JP6265705B2 - Image coding apparatus and image coding method - Google Patents

Description

  The present invention relates to an image encoding device and an image encoding method, and more particularly to a technique suitable for use in encoding an image of each color (RGB) of red (R), green (G), and blue (B). It is.

  Currently, digital devices capable of recording moving images, such as digital video cameras and hard disk recorders, are in widespread use. In these digital devices, in order to efficiently record a moving image having a large amount of information on a recording medium such as a flash memory or a hard disk with a limited capacity, the moving image data is compressed and encoded.

  Various systems have been proposed and standardized as moving picture compression encoding systems. One of the typical moving image compression coding systems is the H.264 coding system. The H.264 encoding method is employed in AVCHD, which is a high-definition recording method for video cameras, and one-segment broadcasting of terrestrial digital broadcasting, and is a widely used video compression encoding method.

  In such an encoding method, a YCbCr signal is mainly used as an input signal. The YCbCr signal is a signal obtained by color space conversion from an RGB signal to a luminance signal Y and two color difference signals Cb and Cr.

  Since the color difference signals Cb and Cr are less visible than the luminance signal Y, there is a feature that even if the resolution is lowered, the visual influence is small. Therefore, the amount of information can be suppressed by down-sampling the color difference signal. As a format in which the amount of information is suppressed, YCbCr 4: 2: 2 format in which the number of pixels of the color difference signal in the horizontal direction is halved, and YCbCr 4: 2: 0 in which the number of pixels in both the horizontal direction and the vertical direction is halved. There are formats.

  In the image encoding apparatus, the use of these YCbCr 4: 2: 2 format and YCbCr 4: 2: 0 format can suppress the amount of code and reduce various processing amounts at the time of encoding. Benefits are gained.

  On the other hand, with the recent increase in resolution and gradation of displays and video cameras, an encoding method for directly encoding RGB signals and providing a higher quality encoded image is also standardized. The RGB signals all have the same number of pixels in each of the RGB planes, and are also referred to as RGB 4: 4: 4 format. As an encoding apparatus that encodes the RGB 4: 4: 4 format, techniques such as Patent Document 1 and Patent Document 2 can be cited.

  There are two methods for encoding the RGB 4: 4: 4 format. One is an encoding method in which a motion vector, an intra prediction mode, a quantization value, and the like can be independently controlled in each of the RGB planes. Another method is a coding method that uses a common motion vector, intra prediction mode, quantized value, and the like for each RGB plane.

One of the different processes in the 4: 2: 0 image signal or 4: 2: 2 image signal encoding process and the 4: 4: 4 image signal encoding process is motion search.
Conventionally, in encoding of a 4: 2: 0 image signal or a 4: 2: 2 image signal, a motion search is generally performed using the luminance signal Y. The luminance signal Y is suitable as a signal for motion search because it can express most of the contour and contrast of the subject and has high coverage.

JP 2010-110006 A JP 2008-172599 A

  In the image encoding device described in the patent document, when the RGB 4: 4: 4 format is encoded, if the calculation of intra prediction mode selection is performed using all the RGB planes, the amount of calculation is very large. Become. In particular, an encoding scheme that selects one common intra prediction mode in units of blocks for each of the RGB planes uses a 4: 2: 0 or 4: 2: 2 format to select one intra prediction mode. The amount of calculation is three times as large as that for intra prediction for the luminance signal. For this reason, there has been a problem that the processing load becomes very large.

  On the other hand, in the 4: 4: 4 format encoding device described in Patent Document 2, a motion search is performed using a part of the signals among the three planes, or a motion search is performed on all the planes. A simple motion vector is selected. For this reason, the following problems exist.

When motion search is performed using all planes, the amount of processing required for motion search is tripled, and the processing cycle and hardware scale increase. In addition, the reference image data read from the reference image memory is required three times, and the memory bus rate is increased. Therefore, it is necessary to provide a high-speed memory bus, and there is a problem that power consumption increases.
The present invention has been made in view of the above-described problems, and an object of the present invention is to be able to reduce the calculation processing load performed when an RGB image is encoded.

The image encoding device of the present invention is an image encoding device that encodes an image in 4: 4: 4 format, and includes image feature detection means for detecting the image feature of each plane of input image data for each block; A plane selection unit that selects a plane suitable for intra prediction mode selection according to the image feature of each plane detected by the image feature detection unit, and an intra prediction unit that performs intra prediction, the intra prediction unit includes: A common intra prediction mode is selected for each plane using image data of the plane selected by the plane selection unit, and an intra prediction mode is not selected from the planes not selected by the plane selection unit. .

  According to the present invention, it is possible to significantly reduce the calculation processing load performed when an RGB image is encoded. This makes it possible to obtain a good encoded image while suppressing an increase in circuit scale.

It is a block diagram which shows the structural example of the image coding apparatus of 1st Embodiment. It is a figure explaining the pixel for reference images used for intra prediction. It is a figure which shows the prediction mode of intra 4x4 prediction. It is a figure explaining the production method of the prediction image in the prediction mode of intra 4x4 prediction. It is a figure explaining the influence on intra prediction when plane selection is performed using feature detection. It is a block diagram which shows the structural example of the image coding apparatus of 2nd Embodiment. It is a block diagram which shows the structural example of the image coding apparatus of 3rd Embodiment. It is a block diagram which shows the structural example of the image coding apparatus of 4th Embodiment.

Hereinafter, preferred embodiments of the present invention will be described with reference to the drawings.
(First embodiment)
FIG. 1 is a block diagram illustrating a configuration example of an image encoding device according to the present embodiment. Here, an example of an encoding apparatus that encodes RGB 4: 4: 4 format composed of each color of red (R), green (G), and blue (B) using the H.264 encoding method is shown. Yes. Moreover, in this embodiment, in order to perform description regarding intra prediction, a block diagram and description regarding inter prediction are omitted.

The encoded image of the RGB 4: 4: 4 format to be encoded is stored in the frame buffer 101.
The feature detection unit 102 reads out the macroblock to be encoded from the frame buffer 101 for each RGB plane, performs image feature detection for each plane, and outputs the result to the plane selection unit 103.

  The feature of the image can be detected, for example, by calculating a dispersion value indicating the variation of the pixels of the macroblock. When the pixel values of the macroblock are P1 to P256, the average value Pmean of the pixels and the variance value var of the pixels can be calculated by the following equations (1) and (2).

  The plane selection unit 103 performs the plane selection most suitable for the intra prediction mode selection from the image feature information of each RGB plane sent from the feature detection unit 102. The selected plane information is output to the intra prediction unit 104. Various plane selection methods are conceivable. For example, when the image feature information is a variance value of a macroblock, a plane having the highest variance value is selected as a plane used for intra prediction mode selection.

  Here, the variance value of the pixel of the macroblock is used as the image feature information, but the image feature information is not limited to this. For example, using the result of edge detection, the plane with the largest edge component is selected. You may select and the value which performed frequency conversion may be used. In addition, although a high-macroblock pixel dispersion value is selected as the plane selection method, the selection method is not limited to this, and a plane having an image feature with a large spatial frequency may be selected.

  The intra prediction unit 104 uses the plane information sent from the plane selection unit 103 to reference the encoded image of the corresponding plane stored in the frame buffer 101 and the corresponding plane stored in the reference frame buffer 113. Read the image. Then, the intra prediction unit 104 performs intra prediction mode selection using the read-out plane encoded image and reference image. By doing in this way, compared with the case where intra prediction mode selection is performed using 3 planes of RGB, the amount of calculation can be significantly reduced.

In the intra prediction of the H.264 encoding method, pixels around a macroblock to be encoded are used as a reference image. The reference image pixels used for intra prediction are shown in FIG.
Intra prediction modes include intra 4 × 4 prediction, intra 8 × 8 prediction, and intra 16 × 16 prediction, and there are a plurality of prediction modes such as a vertical prediction mode, a horizontal prediction mode, and a DC prediction mode, respectively. Here, the intra 4 × 4 prediction mode will be described as an example.

  For the pixels a to p of the 4 × 4 block to be encoded, the surrounding pixels A to M are used as pixels for the predicted image. The pixels A to D are four pixels adjacent in the upward direction of the 4 × 4 block to be encoded, and the pixels E to H are four pixels continuing in the right direction of the pixel D. The pixels I to L are four pixels adjacent in the left direction of the 4 × 4 block to be encoded, and the pixel M is a pixel at a position above the pixel I. As the peripheral pixels A to M, pixels of the original image or pixels of the local decoded image subjected to encoding are used.

  Either pixel can be used when the intra prediction mode is selected, but when generating a difference image between the encoded image output to the orthogonal transform unit 105 and the predicted image, a local decoded image is used as a reference image for generating a predicted image. Must be used. How a predicted image is created using the peripheral pixels differs depending on the prediction mode.

FIG. 3 shows a prediction mode for intra 4 × 4 prediction.
For intra 4 × 4 prediction, nine prediction modes from prediction mode 0 to prediction mode 8 are prepared. FIG. 4 shows a method for creating a predicted image in each prediction mode.

  For example, the prediction mode 0 is a mode for generating a prediction image from pixels A to D adjacent in the vertical direction. For the pixels a to p of the 4 × 4 block to be encoded, the prediction image for the pixels a, e, i, m in the first column is the prediction for the pixels b, f, j, n in the second column. The image is pixel B. The predicted image for the pixels c, g, k, and o in the third column is the pixel C, and the predicted image for the pixels d, h, l, and p in the fourth column is the pixel D.

Moreover, the prediction mode 2 is DC prediction, and the pixel shown by Formula (3) becomes a prediction image with respect to all the pixels a-p.
(A + B + C + D + I + J + K + L + 4) >> 3 (3)
In this way, subtraction is performed between the prediction image generated in each prediction mode and the encoded image to generate pixel difference data. Using this pixel difference data, it is selected which intra prediction mode is used for encoding. For example, the prediction mode 0 to the prediction mode 8 that has the smallest sum of pixel difference values is selected as the intra prediction mode.

  The influence on intra prediction when plane selection is performed using feature detection will be described with reference to FIG. There is an RGB 4: 4: 4: format macroblock 501 to be encoded. The macroblock 501 is divided into R planes, the macroblock 502 is divided into G planes, and the macroblock 503 is divided into B planes. What is a macroblock 504.

  Here, it is assumed that the macroblock 502 of the R plane and the macroblock 503 of the G plane are flat as image features, and the macroblock 504 of the B plane has a strong horizontal correlation as the image features.

  The R-plane macroblock 502 and the G-plane macroblock 503 are flat as image features. In this case, the variance value in the macroblock is a small value. On the other hand, the macroblock 504 of the B plane has a strong correlation in the horizontal direction as a feature of the image. In this case, the variance value in the macroblock is a large value.

  Since the macroblock 502 of the R plane and the macroblock 503 of the G plane are flat, the difference between the pixel difference values of the predicted image and the encoded image hardly occurs regardless of which prediction mode is selected as the intra prediction mode. There is little impact on.

  On the other hand, the macroblock 504 of the B plane has a feature that the correlation in the horizontal direction is strong as a feature of the image. In this case, the pixel difference value between the predicted image and the encoded image tends to increase in the prediction mode from the vertical direction, but the pixel difference value between the prediction image and the encoded image tends to decrease in the prediction mode from the horizontal direction. That is, in the macroblock 504 of the B plane, the influence of the intra prediction mode selection becomes large.

  In this case, the accuracy of the intra prediction mode selection is higher when the intra prediction is performed using the B plane having a large influence on the intra prediction mode selection than the R plane and the G plane having a small influence on the intra prediction mode selection. . That is, by calculating a variance value for each RGB plane of a macroblock to be encoded as a feature of an image and performing intra prediction mode selection using a plane with a high variance value, while maintaining the accuracy of intra prediction mode selection The amount of calculation can be reduced.

  In the intra prediction unit 104, using the determined intra prediction mode, a predicted image is generated using a local decoded image for each of the RGB planes, and a pixel difference is calculated between the predicted image and the encoded image. The image is output to the orthogonal transform unit 105. In addition, the prediction image is output to the motion compensation unit 111 for creating a local decoded image.

The orthogonal transform unit 105 performs orthogonal transform (for example, discrete cosine transform) on the transmitted difference image, generates a transform coefficient, and outputs the transform coefficient to the quantization unit 106.
The quantization unit 106 performs quantization on the received transform coefficient according to the quantization parameter output from the quantization control unit 107. The quantized transform coefficient is sent to the entropy coding unit 108 and the inverse quantization unit 109 for creating a local decoded image.

  The entropy encoding unit 108 performs variable length encoding by performing a zigzag scan, an alternate scan, and the like on the quantized transform coefficients. On the other hand, an encoded stream is generated by adding a variable length encoded encoding method information such as a motion vector, a quantization parameter, and macroblock division information. In addition, the amount of generated code for each macroblock is calculated at the time of encoding and is sent to the quantization control unit 107.

The quantization control unit 107 determines a quantization parameter. The generated code amount is received from the entropy encoding unit 108, the quantization parameter is determined so as to be a target code amount, and the generated code amount is output to the quantization unit 106.
The inverse quantization unit 109 performs inverse quantization on the input quantized transform coefficient, generates a transform coefficient for local decoding, and outputs the transform coefficient to the inverse orthogonal transform unit 110.

The inverse orthogonal transform unit 110 performs the inverse transform of the above-described orthogonal transform (for example, inverse discrete cosine transform) on the input transform coefficient to generate a difference image. The difference image is output to the motion compensation unit 111.
The motion compensation unit 111 generates image data for local decoding by adding the prediction image for intra prediction sent from the intra prediction unit 104 and the difference image sent from the inverse orthogonal transform unit 110. To do. The generated image data is output to the deblocking filter unit 112.

The deblocking filter unit 112 applies a deblocking filter to the input image data. The image after the deblocking filter is stored in the reference frame buffer 113 as a local decoded image.
In this way, an encoded stream and a local decoded image are generated.

  As described above, in this embodiment, image feature detection is performed for each macroblock. Based on the characteristics, one plane is selected from the three RGB planes, and the intra prediction mode is determined using only the selected plane, thereby greatly increasing the calculation amount without reducing the accuracy of selecting the intra prediction mode. Can be reduced.

In the present embodiment, the RGB 4: 4: 4 format has been described as an example of the image format. However, the present invention is not limited to this, and other color spaces may be used.
In addition, although the H.264 encoding method has been described as an example of the image compression encoding method, the encoding method is not limited to this. Any image compression encoding method may be applied as long as it uses intra prediction encoding such as H.265 (HEVC).

(Second Embodiment)
The present embodiment is an image encoding device that performs encoding using intra prediction, as in the first embodiment.
FIG. 6 is a block diagram illustrating a configuration example of the image encoding device according to the present embodiment. Here, the same components as those in the first embodiment are denoted by the same reference numerals, description thereof is omitted, and only differences from the first embodiment will be described in detail.
The difference from the first embodiment is that, in addition to feature detection of an encoded image, plane selection is performed by weighting the ease of selection of each plane in view of the nature of the color space of the 4: 4: 4 format. That is.

  The weight calculation unit 601 receives the color space information of the encoded image, and calculates a weight coefficient corresponding to the color space. This color space information is, for example, information that the encoded image is in the RGB 4: 4: 4 format or the YCbCr 4: 4: 4 format of ITU-R BT.601. Here, as an example, it is assumed that the encoded image is in the RGB 4: 4: 4 format.

  In the case of an RGB signal, the degree of correlation with the luminance signal Y differs in each plane. In general, the luminance signal Y has a characteristic that the visual influence on the coding deterioration is larger than that of the color difference signals Cb and Cr. The stronger the correlation with the luminance signal Y, the larger the visual influence on the coding deterioration. . In consideration of this characteristic, each plane is weighted at the time of plane selection, and the ease of selection is changed.

The weighting factor for the R plane is α, the weighting factor for the G plane is β, the weighting factor for the B plane is γ, and the weight calculation unit 601 calculates these values. It is assumed that the larger the value of these coefficients, the easier the corresponding plane is selected. In the case of the RGB signal, the correlation with the luminance signal Y is strong in the order of G>R> B, and the visual influence on the coding deterioration becomes large. Therefore, the coefficient of the relationship as in Expression (4) is calculated.
β>α> γ (4)

The calculated weighting factors α, β, and γ are output to the plane selection unit 602.
In the plane selection unit 602, the more the luminance signal component is included, the easier it is to select.
Although an example in which an image in the RGB 4: 4: 4 format is input is shown here, when an image in another color space is input, different coefficients are calculated according to the characteristics of the color space. Will do. For example, when the input image is in the YCbCr 4: 4: 4 format and it is desired to perform intra prediction only with the Y signal, the weighting factor of the Y signal is set to 1, and the weighting factor of the Cb and Cr signals is set to 0. It is also possible.

  The plane selection unit 602 receives a variance value as image feature information from the feature detection unit 102. Here, the variance value of the R plane is VAR_R, the variance value of the G plane is VAR_G, and the variance value of the B plane is VAR_B. Also, weighting factors α, β, and γ are received from the weight calculation unit 601.

  A value obtained by multiplying the variance value by a weighting coefficient is an evaluation value for plane selection. The evaluation value of each plane is set to estmR, estmG, and estmB, respectively, and is shown in Expression (5).

The plane selection unit 602 compares estmR, estmG, and estmB, and selects the plane having the largest evaluation value. The selected plane information is output to the intra prediction unit 104, and the same processing as in the first embodiment is performed.
As described above, it is possible to perform intra prediction mode selection with high accuracy in consideration of the influence of visual characteristics by weighting plane selection used for intra prediction mode selection using color space information.

(Third embodiment)
The first embodiment and the second embodiment have described the image coding apparatus that performs coding using intra prediction. In the present embodiment, an image coding apparatus that performs coding by performing motion search will be described.
FIG. 7 is a block diagram illustrating a configuration example of the image encoding device according to the present embodiment. Hereinafter, the operation of each block will be described in the configuration example of the image encoding device of the present embodiment. In the description of the present embodiment, the 4: 4: 4 image signal is described as an example using RGB as the color space, but the present invention is not limited to this and may be another color space.

  An RGB 4: 4: 4 format image signal (hereinafter, RGB signal) input from the outside to the image input unit 701 is stored as input image data in the input image storage area of the memory 704 via the memory interface 705. The memory 704 functions as an input image memory that holds input image data and a reference image memory that holds reference image data.

  The memory interface 705 is an interface that arbitrates read / write requests from each block (image input unit 701, motion search unit 702, motion compensation unit 703) and performs data read / write control on the memory 704.

The memory 704 is a random access memory such as a DRAM, and has an input image storage area for storing input image data and a reference image storage area for storing reference image data.
The input image data stored in the input image storage area of the memory 704 is read out in units of macroblocks via the memory interface 705 and input to the image feature detection unit 711 and the motion search unit 702.

  The image feature detection unit 711 detects the image feature of each RGB plane of the input image data that has been input for each macroblock, and outputs the image feature information to the plane selection unit 712. The image feature is detected by detecting an activity, for example. The activity is determined by a variance indicating how scattered the pixel values in the block are from the block average value. In addition to this, detection of image features may be combined with edge detection, frequency spectrum, and the like.

The plane selection unit 712 compares the image feature information of each plane, selects a plane most suitable for motion search for each macroblock, and outputs the selected plane to the motion search unit 702 as plane selection information. If the image feature information is, for example, an activity, the plane with the highest activity value may be selected.
The present embodiment is not limited to the specific detection method of the image feature detection unit 711 and the specific selection method of the plane selection unit 712.

  The motion search unit 702 performs prediction mode determination and motion search for each macroblock based on the plane selection information from the plane selection unit 712. As described above, all the RGB signal planes are input to the input image data for motion search, and a plane indicated in the plane selection information is selected and used. On the other hand, as the reference image data, only the image data of the plane indicated by the plane selection information is read from the memory 704 via the memory interface 705 and used.

The prediction mode information and motion vector information obtained as described above are output to the motion compensation unit 703.
The motion compensation unit 703 reads the input image data stored in the memory 704 and the reference image data stored in the same way through the memory interface 705, and performs motion compensation prediction processing and local decoding processing in units of macroblocks.

The motion compensation prediction process in the motion compensation unit 703 generates prediction image data from reference image data using the prediction mode information and motion vector information from the motion search unit 702, and calculates a difference value from the input image data as a prediction residual image. The data is output to the orthogonal transformation unit 706 as data.
The local decoding process in the motion compensation unit 703 generates local decoded image data using the prediction residual image data from the inverse orthogonal transform unit 709 and the reference image data input from the memory interface 705. The generated locally decoded image data is stored in the reference image storage area of the memory 704 via the memory interface 705.

The orthogonal transform unit 706 performs orthogonal transform such as DCT on the input prediction residual image data and outputs transform coefficient data to the quantization unit 707.
The quantization unit 707 quantizes the transform coefficient data and outputs the quantized coefficient data to the inverse quantization unit 708 and the entropy coding unit 710.

The inverse quantization unit 708 inversely quantizes the transform coefficient data quantized by the quantization unit 707 and outputs the result to the inverse orthogonal transform unit 709.
The inverse orthogonal transform unit 709 performs inverse orthogonal transform on the transform coefficient data inversely quantized by the inverse quantization unit 708, and outputs prediction residual image data used for local decoding to the motion compensation unit 703.

  The entropy encoding unit 710 performs entropy encoding specified by the encoding method on the transform coefficient data quantized by the quantization unit 707 and encoding information such as a prediction mode and a motion vector for each macroblock. And output to the outside as encoded data. The encoded data output from the entropy encoding unit 710 is output to a transmission medium or recording medium (not shown).

  As described above, in this embodiment, the motion search unit 702 detects the image feature of each plane using the input image data read from the memory 704, and selects the plane most suitable for motion search. I made it. The motion search unit 702 performs a motion search using only the image data of the selected plane.

  Thereby, it is possible to accurately perform the motion search of the 4: 4: 4 image signal without increasing the processing amount of the motion search. In addition, the reference image data read out from the memory 704 by the motion search unit 702 is read out only from the selected plane. Thereby, the reading amount of the image data from the memory can be suppressed.

(Fourth embodiment)
In the present embodiment, an image coding apparatus that performs coding by motion search as in the third embodiment will be described.
FIG. 8 is a block diagram illustrating a configuration example of the image encoding device according to the present embodiment. Hereinafter, the operation of each block will be described in the configuration example of the image encoding device of the present embodiment. In the description of the present embodiment, the 4: 4: 4 image signal is described as an example using RGB as the color space, but the present invention is not limited to this and may be another color space.

  In FIG. 8, an image input unit 701, a motion compensation unit 703, a memory 704, a memory interface 705, an orthogonal transform unit 706, a quantization unit 707, an inverse quantization unit 708, an inverse orthogonal transform unit 709, and an entropy encoding unit 710 are: This corresponds to each part described in FIG.

  The difference between the third embodiment and the fourth embodiment lies in the plane selection information generation method. In the fourth embodiment, image signals in the RGB 4: 4: 4 format input from the outside to the image input unit 701 are stored as input image data in the input image storage area of the memory 704 via the memory interface 705. The It is also output to the image feature detection unit 711.

  The image feature detection unit 711 detects the image feature of each RGB plane of the input image data that has been input for each macroblock, and outputs the image feature information to the plane selection unit 712. The image feature is detected by detecting an activity, for example. The activity is determined by a variance indicating how scattered the pixel values in the block are from the block average value. In addition to this, detection of image features may be combined with edge detection, frequency spectrum, and the like.

The plane selection unit 712 compares the image feature information of each plane, selects a plane most suitable for motion search for each macroblock, and outputs the selected plane to the plane selection information holding unit 801 as plane selection information. If the image feature information is, for example, an activity, the plane with the highest activity value may be selected.
The present invention is not limited to the specific detection method of the image feature detection unit 711 and the specific selection method of the plane selection unit 712.

  The plane selection information holding unit 801 is composed of a memory capable of holding a plurality of pictures of plane selection information for each macroblock, and temporarily holds the plane selection information from the plane selection unit 712. The held plane selection information is output to the motion search unit 702 at a timing when the corresponding picture is processed in the motion search unit 702.

The motion search unit 702 performs prediction mode determination and motion search for each macroblock based on the plane selection information from the plane selection information holding unit 801. As input image data and reference image data for motion search, only image data of a plane indicated by plane selection information is read from the memory 704 via the memory interface 705 and used.
The prediction mode information and motion vector information obtained in this way are output to the motion compensation unit 703.

  As described above, in the fourth embodiment, the image feature of each plane is detected using the input image data from the image input unit 701, the plane most suitable for motion search is selected, and the plane selection is performed. The information is held in the plane selection information holding unit 801. In addition, the motion search unit 702 performs a motion search using only the image data of the selected plane.

  Thereby, it is possible to accurately perform the motion search of the 4: 4: 4 image signal without increasing the processing amount of the motion search. Also, only the selected plane is read out from the input image data and the reference image data read out from the memory 704 by the motion search unit 702. As a result, the amount of image data read from the memory is further suppressed than in the third embodiment.

(Other embodiments)
The present invention can also be realized by executing the following processing. That is, software (computer program) that implements the functions of the above-described embodiments is supplied to a system or apparatus via a network or various computer-readable storage media. Then, the computer (or CPU, MPU, etc.) of the system or apparatus reads out and executes the program.

101 Frame buffer 102 Feature detection unit 103 Plane selection unit 104 Intra prediction unit 105 Orthogonal transformation unit 106 Quantization unit 107 Quantization control unit 108 Entropy coding unit 109 Inverse quantization unit 110 Inverse orthogonal transformation unit 111 Motion compensation unit 112 Deblocking Filter unit 113 Reference frame buffer

Claims (12)

An image encoding device that encodes an image in 4: 4: 4 format,
Image feature detection means for detecting the image feature of each plane of the input image data for each block;
Plane selection means for selecting a plane suitable for intra prediction mode selection according to the image feature of each plane detected by the image feature detection means;
Intra prediction means for performing intra prediction,
The intra prediction unit selects an intra prediction mode common to each plane using the image data of the plane selected by the plane selection unit, and selects an intra prediction mode from the planes not selected by the plane selection unit. An image encoding device characterized by not .   The image coding apparatus according to claim 1, wherein the plane selection unit changes the ease of selecting each plane according to color space information of the coded image.   The image coding apparatus according to claim 2, wherein the plane selection unit selects a plane including a large amount of luminance signal components in each plane.   The image coding apparatus according to claim 1, wherein the image feature detected by the image feature detection unit is a variance value of an image.   The image coding apparatus according to claim 4, wherein the plane selection unit selects a plane having the highest image variance value. An image encoding device that encodes an image in 4: 4: 4 format,
Image feature detection means for detecting the image feature of each plane of the input image data for each block;
A plane selection unit that selects a plane suitable for motion search according to the image feature of each plane detected by the image feature detection unit, and outputs the plane selection information;
Input image memory means for holding input image data;
Reference image memory means for holding reference image data;
A motion search means for performing a motion search,
The motion search means outputs motion vector information common to each plane using the image data of the plane selected by the plane selection means, and outputs motion vector information from the plane not selected by the plane selection means. An image encoding device characterized by not .   The image code according to claim 6, further comprising plane selection information holding means for holding selection information for each block from the plane selection means for a plurality of pictures and outputting it in accordance with the operation of the motion search means. Device. An image encoding method of an image encoding device for encoding an image in 4: 4: 4 format,
An image feature detection step of detecting the image feature of each plane of the input image data for each block;
A plane selection step of selecting a plane suitable for intra prediction mode selection according to the image feature of each plane detected in the image feature detection step;
An intra prediction process for performing intra prediction,
In the intra prediction step, an intra prediction mode common to each plane is selected using the image data of the plane selected in the plane selection step, and an intra prediction mode is selected from the planes not selected in the plane selection step. An image encoding method characterized by not . An image encoding method of an image encoding device for encoding an image in 4: 4: 4 format,
An image feature detection step of detecting the image feature of each plane of the input image data for each block;
A plane selection step of selecting a plane suitable for motion search according to the image feature of each plane detected in the image feature detection step, and outputting as plane selection information;
An input image memory process for holding input image data;
A reference image memory step for holding reference image data;
A motion search step for performing motion search,
In the motion search step, motion vector information common to each plane is output using the image data of the plane selected in the plane selection step, and motion vector information is output from the plane not selected in the plane selection step. An image encoding method characterized by not . A program for causing a computer to execute an image encoding method for encoding an image in 4: 4: 4 format,
An image feature detection step of detecting the image feature of each plane of the input image data for each block;
A plane selection step of selecting a plane suitable for intra prediction mode selection according to the image feature of each plane detected in the image feature detection step;
The computer executes an intra prediction process for performing intra prediction,
In the intra prediction step, an intra prediction mode common to each plane is selected using the image data of the plane selected in the plane selection step, and an intra prediction mode is selected from the planes not selected in the plane selection step. A program characterized by not . A program for causing a computer to execute an image encoding method for encoding an image in 4: 4: 4 format,
An image feature detection step of detecting the image feature of each plane of the input image data for each block;
A plane selection step of selecting a plane suitable for motion search according to the image feature of each plane detected in the image feature detection step, and outputting as plane selection information;
An input image memory process for holding input image data;
A reference image memory step for holding reference image data;
Causing a computer to execute a motion search process for performing motion search,
In the motion search step, motion vector information common to each plane is output using the image data of the plane selected in the plane selection step, and motion vector information is output from the plane not selected in the plane selection step. A program characterized by not .   A computer-readable storage medium storing the program according to claim 10 or 11.

Images