JP2012191352A - Image encoding method, image encoder, image decoding method, image decoder and program thereof - Google Patents

Abstract

PROBLEM TO BE SOLVED: To realize intra-prediction encoding and decoding, in which an intra-prediction error is reduced by noticing correlation between the intra-prediction errors.SOLUTION: The correlation between predictive residuals of similar regions is used, prediction is performed between both the predictive residuals and residual prediction for reducing the predictive residual is introduced to an intra-prediction residual signal. A residual prediction processing section 110 selects an encoded region used for prediction processing to the intra-prediction residual signal based on a prediction direction of intra-prediction to an encoding object region, designates a specified component of a conversion coefficient in the encoded region as a reference signal, calculates a difference with a component of the corresponding conversion coefficient in the encoding object region to generate a residual prediction error signal and sets the residual prediction error signal as an object of encoding.

Description

  The present invention relates to a high-efficiency image signal encoding and decoding technique, and more particularly to an image encoding and image decoding technique for reducing a generated code amount in an intra prediction image encoding process.

  The moving image coding algorithms are roughly classified into interframe coding (inter coding) and intraframe coding (intra coding). Interframe coding is an approach for compressing information by utilizing the correlation in the time direction in a moving image. A typical example is inter-frame prediction using motion compensation. On the other hand, intra-frame coding is an approach for compressing information using intra-frame correlation. In JPEG and MPEG-2, an approach using discrete cosine transform (DCT) was used, and in JPEG 2000, an approach using discrete wavelet transform was taken. H. In H.264 / AVC, prediction in the spatial direction (see Non-Patent Document 1) is performed in addition to the transform coding described above.

  Generally, the coding efficiency of intra coding is lower than that of inter coding, and further improvement in coding efficiency is required. H. As an approach for improving the intra prediction of H.264 / AVC, a method that subdivides eight prediction directions and supports 33 prediction directions has been proposed. This method aims to reduce the prediction error due to the coarseness of the granularity in the prediction direction.

  FIG. 13 is a diagram illustrating a configuration example of a conventional general image encoding device. In the image encoding device in FIG. 13, only the portion particularly related to intra prediction is illustrated. This image encoding device includes an intra prediction processing unit 301 that generates an intra prediction signal from neighboring pixels of a block to be encoded, a transform processing unit 302 that performs orthogonal transform such as DCT, and a quantization coefficient by quantizing the transform coefficient A quantization processing unit 303 that performs inverse quantization on a quantization index, an inverse transform processing unit 305 that performs inverse orthogonal transform such as IDCT, a frame memory 306 that stores a decoded signal of a coded image, An intra prediction information encoding unit 307 that encodes intra prediction information such as a prediction mode, an entropy encoding processing unit 308 that outputs an encoded stream by entropy encoding quantization coefficients and intra prediction information, and calculates a prediction residual signal A subtractor 310 for generating a decoded signal, and an adder 311 for generating a decoded signal.

  FIG. 14 is a flowchart of a conventional image encoding process. Image encoding processing executed by the image encoding device shown in FIG. 13 will be described with reference to FIG.

  The image encoding apparatus repeats the following process for each encoding unit (sometimes referred to as a block) (step S601). First, in step S602, the intra prediction processing unit 301 determines an intra prediction mode based on the encoded adjacent pixels stored in the frame memory 306 and outputs an intra prediction signal for the block to be encoded. To do. The subtractor 310 obtains an intra prediction residual signal from the difference between the input signal of the original image and the intra prediction signal, and outputs it to the conversion processing unit 302.

  In step S603, the transform processing unit 302 performs transform processing of orthogonal transform such as DCT on the prediction residual signal to generate transform coefficients. In step S604, the quantization processing unit 303 performs a quantization process on the transform coefficient to generate a quantization index. In step S605, the entropy encoding processing unit 308 entropy encodes the quantization index.

  On the other hand, in step S606, the inverse quantization processing unit 304 performs an inverse quantization process on the quantization index to generate a decoded value of the transform coefficient. In step S607, the inverse transform processing unit 305 performs an inverse transform process such as IDCT on the decoded value of the transform coefficient, and generates a decoded signal of the intra prediction residual signal. Next, in step S608, the adder 311 generates a decoded signal by adding the intra prediction signal generated in the intra prediction in step S602 to the decoded value of the intra prediction residual signal, and generates a decoded signal. Store in the frame memory 306. The above process is repeated for all the coding unit units, and when the coding for all the coding unit units is completed, the coding process is finished (step S609).

  FIG. 15 is a diagram illustrating a configuration example of a conventional general image decoding apparatus. In the image decoding apparatus in FIG. 15, only the part particularly related to intra prediction is illustrated. This image decoding apparatus receives an encoded stream encoded by the image encoding apparatus shown in FIG. 13 and performs entropy decoding processing section 401 for entropy decoding, and inverse quantization processing section for performing inverse quantization processing on the quantization index. 402, an inverse transform processing unit 403 that performs inverse orthogonal transform such as IDCT, a frame memory 404 that stores a decoded signal of a decoded image, and intra prediction processing with reference to intra prediction information and the decoded image stored in the frame memory 404 The intra prediction processing unit 405 includes an intra prediction information storage unit 406 that stores intra prediction information such as a prediction mode.

  FIG. 16 is a flowchart of a conventional image decoding process. Image decoding processing executed by the image decoding apparatus shown in FIG. 15 will be described with reference to FIG.

  The image decoding apparatus repeats the following processing for each encoding unit (step S701). First, in step S702, the entropy decoding processing unit 401 performs entropy decoding on the encoded data in the encoded stream. The quantization index of the decoding result is output to the inverse quantization processing unit 402, and the index prediction information is stored in the intra prediction information storage unit 406.

  In step S703, the intra prediction processing unit 405 refers to the decoded decoded signal stored in the frame memory 404 based on the intra prediction information stored in the intra prediction information storage unit 406, and performs an intra prediction process. To generate an intra prediction signal.

  In step S704, the inverse quantization processing unit 402 performs an inverse quantization process on the quantization index to generate a decoded value of the transform coefficient. In step S705, the inverse transform processing unit 403 performs an inverse transform process such as IDCT on the decoded value of the transform coefficient to generate a decoded signal of an intra prediction residual signal.

  Next, in step S706, the adder 410 adds the intra prediction signal generated by the intra prediction in step S703 to the decoded value of the intra prediction residual signal, thereby generating a decoded signal. This decoded signal is stored in the frame memory 404 and is output to a playback device or the like. The above process is repeated for all the coding unit units, and when the decoding for all the coding unit units is completed, the decoding process is terminated (step S707).

Supervised by Satoshi Okubo, Impress Standard Textbook Series, revised third edition H.264 / AVC textbook, Impress R & D, Inc., pp. 110-116, published January 1, 2009

  Since all the conventional intra predictions are extrapolation predictions, the prediction error tends to increase as the distance between the reference pixel and the predicted pixel increases. For this reason, although there is a difference in the granularity of the prediction direction, the problem that such extrapolation prediction is mechanistically unsolved remains, and there is room for improvement in improving the coding efficiency.

  The present invention has been made in view of such circumstances, and aims to establish a highly efficient intra coding method by focusing on the correlation between intra prediction errors and reducing intra prediction errors. .

  When intra prediction in the same direction or a similar direction is applied to two regions in an image, it is expected that the intra prediction residuals are similar signals. That is, although a prediction residual is generated, there is a correlation between the generated prediction residuals.

  Therefore, in the present invention, residual prediction is introduced that uses the correlation between prediction residuals in similar regions, performs prediction between both prediction residuals, and reduces the prediction residual.

  That is, in order to solve the above-described problem, the present invention generates a prediction signal using spatial inter-pixel prediction, generates a residual signal between the prediction signal and the original signal, and generates a residual signal for the generated residual signal. In an image coding method for generating a transform coefficient by performing orthogonal transform, an encoded region is specified based on a prediction direction (also referred to as a prediction mode) of intra prediction, and a transform coefficient in the encoded region is further specified. A specific prediction component is designated as a reference signal, a residual prediction error signal is generated as a difference signal with a corresponding transform coefficient component in the encoding target region, and the residual prediction error signal is to be encoded. Is the main feature.

  Further, in the image decoding method for decoding the encoded stream generated by the above image encoding method, the residual prediction error signal in the encoded stream is decoded, the information indicating the prediction direction of the intra prediction is decoded, and Based on the prediction direction, a decoded region is selected, a specific component of the transform coefficient in the decoded region is specified as a reference signal, and the intra prediction residual of the encoding target region is used by using the reference signal. A main feature is that a prediction signal for a signal is generated, and a decoded signal of a decoding target region is generated using the decoded residual prediction error signal and the prediction signal.

  According to the present invention, the amount of codes can be reduced through reduction of intra prediction errors. This is based on the correlation between intra prediction residuals of two regions for two regions to which intra prediction in the same direction or similar direction is applied, and further performs prediction processing between the two intra prediction residuals. This is achieved by reducing the intra prediction residual.

It is a figure which shows the example of to-be-predicted PU and reference PU. It is a figure which shows the structural example of the image coding apparatus which is one Embodiment of this invention. It is a flowchart of an image encoding process. It is a flowchart of the decoding value generation process of the transform coefficient using a residual prediction process. It is a flowchart of a residual prediction process (example 1). It is a flowchart of a residual prediction process (example 2). It is a flowchart of a residual prediction process (example 3). It is a figure which shows the structural example of the image decoding apparatus which is one Embodiment of this invention. It is a flowchart of an image decoding process. It is a flowchart of the decoding value generation process of the transform coefficient using a residual prediction process. It is a figure which shows the hardware structural example in the case of implement | achieving an image coding apparatus using a software program. FIG. 25 is a diagram illustrating a hardware configuration example when an image decoding device is realized using a software program. It is a figure which shows the structural example of the conventional general image coding apparatus. It is a flowchart of the conventional image encoding process. It is a figure which shows the structural example of the conventional common image decoding apparatus. It is a flowchart of the conventional image decoding process.

  Prior to describing specific embodiments of the present invention, first, terms used in this specification will be defined. As described above, the present invention introduces residual prediction that uses the correlation between prediction residuals in similar regions, performs prediction between both prediction residuals, and reduces the prediction residual. Prediction between the prediction residuals is defined as intra prediction residual prediction (IPRP). A rectangular area where intra prediction is performed is a prediction unit (abbreviated as PU), and a rectangular area where orthogonal transformation such as DCT is performed is a transform unit (abbreviated as TU). A TU to be predicted in IPRP is a predicted TU, and a TU that is a reference destination in IPRP is a reference TU. Also, a PU including a predicted TU is a predicted PU, and a PU including a reference TU is a reference PU. The sizes of PU and TU are variable and are defined by the horizontal width and vertical width.

  The IPRP technique includes a reference PU used for residual prediction, a reference TU, a predicted PU, a predicted TU setting, a prediction method setting for an intra prediction residual signal, and an IPRP designation method. Hereinafter, specific processing of IPRP will be described.

  FIG. 1 is a diagram illustrating an example of a predicted PU and a reference PU. FIG. 1A shows an example where the PU size matches the TU size, and FIG. 1B shows the case where the PU size is larger than the TU size in the PU. That is, an example in which the PU includes a plurality of TUs is shown.

[Reference PU and reference TU setting method (when the PU size matches the TU size)]
A case where the size of the PU and the size of the TU in the PU are the same, that is, the case where the PU and the TU correspond one-to-one will be described. In this case, since the sizes of PU and TU are the same, hereinafter, the reference PU and reference TU are collectively referred to as a reference block. Further, the predicted PU and the predicted TU are collectively referred to as a predicted block. Based on the block size and intra prediction direction of the predicted PU, PUs and TUs to be referred to in IPRP are set. As the reference block setting method, for example, the following setting method 1 and setting method 2 can be used.

<Setting method 1>
Neighboring PUs having the same size and the same prediction direction are identified for the PU including the predicted TU, and the TU in the identified PU is set as a reference TU for residual prediction. In the example of FIG. 1A, since PU3 is in the vicinity of PU6 with respect to the predicted PU6 and has the same prediction direction (index = 1), it is selected as a reference PU (reference TU).

  As the neighborhood PU, the nearest neighborhood PU having the shortest distance between the PUs is set. Alternatively, the reference PU can be limited to PUs within a certain range. For example, the range can be specified using the distance from the block to be predicted. When the narrowest range is designated, the reference block is limited to a block adjacent to the predicted block. When the range is limited in this way, residual prediction is not applied unless there is a corresponding block within the range. The inapplicability of residual prediction can be shared between the encoding side and the decoding side without any special additional information. For this reason, selection information indicating application / non-application of residual prediction can be omitted for blocks to which residual prediction is not applied.

<Setting method 2>
Neighboring blocks having the same size and similar prediction direction with respect to the block to be predicted are set as reference blocks for residual prediction. The neighboring blocks are the same as described above. In addition, which prediction direction is considered to be similar is determined separately. For example, H. In the case of H.264, for example, as shown on page 112 of Non-Patent Document 1, average value prediction and eight kinds of direction prediction are prepared. There is also a method that subdivides the prediction direction. For example, together with the direction prediction of the prediction direction expanded to 33 ways, As in the case of H.264, average value prediction can also be used. With respect to the prediction direction in the block to be predicted, 2N + 1 prediction directions composed of 2N prediction directions and the same prediction direction with the smallest deviation of the prediction direction are regarded as similar directions. The parameter N that defines the number of similar directions is given from the outside. Note that when there are no designated number of prediction directions, the number of similar directions is less than 2N + 1. Alternatively, a specific prediction direction is determined as a representative direction, and the other prediction directions are aggregated in a representative direction that minimizes the angle divergence of the prediction directions.

[Reference PU and reference TU setting method (when PU includes multiple TUs)]
An example in which the PU includes a plurality of TUs, that is, a case where the size of the PU is larger than the size of the TU in the PU will be described. Information specifying the position of the TU in the PU is referred to as a PU position. As a method for setting the reference PU and the reference TU, for example, the following setting method 1 and setting method 2 can be used.

<Setting method 1>
First, neighboring PUs having the same size and the same prediction direction are identified for the PU including the predicted TU. Next, the position in the PU of the predicted TU in the predicted PU is identified, and in the reference PU, the TU located at the position in the PU is identified as a reference TU for residual prediction.

  In the example of FIG. 1B, four TUs are included in the predicted PU 6, and the neighboring PU 3 in the same prediction direction is set as the reference PU for the PU 6. Then, the TU 32 in the reference PU 3 having the same PU position as the predicted TU 62 in the predicted PU 6 is used as the reference TU.

  As the neighborhood PU, the nearest neighborhood PU having the shortest distance between the PUs is set. Alternatively, the reference PU can be limited to PUs within a certain range. For example, the range can be specified using the distance from the predicted PU. When the narrowest range is specified, it is a case where the reference PU is limited to a PU adjacent to the predicted PU. When the range is limited in this way, residual prediction is not applied if there is no corresponding PU in the range. The inapplicability of residual prediction can be shared between the encoding side and the decoding side without any special additional information. For this reason, selection information indicating application / non-application of residual prediction can be omitted for PUs to which residual prediction is not applied. When selection information indicating application / non-application of residual prediction is set for each TU, selection indicating application / non-application of residual prediction for all TUs in the PU to which residual prediction is not applied. Information can be omitted.

<Setting method 2>
Neighboring PUs of the same size and similar prediction direction with respect to the predicted PU are used as reference PUs for residual prediction. The neighborhood PU is the same as described above. In addition, which prediction direction is considered to be similar is determined separately. For example, H. In the case of H.264, for example, as shown on page 112 of Non-Patent Document 1, average value prediction and eight kinds of direction prediction are prepared. There is also a method that subdivides the prediction direction. For example, together with the direction prediction of the prediction direction expanded to 33 ways, As in the case of H.264, average value prediction can also be used. With respect to the prediction direction in the predicted PU, 2N + 1 prediction directions composed of 2N prediction directions with the smallest deviation in the prediction direction and the same prediction direction are regarded as similar directions. The parameter N that defines the number of similar directions is given from the outside. Note that when there are no designated number of prediction directions, the number of similar directions is less than 2N + 1. Alternatively, a specific prediction direction is determined as a representative direction, and the other prediction directions are aggregated in a representative direction that minimizes the angle divergence of the prediction directions.

[Intra prediction residual signal prediction method]
As a prediction method of an intra prediction residual signal, for example, any of the following four prediction methods 1 to 4 can be used.

<Prediction method 1 (adaptive transform coefficient prediction)>
In the prediction method 1, the decoded value of the transform coefficient obtained by performing transform coding, quantization, and inverse quantization on the intra prediction residual of each TU (reference TU) in the reference PU is used in the predicted TU. The prediction signal of the corresponding component is regarded, and only the conversion coefficient of the specific component is limited, and the difference between the two coefficients is calculated. At this time, the component of the transform coefficient to be predicted is called a prediction target component. It is assumed that which component is the prediction target component is switched according to the intra prediction direction. Information indicating the position of the prediction target component is stored in a table for each intra prediction direction, and is read as appropriate. Which TU in the reference PU is used as the reference TU is set to a TU having a minimum distance from the predicted TU.

<Prediction method 2 (weighted adaptive transform coefficient prediction)>
In the prediction method 2, the decoding value of the transform coefficient obtained by performing transform coding, quantization, and inverse quantization on the intra prediction residual of each TU (reference TU) in the reference PU is referred to and decoded. A value obtained by multiplying the value by a weighting coefficient is regarded as a prediction signal of a corresponding component in the predicted TU, and is limited to only a conversion coefficient of a specific component, and a difference between both coefficients is calculated. It is assumed that which component is the prediction target component is switched according to the intra prediction direction. Information indicating the position of the prediction target component is stored in a table for each intra prediction direction, and is read as appropriate. The weighting factor shall be given separately. Which TU in the reference PU is used as the reference TU is set to a TU having a minimum distance from the predicted TU.

<Prediction method 3 (weighted adaptive transform coefficient prediction)>
In the prediction method 3, the decoding value of the transform coefficient obtained by performing transform coding, quantization, and inverse quantization on the intra prediction residual of each TU (reference TU) in the reference PU is referred to and decoded. The value obtained by multiplying the value by the weighting coefficient and adding the bias term coefficient is regarded as the prediction signal of the corresponding component in the predicted TU, and is limited to only the conversion coefficient of the specific component, and the difference between the two coefficients is calculated. To do. It is assumed that which component is the prediction target component is switched according to the intra prediction direction. Information indicating the position of the prediction target component is stored in a table for each intra prediction direction, and is read as appropriate. The weighting factor and bias term shall be given separately. Which TU in the reference PU is used as the reference TU is set to a TU having a minimum distance from the predicted TU.

<Prediction method 4 (weighted adaptive transform coefficient prediction)>
In the prediction method 4, a decoding value of a transform coefficient obtained by performing transform coding, quantization, and inverse quantization on the intra prediction residual of each TU (reference TU) in the reference PU is referred to, Only for the prediction target component, a value obtained by multiplying the corresponding decoded value by the weighting coefficient and adding the coefficient of the bias term is regarded as a prediction signal of the corresponding component in the predicted TU, and for other prediction target components Considers the corresponding decoded value as the prediction signal of the corresponding component in the predicted TU, and calculates the difference between the two coefficients for the transform coefficient of the prediction target component. It is assumed that which component is the prediction target component is switched according to the intra prediction direction. Information indicating the position of the prediction target component is stored in a table for each intra prediction direction, and is read as appropriate. The weighting factor and bias term shall be given separately. Which TU in the reference PU is used as the reference TU is set to a TU having a minimum distance from the predicted TU.

  In the prediction method described above, the prediction target can be transmitted as the absolute value of the prediction residual and the prediction residual code information can be transmitted without being the target of the prediction process.

[How to specify residual prediction]
Whether the residual prediction is applied or not is performed in a predetermined unit. Applicability is transmitted using a 1-bit flag (residual prediction flag). The unit to be switched can be a sequence, a frame, or an area within the frame, and this switching unit is determined separately.

  It is also possible to calculate the correlation between the intra prediction reference pixel group in the reference TU and the intra prediction reference pixel group in the predicted TU and to limit the correlation value to a certain threshold value or more. In this case, when the correlation value is less than the threshold value, it is not necessary to transmit the residual prediction flag.

  Alternatively, if the decoded value of the transform coefficient to be referenced is equal to or less than a given threshold value, the residual prediction is not applied, and this is shared as a rule between the encoder / decoder. This eliminates the need to transmit the residual prediction flag when the decoded value of the reference transform coefficient is equal to or less than a given threshold value. As a result, it is possible to omit a code amount necessary for expressing the residual prediction flag. A zero value can be given as an example of the threshold value. Setting a zero value as the threshold means that the residual prediction is omitted when the decoded value of the reference transform coefficient is zero. When the reference signal is zero, the predicted coefficient does not change before and after application of residual prediction. For this reason, it is possible to reduce the amount of code necessary for expressing the residual prediction flag without degrading the encoding performance.

[Configuration Example of Image Encoding Device]
FIG. 2 is a diagram illustrating a configuration example of an image encoding device according to an embodiment of the present invention. The image encoding apparatus in FIG. 2 may include a processing unit that performs inter prediction encoding. However, only a portion related to intra prediction will be illustrated and described here.

  This image encoding device includes an intra prediction processing unit 101 that generates an intra prediction signal from neighboring pixels of a block to be encoded, a transform processing unit 102 that performs orthogonal transform such as DCT, and a quantization coefficient by quantizing the transform coefficient. A quantization processing unit 103 that performs inverse quantization on a quantization index, an inverse transform processing unit 105 that performs inverse orthogonal transform such as IDCT, a frame memory 106 that stores a decoded signal of an encoded image, Intra prediction information encoding unit 107 that encodes intra prediction information such as a prediction method (prediction mode), an entropy encoding processing unit 108 that entropy encodes a quantization index or intra prediction information, and intra prediction that stores intra prediction information An information storage unit 109, a residual prediction processing unit 110 for performing residual prediction, and a prediction residual coefficient are stored. Measurement residual coefficient storage unit 111, transform coefficient storage unit 112 that stores transform coefficients, residual prediction determination unit 113 that determines whether or not to apply residual prediction, and residual prediction selection that outputs a residual prediction selection flag A flag output unit 114; a multiplexing processing unit 115 that multiplexes and outputs an entropy encoding result and a residual prediction selection flag; a subtractor 120 that calculates a prediction residual signal; and an intra prediction residual in a decoded value of the residual prediction error signal. An adder 121 that adds a prediction signal of the difference signal, an adder 122 that generates a decoded signal, and switch units 123 and 124 are provided.

  The intra prediction processing unit 101 receives a decoded signal in the neighborhood area stored in the frame memory 106, performs a spatial prediction process, and outputs a signal obtained by the prediction as an intra prediction signal. The subtractor 120 calculates a difference signal between the input signal and the intra prediction signal output from the intra prediction processing unit 101, and outputs the difference signal as an intra prediction residual signal.

  The conversion processing unit 102 receives the intra prediction residual signal, performs a conversion process such as DCT, and outputs a conversion coefficient. The quantization processing unit 103 receives the transform coefficient output from the transform processing unit 102, performs quantization processing on the transform coefficient, and outputs a quantization index representing a quantized representative value.

  The inverse quantization processing unit 104 receives the quantization index output from the quantization processing unit 103, performs an inverse quantization process, and outputs a decoded value of the transform coefficient. The inverse transform processing unit 105 receives the decoded value of the transform coefficient output from the inverse quantization processing unit 104, performs an inverse transform process, and outputs a decoded value of the intra prediction residual signal.

  The adder 122 generates a decoded signal by adding the intra prediction signal to the decoded value of the intra prediction residual signal. This decoded signal is stored in the frame memory 106.

  The intra prediction information encoding unit 107 encodes, as information for expressing an intra prediction signal, an intra prediction mode that defines the direction of intra prediction and a PU size that indicates the size of a region where intra prediction is performed. The entropy encoding processing unit 108 receives the quantization index output from the quantization processing unit 103 and the intra prediction information output from the intra prediction information encoding unit 107 as input, performs entropy encoding processing, and performs a multiplexing processing unit. The encoded data is multiplexed via 115 and an encoded stream is output.

  The processing of each unit described above is basically the same as the processing of the conventional image encoding device described with reference to FIG. Further, this image encoding device includes the following processing unit and storage unit for image encoding using residual prediction.

  The intra prediction information storage unit 109 stores prediction mode information for intra prediction and a block size for the prediction. The residual prediction processing unit 110 receives the prediction mode information of intra prediction stored in the intra prediction information storage unit 109 and the block size of the prediction, identifies the reference PU, and further converts the conversion coefficient for the intra prediction residual signal. The decoded coefficient stored in the prediction residual coefficient storage unit 111 is input, and a difference value (residual prediction error signal) between the two is output.

  The prediction residual coefficient storage unit 111 stores the decoded value of the transform coefficient for the intra prediction residual signal. Also, the transform coefficient storage unit 112 stores transform coefficients for the intra prediction residual signal.

  The residual prediction determination unit 113 receives the transform coefficient for the intra prediction residual of the encoding target TU and the transform coefficient of the TU stored in the transform coefficient storage unit 112, and determines whether to perform residual prediction. . First, it is determined whether or not an application condition for residual prediction is satisfied. As an application condition, for example, a correlation between a reference pixel group for intra prediction in a reference block and a reference pixel group for intra prediction in a predicted block is calculated, and the correlation value is equal to or greater than a certain threshold. In addition, there is a condition that an appropriate reference PU whose prediction direction matches or is similar to that of the predicted PU exists within a predetermined neighborhood range, but other application conditions may be used.

  If the application condition is not satisfied, the input to the residual prediction selection flag output unit 114 is skipped, the switch units 123 and 124 are switched, and the addition process by the adder 121 is omitted. When the application condition is satisfied, whether or not the residual prediction is applied is determined based on a predetermined evaluation scale such as an RD cost, and when it is determined that the residual prediction is not applied, the residual prediction is performed. In the selection flag output unit 114, the residual prediction selection flag is output as OFF, and the addition processing after the switch units 123 and 124 is omitted.

  On the other hand, if the residual prediction determination unit 113 determines that the residual prediction is applied, the residual prediction selection flag output unit 114 outputs the residual prediction selection flag as ON, and the switch unit 123 Switch to the residual prediction processing unit 110 side, and switch the switch unit 124 so that the adder 121 performs addition processing.

  The residual prediction selection flag output unit 114 outputs the residual prediction selection flag as ON when the residual prediction determination unit 113 determines that the residual prediction is to be performed, and otherwise outputs the residual prediction selection flag. The flag is output as OFF. The multiplexing processing unit 115 multiplexes the residual prediction selection flag and the other encoded streams as one encoded stream and outputs it.

[Image coding processing]
Next, the flow of the image encoding process executed by the image encoding device shown in FIG. 2 will be described according to the flowcharts shown in FIGS.

  FIG. 3 is a flowchart of image encoding processing using residual prediction according to the present embodiment. An input signal of the image encoding device is an image signal to be encoded. In FIG. 3, the processing parts of steps S104 to S111 surrounded by a dotted frame are mainly different from the prior art. When an image signal is input, the image encoding device performs the following processing.

  [Step S101]: The processing up to Step S114 is repeatedly executed for each encoding unit.

  [Step S102]: The intra prediction processing unit 101 sets an intra encoding method (intra prediction mode), and generates an intra prediction signal based on the set intra prediction mode. Information related to the intra coding method is stored in the intra prediction information storage unit 109 and also output to the intra prediction information coding unit 107. The intra prediction signal is sent to the subtracter 120, and the subtractor 120 generates a difference signal (intra prediction residual signal) between the intra prediction signal and the input signal.

  [Step S103]: The transform processing unit 102 receives the intra prediction residual signal as input and performs orthogonal transform to generate transform coefficients. The conversion process is determined in advance, and for example, discrete cosine transform (DCT) is used. The transform coefficient is output to the switch unit 123 and the residual prediction determination unit 113 and is also stored in the transform coefficient storage unit 112.

  [Step S104]: The residual prediction determination unit 113 determines whether to perform residual prediction based on the application conditions of the residual prediction. If the application condition is satisfied, the process proceeds to steps S105 and S107. If the application condition is not satisfied, the process proceeds to step S110.

  [Step S105]: The residual prediction processing unit 110 performs prediction processing on the intra prediction residual signal, generates a prediction signal of the intra prediction residual signal, and a difference signal between the prediction signal and the intra prediction residual signal ( A residual prediction error signal). A decoded coefficient transform value is generated when the residual prediction error signal is quantized and encoded. Details of this processing will be described with reference to FIG.

  [Step S106]: The coding distortion is calculated for the decoded value of the transform coefficient generated in Step S105, and the code amount (or the estimated value) necessary for expressing the decoded value is calculated. An RD cost is calculated as a weighted sum of the coding distortion and the code amount.

  [Step S107] On the other hand, a decoded value of the transform coefficient when the transform coefficient is encoded without performing the residual prediction process is generated by the same process as in Steps S604 to S606 according to the conventional method described in FIG. That is, quantization processing is performed with the transform coefficient as an input, and the quantized transform coefficient is generated as a quantization index. In addition, entropy encoding processing is performed using the quantization index as an input, and converted to a binary sequence. The entropy coding method shall be given separately. Further, inverse quantization processing is performed with the quantization index as an input, and a decoded value of the transform coefficient is generated.

  [Step S108]: The encoding distortion is calculated for the decoded value of the transform coefficient generated in step S107, and the code amount (or the estimated value) necessary for expressing the decoded value is calculated. An RD cost is calculated as a weighted sum of the coding distortion and the code amount.

  [Step S109]: The RD cost output in step S106 is compared with the RD cost output in step S108. If the former is smaller, the process proceeds to step S111. Otherwise, the process proceeds to step S110.

  [Step S110]: The residual prediction selection flag output unit 114 outputs a residual prediction selection flag indicating the use of the intra prediction residual as OFF (binary 0), and the quantization processing unit 103 and the entropy encoding process Coding information for the quantization index of the transform coefficient for the intra prediction residual signal encoded by unit 108 is output. Thereafter, the process proceeds to step S112.

  [Step S111]: The residual prediction selection flag output unit 114 outputs a flag indicating the use of the intra prediction residual as ON (binary number 1), and the quantization processing unit 103 and the entropy encoding processing unit 108 Encoding information for the quantization index of the transform coefficient for the converted residual prediction error signal is output.

  [Step S112]: The inverse transform processing unit 105 receives the decoded value of the transform coefficient as input and performs an inverse transform process to generate a decoded signal of the intra prediction residual signal.

  [Step S113]: The adder 122 receives the decoded value of the intra prediction residual signal and the intra prediction signal, adds them together, generates a decoded signal, and stores it in a predetermined storage area (frame memory 106).

  [Step S114]: The above processing is repeated for all the coding unit units, and when the processing for all the coding unit units is completed, the coding processing ends.

  FIG. 4 is a flowchart of the transform coefficient decoding value generation process using the residual prediction process. Next, the processing in step S105 in FIG. 3 will be described in detail with reference to FIG.

  [Step S201]: The residual prediction processing unit 110 receives the prediction target component of the reference TU, performs a prediction process on the intra prediction residual signal, and generates a prediction signal of the intra prediction residual signal. Further, the prediction target component of the TU to be predicted and the prediction signal are input, and a residual prediction error signal is generated as a difference signal between them. Details of the residual prediction process will be described with reference to three examples with reference to FIGS. 5, 6, and 7.

  [Step S202]: The quantization processing unit 103 performs a quantization process using the residual prediction error signal as an input, and outputs it as a quantization index obtained by the quantization.

  [Step S203]: The entropy encoding processing unit 108 receives the quantization index as input and performs entropy encoding processing to convert it into a binary sequence. The entropy coding method shall be given separately.

  [Step S204]: The inverse quantization processing unit 104 receives the quantization index and performs inverse quantization processing to generate a decoded value of the residual prediction error signal.

  [Step S205]: The adder 121 inputs the decoded value of the residual prediction error signal generated in step S204 and the prediction signal of the intra prediction residual signal generated in step S201, adds them, and decodes the transform coefficient Generate a value.

  FIG. 5 is a flowchart showing the residual prediction process (example 1). In step S201 of FIG. 4, the process shown in FIG. 5 is executed as a first example of the residual prediction process.

  [Step S301]: The residual prediction processing unit 110 reads the size of the predicted PU, the size of the predicted TU, and the prediction mode in the predicted PU from the intra prediction information storage unit 109.

  [Step S302]: The information read in Step S301 is used as an input to identify the reference PU and reference TU. The specific setting method is the above-described [reference block setting method (when the PU size matches the TU size)] and [reference block setting method (when the PU includes a plurality of TUs)]. Use.

  [Step S303]: Using the information read in Step S310 as an input, the prediction target component of the transform coefficient is identified. The specific identification method shall be given separately.

  [Step S304]: The prediction target component in the reference TU is input, and a prediction signal for residual prediction is generated. A specific setting method follows <Prediction method 1> in [Intra prediction residual signal prediction method] described above.

  [Step S305]: The prediction signal generated in step S304 and the prediction target component in the predicted TU are input, and the difference between the two is output.

  FIG. 6 is a flowchart showing a residual prediction process (example 2). In step S201 in FIG. 4, the process shown in FIG. 6 may be executed as a second example of the residual prediction process.

  [Steps S311 to S313]: The same processes as steps S301 to S303 described in FIG. 5 are executed.

  [Step S314]: The residual prediction processing unit 110 reads the weighting coefficient to be multiplied by the prediction target component in the reference TU and the value of the bias coefficient to be added to the prediction target component. The weighting coefficient and the bias coefficient are given in advance or separately determined.

  [Step S315]: The prediction target component in the reference TU is input and set as a prediction signal for residual prediction. The specific setting method follows <Prediction method 2> in [Intra prediction residual signal prediction method].

  [Step S316]: The prediction signal generated in step S315 and the prediction target component in the predicted TU are input, and the difference between the two is output.

  FIG. 7 is a flowchart showing a residual prediction process (example 3). In step S201 in FIG. 4, the process shown in FIG. 7 may be executed as a third example of the residual prediction process.

  [Steps S321 to S323]: The same processing as steps S301 to S303 described in FIG. 5 is executed.

  [Step S324]: The residual prediction processing unit 110 reads the weighting coefficient to be multiplied by the prediction target component in the reference TU and the value of the bias coefficient to be added to the prediction target component. The weighting coefficient and the bias coefficient are given in advance or separately determined.

  In subsequent steps S325 to S331, the residual prediction processing unit 110 receives the prediction target component in the reference TU and generates a prediction signal for residual prediction. A specific setting method follows <Prediction method 4> in [Intra prediction residual signal prediction method]. Note that the processing in the case of <prediction method 3> is almost the same, and therefore the description of the processing method when <prediction method 3> is used will be omitted.

  [Step S325]: The following processing is repeated for all prediction target components of the transform coefficient.

  [Step S326]: It is determined whether or not the prediction target component is a component for performing weighted prediction processing. In the case of the component for which the weighted prediction process is performed, the process proceeds to step S328. Otherwise, the process proceeds to step S327.

  [Step S327]: The prediction target component in the reference TU is set as a prediction signal for residual prediction. Thereafter, the process proceeds to step S330.

  [Step S328]: A weighting coefficient to be multiplied by the prediction target component in the reference TU and a bias coefficient to be added to the prediction target component are read.

  [Step S329]: A value obtained by multiplying the prediction target component in the predicted TU by the weighting coefficient and adding the bias coefficient is set as a prediction signal for residual prediction.

  [Step S330]: The prediction signal set in Step S327 or Step S329 and the prediction target component in the predicted TU are input, and the difference between the two is calculated.

  [Step S331]: The processing from step S325 onward is repeated for all the prediction target components, and when the processing for all the prediction target components is completed, the processing ends.

[Configuration Example of Image Decoding Device]
FIG. 8 is a diagram illustrating a configuration example of an image decoding device according to an embodiment of the present invention. The image decoding apparatus in FIG. 8 may include a processing unit that performs inter prediction decoding. Here, only a portion related to intra prediction will be illustrated and described.

  This image decoding apparatus includes an entropy decoding processing unit 201 that entropy decodes a quantization index and intra prediction information, an inverse quantization processing unit 202 that inversely quantizes the quantization index, and an inverse transformation processing unit that performs inverse orthogonal transformation such as IDCT. 203, a frame memory 204 that stores a decoded signal of a decoded image, an intra prediction processing unit 205 that generates an intra prediction signal from neighboring pixels of the decoding target block, and an intra prediction information storage unit 206 that stores intra prediction information that has been entropy-decoded. , A separation processing unit 207 that separates a residual prediction selection flag from the encoded stream, a residual prediction processing unit 208 that performs residual prediction, a prediction residual coefficient storage unit 209 that stores prediction residual coefficients, and a residual prediction selection flag Such as residual prediction selection flag reading unit 210 for reading and prediction method (prediction mode) An intra prediction information storage unit 211 that stores prediction information, a transform coefficient storage unit 212 that stores transform coefficients, a residual prediction determination unit 213 that determines whether to apply residual prediction, or a decoded value of a residual prediction error signal Are provided with an adder 220 for adding a prediction signal of an intra prediction residual signal, an adder 221 for generating a decoded signal, and switch units 222 and 223.

  The entropy decoding processing unit 201 receives the encoded stream separated by the separation processing unit 207, performs entropy decoding processing, and outputs a quantization index and intra prediction information. The inverse quantization processing unit 202 receives the quantization index output from the entropy decoding processing unit 201, performs an inverse quantization process, and outputs a decoded value of the transform coefficient. The inverse transform processing unit 203 receives the decoded value of the transform coefficient output from the inverse quantization processing unit 202, performs an inverse transform process, and outputs a decoded value of the intra prediction residual signal. The adder 221 adds the intra prediction signal to the decoded value of the intra prediction residual signal to generate a decoded signal. This decoded signal is stored in the frame memory 204.

  The intra prediction processing unit 205 receives a decoded signal in the neighborhood area stored in the frame memory 204, performs a spatial prediction process, and outputs a signal obtained by the prediction process as an intra prediction signal. The intra-prediction information storage unit 206 is information for expressing the intra-prediction signal output from the entropy decoding processing unit 201 (intra-prediction mode that defines the direction of intra-prediction, PU size that represents the size of the region where intra-prediction is performed) ).

  The processing of each unit described above is basically the same as the processing of the conventional image decoding apparatus described with reference to FIG. Further, the image decoding apparatus includes the following processing unit and storage unit for image decoding using residual prediction.

  The separation processing unit 207 receives the encoded stream output from the image encoding device described in FIG. 2 as an input, and separates it into a part including a residual prediction selection flag and another part. The residual prediction processing unit 208 receives the transform coefficient for the intra prediction residual signal and the decoded value of the transform coefficient stored in the prediction residual coefficient storage unit 209, and outputs the difference value (residual prediction error signal) between the two. To do.

  The residual prediction selection flag reading unit 210 receives the encoded stream output from the separation processing unit 207 as an input, and reads the residual prediction selection flag from the portion including the residual prediction selection flag. Further, when the residual prediction selection flag is ON, the residual prediction selection flag reading unit 210 performs processing by the residual prediction processing unit 208, and outputs the inverse quantization processing unit 202 and the residual prediction processing unit 208. Is output to the switch unit 222 by the adder 220. On the other hand, when the residual prediction selection flag is OFF, control indicating that the output of the inverse quantization processing unit 202 is output as it is without adding the output of the inverse quantization processing unit 202 and the output of the residual prediction processing unit 208. Output a signal.

  The intra prediction information storage unit 211 stores prediction mode information for intra prediction and a block size for intra prediction. The transform coefficient storage unit 212 stores a decoded value of the transform coefficient for the intra prediction residual signal.

  The residual prediction determination unit 213 determines whether or not to perform residual prediction by using the transform coefficient for the intra prediction residual of the decoding target TU and the transform coefficient of the TU stored in the transform coefficient storage unit 212 as inputs. First, it is determined whether or not an application condition for residual prediction is satisfied. If the application condition is not satisfied, control indicating that the processing of the residual prediction selection flag reading unit 210 is skipped for the switch unit 223 is performed. Output a signal. Further, a control signal indicating that the output of the inverse quantization processing unit 202 is output as it is without adding the output of the inverse quantization processing unit 202 and the output of the residual prediction processing unit 208 is output to the switch unit 222. To do.

  When the application condition is satisfied, the residual prediction selection flag reading unit 210 reads the residual prediction selection flag, and when the residual prediction selection flag is OFF, the output of the inverse quantization processing unit 202 and the residual prediction A control signal indicating that the output of the inverse quantization processing unit 202 is output as it is without adding the output of the processing unit 208 is output to the switch unit 222. On the other hand, when the residual prediction selection flag is ON, a control signal indicating that the output of the inverse quantization processing unit 202 and the output of the residual prediction processing unit 208 are output by the adder 220 is output to the switch unit. It outputs to 222.

[Image decoding processing]
Next, the flow of the image decoding process executed by the image decoding apparatus shown in FIG. 8 will be described with reference to the flowcharts shown in FIGS.

  FIG. 9 is a flowchart of image decoding processing using residual prediction according to the present embodiment. The input signal of the image decoding apparatus is encoded data of the encoded stream output from the image encoding apparatus shown in FIG. In FIG. 9, the processing part of steps S404 to S408 surrounded by a dotted frame is mainly a part different from the prior art. When the encoded data is input, the image decoding apparatus performs the following processing.

  [Step S401]: The encoded data separated from the encoded stream is input by the separation processing unit 207, and the processing up to step S411 is repeatedly executed for each encoding unit.

  [Step S402]: The entropy decoding processing unit 201 receives the encoded data of the encoding unit unit as input and performs entropy decoding processing on the quantization index and the intra prediction information.

  [Step S403]: The intra prediction processing unit 205 reads an intra coding method (intra prediction mode) from the decoded intra prediction information, and generates an intra prediction signal based on the intra prediction mode.

  [Step S404]: The residual prediction determination unit 213 determines whether or not to perform residual prediction based on the application conditions of residual prediction. If the application condition is satisfied, the process proceeds to step S406. If the application condition is not satisfied, the process proceeds to step S405.

  [Step S405]: The inverse quantization processing unit 202 performs an inverse quantization process on the quantization index, and generates a decoded value of the transform coefficient. Thereafter, the process proceeds to step S409.

  [Step S406]: The residual prediction selection flag reading unit 210 reads a flag (residual prediction selection flag) indicating the use of the intra prediction residual from the encoded data separated by the separation processing unit 207.

  [Step S407]: If the residual prediction selection flag is ON, the process proceeds to step S408. Otherwise, the process proceeds to step S405.

  [Step S408]: Generate a decoded value of the transform coefficient based on the residual prediction. Details of this processing will be described later with reference to FIG.

  [Step S409]: The inverse transformation processing unit 203 receives the decoded value of the transformation coefficient as an input, performs an inverse transformation process, and generates a decoded signal of the intra prediction residual signal.

  [Step S410]: The adder 221 receives the decoded value of the intra prediction residual signal and the intra prediction signal, adds them together, generates a decoded signal, and stores it in a predetermined storage area (frame memory 204).

  [Step S411]: The above processing is repeated for all the coding unit units, and when the processing for all the coding unit units is completed, the decoding processing is terminated.

  FIG. 10 is a flowchart of a transform coefficient decoded value generation process using a residual prediction process at the time of decoding. Next, the process of step S408 in FIG. 9 will be described in detail with reference to FIG.

  [Step S501]: The residual prediction processing unit 208 receives the prediction target component of the reference TU, performs prediction processing on the intra prediction residual signal, and generates a prediction signal of the intra prediction residual signal. Further, the prediction target component of the TU to be predicted and the prediction signal are input, and a residual prediction error signal is generated as a difference signal between them. This residual prediction process is the same as the process of step S201 (FIG. 4) described in the image encoding process, and more specifically, some processes such as those described with reference to FIGS. There are variations.

  [Step S502]: The inverse quantization processing unit 202 performs an inverse quantization process using the quantization index as an input, and generates a decoded value of the residual prediction error signal.

  [Step S503]: The adder 221 inputs the decoded value of the residual prediction error signal generated in step S502 and the prediction signal of the intra prediction residual signal generated in step S501, adds them, and decodes the transform coefficient A value is generated and stored in a predetermined storage area (frame memory 204).

  FIG. 11 shows a hardware configuration example in the case where the image encoding device of FIG. 2 is configured by a computer and a software program. This system includes a CPU 150 that executes a program, a memory 151 such as a RAM that stores programs and data accessed by the CPU 150, and an image signal input unit 152 (disk device) that inputs an image signal to be encoded from a camera or the like. A program storage device 153 in which an image encoding program 154 that is a software program for causing the CPU 150 to execute processing for encoding an input image by this method is stored, and the CPU 150 An encoded stream output unit 155 that outputs encoded data generated by executing the image encoding program 154 loaded in the memory 151 via, for example, a network (a storage unit that stores an encoded stream by a disk device or the like) But may be connected by bus It has become the configuration.

  FIG. 12 shows a hardware configuration example when the image decoding apparatus of FIG. 8 is configured by a computer and a software program. This system receives a CPU 250 that executes a program, a memory 251 such as a RAM that stores programs and data accessed by the CPU 250, and an encoded stream encoded by the image encoding apparatus of FIG. A program storage in which an encoded stream storage unit 252 (which may be an input unit via a network or the like) to be stored and an image decoding program 254 that is a software program for causing the CPU 250 to execute processing for decoding the encoded stream by this method is stored. And a decoded image output unit 255 that outputs a decoded image obtained by decoding the encoded stream to the playback device by the CPU 250 executing the image decoding program 254 loaded in the memory 251. It is configured to be connected by a bus.

  The embodiments of the present invention have been described above with reference to the drawings. However, the above embodiments are merely examples of the present invention, and it is obvious that the present invention is not limited to the above embodiments. is there. Accordingly, additions, omissions, substitutions, and other modifications of the components may be made without departing from the spirit and technical scope of the present invention.

101, 205 Intra prediction processing unit 102 Transform processing unit 103 Quantization processing unit 104, 202 Inverse quantization processing unit 105, 203 Inverse transform processing unit 106, 204 Frame memory 107 Intra prediction information encoding unit 108 Entropy encoding processing unit 109 , 211 Intra prediction information storage unit 110, 208 Residual prediction processing unit 111, 209 Prediction residual coefficient storage unit 112, 212 Transform coefficient storage unit 113, 213 Residual prediction determination unit 114 Residual prediction selection flag output unit 115 Multiplexing Processing unit 120 Subtractor 121, 122, 220, 221 Adder 123, 124, 222, 223 Switch unit 201 Entropy decoding processing unit 206 Intra prediction information storage unit 207 Separation processing unit 210 Residual prediction selection flag reading unit

Claims (14)

Generates an intra prediction signal using spatial inter-pixel prediction, generates a residual signal between the intra prediction signal and the original signal in the encoding target area of the input image, and performs orthogonal transform on the generated residual signal In an image encoding method for performing encoding to generate transform coefficients,
Based on the prediction direction of intra prediction with respect to the encoding target region, an encoded region to be used for prediction processing for the residual signal is selected, and a specific component of a transform coefficient in the encoded region is designated as a reference signal. Process,
A specific component of the transform coefficient in the encoded region, a value obtained by multiplying the specific component by a given weighting factor or a given bias value, and a corresponding transform coefficient in the encoding target region Generating a residual prediction error signal that is a difference signal from the component;
And a process of encoding the residual prediction error signal as an encoding target. The image encoding method according to claim 1,
Determining whether to encode the residual prediction error signal according to a predetermined evaluation scale;
Encoding residual prediction selection information indicating whether to encode the residual prediction error signal according to the determination result,
When it is determined that the residual prediction error signal is to be encoded, an encoding process involving a prediction process is performed on the residual signal, and when it is determined that the residual prediction error signal is not to be encoded, the residual prediction error signal is encoded. An image encoding method, wherein a transform coefficient obtained by orthogonally transforming the residual signal is encoded without performing a prediction process on the difference signal. The image encoding method according to claim 1 or 2,
In the process of designating as the reference signal, when the region serving as a unit of intra prediction is larger than the region to be encoded, it is the same based on the prediction direction of the region serving as the unit of intra prediction including the region to be encoded. Alternatively, a reference region that is a unit of intra prediction that has been encoded in a similar prediction direction is selected, and the reference region in the reference region that is at the same position as the encoding target region in the region that is the intra prediction unit. An image encoding method, wherein an area is selected as an encoded area used for prediction processing for the residual signal. Generates an intra prediction signal using spatial inter-pixel prediction, generates a residual signal between the intra prediction signal and the original signal in the encoding target area of the input image, and performs orthogonal transform on the generated residual signal In an image encoding apparatus that performs encoding to generate transform coefficients by applying
Based on the prediction direction of intra prediction with respect to the encoding target region, an encoded region to be used for prediction processing for the residual signal is selected, and a specific component of a transform coefficient in the encoded region is designated as a reference signal. Means,
A specific component of the transform coefficient in the encoded region, a value obtained by multiplying the specific component by a given weighting factor or a given bias value, and a corresponding transform coefficient in the encoding target region Means for generating a residual prediction error signal that is a difference signal from the component;
An image encoding apparatus comprising: means for encoding the residual prediction error signal as an encoding target. The image encoding device according to claim 4, wherein
Means for determining whether to encode the residual prediction error signal according to a predetermined evaluation scale;
Means for encoding residual prediction selection information indicating whether or not to encode the residual prediction error signal according to the determination result;
When it is determined that the residual prediction error signal is to be encoded, an encoding process involving a prediction process is performed on the residual signal, and when it is determined that the residual prediction error signal is not to be encoded, the residual prediction error signal is encoded. An image encoding apparatus, wherein a transform coefficient obtained by orthogonally transforming the residual signal is encoded without performing a prediction process on the difference signal. In the image coding device according to claim 4 or 5,
The means to designate as the reference signal is the same based on the prediction direction of the region serving as the unit of intra prediction including the coding target region when the region serving as the unit of intra prediction is larger than the coding target region. Alternatively, a reference region that is a unit of intra prediction that has been encoded in a similar prediction direction is selected, and the reference region in the reference region that is at the same position as the encoding target region in the region that is the intra prediction unit. An image coding apparatus, wherein a region is selected as a coded region used for prediction processing for the residual signal. In an image decoding method for generating an intra prediction signal using spatial inter-pixel prediction, and decoding an encoded image from the intra prediction signal and an intra prediction residual signal obtained from input encoded data ,
A specific component of the transform coefficient in the decoded area included in the input encoded data, a value obtained by multiplying the specific component by a given weight coefficient or a given bias value, and a decoding target area Decoding a residual prediction error signal that is a difference signal from the corresponding transform coefficient component in FIG.
Decoding information indicating the prediction direction of intra prediction for a decoding target region, selecting a decoded region based on the prediction direction, and designating a specific component of a transform coefficient in the decoded region as a reference signal; ,
Generating a prediction signal for the intra prediction residual signal of the decoding target region using the reference signal;
An image decoding method comprising: generating a decoded signal of a decoding target region using the residual prediction error signal and a prediction signal for the intra prediction residual signal. The image decoding method according to claim 7, wherein
Decoding residual prediction selection information indicating whether or not the residual prediction error signal has been encoded included in the input encoded data;
When the residual prediction selection information indicates that the residual prediction error signal is encoded, a decoding process with a prediction process is performed on the residual signal, and a code of the residual prediction error signal is obtained. An image decoding method comprising: decoding encoded data relating to a transform coefficient obtained by orthogonally transforming the residual signal without performing prediction processing on the residual signal when it is indicated that the conversion is not performed. In the image decoding method according to claim 7 or 8,
In the process of designating as the reference signal, when the region serving as the unit of intra prediction is larger than the region to be decoded, decoding of the same or similar prediction direction is completed based on the information indicating the prediction direction of the decoded intra prediction. A reference region that is a unit of intra prediction is selected, and a region in the reference region that is at the same position as the position of the decoding target region in the region that is the intra prediction unit is predicted for the residual signal. An image decoding method, characterized in that it is selected as a decoded area to be used for. In an image decoding apparatus that generates an intra prediction signal using spatial inter-pixel prediction and decodes an encoded image from the intra prediction signal and an intra prediction residual signal obtained from input encoded data ,
A specific component of the transform coefficient in the decoded area included in the input encoded data, a value obtained by multiplying the specific component by a given weight coefficient or a given bias value, and a decoding target area Means for decoding a residual prediction error signal which is a difference signal from the corresponding transform coefficient component in FIG.
Means for decoding information indicating a prediction direction of intra prediction for a decoding target region, selecting a decoded region based on the prediction direction, and designating a specific component of a transform coefficient in the decoded region as a reference signal; ,
Means for generating a prediction signal for an intra prediction residual signal in a decoding target region using the reference signal;
An image decoding apparatus comprising: means for generating a decoded signal of a decoding target region using the residual prediction error signal and a prediction signal for the intra prediction residual signal. The image decoding device according to claim 10,
Means for decoding residual prediction selection information indicating whether or not the residual prediction error signal is encoded included in the input encoded data;
When the residual prediction selection information indicates that the residual prediction error signal is encoded, a decoding process with a prediction process is performed on the residual signal, and a code of the residual prediction error signal is obtained. An image decoding apparatus characterized by decoding encoded data relating to a transform coefficient obtained by orthogonally transforming the residual signal without performing prediction processing on the residual signal when it is indicated that the conversion is not performed. The image decoding device according to claim 10 or 11,
The means for designating as the reference signal, when the region that is the unit of intra prediction is larger than the region to be decoded, is decoded in the same or similar prediction direction based on the information indicating the prediction direction of the decoded intra prediction A reference region that is a unit of intra prediction is selected, and a region in the reference region that is at the same position as the position of the decoding target region in the region that is the intra prediction unit is predicted for the residual signal. An image decoding apparatus, wherein the image decoding apparatus is selected as a decoded area to be used.   An image encoding program for causing a computer to execute the image encoding method according to claim 1, 2 or 3.   An image decoding program for causing a computer to execute the image decoding method according to claim 7, 8 or 9.

Images