explanatory diagram showing the processing order of blocks in intra 16 × 16 prediction.
The intra-predicted image used for encoding needs to be generated from the locally decoded image. For example, in order to generate the intra-predicted image of subblock 1 in FIG. 8, the locally decoded image of subblock 0 is necessary.
Therefore, paying attention to the operation of a certain sub-block, the subtraction process in the subtracter 7 in FIG. 1, the integer conversion process in the integer conversion unit 8, the quantization process in the quantization unit 9, the inverse quantization process in the inverse quantization unit 10, In the inverse integer transform process in the inverse integer transform unit 11 and the addition process in the adder 12, a local decoded image of a subblock of the previous subblock number is generated, and an intra predicted image is generated using the local decoded image. Can be started after.

FIG. 11 is an explanatory diagram showing calculation timing of intra 4 × 4 prediction when the operation of the intra prediction mode correction unit 3 is not considered, and FIG. 12 is an intra 8 × when the operation of the intra prediction mode correction unit 3 is not considered. It is explanatory drawing which shows the calculation timing of 8 prediction.
FIG. 13 is an explanatory diagram showing calculation timing of intra 16 × 16 prediction when the operation of the intra prediction mode correction unit 3 is not considered.
For example, focusing on the calculation timing of sub block 1 for intra 4 × 4 prediction, as shown in FIG. 11, when the inverse integer conversion process of the calculation of sub block 0 is completed, the 4 × 4 block of FIG. Since the local decoded image on the left side is obtained, at this point, the intra predicted image generation unit 5 can generate the intra predicted image of the sub-block 1.

  The intra prediction mode correction unit 3 inputs a correction function ON / OFF flag, and when the correction function ON / OFF flag indicates “correction function ON”, the adjacent prediction mode stored in the original image peripheral pixel memory 1 The local decoded image stored in the local decoded image peripheral pixel memory 13 (a local decoded image in a region adjacent to the intra prediction target rectangular region, and the intra predicted image generation unit 5 For example, the sum of absolute differences between the adjacent original image and the locally decoded image is calculated as a difference from the locally decoded image used when the image is generated.

When the intra prediction mode correction unit 3 calculates the sum of absolute differences between the adjacent original image and the local decoded image, the intra prediction mode correction unit 3 compares the sum of absolute differences with a predetermined threshold value input from the outside.
When the sum of absolute differences is smaller than a predetermined threshold, the intra prediction mode correction unit 3 outputs an intra prediction mode selection command determined by the intra prediction mode determination unit 2 to the selector 4.
On the other hand, if the sum of absolute differences is larger than a predetermined threshold value, a DC mode selection command is output to the selector 4.

When the correction function ON / OFF flag indicates “correction function OFF”, the intra prediction mode correction unit 3 calculates the sum of absolute differences between the adjacent original image and the local decoded image, and compares it with a predetermined threshold value. A DC mode selection command is output to the selector 4 without executing the processing.
The correction function ON / OFF flag can be adaptively turned on / off in units of pictures, macroblocks, and subblocks.

FIG. 14 is an explanatory diagram showing calculation timing of intra 4 × 4 prediction when the operation of the intra prediction mode correction unit 3 is considered, and FIG. 15 is an intra 8 × when the operation of the intra prediction mode correction unit 3 is considered. It is explanatory drawing which shows the calculation timing of 8 prediction.
FIG. 16 is an explanatory diagram showing calculation timing of intra 16 × 16 prediction when the operation of the intra prediction mode correction unit 3 is considered.
For example, paying attention to the calculation timing of the sub block 1 of intra 4 × 4 prediction, as shown in FIG. 14, when the inverse integer conversion process of the calculation of sub block 0 is completed, the 4 × 4 block of the 4 × 4 block shown in FIG. Since the left local decoded image is obtained, at this point, the intra prediction mode correction unit 3 can compare the original image adjacent to the sub-block 1 at the position shown in FIG. 3 with the local decoded image.

When the selector 4 receives the instruction for selecting the intra prediction mode from the intra prediction mode correction unit 3, the selector 4 selects the intra prediction mode determined by the intra prediction mode determination unit 2, and selects the intra prediction mode as the intra prediction image generation unit 5. Output to.
On the other hand, when a DC mode selection command is received from the intra prediction mode correction unit 3, the DC mode is output to the intra prediction image generation unit 5.

When receiving the intra prediction mode determined by the intra prediction mode determination unit 2 from the selector 4, the intra prediction image generation unit 5 uses the local decoded image stored in the local decoded image peripheral pixel memory 13 in the intra prediction mode. An intra-prediction image of a rectangular region to be intra-predicted is generated.
On the other hand, when the DC mode is received from the selector 4, an intra-prediction image of a rectangular area to be intra-predicted is generated from the local decoded image stored in the local decoded image peripheral pixel memory 13 in the DC mode.

The selector 6 refers to the inter / intra prediction determination result input from the outside. If the inter / intra prediction determination result indicates that the inter prediction is adopted, the selector 6 selects the inter prediction image input from the outside. Then, the inter prediction image is output to the subtracter 7 and the adder 12.
On the other hand, if the inter / intra prediction determination result indicates that the intra prediction is adopted, the intra predicted image generated by the intra predicted image generation unit 5 is selected, and the intra predicted image is added to the subtractor 7 and added. To the device 12.

The subtracter 7 inputs the original image of the rectangular area to be intra predicted and the image selected by the selector 6 (inter predicted image or intra predicted image), and subtracts the image selected by the selector 6 from the original image. Then, the difference image as the subtraction result is output to the integer conversion unit 8.
When the integer conversion unit 8 receives the difference image from the subtractor 7, the integer conversion unit 8 performs integer conversion on the difference image and outputs the integer conversion coefficient to the quantization unit 9.

When the quantizing unit 9 receives the integer transform coefficient from the integer transforming unit 8, the quantizing unit 9 quantizes the integer transform coefficient using a predetermined quantization parameter, and outputs the quantized transform coefficient to the inverse quantization unit 10.
The quantization unit 9 also outputs the quantized transform coefficient to an encoding unit such as an entropy encoding unit.
Note that the encoding unit performs variable-length encoding processing or arithmetic encoding processing on the quantized transform coefficient output from the quantization unit 9 to generate encoded data (encoded image).

When receiving the quantized transform coefficient from the quantizing unit 9, the inverse quantizing unit 10 inversely quantizes the quantized transform coefficient and outputs the inverse quantized transform coefficient to the inverse integer transform unit 11.
When receiving the inverse quantization transform coefficient from the inverse quantization unit 10, the inverse integer transform unit 11 performs inverse integer transform on the inverse quantization transform coefficient and outputs the inverse integer transform coefficient to the adder 12.
When the adder 12 receives the inverse integer transform coefficient from the inverse integer transform unit 11, the adder 12 adds the inverse integer transform coefficient and the image (inter prediction image or intra prediction image) selected by the selector 6, and the addition result Are stored in the locally decoded image peripheral pixel memory 13 as a locally decoded image of a rectangular region to be intra predicted.

  As is clear from the above, according to the first embodiment, the intra prediction mode is determined from the original image of the rectangular region to be intra predicted and the adjacent original image that is the original image of the region adjacent to the rectangular region. A difference between the intra prediction mode determination unit 2 and the adjacent original image and a locally decoded image of the region adjacent to the rectangular region is obtained. If the difference is smaller than a predetermined threshold, the intra prediction mode determination unit 2 determines When the selected intra prediction mode selection command is output to the selector 4 and the difference is larger than a predetermined threshold, the intra prediction mode correction unit 3 that outputs the DC mode selection command to the selector 4, and the intra prediction mode correction unit 3 When the intra prediction mode selection command is received, the intra prediction mode determined by the intra prediction mode determination unit 2 is selected and the intra prediction mode is selected. A measurement mode is output to the intra-predicted image generation unit 5, and when a DC mode selection command is received from the intra-prediction mode correction unit 3, a selector 4 is provided to output the DC mode to the intra-predicted image generation unit 5. When the image generation unit 5 receives the intra prediction mode determined by the intra prediction mode determination unit 2 from the selector 4, in the intra prediction mode, an intra prediction is performed from a locally decoded image in a region adjacent to the intra prediction target rectangular region. Since the intra prediction image of the rectangular area to be predicted is generated and the DC mode is received from the selector 4, the intra prediction image of the rectangular area to be predicted is generated from the local decoded image in the DC mode. Even if the difference between the original image and the locally decoded image becomes large, it is possible to suppress the deterioration of the image quality of the encoded image. An effect.

  That is, when the difference between the original image and the locally decoded image increases, the intra prediction mode determined by the intra prediction mode determination unit 2 is not necessarily the optimal intra prediction mode, and as a result, the image quality of the encoded image is degraded. However, in the first embodiment, when the difference between the original image and the locally decoded image becomes large, the intra prediction from the locally decoded image is performed in the DC mode instead of the intra prediction mode determined by the intra prediction mode determining unit 2. Since the image is generated, even if the difference between the original image and the locally decoded image becomes large, it is possible to suppress deterioration in the image quality of the encoded image.

  Further, according to the first embodiment, the intra prediction mode correction unit 3 inputs the correction function ON / OFF flag, and the intra prediction mode ON / OFF flag is set only when the correction function ON / OFF flag indicates “correction function ON”. If the intra prediction mode or the DC mode selected by the prediction mode determination unit 2 is selected and the correction function ON / OFF flag indicates “correction function OFF”, the difference between the adjacent original image and the locally decoded image Since the selection command for the DC mode is output to the selector 4 without performing the selection process of calculating the sum of absolute values and comparing with the predetermined threshold value, for example, the intra mode is set according to the image quality status of the encoded image. There is an effect that the operation of the prediction mode correction unit 3 can be adaptively controlled.

Embodiment 2. FIG.
FIG. 17 is a block diagram showing a moving picture coding apparatus according to Embodiment 2 of the present invention. In the figure, the same reference numerals as those in FIG.
The selector 21 receives a luminance 4 × 4 intra prediction mode, a luminance 8 × 8 intra prediction mode, a luminance 16 × 16 prediction mode, and a color difference intra prediction mode from the outside, and an intra prediction mode determination unit is selected from these intra prediction modes. A process of selecting an intra prediction mode corresponding to the prediction block size determined in 2 is performed.

The intra prediction mode correction unit 22 is adjacent to the adjacent original image stored in the original image peripheral pixel memory 1 and the local decoded image stored in the local decoded image peripheral pixel memory 13 (adjacent to the rectangular region targeted for intra prediction). If the difference is smaller than a predetermined threshold, the intra prediction mode selection command determined by the intra prediction mode determination unit 2 is output to the selector 23, and the difference is determined to be a predetermined value. When it is larger than the threshold value, a process of outputting a selection command for the intra prediction mode selected by the selector 21 to the selector 23 is performed.
However, the intra prediction mode correction unit 22 receives the correction function ON / OFF flag, and is determined by the intra prediction mode determination unit 2 only when the correction function ON / OFF flag indicates “correction function ON”. When the intra prediction mode or the intra prediction mode selected by the selector 21 is selected and the correction function ON / OFF flag indicates “correction function OFF”, the difference between the adjacent original image and the locally decoded image is calculated. The process of outputting the selection command of the intra prediction mode selected by the selector 21 to the selector 23 is performed without performing the selection process of calculating and comparing with a predetermined threshold value.

When the selector 23 receives an instruction for selecting the intra prediction mode determined by the intra prediction mode determination unit 2 from the intra prediction mode correction unit 22, the selector 23 selects the intra prediction mode determined by the intra prediction mode determination unit 2 and selects the intra prediction mode. When the prediction mode is output to the intra prediction image generation unit 5 and an instruction for selecting the intra prediction mode selected by the selector 21 is received from the intra prediction mode correction unit 22, the intra prediction mode is output to the intra prediction image generation unit 5. Perform the process.
The selector 21, the intra prediction mode correction unit 22, and the selector 23 constitute a mode selection unit.

In the first embodiment, the intra prediction mode correction unit 3 outputs the DC mode selection command to the selector 4 when the difference between the adjacent original image and the local decoded image is larger than the predetermined threshold value. When the difference between the adjacent original image and the locally decoded image is larger than a predetermined threshold, the prediction mode correction unit 22 sends a selection command for the intra prediction mode (intra prediction mode input from the outside) selected by the selector 21 to the selector 23. You may make it output.
Specifically, it is as follows.

First, in the original image peripheral pixel memory 1, as in the first embodiment, an adjacent original image that is an original image of a region adjacent to a rectangular region that is an intra prediction target is temporarily stored.
That is, the original image peripheral pixel memory 1 includes an adjacent original image that is an original image of an area adjacent to 16 rectangular areas in the 4 × 4 block shown in FIG. 3, and 4 in the 8 × 8 block shown in FIG. 5. An adjacent original image that is an original image of an area adjacent to the rectangular areas and an adjacent original image that is an original image of an area adjacent to the rectangular area in the 16 × 16 block shown in FIG. 7 are temporarily stored. The
Similarly, for the color difference block, an adjacent original image that is an original image of an adjacent region is temporarily stored.

The local decoded image peripheral pixel memory is also adjacent to the 16 pixels in the 4 × 4 block shown in FIG. 3 and the 4 pixels in the 8 × 8 block shown in FIG. The local decoded image of the area being stored, and the local decoded image of the area adjacent to the rectangular area in the 16 × 16 block shown in FIG. 7 are temporarily stored.
Similarly, local decoded images of adjacent areas are temporarily stored for the color difference blocks.

As in the first embodiment, the intra prediction mode determination unit 2 determines an intra prediction mode from the original image in the rectangular area to be intra predicted and the adjacent original image stored in the original image peripheral pixel memory 1.
That is, the intra prediction mode determination unit 2 uses, for example, a plurality of intra prediction modes (for example, nine types of luminance 4 × 4 intra prediction modes, luminance 8 × 8 from the adjacent original images stored in the original image peripheral pixel memory 1. Nine types of intra prediction modes, four types of luminance 16 × 16 intra prediction modes, and four types of color difference intra predictions) are used to generate prediction images, and these prediction images and the original image of the rectangular region to be intra-predicted are obtained. By comparing and selecting the prediction image having the smallest difference from the original image among the plurality of prediction images, the prediction block size and the intra prediction mode related to the prediction image are selected by the selectors 21 and 23 and the intra prediction mode correction unit. 22 for output.

The selector 21 inputs a luminance 4 × 4 intra prediction mode, a luminance 8 × 8 intra prediction mode, a luminance 16 × 16 prediction mode, and a color difference intra prediction mode from the outside, and determines an intra prediction mode from these intra prediction modes. The intra prediction mode corresponding to the prediction block size determined by the unit 2 is selected and output to the selector 23.
For example, if the predicted block size related to the predicted image having the smallest difference from the original image is 4 × 4 blocks, the luminance 4 × 4 intra prediction mode is selected and output to the selector 23, and the predicted block size is 8 × 8. If it is a block, the luminance 8 × 8 intra prediction mode is selected and output to the selector 23.

  When the correction function ON / OFF flag is input and the correction function ON / OFF flag indicates “correction function ON”, the intra prediction mode correction unit 22 is similar to the intra prediction mode correction unit 3 of FIG. An adjacent original image stored in the original image peripheral pixel memory 1 and a local decoded image stored in the local decoded image peripheral pixel memory 13 (a local decoded image in an area adjacent to a rectangular area to be intra predicted). Thus, for example, a sum of absolute differences between the adjacent original image and the local decoded image is calculated as a difference from the local decoded image used when the intra predicted image generation unit 5 generates the intra predicted image.

When the intra prediction mode correction unit 22 calculates the sum of absolute differences between the adjacent original image and the locally decoded image, the intra prediction mode correction unit 22 compares the sum of absolute differences with a predetermined threshold value input from the outside.
When the sum of absolute differences is smaller than a predetermined threshold, the intra prediction mode correction unit 22 outputs an intra prediction mode selection command determined by the intra prediction mode determination unit 2 to the selector 23.
On the other hand, when the sum of absolute differences is larger than a predetermined threshold value, the selection command for the intra prediction mode selected by the selector 21 is output to the selector 23.

When the correction function ON / OFF flag indicates “correction function OFF”, the intra prediction mode correction unit 22 selects the difference absolute value sum of the adjacent original image and the local decoded image and compares it with a predetermined threshold value. The intra prediction mode selection command selected by the selector 21 is output to the selector 23 without executing the process.
The correction function ON / OFF flag can be adaptively turned on / off in units of pictures, macroblocks, and subblocks.

When the selector 23 receives an instruction for selecting an intra prediction mode from the intra prediction mode correction unit 22, the selector 23 selects the intra prediction mode determined by the intra prediction mode determination unit 2, and selects the intra prediction mode as the intra prediction image generation unit 5. Output to.
On the other hand, when an instruction for selecting the intra prediction mode selected by the selector 21 is received from the intra prediction mode correction unit 22, the intra prediction mode selected by the selector 21 is output to the intra prediction image generation unit 5.
Since the subsequent processing contents are the same as those in the first embodiment, description thereof is omitted.

  As is apparent from the above, according to the second embodiment, the intra prediction mode is determined from the original image of the rectangular region to be intra predicted and the adjacent original image that is the original image of the region adjacent to the rectangular region. An intra prediction mode determination unit 2 and a luminance 4 × 4 intra prediction mode, a luminance 8 × 8 intra prediction mode, a luminance 16 × 16 prediction mode, and a color difference intra prediction mode are input from the outside, and from among these intra prediction modes, A selector 21 for selecting an intra prediction mode corresponding to the prediction block size determined by the intra prediction mode determination unit 2 and a difference between the adjacent original image and a locally decoded image in an area adjacent to the rectangular area; When the difference is smaller than the predetermined threshold, the intra prediction mode selection command determined by the intra prediction mode determination unit 2 is selected. When the difference is larger than a predetermined threshold value, the intra prediction mode correction unit 22 that outputs a selection command of the intra prediction mode selected by the selector 21 to the selector 23, and the intra prediction mode correction unit 22 performs intra prediction. When receiving the instruction for selecting the intra prediction mode determined by the mode determination unit 2, the intra prediction mode determined by the intra prediction mode determination unit 2 is selected and the intra prediction mode is output to the intra prediction image generation unit 5. When receiving the instruction for selecting the intra prediction mode selected by the selector 21 from the intra prediction mode correction unit 3, the intra prediction mode selected by the selector 21 is selected and the intra prediction mode is sent to the intra prediction image generation unit 5. And an intra-predicted image generation unit. However, in the intra prediction mode output from the selector 23, the intra prediction image of the rectangular region to be intra predicted is generated from the locally decoded image of the region adjacent to the rectangular region to be intra predicted. Similar to the first embodiment, even if the difference between the original image and the locally decoded image becomes large, there is an effect that it is possible to suppress the image quality deterioration of the encoded image.

  Further, according to the second embodiment, when the difference between the adjacent original image and the locally decoded image is larger than the predetermined threshold value, the intra prediction mode selected by the selector 21 (inputted from the outside) is used instead of the fixed DC mode. Intra prediction mode) is selected, for example, an intra prediction image can be generated in a more appropriate intra prediction mode in accordance with the image quality of the encoded image. As compared with the first embodiment, the image quality deterioration of the encoded image can be further suppressed.

  In addition, according to the second embodiment, the intra prediction mode correction unit 22 inputs the correction function ON / OFF flag, and the intra prediction mode correction unit 22 receives the correction function ON / OFF flag only when the correction function ON / OFF flag indicates “correction function ON”. If the intra prediction mode selected by the prediction mode determination unit 2 or the intra prediction mode selected by the selector 21 is selected and the correction function ON / OFF flag indicates “correction function OFF”, the adjacent source mode is selected. The intra prediction mode selection command selected by the selector 21 is output to the selector 23 without performing the selection process of calculating the difference absolute value sum of the image and the locally decoded image and comparing it with a predetermined threshold. Therefore, for example, the effect of being able to adaptively control the operation of the intra prediction mode correction unit 22 according to the image quality situation of the encoded image Unlikely to.

Embodiment 3 FIG.
FIG. 18 is a block diagram showing a moving picture coding apparatus according to Embodiment 3 of the present invention. In the figure, the same reference numerals as those in FIG.
The intra prediction mode correction unit 31 is adjacent to the adjacent original image stored in the original image peripheral pixel memory 1 and the local decoded image stored in the local decoded image peripheral pixel memory 13 (the intra prediction target rectangular region). A process of obtaining a difference from the local decoded image of the region, comparing the difference with a predetermined threshold, and outputting the comparison result to the inter prediction forcing unit 32 is performed.
However, the intra prediction mode correction unit 31 inputs the correction function ON / OFF flag, and performs the above comparison process only when the correction function ON / OFF flag indicates “correction function ON”, and the correction is performed. If the function ON / OFF flag indicates “correction function OFF”, the operation is stopped without performing the above comparison processing.

When the comparison result output from the intra prediction mode correction unit 31 indicates that the difference between the adjacent original image and the local decoded image is smaller than a predetermined threshold, the inter prediction forcing unit 32 receives the inter / intra signal input from the outside. If the prediction determination result indicates that the inter prediction is adopted, an inter prediction image selection command input from the outside is output to the selector 33, and the inter / intra prediction determination result indicates that the intra prediction is adopted. If so, a process of outputting an instruction for selecting an intra predicted image generated by the intra predicted image generation unit 5 to the selector 33 is performed.
On the other hand, when the difference between the adjacent original image and the locally decoded image indicates that the difference is larger than a predetermined threshold, the selection command for the inter prediction image input from the outside regardless of the inter / intra prediction determination result input from the outside Is output to the selector 33.
However, if the intra prediction mode correction unit 31 stops the operation without performing the comparison process, the inter prediction forcing unit 32 may be external if the inter / intra prediction determination result indicates that the inter prediction is adopted. Is output to the selector 33, and if the inter / intra prediction determination result indicates that the intra prediction is adopted, the intra predicted image generated by the intra predicted image generation unit 5 is used. The process of outputting the selection command to the selector 33 is performed.

When the selector 33 receives the intra prediction image selection command from the inter prediction forcing unit 32, the selector 33 selects the intra prediction image generated by the intra prediction image generation unit 5 and outputs the selected image to the subtracter 7 and the adder 12. When an inter-predicted image selection command is received from the unit 32, an inter-predicted image input from the outside is selected and output to the subtracter 7 and the adder 12.
The intra prediction mode correction unit 31, the inter prediction forcing unit 32, and the selector 33 constitute image selection means.

In the first embodiment, the intra prediction mode correction unit 3 outputs the DC mode selection command to the selector 4 when the difference between the adjacent original image and the local decoded image is larger than the predetermined threshold. When the difference between the adjacent original image and the locally decoded image is larger than a predetermined threshold, the prediction forcing unit 32 issues an instruction to select an inter prediction image input from the outside regardless of the inter / intra prediction determination result input from the outside. You may make it output to the selector 33. FIG.
Specifically, it is as follows.

  The intra prediction mode correction unit 31 inputs a correction function ON / OFF flag, and when the correction function ON / OFF flag indicates “correction function ON”, similarly to the intra prediction mode correction unit 3 of FIG. An adjacent original image stored in the original image peripheral pixel memory 1 and a local decoded image stored in the local decoded image peripheral pixel memory 13 (a local decoded image in an area adjacent to a rectangular area to be intra predicted). Thus, for example, a sum of absolute differences between the adjacent original image and the local decoded image is calculated as a difference from the local decoded image used when the intra predicted image generation unit 5 generates the intra predicted image.

When the intra prediction mode correction unit 31 calculates the sum of absolute differences between the adjacent original image and the locally decoded image, the intra prediction mode correction unit 31 compares the sum of absolute differences with a predetermined threshold value input from the outside, and the comparison result is inter-predicted forcibly. To the unit 32.
When the correction function ON / OFF flag indicates “correction function OFF”, the intra prediction mode correction unit 31 stops the operation without performing the above comparison process.
The correction function ON / OFF flag can be adaptively turned on / off in units of pictures, macroblocks, and subblocks.

The inter prediction forcing unit 32, when the comparison result output from the intra prediction mode correction unit 31 indicates that the difference between the adjacent original image and the local decoded image is smaller than a predetermined threshold, If the intra prediction determination result indicates that inter prediction is adopted, an inter prediction image selection command input from the outside is output to the selector 33.
If the inter / intra prediction determination result indicates that the intra prediction is adopted, the intra prediction image selection command generated by the intra prediction image generation unit 5 is output to the selector 33.
On the other hand, when the comparison result output from the intra prediction mode correction unit 31 indicates that the difference between the adjacent original image and the locally decoded image is greater than a predetermined threshold, the inter / intra prediction determination result input from the outside Regardless, an inter prediction image selection command input from the outside is output to the selector 33.

However, the inter prediction forcing unit 32 performs inter / intra prediction when the intra prediction mode correction unit 31 stops the operation without performing the comparison process (when the comparison result is not output from the intra prediction mode correction unit 31). If the determination result indicates that inter prediction is adopted, an inter prediction image selection command input from the outside is output to the selector 33.
If the inter / intra prediction determination result indicates that the intra prediction is adopted, the intra prediction image selection command generated by the intra prediction image generation unit 5 is output to the selector 33.

When the selector 33 receives an instruction for selecting an intra prediction image from the inter prediction forcing unit 32, the selector 33 selects the intra prediction image generated by the intra prediction image generation unit 5 and outputs it to the subtracter 7 and the adder 12.
On the other hand, when an inter prediction image selection command is received from the inter prediction forcing unit 32, an inter prediction image input from the outside is selected and output to the subtracter 7 and the adder 12.
Since the subsequent processing contents are the same as those in the first embodiment, description thereof is omitted.

  As apparent from the above, according to the third embodiment, the difference between the adjacent original image stored in the original image peripheral pixel memory 1 and the local decoded image stored in the local decoded image peripheral pixel memory 13 , The difference is compared with a predetermined threshold, and the comparison result output to the inter prediction forcing unit 32 and the comparison result output from the intra prediction mode correction unit 31 are the adjacent original images. When the difference between the local decoded image and the local decoded image is smaller than a predetermined threshold, if the inter / intra prediction determination result input from the outside indicates that the inter prediction is adopted, the inter prediction input from the outside If an image selection command is output to the selector 33 and the inter / intra prediction determination result indicates that the intra prediction is adopted, the intra predicted image When the selection command for the intra prediction image generated by the generation unit 5 is output to the selector 33 and indicates that the difference between the adjacent original image and the local decoded image is larger than a predetermined threshold, the inter / intra signal input from the outside An inter prediction forcing unit 32 that outputs an inter prediction image selection command input from the outside to the selector 33 regardless of the prediction determination result is provided, and the selector 33 issues an intra prediction image selection command from the inter prediction forcing unit 32. When received, the intra prediction image generated by the intra prediction image generation unit 5 is selected and output to the subtractor 7 and the adder 12, and when an inter prediction image selection command is received from the inter prediction forcing unit 32, it is input from the outside. Since the inter prediction image to be selected is selected and output to the subtracter 7 and the adder 12, the same as in the first embodiment , Even if the difference of the original image and the local decoded image is increased, an effect capable of suppressing image quality degradation of the encoded image.

  Further, according to the third embodiment, only when the intra prediction mode correction unit 31 inputs the correction function ON / OFF flag and the correction function ON / OFF flag indicates “correction function ON”, the difference And the threshold comparison result are output to the inter prediction forcing unit 32, and if the correction function ON / OFF flag indicates “correction function OFF”, the operation is stopped without performing the comparison result. Therefore, for example, there is an effect that the operation of the intra prediction mode correction unit 31 can be adaptively controlled according to the image quality of the encoded image.

  DESCRIPTION OF SYMBOLS 1 Original image periphery pixel memory, 2 Intra prediction mode determination part (intra prediction mode determination means), 3 Intra prediction mode correction | amendment part (mode selection means), 4 Selector (mode selection means), 5 Intra prediction image generation part (Intra prediction) (Image generation means), 6 selector, 7 subtractor (local decoded image generation means), 8 integer conversion section (local decoded image generation means), 9 quantization section (local decoded image generation means), 10 inverse quantization section (local) (Decoded image generation means), 11 inverse integer transform section (local decoded image generation means), 12 adder (local decoded image generation means), 13 local decoded image peripheral pixel memory, 21 selector (mode selection means), 22 intra prediction mode Correction unit (mode selection unit), 23 selector (mode selection unit), 31 intra prediction mode correction unit (image selection unit), 32 Inter prediction forcing part (image selection means), 33 selector 33 (image selection means).

Claims (6)

  Intra prediction mode determination means for determining an intra prediction mode from an original image of a rectangular area to be intra-predicted and an original image of an area adjacent to the rectangular area, and the adjacent original image and the rectangular area When the difference between the adjacent region and the locally decoded image is obtained and the difference is smaller than a predetermined threshold, the intra prediction mode determined by the intra prediction mode determination unit is selected, and the difference is larger than the predetermined threshold. In this case, when an intra prediction mode is selected by the mode selection unit that selects a DC mode for generating a prediction image using an average value of pixel values and the mode selection unit, in the intra prediction mode, from the local decoded image When the intra prediction image of the rectangular area is generated and the DC mode is selected by the mode selection unit, Intra prediction image generation means for generating an intra prediction image of the rectangular area from the locally decoded image, and a difference between the original image of the rectangular area and the intra prediction image generated by the intra prediction image generation means, Local decoded image generating means for outputting a quantized transform coefficient, dequantizing the transform coefficient, adding the inverse quantized transform coefficient and the intra-predicted image, and generating a locally decoded image of the rectangular region A video encoding apparatus comprising:   The mode selection means inputs a flag indicating whether or not the intra prediction mode is corrected, and the intra prediction mode or DC mode selection processing determined by the intra prediction mode determination means only when the flag indicates that there is correction. 2. The moving picture encoding apparatus according to claim 1, wherein the DC mode is selected when the flag indicates no correction.   Intra prediction mode determination means for determining an intra prediction mode from an original image of a rectangular area to be intra-predicted and an original image of an area adjacent to the rectangular area, and the adjacent original image and the rectangular area When the difference between the adjacent region and the locally decoded image is obtained and the difference is smaller than a predetermined threshold, the intra prediction mode determined by the intra prediction mode determination unit is selected, and the difference is larger than the predetermined threshold. In this case, mode selection means for selecting an intra prediction mode input from the outside, and intra prediction image generation for generating an intra prediction image of the rectangular area from the locally decoded image in the intra prediction mode selected by the mode selection means Means, the original image of the rectangular area and the intra predicted image generated by the intra predicted image generating means. Quantize the minutes, output the quantized transform coefficient, dequantize the transform coefficient, add the inverse quantized transform coefficient and the intra prediction image, A moving image coding apparatus comprising local decoded image generation means for generating.   The mode selection means inputs a flag indicating whether or not the intra prediction mode is corrected, and is input from the intra prediction mode determined by the intra prediction mode determination means or from outside only when the flag indicates that there is correction. 4. The moving image encoding apparatus according to claim 3, wherein intra prediction mode selection processing is performed, and if the flag indicates no correction, an intra prediction mode input from the outside is selected.   Intra prediction mode determination means for determining an intra prediction mode from an original image of a rectangular area to be intra-predicted and an adjacent original image that is an original image of an area adjacent to the rectangular area, and determined by the intra prediction mode determination means In intra prediction mode, intra prediction image generation means for generating an intra prediction image of the rectangular area from a local decoded image of an area adjacent to the rectangular area, and a difference between the adjacent original image and the local decoded image When the difference is smaller than a predetermined threshold, if the inter / intra prediction determination result input from the outside indicates that the inter prediction is adopted, the inter prediction image input from the outside is selected, and the inter prediction image is selected. / If the intra prediction determination result indicates that the intra prediction is adopted, the intra prediction image generating means generates When the selected intra predicted image is selected and the difference is larger than a predetermined threshold, regardless of the inter / intra prediction determination result, the image selecting means for selecting the inter image, the original image of the rectangular area, and the image The difference between the images selected by the selection means is quantized and the quantized transform coefficient is output, and the transform coefficient is dequantized, and the inverse quantized transform coefficient and the image selected by the image selecting means are selected. And a local decoded image generating means for generating a local decoded image of the rectangular area.   The image selection means inputs a flag indicating whether or not the intra prediction mode is corrected, and the intra prediction is performed according to the difference and threshold comparison result and the inter / intra prediction determination result only when the flag indicates that there is correction. If the intra prediction image or the inter image generated by the image generation unit is selected and the flag indicates no correction, the intra prediction generated by the intra prediction image generation unit according to the inter / intra prediction determination result 6. The moving image encoding apparatus according to claim 5, wherein an image or an inter image is selected.

Images