WO2014155451A1 - Image coding device and image coding method - Google Patents

Abstract

Provided is an image coding device (100) for coding an inputted image, comprising: a specified region determination unit (103) which determines, for each determination block which is formed from a plurality of pixels and which is included in the inputted image, whether the determination block is a specified region which includes a text character or line art; and an orthogonal transform unit (105) which, by selectively carrying out, on a transform processing unit which is adaptively selected from a plurality of transform processing units, an orthogonal transform of the determination block, outputs a residual coefficient. When it is determined by the specified region determination unit (103) that the determination block is the specified region, the orthogonal transform unit (105) selectively carries out, on each of the transform processing units which are consistently set to a 4×4-pixel block size, the orthogonal transform of the determination block.

Description

Image coding apparatus and image coding method

The present disclosure relates to an image encoding device that can selectively perform orthogonal transform in an encoding operation.

In recent years, with the development of multimedia applications, it has become common to handle all media information such as images, sounds and texts in a unified manner. Also, since a digitized image has a huge amount of data, an image information compression technique is indispensable for storage and transmission.

On the other hand, standardization of compression technology is also important for interoperating compressed image data. For example, as a standard for image compression technology, H.264 of ITU-T (International Telecommunication Union, Telecommunication Standardization Division). 261, H.H. 263, H.M. H.264 and ISO / IEC (International Organization for Standardization) MPEG-1, MPEG-3, MPEG-4, MPEG-4AVC, and the like. At present, standardization activities for a next-generation image coding method called HEVC (High Efficiency Video Coding) in cooperation with ITU-T and ISO / IEC are in progress.

In such image coding, each picture to be coded is divided into coding unit blocks, and the amount of information is compressed by reducing redundancy in the time direction and the space direction for each block. In inter-frame predictive coding for the purpose of reducing temporal redundancy, motion is detected and a predicted image is created in block units with reference to the front or rear picture, and the obtained predicted image and encoding target are obtained. The difference image with the block of is acquired. In addition, in the intra prediction encoding for the purpose of reducing spatial redundancy, a prediction image is generated from pixel information of surrounding encoded blocks, and the obtained prediction image and the block to be encoded are Get the difference image.

* Perform orthogonal transform such as discrete cosine transform and quantization on the obtained difference image. In orthogonal transform, luminance component information and color difference component information are converted into frequency component information. In general, the higher the frequency component, the more difficult it is for human eyes to recognize. For this reason, in general, quantization is performed such that the higher the frequency region is, the wider the quantization range is, the information that is difficult to be recognized by the human eye is omitted, and the deterioration is difficult to understand by the human eye. Furthermore, the amount of information is compressed by generating a code string using variable length coding.

In HEVC (see Non-Patent Document 1), in the above-described orthogonal transform, there are four candidates for the block size to be orthogonally transformed: 4 × 4 pixels, 8 × 8 pixels, 16 × 16 pixels, and 32 × 32 pixels. When determining the block size, one block size is selected from among four candidates by cost determination within a range that can be divided by the quadtree structure for each block to be encoded.

This contributes greatly to the improvement of coding efficiency by adaptively switching the size of orthogonal transform according to the nature of the image. Also, only when a size of 4 × 4 pixels is selected, it is possible to select a method of quantizing a difference image as it is without performing orthogonal transformation.

In the above-described quantization process, quantization is performed with a quantization width determined by multiplying the quantization parameter set for each block and the quantization matrix set for each picture.

In HEVC, as described above, there are four types of block sizes for orthogonal transformation and the division method is complicated. Therefore, when one of the sizes is selected, the amount of calculation processing required for orthogonal transformation is H.264. More than H.264.

In addition, when the orthogonal transformation target block is uniquely determined in order to simply reduce the calculation processing amount, depending on the type of image, there is a risk that subjective image quality may be deteriorated due to noise generated by the orthogonal transformation and quantization. .

Therefore, the present disclosure provides an image encoding device and an image encoding method capable of improving subjective image quality and reducing the amount of calculation processing.

An image encoding device according to the present disclosure is an image encoding device that encodes an input image, and for each determination block including a plurality of pixels included in the input image, the determination block includes a specific area including a character or a line drawing A residual coefficient is output by selectively performing orthogonal transform of the determination block in a determination unit that determines whether or not and a transform processing unit adaptively selected from a plurality of transform processing units. An orthogonal transform unit, and when the determination unit determines that the determination block is a specific area, the orthogonal transform unit always orthogonalizes the determination block for each conversion processing unit set to a block size of 4 × 4 pixels. Perform the conversion selectively.

According to the image coding apparatus according to the present disclosure, it is possible to improve the subjective image quality and reduce the amount of calculation processing.

FIG. 1 is a block diagram showing an image coding apparatus according to Embodiment 1. FIG. 2 is a flowchart showing determination of a specific area and orthogonal transform processing according to the first embodiment. FIG. 3 is a block diagram showing an image coding apparatus according to Embodiment 2. FIG. 4 is a flowchart showing determination of a specific area and orthogonal transformation processing according to the second embodiment. FIG. 5 is a flowchart illustrating specific area determination and orthogonal transform processing according to a modification of the second embodiment. FIG. 6 is a block diagram showing an image coding apparatus according to Embodiment 3. FIG. 7 is a flowchart showing determination of a specific region, orthogonal transform processing, and setting of a quantization parameter according to the third embodiment. FIG. 8 is a flowchart showing the setting of the quantization parameter according to the third embodiment. FIG. 9 is a block diagram showing an image coding apparatus according to Embodiment 4. FIG. 10 is a flowchart showing specific area determination, orthogonal transform processing, and quantization parameter setting according to the fourth embodiment. FIG. 11 is a flowchart illustrating specific area determination, orthogonal transform processing, and quantization parameter setting according to a modification of the fourth embodiment.

Hereinafter, embodiments will be described in detail with reference to the drawings as appropriate. However, more detailed explanation than necessary may be omitted. For example, detailed descriptions of already well-known matters and repeated descriptions for substantially the same configuration may be omitted. This is to avoid the following description from becoming unnecessarily redundant and to facilitate understanding by those skilled in the art.

In addition, the inventors provide the accompanying drawings and the following description in order for those skilled in the art to fully understand the present disclosure, and are not intended to limit the subject matter described in the claims. Absent.

(Embodiment 1)
Hereinafter, the first embodiment will be described with reference to FIGS. 1 and 2.

[Description of Processing of Entire Image Encoding Device]
FIG. 1 is a block diagram of an image coding apparatus 100 according to the present embodiment. The image encoding apparatus 100 divides an image input in units of pictures into blocks and performs an encoding process in units of blocks to generate a code string.

As illustrated in FIG. 1, the image encoding device 100 includes a picture memory 101, a block division unit 102, a specific area determination unit 103, a difference calculation unit 104, an orthogonal transformation unit 105, a quantization unit 106, An inverse quantization unit 107, an inverse orthogonal transform unit 108, an addition operation unit 109, a predicted image generation unit 110, and a code string generation unit 111 are provided.

The picture memory 101 stores the input image signal in units of pictures. When the picture memory 101 receives a read command from the block division unit 102, the picture memory 101 outputs an image signal related to the read command.

The input image signal is image data indicating a still image on a paper surface such as a newspaper or a magazine. Alternatively, the input image signal may be moving image data indicating a video in which characters or line drawings are multiplexed, for example, a video including subtitles.

The block dividing unit 102 divides the image signal input from the picture memory 101 into blocks of, for example, 64 × 64 pixels called a coding unit (CU) that is an encoding processing unit.

The block dividing unit 102 further divides the CU into, for example, 8 × 8 pixel blocks called prediction units (PUs) that are processing units for predictive image generation.

In the above description, the CU describes a block of 64 × 64 pixels, but a block size of 32 × 32 pixels, 16 × 16 pixels, or 8 × 8 pixels may be used.

In the above description, the PU has been described as an 8 × 8 pixel block. However, for example, another size such as 8 × 16 pixels or 8 × 4 pixels may be used. The PU has the same size as the block size of the CU, or a size obtained by dividing the CU into two or four.

The specific area determination unit 103 determines, for each determination block including a plurality of pixels included in the input image, whether or not the determination block is a specific area including a character or a line drawing. Specifically, the specific area determination unit 103 acquires an image feature amount from the encoding target picture output from the block division unit 102 and determines whether or not the predetermined area is a specific area for each predetermined determination block.

It should be noted that the specific area refers to an area including characters or line drawings. In an area drawn with a character or line drawing, an edge is likely to occur between the character or line drawing and the background image. Therefore, the specific area is also an area including an edge.

For the determination block determined as the specific area, 4 × 4 pixels are always selected as a block size (TU size) called a transform unit (TU) that is a unit of orthogonal transformation. The TU size includes blocks of 32 × 32 pixels, 16 × 16 pixels, 8 × 8 pixels, and 4 × 4 pixels. The TU size of the determination block determined not to be the specific region is selected from the above blocks of 32 × 32 pixels to 4 × 4 pixels, which has high encoding efficiency such as orthogonal transform and quantization.

Note that the TU has the same size as the CU block size or a smaller block size. The subsequent processing is performed in units of blocks of CU, PU, and TU depending on the processing content.

Details of the specific area determination will be described later.

The difference calculation unit 104 generates a difference image signal that is a difference value between the PU unit image signal input from the specific region determination unit 103 and the PU unit prediction image signal input from the prediction image generation unit 110. The difference calculation unit 104 outputs the generated difference image signal to the orthogonal transformation unit 105.

The orthogonal transform unit 105 outputs a residual coefficient signal by selectively performing orthogonal transform of the determination block in a transform processing unit adaptively selected from a plurality of transform processing units. Specifically, the orthogonal transform unit 105 generates a residual coefficient signal by performing orthogonal transform on the difference image signal input from the difference calculation unit 104 in units of TUs.

Note that the orthogonal transform unit 105 can select whether or not to perform orthogonal transform when the transform processing unit (TU) is 4 × 4 pixels. That is, the orthogonal transform unit 105 selectively performs orthogonal transform when the TU size is 4 × 4 pixels.

For example, when the TU size is 4 × 4 pixels, the orthogonal transform unit 105 can generate and output a residual coefficient signal by performing orthogonal transform on the difference image signal (operation A). Alternatively, when the TU size is 4 × 4 pixels, the orthogonal transform unit 105 can output the difference image signal as a residual coefficient signal without performing orthogonal transform on the difference image signal (operation B). For example, the orthogonal transform unit 105 selects one of the operation A and the operation B that has better coding efficiency based on a predetermined cost determination.

When the specific region determination unit 103 determines that the determination block is a specific region, the orthogonal transform unit 105 always orthogonalizes the determination block for each transform processing unit (TU) set to a block size of 4 × 4 pixels. Residual coefficients are output by selectively performing the transformation. That is, the orthogonal transform unit 105 determines whether or not to perform orthogonal transform for each TU included in the determination block. For each TU, the orthogonal transform unit 105 generates and outputs a residual coefficient signal by performing orthogonal transform on the difference image signal when performing (i) orthogonal transform, and (ii) when not performing orthogonal transform. The difference image signal is output as a residual coefficient signal as it is.

Further, when the specific area determination unit 103 determines that the determination block is not the specific area, the orthogonal transform unit 105 uses the determination block in a conversion processing unit selected from a plurality of predetermined conversion processing units (TU). The residual coefficient is output by selectively performing the orthogonal transformation. Specifically, when it is determined that the determination block is not a specific region, the orthogonal transform unit 105 selects one block from blocks of 32 × 32 pixels to 4 × 4 pixels. Note that when a 4 × 4 pixel block is selected as a transform processing unit, the orthogonal transform unit 105 can select whether or not to perform orthogonal transform as described above.

The quantization unit 106 quantizes the residual coefficient signal input from the orthogonal transform unit 105 in units of TUs. At this time, the quantization unit 106 generates a quantized residual coefficient signal by quantizing the residual coefficient signal according to a quantization value (quantization parameter) and a quantization matrix set in units of CUs. .

The inverse quantization unit 107 generates a reconstructed residual coefficient signal by performing inverse quantization on the quantized residual coefficient signal input from the quantization unit 106 in units of TUs. The inverse quantization unit 107 outputs the generated reconstructed residual coefficient signal to the inverse orthogonal transform unit 108.

The inverse orthogonal transform unit 108 generates a reconstructed difference image signal by performing inverse orthogonal transform on the reconstructed residual coefficient signal input from the inverse quantization unit 107 in units of TUs. Then, the inverse orthogonal transform unit 108 outputs the generated reconstructed difference image signal to the addition operation unit 109.

The inverse orthogonal transform unit 108 can select whether to perform inverse orthogonal transform when the TU size is 4 × 4 pixels. Specifically, when the reconstructed residual coefficient signal is input from the inverse quantization unit 107, the inverse orthogonal transform unit 108 is obtained by orthogonally transforming the reconstructed residual coefficient signal in the orthogonal transform unit 105. By performing inverse orthogonal transform on the reconstructed residual coefficient signal, a reconstructed differential image signal is generated and output. Further, when the reconstructed residual coefficient signal is obtained without performing orthogonal transform in the orthogonal transform unit 105, the inverse orthogonal transform unit 108 reconstructs the reconstructed residual coefficient signal as it is without performing inverse orthogonal transform. The difference image signal is output to the addition operation unit 109.

The addition operation unit 109 generates a reconstructed image signal by adding the reconstructed difference image signal input from the inverse orthogonal transform unit 108 and the predicted image signal input from the predicted image generation unit 110 in units of PUs. To do.

The predicted image generation unit 110 uses the reconstructed image signal input from the addition operation unit 109 to the PU unit image signal input from the specific region determination unit 103, and performs intra-screen prediction (intra prediction) or PU unit. A prediction image is generated by performing inter-screen prediction (inter prediction). When using inter-screen prediction, the predicted image generation unit 110 uses a reconstructed image signal of a past picture that has already been encoded. On the other hand, when using intra prediction, the prediction image generation unit 110 uses a reconstructed image signal of the same picture that has already been encoded and is adjacent to the PU to be encoded. Note that if the input image input to the image coding apparatus 100 is a still image composed of only one picture, there is no past picture, and therefore only intra prediction is always used.

The code string generation unit 111 performs variable-length coding and arithmetic on the quantized residual coefficient signal, the quantization matrix signal, and other encoded information signals necessary for decoding processing input from the quantization unit 106. A code string is generated by encoding.

[Specific area determination unit and orthogonal transform unit]
Here, the processing performed by the specific area determination unit 103 and the orthogonal transform unit 105 will be specifically described with reference to FIG. FIG. 2 is a flowchart showing determination of a specific region and orthogonal transformation processing according to the present embodiment.

First, the specific area determination unit 103 determines, for each determination block including a plurality of pixels included in the input image, whether the determination block is a specific area including a character or a line drawing (S110). Specifically, the specific area determination unit 103 calculates an image feature amount from a determination block included in the encoding target block (CU) output from the block division unit 102, and determines whether the determination block is a specific area. Determine whether or not.

Here, the specific area determination unit 103 calculates, for example, information based on a luminance component as an image feature amount. The luminance component of a character or line drawing is generally concentrated on high luminance and low luminance, and the luminance changes extremely in a small range. Therefore, the inclination of the luminance component between adjacent pixels tends to increase.

The specific area determination unit 103 calculates, for example, a luminance difference between adjacent pixels as an image feature amount. The specific area determination unit 103 indicates that the determination block is a specific area when the ratio of pixels in the determination block and the ratio of pixels having a large luminance difference between adjacent pixels is equal to or greater than a predetermined ratio. judge.

Specifically, the specific area determination unit 103 calculates a difference in luminance value between adjacent pixels. Then, the specific area determination unit 103 determines whether or not the calculated difference is greater than a predetermined first threshold value. When it is determined that the pixel is larger than the first threshold, the adjacent pixel is determined to be a pixel having a large luminance difference. The specific area determination unit 103 compares the difference between the luminance values with the first threshold value for all the pixels included in the determination block. The first threshold value is, for example, a value that is 50% or more of the difference between the minimum value and the maximum value of the luminance values.

Furthermore, the specific area determination unit 103 calculates a ratio of pixels determined to be pixels having a large luminance difference in the determination block. Then, the specific area determination unit 103 determines whether or not the calculated ratio is greater than or equal to a predetermined ratio (second threshold). For example, the specific area determination unit 103 determines whether or not the pixel ratio is equal to or greater than the second threshold. When the pixel ratio is equal to or greater than the second threshold, the specific area determination unit 103 determines that the determination block is a specific area. The second threshold is a value of 20% or more, for example.

In the above description, the luminance component is used as the reference for the image feature amount. However, any reference may be used as long as it is information calculated from the picture to be encoded.

Further, the specific area determination unit 103 sets the size of the determination block to 8 × 8 pixels, for example. In the present embodiment, as will be described later, a determination block determined to be a specific region is orthogonally converted in units of 4 × 4 pixel conversion processing. For this reason, the size of the determination block may be 4 × 4 pixels or more. However, when the determination block is 4 × 4 pixels, the number of pixel samples per determination block is small, and erroneous determination increases due to noise. Therefore, the determination block for determining the specific area is preferably a block composed of 8 × 8 pixels.

When it is determined that the determination block is a specific area (Yes in S110), the specific area determination unit 103 always selects a 4 × 4 pixel block as the TU size (S120). On the other hand, when it is determined that the determination block is not the specific area (No in S110), the specific area determination unit 103 selects any one of 32 × 32 pixels to 4 × 4 pixels as the TU size (S130). The TU size is determined based on, for example, the encoding target block. For example, a small TU size is selected for a complex image and a large TU size is selected for a simple image.

The orthogonal transform unit 105 selectively performs orthogonal transform of the determination block for each selected TU (S140). For example, when the orthogonal transformation unit 105 selects to perform orthogonal transformation when the TU size is 4 × 4 pixels, the orthogonal transformation unit 105 performs orthogonal transformation on the difference image signal to generate and output a residual coefficient signal. . In addition, when the TU size is 4 × 4 pixels and the orthogonal transform unit 105 selects not to perform the orthogonal transform, the orthogonal transform unit 105 outputs the difference image signal as it is as a residual coefficient signal. Further, when the TU size is not 4 × 4 pixels, the orthogonal transform unit 105 generates and outputs a residual coefficient signal by performing orthogonal transform on the difference image signal. The orthogonal transform unit 105 receives the TU output from the specific region determination unit 103, and determines whether or not the TU size is 4 × 4 pixels.

In the above description, in the process of selecting 4 × 4 pixels as the TU size, the specific area determination unit 103 determines the TU size. However, the present invention is not limited to this. For example, after the block division unit 102 determines the TU size, the specific area determination unit 103 may determine 4 × 4 pixels by overwriting the TU size. In the above description, the block division unit 102 determines each block size. However, the CU and PU sizes may be determined according to the TU size determined by the specific area determination unit 103.

In the above description, the determination block size is 8 × 8 pixels, but other predetermined sizes may be used as long as the size is 4 × 4 pixels or more. This predetermined size may be set depending on the number of pixels occupying one character in the acquired input image.

Specifically, the specific area determination unit 103 may determine the determination block size so that one character in the input image fits in one determination block. The size of characters in the input image depends on the screen resolution of the input image. For example, when encoding a magazine or a newspaper, when the screen resolution is determined, the character size is often determined naturally. Therefore, by setting the determination block according to the screen resolution, it is possible to fit one character into one determination block.

Therefore, for example, the specific area determination unit 103 may determine the determination block size according to the screen resolution of the input image. Specifically, the specific area determination unit 103 determines the determination block size to be 8 × 8 pixels when the screen resolution of the input image is 1920 × 1080 (full HD) or more and 3840 × 2160 (4K2K) or less. May be. Thereby, it is possible to fit one character in the determination block of 8 × 8 pixels. In this case, the number of determination blocks including edges can be reduced.

Further, the specific area determination unit 103 may determine the determination block size to be a non-fixed size with respect to the input image, such as a CU size. That is, the specific area determination unit 103 may set the size of the determination block to N × N pixels (N is an integer of 4 or more). As an example, N is 8.

As described above, the TU size of the determination block determined to be a specific region by the specific region determination unit 103 is always limited to a 4 × 4 pixel block. In the specific area, there are many high-frequency components around the character or line drawing. Since orthogonal transformation and quantization tend to lose high-frequency component information, noise is likely to occur in the vicinity of a character or line drawing.

Therefore, by performing orthogonal transformation and quantization using a small block size of 4 × 4 pixels, noise can be generated in a small block. Therefore, the subjective image quality for the character or line drawing area in the encoding target picture can be improved as compared with the case where the block size is large.

Also, there are usually 4 types of TU sizes, which are determined by a complicated division method. For this reason, when determining the TU size, the amount of calculation processing increases. However, by determining the specific area, the processing block can be omitted by always uniquely setting the TU size of the determination block determined to be the specific area as a 4 × 4 pixel block. Therefore, the amount of calculation processing can be greatly reduced for input images with more characters or line drawings.

[Summary]
The image encoding apparatus according to the present embodiment is an image encoding apparatus 100 that encodes an input image, and for each determination block including a plurality of pixels included in the input image, the determination block displays a character or a line drawing. By selectively performing orthogonal transform of the determination block in a specific region determination unit 103 that determines whether or not the specific region is included, and a transform processing unit adaptively selected from a plurality of transform processing units, An orthogonal transform unit 105 that outputs a residual coefficient. When the specific region determination unit 103 determines that the determination block is a specific region, the orthogonal transform unit 105 is always set to a block size of 4 × 4 pixels. For each transform processing unit, the orthogonal transform of the determination block is selectively performed.

As described above, in the determination block determined to be the specific area, orthogonal transformation is always performed with a size of 4 × 4 pixels, and in the determination block determined not to be the specific area, a plurality of predetermined TU sizes are determined. Orthogonal transformation is performed with a TU size selected from among them. At this time, since the specific region has many high-frequency components, it is a region where noise is likely to occur due to quantization. However, since the generated noise is suppressed to a small size of 4 × 4 pixels, subjective image quality can be improved.

Further, when it is determined that the determination block is a specific area, 4 × 4 pixels are always uniquely determined as the TU size of the determination block, so that the processing required to determine the TU size can be omitted, The amount of processing can be reduced. In this way, by performing orthogonal transformation with an appropriate size, high image quality of the specific area is realized, and the TU size can be uniquely determined in the specific area, so that the amount of calculation processing can be reduced.

(Embodiment 2)
Hereinafter, the second embodiment will be described with reference to FIGS. 3 and 4.

[Description of Processing of Entire Image Encoding Device]
FIG. 3 is a block diagram of image coding apparatus 200 according to the present embodiment. The image coding apparatus 200 divides an image input in units of pictures into blocks and performs a coding process in units of blocks to generate a code string. Note that the second embodiment will be described with a focus on differences from the first embodiment, and the description of the same configuration may be omitted.

As illustrated in FIG. 3, the image coding apparatus 200 according to the present embodiment is different from the image coding apparatus 100 illustrated in FIG. 1 in that the specific area determination unit 103 and the orthogonal transform unit 105 are replaced with a specific area determination unit. The difference is that the unit 203 and the orthogonal transform unit 205 are provided.

The specific area determination unit 203 outputs, to the orthogonal transform unit 205, a result of determining whether or not the determination block is a specific area in addition to the operation shown in the first embodiment.

When the specific region determination unit 203 determines that the determination block is not a specific region, the orthogonal transform unit 205 always executes orthogonal conversion of the determination block. Specifically, the orthogonal transform unit 205 receives the determination result from the specific region determination unit 203 and determines whether or not to perform orthogonal transform based on the received determination result.

In HEVC, whether or not to perform orthogonal transformation can be selected only when the TU size is 4 × 4 pixels. However, in the present embodiment, when the specific area determination unit 203 determines that the determination block is not the specific area, even if a 4 × 4 pixel block is selected as the TU size, the orthogonal transform unit 205 always performs orthogonal transformation.

On the other hand, when the specific area determination unit 203 determines that the determination block is the specific area, 4 × 4 pixels are always selected as the TU size. In this case, the orthogonal transform unit 205 can selectively execute orthogonal transform.

[Specific area determination unit and orthogonal transform unit]
Here, the processing in the specific area determination unit 203 and the orthogonal transform unit 205 according to the present embodiment will be specifically described with reference to FIG. FIG. 4 is a flowchart showing determination of a specific area and orthogonal transformation processing according to the present embodiment.

Here, the processing (S110 to S130) until the TU size is determined is the same as that shown in FIG.

When it is determined that the determination block is a specific area (Yes in S110), 4 × 4 pixels are always selected as the TU size of the determination block (S120). At this time, the orthogonal transform unit 205 selectively performs orthogonal transform on the determination block for each 4 × 4 pixel block (S140).

For example, when the difference between the reconstructed image obtained when orthogonal transformation is performed and the input image is smaller than when orthogonal transformation is not performed, the orthogonal transformation unit 205 selects execution of orthogonal transformation. On the other hand, the orthogonal transform unit 205 selects not to perform orthogonal transform when the difference between the reconstructed image obtained when not performing orthogonal transform and the input image is smaller than when performing orthogonal transform. As described above, when it is determined that the determination block is the specific region, the orthogonal transform unit 205 selectively performs orthogonal transform in the same manner as in the first embodiment.

On the other hand, if it is determined that the determination block is not a specific area (No in S120), any one of 4 × 4 pixels to 32 × 32 pixels is selected as the TU size of the determination block. In this case, the orthogonal transform unit 205 always performs orthogonal transform with the selected TU size (S250).

For example, when 4 × 4 pixels are selected as the TU size, the orthogonal transform unit 205 can normally select whether to perform orthogonal transform. However, determining whether or not to perform orthogonal transformation in all 4 × 4 pixel blocks requires a large amount of calculation processing.

On the other hand, the orthogonal transform unit 205 according to the present embodiment always performs orthogonal transform even when it is determined that the determination block is not a specific region, even if a TU size of 4 × 4 pixels is selected. To do. Thereby, the amount of calculation processing required to determine whether or not to perform orthogonal transform can be reduced.

The determination block determined not to be a specific area is an area that does not include characters or line drawings, such as a natural image. For this reason, even if a high frequency component is lost by orthogonal transformation and quantization, it is not conspicuous subjectively. That is, subjective image quality does not deteriorate.

On the other hand, since the area determined to be a specific area is an area including characters or line drawings, as described above, when high frequency components are lost, noise is noticeable and subjective image quality deteriorates. There is a case. For this reason, when the subjective image quality is better when the orthogonal transform is not performed, the subjective image quality can be improved by selecting the orthogonal transform unit 205 not to perform the orthogonal transform.

As described above, when the determination block is a specific region including a character or a line drawing, the TU size of the determination block can always be uniquely set as a 4 × 4 pixel block by the specific region determination unit 203 and the orthogonal transform unit 205. Furthermore, it can be operated by switching whether or not to perform orthogonal transform only on a specific region.

HEVC can switch whether or not to perform orthogonal transformation in a 4 × 4 pixel block, which is effective in improving image quality. However, when switching whether or not to perform orthogonal transform on all 4 × 4 pixel blocks, the amount of processing increases significantly.

Therefore, as in the present embodiment, switching between whether or not to perform orthogonal transformation is performed only when the specific region has a large effect, and selective orthogonal transformation is performed without significantly increasing the processing amount. Suitable orthogonal transformation and quantization processing can be performed. As a result, high coding efficiency can be obtained for a specific area in the picture to be coded, and improvement in the subjective image quality of the image after coding and increase in the amount of calculation processing can be suppressed.

Also, in a region that is not a specific region, the frequency component spreads without concentrating on the specific component. Therefore, encoding efficiency can be improved by always performing orthogonal transform.

[Summary]
In the image coding apparatus according to the present embodiment, orthogonal transform section 205 always performs orthogonal transform of a determination block when it is determined by specific area determination section 203 that the determination block is not a specific area.

Thus, orthogonal transformation is always executed when it is determined that the region is not a specific region, so that it is possible to reduce the amount of processing required to determine whether or not to perform orthogonal transformation. On the other hand, if it is determined that the region is a specific region, orthogonal transformation can be selectively performed so that the subjective image quality is improved. In this way, whether or not to perform orthogonal transform can be switched adaptively only when necessary, so that it is possible to realize improvement in subjective image quality and improvement in coding efficiency while suppressing the amount of calculation processing.

For example, as illustrated in FIG. 5, even when the determination block is determined to be a specific region (Yes in S110), the orthogonal transform unit 205 performs orthogonal transform on the determination block for each 4 × 4 pixel block. You may do it. FIG. 5 is a flowchart illustrating specific area determination and orthogonal transform processing according to a modification of the present embodiment.

As shown in FIG. 5, whether the determination block is determined to be a specific area (Yes in S110) or the determination block is determined not to be a specific area (No in S110), The orthogonal transform unit 205 always performs orthogonal transform of the determination block for each set transform processing unit (S250). That is, regardless of the determination result of the specific area determination unit 203, the orthogonal transform unit 205 may always execute the orthogonal transform. In other words, regardless of whether or not the determination block is a specific region, the orthogonal transform unit 205 may always perform orthogonal transform.

Thus, in the image coding apparatus according to the modification of the present embodiment, the orthogonal transform unit 205 always has 4 × 4 pixels when the specific region determination unit 203 determines that the determination block is a specific region. The orthogonal transform of the determination block is always executed for each transform processing unit set to the block size.

This makes it possible to reduce the processing required to determine whether or not to perform orthogonal transformation with a 4 × 4 pixel block, thereby reducing the amount of calculation processing.

(Embodiment 3)
Hereinafter, Embodiment 3 will be described with reference to FIGS.

[Description of Processing of Entire Image Encoding Device]
FIG. 6 is a block diagram of an image coding apparatus 300 according to the present embodiment. The image encoding device 300 divides an image input in units of pictures into blocks, and performs an encoding process in units of blocks to generate a code string. Note that the third embodiment will be described with a focus on differences from the first embodiment, and the description of the same configuration may be omitted.

As illustrated in FIG. 6, the image coding apparatus 300 according to the present embodiment includes a specific area determination unit 303 instead of the specific area determination unit 103 as compared to the image coding apparatus 100 illustrated in FIG. 1. The difference is that a quantization parameter setting unit 312 is newly provided.

It should be noted that the specific area determination unit 303 outputs the result of determining whether or not the determination block is a specific area to the quantization parameter setting unit 312 in addition to the operation described in the first embodiment.

The quantization parameter setting unit 312 sets a quantization parameter used by the quantization unit 106 using the determination result output from the specific region determination unit 303.

Specifically, the quantization parameter setting unit 312 sets a quantization parameter in a predetermined processing unit based on a predetermined criterion. For example, the quantization parameter setting unit 312 sets the quantization parameter in a predetermined processing unit based on rate control or the like. Specifically, the quantization parameter setting unit 312 sets the quantization parameter so that the generated code string approaches a predetermined bit rate. Here, the predetermined processing unit is a processing unit in which the quantization parameter can be changed. For example, the predetermined processing unit is a determination block.

Further, the quantization parameter setting unit 312 resets the set quantization parameter to a larger value when the determination block is a specific region. In other words, when the determination block is a specific region, the quantization parameter setting unit 312 sets Q1 that is larger than the quantization parameter value Q2 set when it is determined that the determination block is not the specific region, Set the quantization parameter.

Then, the quantization parameter setting unit 312 outputs the set quantization parameter to the quantization unit 106.

The quantization unit 106 quantizes the residual coefficient signal input from the orthogonal transform unit 105 in units of TUs. At this time, the quantization unit 106 generates a quantization residual coefficient signal by performing quantization using the quantization value (quantization parameter) set by the quantization parameter setting unit 312 and the quantization matrix. . Details of the quantization unit 106 will be described later.

The inverse quantization unit 107 applies the quantization value (quantization value) used when the quantization residual coefficient signal input from the quantization unit 106 is quantized by the quantization unit 106 to the quantization residual coefficient signal. The quantization residual coefficient signal is output to the inverse orthogonal transform unit 108 by performing inverse quantization using the quantization parameter) and the quantization matrix.

[Specific area determination unit, orthogonal transform unit, and quantization parameter setting unit]
Subsequently, processing in the specific region determination unit 303, the orthogonal transform unit 105, and the quantization parameter setting unit 312 according to the present embodiment will be specifically described with reference to FIGS. FIG. 7 is a flowchart showing specific area determination, orthogonal transform processing, and quantization parameter setting according to the present embodiment.

Here, the determination of the TU size (S110 to S130) and the orthogonal transformation (S140) are the same as those in FIG.

If it is determined that the determination block is a specific area (Yes in S110), the quantization parameter setting unit 312 sets the quantization parameter to Q1 (S360).

On the other hand, when it is determined that the determination block is not the specific region (No in S110), the quantization parameter setting unit 312 sets the quantization parameter to Q2 (S370).

Quantization parameters are set for each determination block, for example. The quantization parameter setting unit 312 sets the quantization parameter according to whether or not the determination block is a specific region. For example, when the determination block is a specific region, the quantization parameter used for quantization of the determination block is set to Q1, and when the determination block is not the specific region, the quantization parameter used for quantization of the determination block is set to Q2.

Q1 is set to a value equal to or greater than Q2. Q2 is a value set when it is not determined to be a specific area, and is a value determined based on rate control or the like. On the other hand, the quantization parameter setting unit 312 sets Q1 that is larger than Q2 as a quantization parameter by adding an offset to Q2 determined based on rate control or the like. Thereby, the determination block determined to be the specific region can be roughly quantized.

When the determination block is a specific area, the TU size is always set to 4 × 4 pixels. As a result, high image quality is achieved in the specific area, but since there are many TUs, the overhead increases and the code amount increases. Thus, by setting the quantization parameter for the specific region to a value larger than the quantization parameter for the region that is not the specific region, it is possible to suppress an increase in the code amount. Since the specific area is a small block with a TU size of 4 × 4 pixels, the prediction direction is easy to hit and the prediction accuracy is high. For example, even if the input image to be encoded is a complex image such as a natural image, the TU is a simple block image because the TU size is a small block size of 4 × 4 pixels. For this reason, since prediction accuracy becomes high and a residual component can be reduced, even if it coarsely quantizes, an image quality does not deteriorate easily.

In the above description, although the quantization parameter is set for each determination block, the size of the predetermined processing unit that is the processing unit for setting the quantization parameter may be any standard. For example, the predetermined processing unit may be composed of a plurality of determination blocks. For example, the predetermined processing unit may be a CU.

In this case, the predetermined processing unit includes at least one determination block. That is, the predetermined processing unit is larger than the size of the determination block. As described above, when a plurality of determination blocks are collected, the quantization parameter setting unit 312 sets the quantization parameter based on the ratio of the determination blocks determined to be a specific area within a predetermined processing unit. Specifically, the quantization parameter setting unit 312 is determined based on a predetermined criterion when the ratio of the determination block determined as a specific region within a predetermined processing unit is larger than a predetermined threshold. Set the quantization value to a larger value.

For example, it is assumed that the size of a predetermined processing unit is 32 × 32 pixels, and 16 8 × 8 pixel determination blocks are included therein. At this time, when it is determined that there are 8 or more determination blocks determined to be the specific region among the 16 determination blocks, the quantization parameter setting unit 312 has 7 or less determination blocks determined to be the specific region. Then, the quantization parameter is increased as compared with the case where it is determined.

Here, an example of a quantization parameter setting method will be described with reference to FIG. FIG. 8 is a flowchart showing the setting of the quantization parameter according to the present embodiment.

First, the quantization parameter setting unit 312 sets the quantization parameter to Q2 based on a predetermined criterion (S361). For example, the quantization parameter setting unit 312 determines the quantization parameter based on rate control or the like.

Next, the quantization parameter setting unit 312 determines whether or not the ratio of the determination block determined as the specific area within the predetermined processing unit is larger than a predetermined threshold (S362).

When the ratio of the determination blocks determined as the specific region is larger than the threshold (Yes in S362), the quantization parameter setting unit 312 sets Q1 that is a value larger than Q2 as a quantization parameter by adding an offset to Q2. (S363). When the ratio of the determination blocks determined as the specific region is equal to or less than the threshold (No in S362), the quantization parameter setting unit 312 outputs the quantization parameter set in Q2 to the quantization unit 106 as it is.

As described above, the quantization parameter setting unit 312 uses the case where the ratio of the determination blocks determined as the specific area is equal to or less than the threshold as the quantization parameter when the ratio of the determination blocks determined as the specific area is larger than the threshold. Set a value larger than the quantization parameter.

[Summary]
Image coding apparatus 300 according to the present embodiment sets a quantization parameter of a determination block that is determined to be a specific region by specific region determination unit 303 when it is determined that the determination block is not a specific region. A quantization parameter setting unit 312 for setting a value larger than the quantization parameter to be set, and a quantization unit 106 for quantizing the residual coefficient using the quantization value set by the quantization parameter setting unit 312.

As a result, the quantization parameter of the determination block determined to be the specific region is set to a value larger than the quantization parameter set when the determination block is determined not to be the specific region. The amount of codes can be reduced. In addition, since the determination block determined as the specific area is selected as a TU size of 4 × 4 pixels, that is, a small block, deterioration in image quality is not conspicuous. Therefore, it is possible to suppress deterioration of the subjective image quality of the specific area and reduce the code amount.

In addition, image coding apparatus 300 according to the present embodiment is a specific region in a predetermined processing unit when a predetermined processing unit that is a processing unit for setting a quantization value is equal to or larger than the size of the determination block. The quantization parameter when the ratio of the determination block determined to be greater than a predetermined threshold is greater than the quantization parameter when the ratio of the determination block determined as a specific area within a predetermined processing unit is equal to or less than the threshold. You may provide the quantization parameter setting part 312 set to a big value, and the quantization part 106 which quantizes a residual coefficient using the quantization value set by the quantization parameter setting part 312.

As a result, the quantization parameter of the region having a large ratio of the specific region is set to a larger value than the quantization parameter of the region having a small ratio of the specific region, so that the code amount of the region having a large ratio of the specific region can be reduced. .

(Embodiment 4)
Hereinafter, the fourth embodiment will be described with reference to FIGS. 9 and 10.

Note that the fourth embodiment is an embodiment in which the second embodiment and the third embodiment described above are combined.

[Description of Processing of Entire Image Encoding Device]
FIG. 9 is a block diagram of an image encoding device 400 according to a modification of the embodiment. The image coding apparatus 400 divides an image input in units of pictures into blocks, and performs a coding process in units of blocks to generate a code string. In the fourth embodiment, differences from the second and third embodiments will be mainly described, and the description of the same configuration may be omitted.

As shown in FIG. 9, image coding apparatus 400 differs from image coding apparatus 300 according to Embodiment 3 shown in FIG. 6 in that it includes orthogonal transform section 205 instead of orthogonal transform section 105. Yes. The orthogonal transform unit 205 has the same configuration as the orthogonal transform unit 205 according to the second embodiment.

[Specific area determination unit, orthogonal transform unit, and quantization parameter setting unit]
Subsequently, processing in the specific region determination unit 303, the orthogonal transform unit 205, and the quantization parameter setting unit 312 according to the present embodiment will be described with reference to FIG. FIG. 10 is a flowchart illustrating specific area determination, orthogonal transform processing, and quantization parameter setting according to a modification of the embodiment.

As shown in FIG. 10, the processing until the TU size is determined (S110 to S130), the selective execution of orthogonal transform (S140), and the execution of orthogonal transform (S250) are described in the second embodiment. It is the same as FIG. The quantization parameter setting (S360 and S370) is the same as that in FIG. 8 described in the third embodiment.

[Summary]
As described above, in image coding apparatus 400 according to the present embodiment, orthogonal transform is selected for each block of 4 × 4 pixels selected as a TU size for a determination block determined to be a specific region. Is executed automatically. Further, the quantization parameter used for quantization of the determination block determined to be the specific region is set to Q1 which is a value larger than the quantization parameter value Q2 set when it is determined that the determination block is not the specific region. The

Thus, for example, when orthogonal transformation is not performed on a 4 × 4 pixel block, the residual coefficient to be quantized is a value in the spatial domain, not the frequency domain. Specifically, the residual coefficient indicates a difference in luminance value. Since quantization is performed in the spatial domain, the effect of losing high-frequency components does not appear and image quality deterioration can be suppressed.

Also, since the specific area is an area including a character or a line drawing, the accuracy of intra prediction is better than that of a natural image, and the residual coefficient often has a small value. For this reason, although the number of TU blocks increases, the code amount of one TU can be reduced, and the code amount as a whole can also be suppressed. At this time, since the quantization parameter for the determination block determined as the specific region is increased, the code amount of one TU can be further reduced.

Also, since the TU size is a small block of 4 × 4 pixels, the prediction direction is easy to hit and the prediction accuracy is high. For example, even when the input image to be encoded is a complex image such as a natural image, the TU is a small block size of 4 × 4 pixels. Increases accuracy.

On the other hand, orthogonal transformation is always executed in a determination block that is determined not to be a specific region. The determination block determined not to be the specific region is a complex image such as a natural image. For this reason, even if high frequency components are lost due to orthogonal transformation and quantization, subjective deterioration of image quality is suppressed. Therefore, it is possible to reduce the processing amount required to determine whether or not to perform orthogonal transformation without degrading subjective image quality.

As described above, according to the image coding apparatus 400 according to the present embodiment, it is possible to improve the subjective image quality and reduce the amount of calculation processing.

As in the modification of the second embodiment, for example, as illustrated in FIG. 11, even when the determination block is determined to be a specific region (Yes in S110), the orthogonal transform unit 205 is 4 × Orthogonal transformation may always be performed on the determination block every four pixel blocks. FIG. 11 is a flowchart illustrating specific area determination and orthogonal transform processing according to a modification of the present embodiment.

As shown in FIG. 11, whether the determination block is determined to be a specific area (Yes in S110) or the determination block is determined not to be a specific area (No in S110), The orthogonal transform unit 205 always performs orthogonal transform of the determination block for each set transform processing unit (S250). That is, regardless of the determination result of the specific area determination unit 203, the orthogonal transform unit 205 may always execute the orthogonal transform. In other words, regardless of whether or not the determination block is a specific region, the orthogonal transform unit 205 may always perform orthogonal transform.

As a result, similar to the modification of the second embodiment, it is possible to reduce the processing required to determine whether or not to perform orthogonal transformation with a 4 × 4 pixel block, so that the amount of calculation processing can be reduced. it can.

(Other embodiments)
As described above, Embodiments 1 to 4 have been described as examples of the technology disclosed in the present application. However, the technology in the present disclosure is not limited to this, and can also be applied to an embodiment in which changes, replacements, additions, omissions, and the like are appropriately performed. In addition, it is possible to combine the components described in the first to fourth embodiments to form a new embodiment.

For example, in the modification of the second embodiment and the modification of the fourth embodiment, the orthogonal transform unit 205 determines the determination block for each set transform processing unit regardless of whether the determination block is a specific region. Although an example in which orthogonal transformation is always executed has been described, this is not restrictive. For example, contrary to Embodiment 2, when it is determined that the determination block is a specific region (Yes in S110), the orthogonal transform unit 205 always executes orthogonal transform and the determination block is not the specific region. (No in S110), the orthogonal transformation may be selectively executed. Even in this case, if it is determined that the determination block is a specific area, the processing required to determine whether or not to perform orthogonal transformation can be reduced, so that the amount of calculation processing can be reduced. it can.

In Embodiments 1 to 4, the HEVC standard has been described as the encoding standard used by the image encoding apparatus according to each embodiment. The encoding standard only needs to selectively perform orthogonal transformation. Therefore, the encoding standard is not limited to the HEVC standard.

Furthermore, the processing described in the above embodiment is performed by recording a program having a function equivalent to each unit included in the image encoding device described in the above embodiment on a recording medium such as a flexible disk. Can be easily implemented in an independent computer system. The recording medium is not limited to a flexible disk, but can be similarly implemented as long as it can record a program, such as an optical disk, an IC card, and a ROM (Read Only Memory) cassette.

In addition, a function equivalent to each unit included in the image encoding device shown in the above embodiment may be realized as an LSI which is an integrated circuit. These may be integrated into one chip so as to include a part or all of them. An LSI may also be called an IC, a system LSI, a super LSI, or an ultra LSI depending on the degree of integration.

Further, the method of circuit integration is not limited to LSI, and implementation with a dedicated circuit or a general-purpose processor is also possible. An FPGA (Field Programmable Gate Array) that can be programmed after manufacturing the LSI or a reconfigurable processor that can reconfigure the connection and setting of circuit cells inside the LSI may be used.

Further, if integrated circuit technology that replaces LSI or the like appears due to progress in semiconductor technology or other derived technology, the functional blocks may naturally be integrated using this technology.

Specifically, each component (the picture memory 101, the block division unit 102, the specific area determination units 103, 203, and 303, and the difference included in the image encoding devices 100, 200, 300, and 400 according to the present disclosure Operation unit 104, orthogonal transform units 105 and 205, quantization unit 106, inverse quantization unit 107, inverse orthogonal transform unit 108, addition operation unit 109, predicted image generation unit 110, and code string generation unit 111 and a quantization parameter setting unit 312) are programs executed on a computer including a CPU (Central Processing Unit), a RAM (Random Access Memory), a ROM, a communication interface, an I / O port, a hard disk, a display, and the like. It may be realized by software such as It may be implemented in hardware such as circuit.

Further, at least a part of the functions of the image encoding device or its modification according to the above embodiment may be combined.

As described above, the embodiments have been described as examples of the technology in the present disclosure. For this purpose, the accompanying drawings and detailed description are provided.

Accordingly, among the components described in the attached drawings and detailed description, not only the components essential for solving the problem, but also the components not essential for solving the problem in order to exemplify the above technique. May also be included. Therefore, it should not be immediately recognized that these non-essential components are essential as those non-essential components are described in the accompanying drawings and detailed description.

In addition, since the above-described embodiment is for illustrating the technique in the present disclosure, various modifications, replacements, additions, omissions, and the like can be performed within the scope of the claims or an equivalent scope thereof.

The present disclosure is useful, for example, for an image encoding device that inputs a paper such as a newspaper or a magazine as image data of a still image and outputs it as a still image code string by performing an encoding process. In addition, the present invention is useful as an image encoding device that inputs video in which characters or figures are multiplexed as moving image data and outputs it as a moving image code string by performing encoding processing.

100, 200, 300, 400 Image coding apparatus 101 Picture memory 102 Block division unit 103, 203, 303 Specific area determination unit 104 Difference calculation unit 105, 205 Orthogonal transformation unit 106 Quantization unit 107 Inverse quantization unit 108 Inverse orthogonal transformation Unit 109 addition calculation unit 110 prediction image generation unit 111 code string generation unit 312 quantization parameter setting unit

Claims (8)

1. An image encoding device for encoding an input image,
   A determination unit that determines, for each determination block including a plurality of pixels included in the input image, whether the determination block is a specific region including a character or a line drawing;
   An orthogonal transformation unit that outputs a residual coefficient by selectively performing orthogonal transformation of the determination block in a transformation processing unit adaptively selected from a plurality of transformation processing units;
   When the determination unit determines that the determination block is the specific area, the orthogonal conversion unit performs orthogonal conversion of the determination block for each conversion processing unit set to a block size of 4 × 4 pixels. An image encoding device that is selectively performed.
2. The image encoding device according to claim 1, wherein the orthogonal transform unit always performs orthogonal transform of the determination block when the determination unit determines that the determination block is not the specific region.
3. When the determination unit determines that the determination block is the specific area, the orthogonal conversion unit performs orthogonal conversion of the determination block for each conversion processing unit set to a block size of 4 × 4 pixels. The image encoding apparatus according to claim 1 or 2, which is necessarily executed.
4. The image encoding device according to any one of claims 1 to 3, wherein the determination unit further sets the size of the determination block to N × N pixels (N is an integer of 4 or more).
5. The image encoding device according to claim 4, wherein N is 8. 5.
6. The image encoding device further includes:
   A setting unit that sets a quantization parameter of a determination block determined to be the specific region by the determination unit to a value larger than a quantization parameter that is set when the determination block is determined not to be the specific region When,
   6. The image encoding device according to claim 1, further comprising a quantization unit that quantizes the residual coefficient using a quantization value set by the setting unit.
7. The image encoding device further includes:
   When a predetermined processing unit that is a processing unit for setting a quantization value is equal to or larger than the size of the determination block, a ratio of the determination block determined to be the specific area in the predetermined processing unit is determined in advance. A setting unit that sets a quantization parameter when larger than the threshold value to a value larger than the quantization parameter when the ratio of the determination blocks determined as the specific region within the predetermined processing unit is equal to or less than the threshold value;
   6. The image encoding device according to claim 1, further comprising a quantization unit that quantizes the residual coefficient using a quantization value set by the setting unit.
8. An image encoding method for encoding an input image, comprising:
   For each determination block consisting of a plurality of pixels included in the input image, determine whether the determination block is a specific region including a character or a line drawing,
   By selectively executing orthogonal transformation of the determination block in a transformation processing unit adaptively selected from a plurality of transformation processing units, a residual coefficient is output,
   In the selective execution of the orthogonal transform,
   An image encoding method for selectively performing orthogonal transform of the determination block for each transform processing unit always set to a block size of 4 × 4 pixels when it is determined that the determination block is the specific region.
   

Images