JP2009284275A - Image encoding method, image decoding method, image encoder, image decoder, image encoding program, image decoding program, and recording medium recording programs and readable by computer - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a new technology for achieving in-screen prediction encoding with higher compression efficiency when an image is encoded by using in-screen prediction. <P>SOLUTION: A prediction straight line used for generating a prediction signal of a prediction pixel is generated based on coordinates of the prediction pixel and a prediction angle set at any angle. Intersection coordinates between the generated prediction straight line and a line segment generated by a reference pixel are calculated. A distance between the calculated intersection coordinate and a center coordinate of the reference pixel positioned at the intersection coordinate is calculated. A ratio value of the calculated distance and a distance between the reference pixels is calculated. The prediction signal of the prediction pixel is generated based on the calculated ratio value and values of the plurality of reference pixels including the reference pixel positioned at the intersection coordinate. In this configuration, the prediction angle can be set at any angle, and thus, the number of prediction modes can be increased. Accordingly, the in-screen prediction encoding with higher compression efficiency can be achieved. <P>COPYRIGHT: (C)2010,JPO&INPIT

Description

  The present invention relates to an image encoding method and apparatus for encoding an image using intra prediction, an image decoding method and apparatus for decoding encoded data encoded by the image encoding technique, and the image Image encoding program used for realizing encoding method, computer-readable recording medium recording the program, image decoding program used for realizing image decoding method, and computer-readable recording medium recording the program And about.

  Intra-screen predictive coding that performs image signal prediction within the same screen cannot achieve greater compression efficiency than inter-screen predictive coding that performs image signal prediction between different screens. Therefore, an intra prediction encoding method with higher compression efficiency is desired.

  In-screen prediction is based on the video coding international standard H.264. H.264 / MPEG-4 AVC (see, for example, Non-Patent Document 1), which can remove high-frequency components that are likely to occur at the edge using the correlation between adjacent pixels.

  In-screen prediction is performed in units of blocks in which several pixels are combined. In H.264 / MPEG-4 AVC FRExt (see, for example, Non-Patent Document 2), 16 × 16 macroblocks are divided into three types of block sizes of 4 × 4, 8 × 8, and 16 × 16 for luminance signals. Divide and make predictions.

  For these luminance signals, 9 types of prediction modes can be applied to 4 × 4 and 8 × 8 blocks, and 4 types of prediction modes can be applied to 16 × 16 blocks. The prediction mode and block size with the highest coding efficiency are determined by calculating the Lagrangian cost.

  In the 4 × 4 and 8 × 8 blocks, in addition to generating a prediction signal by referring to one type of DC component, the prediction angle range of 45 ° to 206.57 ° is divided into eight types at unequal angles. Generation of a prediction signal is realized by referring to each decoded pixel value / inter-pixel center value that intersects with a straight line generated when dividing.

  FIG. 15 shows nine types of prediction modes 0 to 8 applicable to a 4 × 4 block used for encoding a luminance signal and their prediction angles.

  Here, in the prediction modes 0 to 2, the values of the pixels A, B, C, and D of the block located on the encoding target block defined in FIG. 16A and the definition in FIG. Using the values of the pixels I, J, K, and L of the block located to the left of the encoding target block, the values of the pixels a to p of the encoding target block are predicted.

  In the prediction modes 3 to 8, the values of these pixels A, B, C, D, I, J, K, and L and the block located at the upper left of the encoding target block defined in FIG. As shown in FIG. 16 (b), based on the value of the pixel X possessed and the values of the pixels E, F, G, and H of the block located at the upper right of the encoding target block defined in FIG. 16 (a). In addition, values X ′ and A ′ to L ′ associated with the values of the pixels X and A to L are calculated, and the block located at the lower left of the encoding target block defined in FIG. A value M ′ associated with the value of the pixel M is calculated, and further, values A ″ to F ″ and I ″ to L ″ associated with the center between the pixels are calculated, and those values X ′ are calculated. , A ′ to M ′, A ″ to F ″, and I ″ to L ″ are used to predict the values of the pixels a to p included in the encoding target block.

  That is, as shown in FIG. 15, in the prediction mode 0, the pixels a to p of the encoding target block are copied by copying the values of the pixels A, B, C, and D in the direction (vertical direction) of the prediction angle 90 °. In the prediction mode 1, the values of the pixels a to p of the block to be encoded are copied by copying the values of the pixels I, J, K, and L in the direction of the prediction angle 180 ° (horizontal direction). In prediction mode 2, the values of pixels a to p of the encoding target block are predicted according to the value of (A + B + C + D + I + J + K + L + 4) / 8.

  In prediction mode 3, the values of the pixels a to p of the block to be encoded are predicted by copying the corresponding values calculated in association with the pixel values in the direction of the prediction angle 45 °. The values of the pixels a to p of the encoding target block are predicted by copying the corresponding values calculated in association with the pixel values in the direction of the prediction angle 135 °.

  In prediction mode 5, the corresponding value calculated in the form of being associated with the pixel value and the corresponding value calculated in the form of being associated with the center between the pixels are copied in the direction of the prediction angle 116.57 ° to thereby encode the block to be encoded. In the prediction mode 6, the corresponding value calculated in association with the pixel value and the corresponding value calculated in association with the center between the pixels are predicted at a prediction angle of 153.43 °. The values of the pixels a to p of the encoding target block are predicted by copying in the direction. In the prediction mode 7, the corresponding value calculated in the form corresponding to the pixel value and the corresponding value calculated in the form corresponding to the center between the pixels. Values are copied in the direction of the prediction angle 63.43 ° to predict the values of the pixels a to p of the encoding target block. In the prediction mode 8, the corresponding values and pixels calculated in association with the pixel values Center The value of the pixel a~p with the encoding target block by copying the corresponding value calculated by the form to be associated with the direction of the prediction angle 206.57 ° predicting.

In this way, in the prior art, when intra prediction encoding is performed, in addition to generating a prediction signal by referring to one type of DC component in 4 × 4 and 8 × 8 blocks, , A prediction signal is generated by referring to each decoded pixel value / inter-pixel center value that intersects with a straight line generated when the prediction angle range of 45 ° to 206.57 ° is divided into eight types at unequal angles. Realized.
ITU-T Rec. H.264 | ISO / IEC 14496-10 version 1 (2003) ITU-T Rec. H.264 | ISO / IEC 14496-10 version 4 (2005)

  Thus, in the conventional intra prediction method, there are only 8 types of prediction directions in 4 × 4 and 8 × 8 blocks. In other words, in the conventional intra-screen prediction method, the prediction angle = 90 °, 180 °, 45 °, 135 °, 116.57 °, 153.43 °, in accordance with copying of the pixel value and the inter-pixel center value. There are only 8 types of prediction directions of 63.43 ° and 206.57 °.

  From this, for example, as shown in FIG. 17, the encoding target block is strong at an angle of 125.78 ° between the prediction mode 4 (prediction angle = 135 °) and the prediction mode 5 (prediction angle = 116.57 °). When there is a correlation, there is a problem that the prediction residual signal becomes large if the intra-screen prediction method of the conventional method is followed.

  That is, when the encoding target block has a pixel value as shown in FIG. 17 and prediction mode 4 is used, a prediction residual signal as shown in FIG. In this case, the prediction residual signal as shown in FIG. 18B is obtained, and the prediction residual signal becomes large.

  As described above, if the conventional intra-screen prediction method is followed, the prediction residual signal may become large, and therefore, the DCT coefficient may increase for expressing the prediction residual signal, and the code amount may increase. There is a problem.

  The present invention has been made in view of such circumstances, and provides a new image encoding technique that realizes intra-screen predictive encoding with higher compression efficiency when an image is encoded using intra-screen prediction. With the goal.

[1] Configuration of the Image Encoding Device of the Present Invention To achieve the above object, the image encoding device of the present invention employs a configuration in which an image is encoded using intra prediction. (B) a prediction line generation unit that generates a prediction line used for generating a prediction signal of a prediction pixel based on the coordinates of the prediction pixel and a prediction angle set to an arbitrary angle; and (b) a prediction line generation unit Intersection coordinate calculation means for calculating the intersection coordinates between the generated predicted straight line and the line segment generated by the reference pixel; (c) the intersection coordinates calculated by the intersection coordinate calculation means and the center coordinates of the reference pixel located at the intersection coordinates; A ratio value calculating unit that calculates a ratio value between the calculated distance and the reference pixel distance; and (d) a ratio value calculated by the ratio value calculating unit and an intersection coordinate calculating unit Reference pixel located at the calculated intersection coordinates Prediction signal generation means for generating a prediction signal of a prediction pixel based on the values of some or all of the reference pixels including: (e) encoding based on the prediction signal of the prediction pixel generated by the prediction signal generation means A prediction angle selection unit that calculates a cost and specifies a prediction angle that minimizes the encoding cost, thereby selecting a prediction angle used for in-screen prediction from a plurality of prediction angles set to an arbitrary angle. With.

  When adopting this configuration, in order to reduce the code amount of the prediction angle, the prediction mode number is defined in association with the prediction angle, and the prediction mode number associated with the prediction angle selected by the prediction angle selection unit is encoded. Prediction mode number encoding means may be provided.

  The image encoding method of the present invention realized by the operation of each of the above processing means can also be realized by a computer program, and this computer program is provided by being recorded on an appropriate computer-readable recording medium. The present invention is realized by being provided via a network, installed when the present invention is implemented, and operating on a control means such as a CPU.

[2] Configuration of Image Decoding Device of Present Invention Upon receiving the image encoding device of the present invention, the image decoding device of the present invention decodes encoded data of an image encoded using intra prediction. (A) Predictive angle decoding in which the image encoding side selected from a plurality of prediction angles set to an arbitrary angle decodes the encoded data for the prediction angle used for the intra-screen prediction. (B) a prediction line generation unit that generates a prediction line used by the image encoding side to generate a prediction signal of the prediction pixel based on the coordinates of the prediction pixel and the prediction angle decoded by the prediction angle decoding unit; (C) intersection coordinate calculation means for calculating the intersection coordinates between the prediction line generated by the prediction line generation means and the line segment generated by the reference pixel; (d) the intersection coordinates calculated by the intersection coordinate calculation means and the intersection coordinates. Reference pixel located at A ratio value calculating means for calculating a distance between the center coordinates of the two and calculating a ratio value between the calculated distance and a reference inter-pixel distance; and (e) the ratio value calculated by the ratio value calculating means and the intersection coordinates. Prediction signal restoration means for restoring the prediction signal of the prediction pixel generated by the image encoding side based on the values of some or all reference pixels including the reference pixel located at the intersection coordinates calculated by the calculation means; ) Decoded image that decodes the encoded data of the prediction residual signal encoded by the image encoding side and generates a decoded image signal based on the decoded prediction residual signal and the prediction signal restored by the prediction signal restoration means Signal generating means.

  The image decoding method of the present invention realized by the operation of each of the above processing means can also be realized by a computer program, which is provided by being recorded on a suitable computer-readable recording medium, The present invention is realized by being provided via a network, installed when executing the present invention, and operating on a control means such as a CPU.

[3] Processing of the Present Invention In the present invention configured as described above, when encoding an image using intra prediction, the image encoding device first sets the coordinates of the predicted pixel and an arbitrary angle. Based on the predicted angle, a prediction straight line used to generate a prediction signal of the prediction pixel is generated.

  For example, as shown in FIG. 1, when encoding using 4 × 4 blocks as an encoding target using intra prediction, based on a prediction angle θ set to an arbitrary angle shown in the figure, When the prediction signal of the prediction pixel indicated by PredPixel shown in the figure is generated, the prediction straight line PredLine shown in the figure is generated based on the center coordinates of the prediction pixel PredPixel and the prediction angle θ.

  Subsequently, intersection coordinates between the generated predicted straight line and a line segment generated by the reference pixel are calculated.

  For example, as shown in FIG. 1, a predicted straight line out of a horizontal reference line RefLine generated by reference pixels X and A to H and a vertical reference line RefLine generated by reference pixels X and I to P is predicted. When PredLine crosses the vertical reference line RefLine on the reference pixel J, the intersection coordinates are calculated.

  Subsequently, a distance between the calculated intersection coordinates and the center coordinates of the reference pixels located at the intersection coordinates is calculated, and a ratio value between the calculated distance and the reference pixel distance is calculated.

  For example, as shown in FIG. 1, “1.0−h”, which is a ratio value between the distance between the intersection coordinates and the center coordinates of the reference pixel J located at the intersection coordinates and the distance between the reference pixels, is calculated. Here, in FIG. 1, it is assumed that the distance between the reference pixels is 1, and the distance between the intersection coordinates and the center coordinates of the reference pixel I adjacent to the reference pixel X and the reference pixel distance are It is assumed that the ratio value of “h” is “h”.

  Subsequently, based on the calculated ratio value and the values of some or all of the plurality of reference pixels including the reference pixel located at the intersection coordinates, between reference pixels associated with the intersection coordinates serving as the prediction signal of the prediction pixel A prediction signal of a prediction pixel is generated by calculating a value (a prediction signal is generated by copying this value).

For example, as shown in FIG. 1, using the value γ of the reference pixel J located at the intersection coordinates and the value β of the reference pixel I adjacent to the intersection coordinates and the one closer to the reference pixel X,
V = (1−h) β + hγ... The prediction signal V of the prediction pixel PredPixel is generated according to the linear interpolation equation (1).

  Here, instead of using such a linear interpolation formula, values of some or all of the plurality of reference pixels (X, A to H, I to P) including the reference pixel J located at the intersection coordinates are used. The prediction signal V of the prediction pixel PredPixel may be generated according to the cubic interpolation formula.

  At this time, it may be possible to refer to pixels M to P below the reference pixel L that have not been used in the prior art. In this case, if these pixels M to P correspond to decoded blocks, the value thereof is used. Is a reference pixel value, and if it has not been decoded, the value of the decoded reference pixel L is extrapolated to obtain a reference pixel value.

  In this way, in the present invention, in order to achieve the object of suppressing the prediction residual signal, which is one of the causes for increasing the DCT coefficient, the prediction signal of the prediction pixel using the prediction angle set to an arbitrary angle. To generate.

  In the present invention, the encoding cost is calculated based on the prediction signal of the prediction pixel generated in this way, and the prediction angle that minimizes the encoding cost can be specified, and can be set to an arbitrary angle. As a result, the prediction angle used for intra prediction is selected from the multiple prediction angles that are set more than in the prior art, thereby realizing intra prediction encoding with a smaller prediction residual signal and code amount. To do.

  Upon receiving the encoded data generated by the image encoding device, the image decoding device uses the prediction angle used for intra prediction by the image encoding side selected from a plurality of prediction angles set to an arbitrary angle. The encoded data is decoded, and based on the coordinates of the prediction pixel and the decoded prediction angle, the image encoding side generates a prediction line used for generating the prediction signal of the prediction pixel. Subsequently, the intersection coordinates between the generated predicted straight line and the line segment generated by the reference pixel are calculated, and the distance between the calculated intersection coordinates and the center coordinate of the reference pixel located at the intersection coordinates is calculated. Then, a ratio value between the calculated distance and the reference inter-pixel distance is calculated. Subsequently, the prediction signal of the prediction pixel generated by the image encoding side is restored based on the calculated ratio value and the values of some or all of the reference pixels including the reference pixel located at the intersection coordinates. Then, the encoded data of the prediction residual signal encoded by the image encoding side is decoded, and a decoded image signal is generated based on the decoded prediction residual signal and the restored prediction signal.

  As described above, according to the present invention, it is possible to refer to an inter-pixel value that cannot be dealt with by the conventional intra-screen prediction, thereby enabling efficient intra-screen prediction for an image having a minute angle. Can be realized. As a result, according to the present invention, the intra prediction encoding efficiency can be greatly improved.

  Hereinafter, the present invention will be described in detail according to embodiments.

  FIG. 2 shows an example of an embodiment of the intra prediction encoding apparatus 10 having the present invention.

  The intra prediction encoding apparatus 10 according to the present invention encodes a processing target image using intra prediction using a still image as a processing target or an intra encoding target image included in a video as a processing target. As shown in FIG. 2, the prediction mode selection unit 101, the prediction residual signal generation unit 102, the encoding unit 103, the cost calculation unit 104, the cost minimum value storage unit 105, and the cost minimum value initial stage Comprises a conversion unit 106, a cost determination unit 107, an optimal prediction mode number storage unit 108, a mode selection end determination unit 109, an optimal prediction mode number encoding unit 110, and an in-screen prediction encoding unit 111.

Here, if the prediction mode number is represented by n (0 ≦ n <MaxPred) with the 4 × 4 block as an encoding target, the intra prediction encoding apparatus 10 of the present invention, for example, by intra prediction. The predicted angle θ when the predicted angle range of 45 ° to 225 ° is divided into an equal number of divisions at an equal angle,
θ = 45 + 180 n / MaxPred [°]... For example, when the predicted angle range of 45 ° to 225 ° is divided into 16, the predicted angle is set as shown in FIG.

  The prediction mode selection unit 101 selects a prediction mode for intra prediction encoding (including a prediction mode using a DC component, and is associated with a prediction angle for other prediction modes) for the current block. To do. The prediction residual signal generation unit 102 generates a prediction signal of the image signal of the encoding target block based on the prediction mode selected by the prediction mode selection unit 101, and the difference between the image signal and the generated prediction signal To generate a prediction residual signal.

  The encoding unit 103 encodes the prediction residual signal generated by the prediction residual signal generation unit 102 (becomes intra-screen prediction encoding) and encodes the number of the prediction mode selected by the prediction mode selection unit 101. To do. The cost calculation unit 104 calculates a Lagrangian cost based on the encoding result of the encoding unit 103. The cost minimum value storage unit 105 stores the minimum value of the Lagrangian cost calculated by the cost calculation unit 104. The cost minimum value initialization unit 106 writes an initial value of a Lagrangian cost indicating a large value in the cost minimum value storage unit 105 at the time when the intra-frame prediction encoding of the current block is started.

  The cost determining unit 107 compares the Lagrangian cost calculated by the cost calculating unit 104 with the Lagrangian cost stored in the minimum cost storage unit 105, and if the cost calculated by the cost calculating unit 104 is smaller, the cost determining unit 107 The Lagrangian cost stored in the minimum cost storage unit 105 is updated according to the Lagrangian cost calculated by the calculation unit 104, and the prediction stored in the optimum prediction mode number storage unit 108 according to the prediction mode selected by the prediction mode selection unit 101. Update the mode number. The optimum prediction mode number storage unit 108 stores the number of the optimum prediction mode that is updated by the cost determination unit 107.

  The mode selection end determination unit 109 instructs the prediction mode selection unit 101 to select the next prediction mode when the process of the cost determination unit 107 ends, so that all prediction modes are selected. In addition to controlling, when selection of all prediction modes is completed, the optimal prediction mode number encoding unit 110 and the intra prediction encoding unit 111 are instructed to perform encoding.

  When there is an encoding instruction from the mode selection end determination unit 109, the optimal prediction mode number encoding unit 110 reads the optimal prediction mode number from the optimal prediction mode number storage unit 108 and encodes it. When there is an encoding instruction from the mode selection end determination unit 109, the intra-screen prediction encoding unit 111 reads the optimal prediction mode number from the optimal prediction mode number storage unit 108, and the optimal prediction mode number indicated by the read prediction mode number is read out. Based on the prediction mode, a prediction signal of the image signal of the encoding target block is calculated, and a prediction residual signal is generated by calculating a difference between the image signal and the calculated prediction signal. Turn into.

  4 to 6 show flowcharts executed by the intra prediction encoding apparatus 10 of the present invention. Next, processing executed by the intra prediction encoding apparatus 10 of the present invention configured as shown in FIG. 2 will be described according to this flowchart.

The intra prediction encoding apparatus 10 of the present invention, when an encoding target block to be processed for intra prediction encoding is given, is first described in step S101, as will be described later, as shown in the flowchart of FIG. The minimum value L _min of the Lagrangian cost L _cost used in step S107 is initialized with a sufficiently large value (+ ∞).

Subsequently, in steps S102 to S109 described below, a loop process for detecting a prediction mode number n that minimizes the Lagrangian cost L _cost is executed. Here, in this flowchart, it is assumed that the total number of prediction modes is MaxPred.

  That is, when one unselected prediction mode number n is selected in step S102, a prediction residual signal is subsequently generated in step S103 according to the process of the flowchart described later with reference to FIG.

  Subsequently, in step S104, the prediction mode number n selected in step S102 is encoded. In subsequent step S105, the prediction residual signal generated in step S103 is DCT-quantized and encoded.

  Here, in step S104, for example, a 1-bit flag determined from a prediction signal from an adjacent block is used, and when prediction fails, the code of the prediction mode number n is performed by performing fixed length or variable length encoding. To do. In step S105, the prediction residual signal is encoded using a conventional method.

Subsequently, in step S106, a conventional method (for example, H.264 Reference Software JM ver. 13.2 Encoder _, http: // iphome) is calculated from the error between the number of bits generated by the encoding in steps S104 and S105 and the original image. Lagrange cost L _cost is calculated using .hhi.de / suehring / tml / _, 2008).

Subsequently, in step S107,
L _cost <L _min
By determining whether or not is established, it is determined whether or not the Lagrange cost L _cost calculated in step S106 is minimum.

When it is determined that the Lagrangian cost L _cost calculated in step S106 is _less than the minimum Lagrangian cost L _min according to the determination process in step S107, the process proceeds to step S108, and the minimum Lagrangian cost is calculated according to the Lagrangian cost L _cost calculated in step S106. Update L _min .

On the other hand, according to the judgment processing in step S107, when the Lagrangian cost L _cost calculated in step S106, it is determined that no less than the minimum Lagrangian cost L _min does not perform processing in step S108, updating processing of the minimum Lagrangian cost L _min is Not performed.

The above processing (steps S102 to S109) is repeated within the range of 0 ≦ n <MaxPred−1, thereby identifying the optimal prediction mode number n that realizes the minimum Lagrangian cost L _min .

  Subsequently, in step S110, the identified optimum prediction mode number is encoded.

  Subsequently, in step S111, a prediction residual signal in the optimal prediction mode is generated according to the processing of the flowchart described with reference to FIG.

  Finally, in step S112, the prediction residual signal generated in step S111 is DCTed, quantized, and encoded to execute intra-frame predictive encoding of the current block.

  Next, the process executed in step S103 of the flowchart of FIG. 4 will be described in detail according to the flowchart of FIG.

  In step S103, as shown in the flowchart of FIG. 5, first, in step S201, according to the prediction mode number n selected in step S102, the prediction mode indicated by the prediction mode number n uses a DC component. It is determined whether the prediction mode or the prediction mode having the prediction angle.

When it is determined that the prediction mode indicated by the prediction mode number n selected in step S102 is a prediction mode having a prediction angle according to the determination process in step S201, the process proceeds to step S202, and the prediction mode number selected in step S102. A predicted angle θ is generated from n. The prediction mode number n and the prediction angle θ are, for example, according to the equation (2) described above.
θ = 45 + 180n / MaxPred [°]
Thus, the prediction angle θ associated with the prediction mode number n selected in step S102 is generated according to this equation.

  Subsequently, in steps S203 to S210 described below, a loop process for calculating a prediction signal is executed for each prediction pixel PredPixel (pixels a to p shown in FIG. 15) of the encoding target block.

  That is, in step S203, one unselected prediction pixel PredPixel is selected from the prediction pixels PredPixel of the encoding target block.

  Subsequently, in step S204, a prediction straight line PredLine as shown in FIG. 1 is generated from the coordinates (i, j) of the prediction pixel PredPixel selected in step S202 and the prediction angle θ generated in step S202.

That is,
y = (xi) tan [theta] + j (3) A predicted straight line PredLine expressed by equation (3) is generated. Here, it is assumed that the center coordinate of the prediction pixel PredPixel at the upper left end in the encoding target block is set as the origin, and the value of j becomes smaller every time raster scanning is performed in the y-axis downward direction.

  Subsequently, in step S205, as described with reference to FIG. 1, the intersection coordinates of the prediction line PredLine generated in step S204 and the reference line RefLine generated by the reference pixel are calculated, and the ratio value h described in FIG. (A value obtained by dividing the distance between the intersection coordinates and the center coordinates of the adjacent reference pixels closer to the reference pixel X by the distance between the reference pixels).

  Subsequently, in step S206, values of reference pixels located around the intersection coordinates are acquired. For example, when the reference pixels α, β, γ, and δ located around the intersection coordinates are defined in the form shown in FIG. 7 and the prediction signal V of the prediction pixel PredPixel is calculated according to cubic interpolation. Obtains the values of the reference pixels α, β, γ, δ located around the intersection coordinates, and calculates the prediction signal V of the prediction pixel PredPixel according to linear interpolation, the reference pixels located around the intersection coordinates The values of β and γ are acquired.

  Subsequently, in step S207, the inter-reference pixel values associated with the intersection coordinates serving as the prediction signal of the prediction pixel PredPixel using an interpolation method of linear interpolation or cubic interpolation (other interpolation methods can also be used). By calculating (a prediction signal is generated by copying this), the real value V of the prediction signal of the prediction pixel PredPixel selected in step S202 is calculated.

For example, when the linear interpolation method is used, the linear interpolation formula (1) described above is used.
V = (1-h) β + hγ (...) The real value V of the prediction signal of the prediction pixel PredPixel is calculated according to the equation (1). For example, when the cubic interpolation method using the values of the reference pixels α, β, γ, and δ is used, the following cubic interpolation equation V = αW (h−1) + βW (h) + γW (1-h) + ΔW (2-h)
W (l) = (w + 2) | l | ³ − (w + 3) | l | ² +1 | l | <1
= W | l | ³ -5w | l | ² + 8w | l | -4w 1 ≦ | l | <2
= 0 2 ≦ | l |
Where w: weighting factor
l: Distance between reference pixels
In accordance with the equation (4), the real value V of the prediction signal of the prediction pixel PredPixel is calculated.

Subsequently, in step S208, the calculated real value V is converted to an integer by rounding off, and then clipped to the pixel depth (if the integer value is less than or equal to the representation range of the bit depth D of the image before encoding, V = 0) If it is equal to or greater than the expression range, V = D _max is clipped), and in step S209, the clipped value is used as the prediction signal of the prediction pixel PredPixel selected in step S202.

  The above processing (steps S202 to S210) is repeated for the prediction pixel PredPixel included in the encoding target block, whereby the prediction mode indicated by the prediction mode number n selected in step S102 is a prediction mode having a prediction angle. The prediction signals of all the prediction pixels PredPixel (pixels a to p shown in FIG. 15) of the encoding target block are generated.

  On the other hand, when it is determined that the prediction mode indicated by the prediction mode number n selected in step S102 is a prediction mode using a DC component according to the determination process in step S201, the process proceeds to step S211 and the reference pixels A to D, I The DC component is calculated as the average of the values of ~ L (shown in FIG. 1).

  Subsequently, in step S212, the calculated DC component is used as a prediction signal for all prediction pixels PredPixel (pixels a to p shown in FIG. 15) of the encoding target block, so that the prediction mode number n selected in step S102 is obtained. When the prediction mode pointed to by is a prediction mode using a DC component, prediction signals of all prediction pixels PredPixel (pixels a to p shown in FIG. 15) of the encoding target block are generated.

  In this manner, when the prediction signal for the image signal of the encoding target block is generated according to the processing of step S201 to step S212, the process proceeds to step S213, and a difference value between the image signal and the prediction signal is calculated. Thus, a prediction residual signal is generated, and thereby the process of step S103 in the flowchart of FIG. 4 is terminated.

  Next, the process executed in step S111 of the flowchart of FIG. 4 will be described in detail according to the flowchart of FIG.

  In step S111, as shown in the flowchart of FIG. 6, first, in step S301, the prediction mode in which the optimum prediction mode uses a DC component according to the optimum prediction mode number n specified by the loop 1 process. Or a prediction mode having a prediction angle.

  When it is determined that the optimal prediction mode is a prediction mode having a prediction angle according to the determination process of step S301, the process proceeds to step S302, and the same process as step S202 of the flowchart of FIG. A prediction angle θ is generated from the prediction mode number n.

  Subsequently, in steps S303 to S310 below, the same processing as in steps S203 to S210 in the flowchart of FIG. 5 is executed, so that the prediction mode indicated by the optimal prediction mode number n is a prediction mode having a prediction angle. In this case, prediction signals for all prediction pixels PredPixel (pixels a to p shown in FIG. 15) of the encoding target block are generated.

  On the other hand, when it is determined that the optimal prediction mode is a prediction mode using a DC component according to the determination process in step S301, the same processes as in steps S211 and S212 in the flowchart of FIG. 5 are executed in steps S311 and S312. Thus, when the prediction mode indicated by the optimal prediction mode number n is a prediction mode using a DC component, the prediction signals of all the prediction pixels PredPixel (pixels a to p shown in FIG. 15) of the encoding target block are obtained. Generate.

  In this way, when the prediction signal for the image signal of the encoding target block is generated according to the processing in steps S301 to S312, the process proceeds to step S313, and the same processing as step S213 in the flowchart of FIG. 5 is executed. Thus, a prediction residual signal in the optimal prediction mode is generated, and thereby the processing of step S111 in the flowchart of FIG. 4 is finished.

  In this way, the intra prediction encoding apparatus 10 of the present invention generates a prediction signal using a prediction angle set to an arbitrary angle when encoding an encoding target block using intra prediction. In the conventional technique, only eight types of prediction directions as shown in FIG. 15 exist, and the intra-screen predictive coding is realized based on the prediction signal generated as described above. Compared to the above, since the number of prediction directions can be greatly increased, the prediction accuracy can be improved, and the prediction residual signal can be greatly reduced.

  For example, as shown in FIG. 8A (the same as that shown in FIG. 17), the current block is encoded in the conventional prediction mode 4 (prediction angle = 135 °) and the conventional prediction mode 5 (prediction angle = 116). In the case where there is a strong correlation at an intermediate angle of 125.78 ° with respect to .57 °), if the intra-screen prediction method of the conventional method is followed, FIG. 8B (the same as that shown in FIG. 18) As shown in FIG. 4, the prediction residual signal becomes large, but the intra prediction encoding apparatus 10 of the present invention can also realize a prediction mode with a prediction angle of 125.78 °. Therefore, as shown in FIG. 8C, the prediction residual signal can be greatly reduced.

  That is, the sum of the absolute values of the prediction residuals in the conventional prediction mode 4 shown on the left side of FIG. 8B is “502”, and the prediction residuals in the conventional prediction mode 5 shown on the right side of FIG. While the sum of absolute values is “595”, the sum of absolute values of prediction residuals in the prediction mode with a prediction angle of 125.78 ° shown in FIG. Can be greatly reduced. In FIGS. 8B and 8C, the prediction signal is generated according to the linear interpolation equation shown in the above equation (1).

  FIG. 9 illustrates an embodiment of the intra-screen predictive decoding device 20 of the present invention that decodes the encoded data generated by the intra-screen predictive encoding device 10 of the present invention configured as shown in FIG.

  As shown in this figure, the intra prediction decoding apparatus 20 according to the present invention is provided with a prediction mode for decoding the encoded data generated by the intra prediction encoding apparatus 10 according to the present invention configured as shown in FIG. A number decoding unit 201, a prediction signal restoration unit 202, a prediction residual signal decoding unit 203, and a decoded image signal generation unit 204 are provided.

  This prediction mode number decoding part 201 decodes prediction mode number by decoding the encoding data of the prediction mode number which the intra prediction encoding apparatus 10 of this invention produced | generated. The prediction signal restoration unit 202 restores the prediction signal generated by the intra prediction encoding apparatus 10 of the present invention based on the prediction mode indicated by the prediction mode number decoded by the prediction mode number decoding unit 201. The prediction residual signal decoding unit 203 decodes the encoded data of the prediction residual signal generated by the intraframe prediction encoding apparatus 10 of the present invention, and performs inverse quantization and inverse DCT on the prediction residual signal. Is decrypted. The decoded image signal generation unit 204 generates a decoded image signal by adding the prediction signal restored by the prediction signal restoration unit 202 and the prediction residual signal decoded by the prediction residual signal decoding unit 203.

  10 and 11 show flowcharts executed by the intra prediction decoding apparatus 20 of the present invention. Next, processing executed by the intra prediction decoding apparatus 20 of the present invention configured as shown in FIG. 9 will be described according to this flowchart.

  When the intraframe prediction decoding apparatus 20 of the present invention receives the encoded data generated by the intraframe prediction encoding apparatus 10 of the present invention, as shown in the flowchart of FIG. The prediction mode number is decoded by decoding the encoded data of the number.

  Subsequently, in step S402, the prediction signal generated by the intra prediction encoding apparatus 10 of the present invention is restored according to the processing of the flowchart described in FIG.

  Subsequently, in step S403, the encoded data of the prediction residual signal is decoded, and the prediction residual signal is decoded by inverse quantization and inverse DCT.

  Finally, in step S404, a decoded image signal is generated by adding the prediction signal restored in step S402 and the prediction residual signal decoded in step S403, and the process ends.

  Next, the process executed in step S402 of the flowchart of FIG. 10 will be described in detail according to the flowchart of FIG.

  In the process of step S402, as shown in the flowchart of FIG. 11, first, in step S501, according to the prediction mode number n decoded in step S401, the prediction mode indicated by the prediction mode number n uses a DC component. It is determined whether the prediction mode or the prediction mode having the prediction angle.

  When it is determined that the prediction mode indicated by the prediction mode number n decoded in step S401 is a prediction mode having a prediction angle according to the determination process in step S501, the process proceeds to step S502, and step S202 in the flowchart of FIG. By executing the same process, the prediction angle θ is generated from the prediction mode number n decoded in step S401.

  Subsequently, in steps S503 to S510 below, the same processing as in steps S203 to S210 in the flowchart of FIG. 5 is executed, so that the prediction mode indicated by the prediction mode number n decoded in step S401 has a prediction angle. In the prediction mode, the prediction signals of all the prediction pixels PredPixel included in the decoding target block generated by the intra prediction encoding apparatus 10 of the present invention are restored.

  On the other hand, when it is determined that the prediction mode indicated by the prediction mode number n decoded in step S401 is a prediction mode using a DC component in accordance with the determination process in step S501, in steps S511 and S512, the steps of the flowchart of FIG. When the prediction mode indicated by the prediction mode number n decoded in step S401 is a prediction mode using a DC component, the intra-frame prediction encoding apparatus 10 of the present invention performs the same processes as in S211 and S212. The prediction signals of all the prediction pixels PredPixel of the generated decoding target block are restored.

  In this way, a prediction signal for the image signal of the decoding target block is generated in accordance with the processing of step S501 to step S512, thereby ending the processing of step S402 in the flowchart of FIG.

  In this way, the intra prediction decoding apparatus 20 according to the present invention is generated by the intra prediction encoding apparatus 10 according to the present invention having a configuration in which a prediction signal is generated using a prediction angle set to an arbitrary angle. The encoded data is decoded.

  Next, the results of experiments conducted to verify the effectiveness of the present invention will be described. In this experiment, the image to be encoded is divided into 16 × 16 macroblocks, and further, the 16 × 16 macroblock is divided into 4 × 4 blocks to generate intra prediction values for the 4 × 4 blocks. I went there. The predicted angle was set by dividing an angle range of 45 ° to 225 ° into equal angles, and the in-screen predicted value was calculated by linear interpolation.

  In the present invention, in the intra prediction encoding, the prediction accuracy is improved by increasing the number of prediction modes, thereby reducing the DCT coefficient, and as a result necessary for encoding the DCT coefficient. As a result, the amount of bits can be reduced and the encoding efficiency can be improved. On the other hand, since the number of prediction modes is increased, the bit amount necessary for encoding the prediction mode number increases. In this experiment, the experiment was conducted in consideration of this point.

  If the contents of this experiment are described, the present invention is also described in H.264. Similarly to H.264, the prediction residual obtained by subtracting each prediction value from the encoding target image is quantized using DCT and a quantization matrix, and the converted DCT coefficient group is converted into a CAVLC format (revised version H .264 / AVC textbook (ISBN4-8443-2204-4) .p.147 See Table 6-7).

  H. In H.264, variable length coding is adaptively performed on the numerical value group stored here using information on adjacent blocks. In this experiment, for the sake of simplicity, variable length coding is performed on these numerical groups by the following method. Therefore, in order to enable objective data comparison, the H.C. Even in the H.264 system (the H.264 system is called because the algorithm is partially changed from the original H.264), variable length coding is performed by the following method.

・ TotalCoeff Absolute value starting from 0 (0 = 1bit, 1 = 2bits, 2 = 3bits, ...)
・ TrailingOnes Absolute value starting from 0 (0 = 1bit, 1 = 2bits, 2 = 3bits, ...)
・ Trailing-ones-sigh-flag plus / minus sign ・ level plus / minus sign + 1 absolute value (1 = 2bits, -1 = 2bits, 2 =
(3bits, -2 = 3bits, ...)
・ Total-zeros revised H.264 / AVC textbook. P.302,303 Refer to Table 13-17, 13-18 ・ run-before revised H.264 / AVC textbook. P.303,304 Refer to Table 13-20. In the H.264 system, the original H.264 standard is used. Similarly to H.264, nine patterns of predicted values are generated. H. In the H.264 system, the original H.264 standard is used. As in H.264, the prediction direction is predicted using the prediction direction in the adjacent upper and left 4 × 4 blocks, and when the prediction value matches the prediction value, it is represented by a 1-bit flag. If they are different, the prediction direction is corrected by a 3-bit correction value in addition to the 1-bit flag. On the other hand, in the method according to the present invention, the prediction value is generated by the number corresponding to the division number. In addition, since the prediction direction is encoded with a fixed length, all 4 × 4 blocks require 4 bits for 16 divisions and 5 bits for 32 divisions.

  In this experiment, H.C. In the method of H.264, 9 patterns are calculated, and in the method of the present invention, the square error between the number of bits required for encoding and the encoding target image is calculated by the number of patterns corresponding to the number of divisions. The experiment was conducted by performing the conversion.

  In FIG. 12, “HARBOUR CIF (352 × 288)” is assumed to be an encoding target image. FIG. 13 shows RD (Rate-Distortion) curves of experimental results when encoding is performed using the H.264 system, the present invention system of 16 divisions, the present invention system of 32 divisions, and the present invention system of 64 divisions. FOREMAN CIF (352 × 288) ”as an encoding target image. The RD curve of the experimental result at the time of encoding by the H.264 system, the present invention system of 16 divisions, the present invention system of 32 divisions, and the present invention system of 64 divisions is illustrated.

  The vertical axis of the RD curve indicates PSNR. PSNR is a logarithm of the error between the image before encoding and the image after encoding. The higher this value, the smaller the difference from the image before encoding, and the more faithfully the image is reproduced. Indicates that it is done. In general, a numerical value of about PSNR = 35 to 38 is often used for a luminance signal.

  The horizontal axis of the RD curve indicates the number of bits generated when an image is encoded using an arbitrary encoding method. That is, when comparing the performance of different encoding methods, when encoding is performed with the same PSNR (image clarity), it can be considered that the method with a smaller number of bits is superior.

  12 and FIG. H.264 compliant encoding method is used only for the prediction method in the intra prediction. H.264-compliant 9-pattern prediction method is changed to a method for increasing the number of divisions in the present invention (16 divisions, 32 divisions, 64 divisions); This is a comparison with the prediction method (H.264 method) of 9 patterns based on H.264. In the method according to the present invention, the average value (DC component) of the peripheral reference pixels is used as the predicted value with the exception of the θ = 45 ° direction in all the division numbers.

  In the RD curve shown in FIG. 12, when PSNR = 36 is focused on, encoding is performed using 64 divisions with the highest encoding efficiency. Compared with the H.264 system, it was confirmed that the code amount could be reduced by about 7.9%, and the effectiveness of the present invention could be verified.

  FIG. 14 shows the experimental results comparing the prediction residuals that occur when the prediction with the smallest prediction residual is performed.

  Here, Sum of absolute difference in FIG. 14 represents an average value of 60 frames in the total sum of predicted residual absolute values calculated by the following procedure.

  (1) All images before encoding are divided into 16 × 16 macroblocks, and further, 16 × 16 macroblocks are divided into 4 × 4 blocks, and 4 × 4 blocks are sequentially selected according to a prescribed access order.

  (2) For each 4 × 4 block, the following all methods and all prediction directions are calculated using RefPixel, and the prediction residual generated when the prediction residual is predicted in the direction that minimizes the prediction residual for each method. Absolute values are calculated (pixel values of all the encoding target images are used for RefPixel).

・ H. In-screen prediction method based on H.264 • Method of dividing 45 ° to 225 ° into 16 equal angles and calculating the predicted angle θ = 45 ° as the average value of adjacent pixels (Div16)
・ A method of dividing 45 ° to 225 ° into 32 equiangular angles and calculating the predicted angle θ = 45 ° as the average value of adjacent pixels (Div32)
・ A method of dividing 45 ° to 225 ° into 64 equiangular angles and calculating the predicted angle θ = 45 ° as the average value of adjacent pixels (Div64)
A method of dividing 45 ° to 225 ° into 128 equiangular angles and calculating the predicted angle θ = 45 ° as an average value of adjacent pixels (Div128)
A method of dividing 45 ° to 225 ° into 256 equiangular angles and calculating the predicted angle θ = 45 ° as an average value of adjacent pixels (Div256)
A method of dividing 45 ° to 225 ° into 512 equiangular angles and calculating the predicted angle θ = 45 ° as an average value of adjacent pixels (Div512)
(3) The sum of predicted residual absolute values (Sum of absolute difference) calculated in step 2 is calculated for each method per image.

  H. H.264. Then, since DCT is performed on the prediction residual when encoding, the prediction residual and the code amount are not necessarily proportional, but the possibility that the code amount can be reduced when the prediction residual is small is high. H. As a prediction residual calculated when using the intra-screen prediction based on H.264, the reduction rate of the prediction residual for each division number in the present invention is shown in Gain in FIG. The effectiveness of the present invention could be verified from the experimental data of FIG.

  In the present invention, in the intra prediction encoding, the prediction accuracy is improved by increasing the number of prediction modes, thereby reducing the DCT coefficient, and as a result necessary for encoding the DCT coefficient. As a result, the amount of bits can be reduced and the encoding efficiency can be improved. On the other hand, since the number of prediction modes is increased, the bit amount necessary for encoding the prediction mode number increases.

  Therefore, as shown in FIG. 12 and FIG. 13, the RD curve indicating the coding efficiency slows down as the efficiency increases as the number of divisions increases. Will occur. From the experimental results shown in FIGS. 12 and 13, it is understood that it is appropriate to use the number of divisions of 32 and 64 when PSNR = 35 to 38, which is generally used, is used.

  In the embodiment described above, the prediction is performed using the prediction angle divided into equal angles according to the above-described equation (2). However, the prediction can be performed using the angles divided into unequal angles. It is. For example, H.M. The prediction using the angle divided into the unequal angles similar to H.264 / MPEG4 AVC is performed by using the table of the prediction angle (unequal angle) of each prediction mode as shown in FIG. It can be realized by holding. H. If the angle subdivided more than H.264 / MPEG4 AVC is also known by the encoder / decoder, it can be easily realized without overhead.

  In addition, although not described in the above-described embodiment, a prediction angle appearance probability distribution LUT (look-up table) that can efficiently reduce a prediction residual by calculation in advance with respect to an encoding target image. The prediction residual can be further reduced by constructing.

  When this prediction angle appearance probability distribution LUT is constructed, a prediction signal can be generated on the decoder side by adding this prediction angle appearance probability distribution LUT to the bitstream. The prediction angle appearance probability distribution LUT can be realized by encoding the prediction angle maximum division number N (= MaxPred) with x bits and constructing N prediction angles with fixed-point y bits and adding overhead of x + Ny bits.

It is a figure explaining the process of this invention. It is an apparatus block diagram of the intra-screen prediction encoding apparatus of this invention. It is a figure which shows an example of the setting of a prediction angle. It is a flowchart which the intra prediction encoding apparatus of this invention performs. It is a flowchart which the intra prediction encoding apparatus of this invention performs. It is a flowchart which the intra prediction encoding apparatus of this invention performs. It is explanatory drawing of the reference pixel located in the periphery of an intersection coordinate. It is a figure explaining the effectiveness of the present invention. It is an apparatus block diagram of the intra-screen prediction decoding apparatus of this invention. It is a flowchart which the intra prediction prediction apparatus of this invention performs. It is a flowchart which the intra prediction prediction apparatus of this invention performs. It is an experimental data of the experiment conducted in order to verify the effectiveness of this invention. It is an experimental data of the experiment conducted in order to verify the effectiveness of this invention. It is an experimental data of the experiment conducted in order to verify the effectiveness of this invention. It is explanatory drawing of the prediction angle used with the conventional intra-screen prediction method. It is explanatory drawing of the pixel value of the reference pixel used with the conventional prediction method in a screen. It is a figure which shows an example of the image signal of an encoding object block. It is explanatory drawing of the conventional prediction method in a screen.

Explanation of symbols

DESCRIPTION OF SYMBOLS 10 Intra prediction encoding apparatus 101 Prediction mode selection part 102 Prediction residual signal generation part 103 Encoding part 104 Cost calculation part 105 Cost minimum value memory | storage part 106 Cost minimum value initialization part 107 Cost determination part 108 Optimal prediction mode number memory | storage Unit 109 mode selection end determination unit 110 optimum prediction mode number encoding unit 111 intra-screen prediction encoding unit

Claims (10)

An image encoding method executed by an image encoding device that encodes an image using intra prediction,
A process of generating a prediction line used for generating a prediction signal of the prediction pixel based on the coordinates of the prediction pixel and the prediction angle set to an arbitrary angle;
Calculating intersection coordinates of the predicted straight line and a line segment generated by a reference pixel;
Calculating a distance between the intersection coordinates and a center coordinate of a reference pixel located at the intersection coordinates, and calculating a ratio value between the calculated distance and a reference inter-pixel distance;
Generating a prediction signal of a prediction pixel based on the ratio value and the value of a part or all of the reference pixels including the reference pixel located at the intersection coordinates,
A characteristic image encoding method. The image encoding method according to claim 1,
By calculating the encoding cost based on the prediction signal of the generated prediction pixel and specifying the prediction angle that minimizes the encoding cost, the screen can be selected from a plurality of prediction angles set to an arbitrary angle. Providing a process of selecting a prediction angle used for intra prediction,
A characteristic image encoding method. The image encoding method according to claim 2, wherein
Defining a prediction mode number associated with a prediction angle and encoding a prediction mode number associated with the selected prediction angle;
A characteristic image encoding method. An image decoding method executed by an image decoding device that decodes encoded data of an image encoded using intra prediction,
A process of decoding the encoded data for the prediction angle used by the image encoding side selected from the plurality of prediction angles set to an arbitrary angle for intra prediction;
Based on the coordinates of the prediction pixel and the decoded prediction angle, the process of generating a prediction line used by the image encoding side to generate the prediction signal of the prediction pixel;
Calculating intersection coordinates of the predicted straight line and a line segment generated by a reference pixel;
Calculating a distance between the intersection coordinates and a center coordinate of a reference pixel located at the intersection coordinates, and calculating a ratio value between the calculated distance and a reference inter-pixel distance;
Restoring the prediction signal of the prediction pixel generated by the image encoding side based on the ratio value and the value of a part or all of the reference pixels including the reference pixel located at the intersection coordinates;
Decoding encoded data of a prediction residual signal encoded by the image encoding side, and generating a decoded image signal based on the decoded prediction residual signal and the restored prediction signal,
A characteristic image decoding method. An image encoding device that encodes an image using intra prediction,
Means for generating a prediction line used for generating a prediction signal of the prediction pixel based on the coordinates of the prediction pixel and a prediction angle set to an arbitrary angle;
Means for calculating intersection coordinates between the predicted straight line and a line segment generated by a reference pixel;
Means for calculating a distance between the intersection coordinates and a center coordinate of a reference pixel located at the intersection coordinates, and calculating a ratio value between the calculated distance and a reference pixel distance;
Generating a prediction signal of a prediction pixel based on the ratio value and the value of a part or all of the reference pixels including the reference pixel located at the intersection coordinates,
An image encoding device. An image decoding apparatus for decoding encoded data of an image encoded using intra prediction,
Means for decoding the encoded data for the prediction angle used by the image encoding side selected from the plurality of prediction angles set to an arbitrary angle for intra prediction;
Based on the coordinates of the prediction pixel and the decoded prediction angle, the image encoding side generates a prediction line used for generating the prediction signal of the prediction pixel;
Means for calculating intersection coordinates between the predicted straight line and a line segment generated by a reference pixel;
Means for calculating a distance between the intersection coordinates and a center coordinate of a reference pixel located at the intersection coordinates, and calculating a ratio value between the calculated distance and a reference pixel distance;
Means for restoring the prediction signal of the prediction pixel generated by the image encoding side based on the ratio value and the value of a part or all of the reference pixels including the reference pixel located at the intersection coordinates;
Means for decoding encoded data of a prediction residual signal encoded by the image encoding side, and generating a decoded image signal based on the decoded prediction residual signal and the restored prediction signal;
A featured image decoding apparatus.   An image encoding program for causing a computer to execute the image encoding method according to any one of claims 1 to 3.   A computer-readable recording medium on which an image encoding program for causing a computer to execute the image encoding method according to any one of claims 1 to 3 is recorded.   An image decoding program for causing a computer to execute the image decoding method according to claim 4.   A computer-readable recording medium on which an image decoding program for causing a computer to execute the image decoding method according to claim 4 is recorded.

Images