JP2010098633A - Prediction coding apparatus, and prediction coding method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To prevent deterioration of a coded image quality in divided areas caused by discontinuous images on image dividing borders while one screen is divided uniformly by a prescribed method to code a plurality of divided images by a plurality of image coding apparatuses in parallel and independently. <P>SOLUTION: An image coding apparatus includes a screen dividing means, a divided area reversing means, a prediction mode control means, and a prediction reference area control means. The screen dividing means divides one screen into areas so that the borders overlap with each other. The divided area reversing means turns the divided areas so that the overlapping areas can be used for in-screen prediction processing. The prediction mode control means controls an in-screen prediction mode on the borders. Thus, efficiency for in-screen prediction coding on the screen dividing borders can be improved, making it possible to prevent the dividing borders from appearing when the divided images are combined on the decoding side. <P>COPYRIGHT: (C)2010,JPO&INPIT

Description

  In the present invention, in order to encode a moving image signal having a large number of pixels, a moving image is divided and encoded in parallel using N encoding circuits, and the output N encoded streams are output. The present invention relates to a parallel encoding apparatus to be integrated and a method thereof.

  In recent years, with the increase in the size of display devices, AV equipment for encoding / decoding high-definition image signals such as high definition television (HDTV) signals is required. In the future, the need to handle ultra high-definition image signals such as 4K2K size is also expected. At that time, since the number of pixels increases, a very high-speed encoding process is required, so that it is difficult to realize and an extremely expensive device. Therefore, the system can be configured at low cost by dividing the HDTV signal into a plurality of image signals and encoding them in parallel with the plurality of encoding LSIs using an inexpensive encoding LSI that is currently popular. can do.

  The operation of the conventional system will be described with reference to FIGS.

  FIG. 10 is a block diagram showing a configuration of a conventional moving picture coding system described in Patent Document 1. In FIG. In FIG. 10, the moving image encoding apparatus 1000 includes a screen dividing apparatus 1010, first to fourth divided image encoding apparatuses 1020-1 to 1020-4, an encoded stream integrating apparatus 1030, and a control apparatus 1040. An HDTV signal as shown in FIG. 12 is input to the moving image encoding apparatus 1000 as an input moving image signal VIDEO. This HDTV signal is a 4: 2: 2 format interlace signal in which the luminance signal has a horizontal 1920 pixels × vertical 1080 lines and the color difference signal has a horizontal 960 pixels × vertical 1080 lines.

  The screen dividing device 1010 changes the size of the input moving image signal VIDEO described below. The screen dividing device 1010 first converts each frame image of the input moving image signal VIDEO into a luminance signal of 1440 pixels in the horizontal direction and a color difference signal to 720 pixels in the horizontal direction by filtering. According to the MPEG2 standard, it is determined that the number of lines in the vertical direction must be a multiple of 32 for interlaced images. Therefore, the screen dividing device 1010 then adds 8 lines of dummy data below the image for both the luminance signal and the color difference signal to obtain 1088 lines. After the size changing process as described above, the screen dividing device 1010 converts the input moving image signal VIDEO of one screen into a luminance signal having a structure of 720 pixels by 544 lines as shown in FIG. Divide into four areas A, B, C, and D. Then, the screen dividing device 1010 outputs the image signal of each divided region to the first to fourth divided image encoding devices 1020-1 to 1020-4. The image signal is divided into four areas A, B, C, and D (hereinafter, divided image signals VIDEO-A, VIDEO-B, VIDEO-C, and VIDEO-D). Each of the divided image signals VIDEO-A, -B, -C, and -D includes a plurality of macro blocks. Each macroblock includes a plurality of image signals. For example, if the size of one macroblock is 16 pixels × 16 pixels, one divided image signal is composed of 720/16 × 544/16 = 45 × 34 macroblocks.

  The first to fourth divided image encoding devices 1020-1 to 1020-4 encode the input divided image signals VIDEO-A, -B, -C, -D according to MPEG2, respectively, and a video stream integrating device. Output to 1040. At that time, in each divided image encoding device 1020-i (i = 1 to 4), for each macroblock constituting the corresponding divided image signal VIDEO-A, -B, -C, -D, a division boundary is set. Encoding conditions are determined according to the positional relationship between and. At that time, each divided image encoding device 1020-i (i = 1 to 4) determines an encoding macroblock type and a quantization scale code (Q scale code) according to the positional relationship with the divided boundaries. .

  The first to fourth divided image encoding devices 1020-1 to 1020-4 have the same configuration, and the configuration of each divided image encoding device 1020-i (i = 1 to 4) is shown in FIG. The divided image encoding device 1020-i (i = 1 to 4) is configured as a general MPEG2-MP @ ML encoder that does not require high performance.

  A divided image encoding device 1020-i (i = 1 to 4) illustrated in FIG. 11 includes a screen rearranging unit 1121, a subtractor 1122, a discrete cosine transform processing unit (DCT processing unit) 1123, a quantization unit 1124, a variable length An encoding unit 1125, a buffer 1126, a rate control unit 1127, an inverse quantization unit 1128, an inverse DCT processing unit 1129, an adder 1130, a frame memory 1131, a motion compensation prediction unit 1132, and a divided image encoding device internal control unit 1133. Have.

  A basic operation of the divided image encoding device 1020-i (i = 1 to 4) shown in FIG. 11 will be described.

  The divided image encoding device 1020-i (i = 1 to 4) converts each divided image signal for each picture sequentially input from the screen dividing device 1010 from 4: 2: 2 format in a format conversion unit (not shown). Convert to 4: 2: 0 format. Then, the screen rearrangement unit 1121 rearranges the image signals in accordance with the encoded picture type, and the DCT processing unit 1123 performs DCT processing in the encoding order.

  When the picture type is a P or B picture and the macroblock type is not an intra macroblock, the subtractor 1122 subtracts the signal of the motion compensation prediction unit 1132 from the signal rearranged by the screen rearrangement unit 1121 to detect a prediction error. To do. The DCT processing unit 1123 performs DCT processing on the prediction error and calculates a DCT coefficient. The quantization unit 1124 quantizes the D11T coefficient obtained by the DCT processing unit 1123, the variable length coding unit 1125 performs variable length coding of the quantization result together with the motion vector and the coding mode information, and the buffer 1126 stores the result. To do. When the picture type is I picture or P picture, it is necessary to use the picture later as a reference screen for motion compensation prediction. Therefore, in addition to the processing in the screen rearrangement unit 1121, the subtractor 1122, the DCT processing unit 1123, the quantization unit 1124, the variable length coding unit 1125, and the buffer 1126 described above, the following processing similar to the processing performed by the decoding device Is done.

  The inverse quantization unit 1128 inversely quantizes the quantization result and calculates a DCRT coefficient corresponding to the DCT coefficient before being input to the quantization unit 1124. The inverse DCT unit 1129 performs inverse DCT processing on the inverse quantization result and calculates a value corresponding to a prediction error before being input to the DCT processing unit 1123. The adder 1130 adds the result of the DCT processing unit 1123 and the result of the motion compensation prediction unit 1132 to calculate a signal corresponding to the output of the screen rearranging unit 1121 before being input to the subtractor 1122, and the addition result Is stored in the frame memory 1131. The motion compensation prediction unit 1132 predicts motion compensation for the image stored in the frame memory 1131, and the result is input to the adder 1130 and the subtractor 1122. The processing of each unit of the divided image encoding device A20-i (i = 1 to 4) described above is controlled by the control unit 1133 in the divided image encoding device A20-i (i = 1 to 4).

  In the divided image encoding apparatus 1020-i (i = 1 to 4) performing the above-described operation, the encoding macroblock type and the quantization scale code are determined by the following method.

  The divided image encoding device 1020i (i = 1 to 4) determines the degree of intra conversion for the encoded macroblock type according to the position from the division boundary when the input moving image signal VIDEO is divided. The macroblock that is in contact with the division boundary when the input moving image signal VIDEO is divided and the encoded macroblock type of the next macroblock are forcibly set to the intra macroblock. It should be noted that the macroblocks that are in contact with the division boundary can be expanded to a minimum of one macroblock and, if necessary, several macroblocks. For each macroblock in the direction away from the division boundary of the input moving image signal VIDEO, the intra macroblock is forced to have a predetermined distribution and a predetermined degree of intra-motion while sequentially decreasing the ratio (%) of forcibly setting the intra-macroblock. Go on. In this control, the same processing is performed on the division boundaries in the vertical and horizontal directions. This process is for a P picture, but the same process is performed for a B picture. However, since all macroblocks of the I picture are intra macroblocks, this processing is not substantially performed.

  Determination of the quantization scale code in the divided image encoding device 1020-i (i = 1 to 4) is performed by the control unit 1033 controlling each unit of the divided image encoding device 1020-i (i = 1 to 4). It is.

  The quantization unit 1124 in the divided image encoding device 1020-i (i = 1 to 4) includes a quantization scale code (Q scale code) or a quantization scale of a macroblock that sandwiches the division boundary with respect to the input moving image signal VIDEO. Lower (Q scale). The quantizing unit 1124 also sets each Q scale code or Q scale so that the Q scale code or Q scale to be lowered gradually decreases with increasing distance from the division boundary for macroblocks in columns far from the division boundary. Lower. On the other hand, in a macroblock that is some distance away from the division boundary, the quantization unit 1124 increases the Q scale code or Q scale.

  The video stream integration device 1030 synthesizes four MPEG2-MP @ ML video streams output from the divided image encoding devices 1020-1 to 1020-4 to generate an MPEG2-MP @ HL video stream. In 720 × 544, 30 frames / second, the number of samples slightly exceeds the upper limit of MP @ ML, but here it is written as MP @ ML for convenience. That is, the video stream integration device 1040 decomposes four MPEG2-MP @ ML video streams in units of slices and reconstructs them into one video stream, thereby converting one MPEG2-MP @ HL video stream. obtain.

The control device 1040 controls each component so that the moving image encoding device A00 performs the desired operation described above.
JP 2003-348597 A

  In the conventional system, one screen is uniformly divided by a predetermined method, and a plurality of image coding apparatuses independently code the divided images in parallel. For this reason, the image becomes discontinuous at the division boundary of the image, and as a result, the problem that the quality of the encoded image in the divided region is lowered is encountered. In order to solve this problem, in the conventional configuration implemented in Patent Document 1, pixel block discontinuity at the division boundary is achieved by macroblock type determination control at the division boundary and rate control by Q scale control at the division boundary. However, when the picture type is an I-picture, the above-described control is not substantially applied, and the problem that the quality of the encoded image at the division boundary is deteriorated again. Encounter.

  In recent years, the number of AV devices that employ MPEG4-AVC / H.264 coding has increased, and even in intra-screen predictive coding employed by this coding standard, A similar problem of reduced quality of the digitized image is encountered.

  FIG. 13 shows an example in which the input image is divided into four. In the intra-screen prediction process in MPEG4-AVC / H.264, the upper and left macroblocks adjacent to the encoding target macroblock perform the prediction process using the pixel information. For the macroblock located at the boundary of the frame, since there is no pixel outside the boundary, the prediction process is performed using only the portion where the pixel exists. Accordingly, since the macroblocks in the hatched portion of the divided image divided into four do not have adjacent macroblock information, the coding efficiency is lowered and the image quality is deteriorated. For this reason, when the decoded four-divided images are combined again, a division boundary appears.

  The present invention solves the above-described conventional problems. When a frame is divided and encoded for a high-definition moving image signal higher than an HDTV signal, the encoded image quality generated near the division boundary is reduced. It is an object of the present invention to provide a predictive coding apparatus that can perform coding without reducing or eliminating discontinuities and substantially reducing the image quality of the entire coded image.

The present invention has been made in consideration of the above points,
The predictive coding apparatus according to the first aspect of the present invention includes divided region inversion means. In this configuration, when the input frame signal is divided into N regions and input to N systems of encoding means, the screen is divided in consideration of the reference pixel region necessary for the intra prediction process, and the divided frames are divided. By flipping up, down, left and right, it is possible to improve the intra prediction encoding efficiency at the division boundary.

  The predictive coding apparatus according to the invention of claim 2 includes a prediction mode control means. In this configuration, it is possible to reduce the bit amount of the intra prediction mode information at the division boundary to be inserted into the encoded stream, and to improve the intra prediction encoding efficiency at the division boundary.

  A predictive coding apparatus according to a third aspect of the present invention includes divided image rotation means. In this configuration, intra prediction encoding from all directions can be performed, the optimal prediction direction of the divided image can be selected for improving the encoding efficiency, and the intra prediction encoding efficiency at the division boundary can be improved. It becomes possible.

  According to a fourth aspect of the present invention, there is provided a predictive coding apparatus comprising predictive reference region control means. In this configuration, by obtaining the reference pixel area necessary for intra prediction encoding from the macroblock information at the same position in the previous frame, the intra prediction encoding efficiency at the division boundary without encoding the reference pixel area. Can be improved.

  According to the predictive encoding device of the present invention, when an image is divided and encoded in parallel, intra-screen predictive encoding can be performed without impairing pixel discontinuity at the division boundary. Predictive coding efficiency can be improved.

  Embodiments of the present invention will be described below with reference to the drawings.

(Embodiment 1)
The embodiments of the present invention will be described with reference to FIGS.

  FIG. 1 is a block diagram of a predictive coding apparatus 100 according to Embodiment 1 of the present invention.

  An image signal input from an image input unit (not shown) is held in the frame buffer 170. In the present embodiment, the input image size will be described by taking a 4K2K image signal (4096 pixels × 2048 pixels) as an example. As shown in FIG. 1, the predictive encoding device 100 encodes an image signal input to the frame buffer 170. The image signal input to the buffer 170 is divided into N divided image signals by the screen dividing unit 110 and held in the frame buffer 150-i (i = 1 to 4). FIG. 4 shows a divided image divided into N divided images by the screen dividing means 110. In FIG. 4, the size of the divided images divided into N is divided into one macroblock size larger both vertically and horizontally, and each divided image 401-i (i = 1 to 4) is overlaid between adjacent divided images. Has an area to wrap. The divided divided images 401-i (i = 1 to 4) are divided by the divided region inversion means 160-i (i = 1 to 4) so that the overlapping regions between the adjacent divided images are positioned on the left and the upper. It is flipped up and down, left and right, or up and down and left and right.

  In FIG. 5, the divided images 401-i (i = 1 to 4) divided into N pieces by the screen dividing means 110 are divided into the divided images 5001 reversed by the divided area reversing means 160-i (i = 1 to 4). -i (i = 1 to 4). Next, the inverted divided images 501-i (i = 1 to 4) are encoded by the divided image encoding means 102-i (i = 1 to 4).

  FIG. 2 shows processing blocks of the divided image encoding means 102-i (i = 1 to 4).

  The divided image encoding means 102-i (i = 1 to 4) shown in FIG. 2 includes a subtractor 221, an orthogonal transformation unit 222, a quantization unit 223, a variable length coding unit 224, a buffer 225, a rate control unit 226, An inverse quantization unit 227, an inverse orthogonal transform unit 228, an adder 229, a deblocking filter unit 2210, a frame memory 2211, a motion compensation prediction unit 2212, an intra prediction unit 2213, and a prediction mode control unit 2214 are included.

  The basic operation of the divided image encoding means 102-i (i = 1 to 4) shown in FIG. 2 will be described. The divided image encoding unit 102-i (i = 1 to 4) is an orthogonal transform unit for each divided image signal for each picture sequentially input from the divided region inversion unit 160-i (i = 1 to 4). At 222, orthogonal transform processing is performed. When the picture type is a P-picture and the macroblock type is not an intra macroblock, the subtractor 2221 subtracts the image signal of the motion compensated prediction unit 2212 from the divided image signal to obtain a prediction error from the motion compensated prediction image. To detect. The orthogonal transform unit 222 calculates a transform coefficient by subjecting the prediction error to an orthogonal transform process. The quantization unit 223 quantizes the transform coefficient obtained by the orthogonal transform unit 222, the variable length coding unit 224 performs variable length coding of the quantization result together with the motion vector and coding mode information, and the buffer 225 stores the result. To do.

  When the picture types are I-picture and P-picture, it is necessary to use the picture later as a reference screen for motion compensation prediction. For this reason, the following processing similar to the processing performed by the decoding apparatus is performed in addition to the processing in the subtractor 221, the orthogonal transformation unit 222, the quantization unit 223, the variable length coding unit 224, and the buffer 225 described above. The inverse quantization unit 227 inversely quantizes the quantization result and calculates an orthogonal transform coefficient before being input to the quantization unit 223. The inverse orthogonal transform unit 228 performs an inverse orthogonal transform process on the inverse quantization result and calculates a value corresponding to the prediction error before being input to the orthogonal transform unit 222. If it is an inter-macroblock, the adder 229 adds the result of the orthogonal transform unit 222 and the result of the motion compensation prediction unit 2212 to calculate a signal corresponding to the output of the divided image signal before being input to the subtractor 221. The result is stored in the frame memory 2211. In the case of an intra macroblock, the adder 229 adds the result of the orthogonal transformation unit 222 and the result of the in-screen prediction unit 2213, and calculates a signal corresponding to the output of the divided image signal before being input to the subtractor 221. The addition result is stored in the frame memory 2211.

  If the image stored in the frame memory 2211 is an inter macroblock, motion compensation is predicted by the motion compensation prediction unit 2212, and the result is input to the adder 229 and the subtractor 221. If it is an intra macroblock, the intra prediction unit 2213 performs intra prediction with adjacent pixels, and the result is input to the adder 229 and the subtractor 221.

  The processing of the intra-screen prediction unit 2213 described above is controlled by the prediction mode control means 2214.

  The in-screen prediction process will be described with reference to FIGS. In FIG. 6, intra prediction of the encoding target macroblock MBx is performed using pixel information of the left adjacent macroblock MBa and the upper adjacent macroblock MBb of the encoding target macroblock MBx. Residual information of the encoding target MBx is generated by the prediction mode from eight directions. In FIG. 7, Block 1 indicates that the vertical prediction mode (= 0) and Block 2 indicates that the horizontal prediction mode (= 1) is performed. It is defined in the MPEG4-AVC / H.264 coding standard that the prediction mode of a macroblock at an image region boundary having no adjacent macroblock is determined to be a DC prediction mode (= 2). In the DC prediction mode, residual information is generated using an average value of existing adjacent pixels as a predicted value. In addition, in the DC prediction mode at the upper left image region boundary in which no adjacent pixel exists, residual information is generated with 128 as a predicted value. For this reason, the predictive coding efficiency around the image area has been reduced. However, in the present embodiment, the divided area inversion means 160-i (i = 1 to 4) is used to divide the macro area into a size larger by 1 macroblock size both vertically and horizontally and so that the extended area is positioned on the left and top of the divided image area. , Up and down, left and right, or up and down and left and right. Since the extended region can be used for intra prediction processing, in FIG. 6, the encoding mode other than DC prediction can be selected as the prediction mode for the left and upper blocks of the encoding target macroblock MBx. Efficiency can be improved. Also, when encoding the intra prediction mode, the prediction mode of the macroblock to be encoded is encoded based on the prediction mode of the macroblock adjacent to the left and above so that the information bits of the prediction mode are not generated as much as possible. Is done. When the prediction mode of the encoding target macroblock matches the minimum value of the prediction mode of neighboring neighboring macroblocks, no extra bits are generated. However, when a prediction mode larger than neighboring neighboring macroblocks occurs, 3-bit encoded information rem_intra4x4_pred_mode is inserted into the encoded stream. Therefore, the prediction mode control unit 2214 controls the in-screen prediction unit 2213 so that a prediction mode smaller than the DC prediction mode is generated for a block inside the boundary block in one macroblock region expanded when the image is divided. . The prediction modes smaller than the DC prediction mode are specifically 0: vertical prediction, 1: horizontal prediction, and 2: DC prediction.

  Next, in FIG. 3, the divided area rotating means F60-i (i = 1 to 4) will be described. FIG. 3 replaces the divided region inversion means 160-i (i = 1 to 4) in FIG. The above-described divided region inversion means 160-i (i = 1 to 4) inverts each divided image up and down and / or left and right so that the divided image region is positioned at the upper left. However, in some cases, encoding efficiency may be improved by selecting the encoding direction with the highest encoding efficiency after performing all prediction directions. Therefore, as shown in FIG. 8, the divided region rotating means 360-i (i = 1 to 4) can perform the intra prediction encoding by rotating the divided image region every 90 degrees.

  Next, the prediction reference area control means will be described with reference to FIG.

  The reference image area used for the intra-screen prediction process described above uses pixel information of different divided image areas in the same frame. Therefore, since it is necessary to reconstruct the reference pixel area at the time of decoding from the encoded data of the same frame, it is necessary to perform the encoding including the extended pixel area at the time of encoding. Therefore, by acquiring the prediction reference area from the macroblock information at the same position of the previous frame, that is, by using already encoded frame information, it is not necessary to encode an extra prediction reference area. In FIG. 9, a hatched area 901-1 is an encoding area that should be encoded originally, and the 901-2 area expanded by one macroblock is required for the intra prediction process. By acquiring the region from the 901-1 region of the previous frame (N-1), if only the shaded portion 9014-4 region is encoded, the reference pixel region at the time of decoding is the same as that of the previous frame that has already been decoded. Reconstructed pixel information can be used.

  In the above-described embodiment, H.P. However, the present invention is not limited to this, and the present invention is not limited to this. In the case where encoding processing is performed using various encoding schemes in which a plurality of pictures can be used as reference pictures. Widely applicable.

  Further, in the above-described embodiment, a case has been described in which each processing unit is configured by hardware and the hardware is configured as an encoding processing terminal. However, the present invention is not limited to this, and each processing unit may be configured by hardware, and the hardware may be configured as one LSI.

  In addition, it may be configured as a built-in program for controlling the arithmetic processing means of the dedicated digital signal processor in the LSI by software processing.

  In addition, it may be configured as a program for controlling general-purpose arithmetic processing means in the personal computer by software processing.

  In this case, the program of the arithmetic processing means may be provided by being recorded on various recording media such as an optical disk, a magnetic disk, a memory card, an instruction memory in an LSI, or provided via a network such as the Internet. You may make it do.

  The predictive coding apparatus according to the present invention is, for example, H.264. It is useful as an imaging system that performs encoding processing of high-definition images higher than HDTV using H.264 encoding.

The block diagram of the prediction encoding apparatus in embodiment of this invention The block diagram of the division | segmentation image encoding means in embodiment of this invention The block diagram of the prediction encoding apparatus in embodiment of this invention Schematic explaining the image divided | segmented by the image division | segmentation means in embodiment of this invention Schematic explaining the image reversed by the divided area reversing means in the embodiment of the present invention Schematic explaining the in-screen prediction process Schematic explaining the prediction mode of in-screen prediction processing Schematic explaining the image rotated by the divided area rotation means in the embodiment of the present invention Schematic explaining operation | movement of the reference area control means in embodiment of this invention Block diagram for explaining a conventional video encoding device Block diagram for explaining conventional divided image encoding means Schematic explaining an image divided by a conventional screen dividing device Schematic explaining macro flock information encoded by a conventional divided image encoding device

Explanation of symbols

DESCRIPTION OF SYMBOLS 100 Predictive encoding apparatus 110 Screen division | segmentation means 120 Division | segmentation image coding means 130 Encoding stream integration means 140 Prediction mode control means 150 Frame buffer 160 Division area inversion means 170 Prediction reference area control means 180 Frame buffer 190 Recording apparatus 221 Subtractor 222 Orthogonal transformation unit 223 Quantization unit 224 Variable length coding unit 225 Buffer 226 Rate control unit 227 Inverse quantization unit 228 Inverse orthogonal transformation unit 229 Adder 2210 Deblock filter unit 2211 Frame memory 2212 Motion compensation prediction unit 2213 Intra-screen prediction unit 2214 Prediction Mode Control Unit 360 Divided Area Rotating Unit 401 Divided Area Image 501 Divided Area Inverted Image 801 Divided Area Rotated Image 901 Encoding Target Divided Area 1000 Image Encoding Device 1010 Screen Dividing Device Device 1020 divided image encoding device 1030 encoded stream integration device 1040 control device 1050 video tape recording device 1121 screen rearrangement unit 1122 subtractor 1123 DCT unit 1124 quantization unit 1125 variable length encoding unit 1126 buffer 1127 rate control unit 1128 inverse Quantization unit 1129 Inverse DCT unit 1130 Adder 1131 Frame memory 1132 Motion compensation prediction unit 1133 Control unit

Claims (7)

Screen dividing means for dividing an input image into N regions and generating N divided image signals corresponding to the regions;
Divided region inversion means for inverting the generated N divided image signals vertically and / or horizontally;
N divided image encoding means for encoding each of the generated N divided inverted image signals to generate N encoded image signals;
A predictive code, comprising: an integration unit that integrates N encoded image signals generated by the plurality of divided image encoding units and generates one encoded image signal corresponding to the input image; Device. The prediction encoding apparatus according to claim 1, further comprising prediction mode control means for controlling a prediction mode of intra prediction processing in the divided image encoding means. The divided area inversion means is a divided area rotation means for rotating the generated N divided image signals in all directions every 90 degrees and inputting them to the divided image encoding means. The predictive encoding device according to 1 or 2. The predictive coding apparatus according to any one of claims 1 to 3, further comprising a predictive reference region control unit that controls a predictive reference region necessary for intra-screen prediction processing in the divided image encoding unit. A screen dividing step of dividing the input image into N regions and generating N divided image signals corresponding to the regions;
A divided region inversion step of inverting the generated N divided image signals vertically and / or horizontally;
A divided image encoding step of encoding each of the generated N divided inverted image signals to generate N encoded image signals;
A prediction code comprising: an integration step of integrating the N encoded image signals generated in the plurality of divided image encoding steps and generating one encoded image signal corresponding to the input image; Method. On the computer,
A screen dividing step of dividing the input image into N regions and generating N divided image signals corresponding to the regions;
A divided region inversion step of inverting the generated N divided image signals vertically and horizontally;
A divided image encoding step of encoding each of the generated N divided inverted image signals to generate N encoded image signals;
A prediction step, wherein the N encoded image signals generated in the plurality of divided image encoding steps are integrated, and an integrated step of generating one encoded image signal corresponding to the input image is executed. Encoding method program. A screen dividing step of dividing the input image into N regions and generating N divided image signals corresponding to the regions;
A divided region inversion step of inverting the generated N divided image signals vertically and horizontally;
A divided image encoding step of encoding each of the generated N divided inverted image signals to generate N encoded image signals;
A prediction code comprising: an integration step of integrating N encoded image signals generated in the plurality of divided image encoding steps and generating one encoded image signal corresponding to the input image; A recording medium on which a program for the conversion method is recorded.

Images