KR101369161B1 - Prediction Direction Change Method and Apparatus and Video Encoding/Decoding Method and Apparatus - Google Patents

Abstract

The present invention relates to an apparatus and method for predicting direction switching and an image encoding / decoding apparatus and method using the same.

The present invention is a device for switching the prediction direction when predicting the current block of the image for image coding, the quantization frequency coefficient of the previous block coded before the current block is stored if input, and if the current block is input of the previous block If the quantization frequency coefficients are all '0', the reference prediction direction is determined as the prediction direction of the current block, and when one or more coefficients of the quantization frequency coefficients of the previous block are not '0', the coding cost of the plurality of prediction direction candidates The present invention provides a prediction direction switching device, wherein the prediction direction having the minimum encoding cost is determined as the prediction direction of the current block.

According to the present invention, when encoding or decoding an image, the accuracy of prediction can be increased, and the encoding efficiency of the image can be improved.

Image, encoding, decoding, intra, prediction, rectangular, direction, transition

Description

Prediction Direction Change Method and Apparatus and Video Encoding / Decoding Method and Apparatus}

The present invention relates to an apparatus and method for predicting direction switching and an image encoding / decoding apparatus and method using the same. More specifically, the present invention relates to an apparatus and method for improving the encoding efficiency of an image by increasing the accuracy of prediction by predicting by differently predicting a prediction direction for each block to be encoded in encoding or decoding an image.

Moving Picture Experts Group (MPEG) and Video Coding Experts Group (VCEG) have developed superior video compression techniques superior to the existing MPEG-4 Part 2 and H.263 standards. This new standard is called H.264 / AVC (Advanced Video Coding) and is jointly announced as MPEG-4 Part 10 AVC and ITU-T Recommendation H.264. In this H.264 / AVC (abbreviated as 'H.264'), a spatial prediction coding method which is different from the international standard related to the conventional video coding such as MPEG-1, MPEG-2 and MPEG- Predictive Coding) method.

In the conventional moving image encoding method, "intra prediction" is used for the coefficient value transformed in the discrete cosine transform domain (DCT domain), thereby seeking to increase the coding efficiency. As a result, Resulting in deterioration in the subjective image quality of the transmission bit rate band. However, H.264 adopts a coding method based on Spatial Intra Prediction in a Spatial Domain instead of a Transform Domain.

An encoder using a coding method based on the existing spatial intraprediction predicts block information to be currently encoded from the information of a previous block that has been already encoded and reproduced and calculates a difference ) Information and transmits it to a decoder. At this time, parameters necessary for predicting the block information may be transmitted to the decoder, or the encoder and the decoder may be synchronized to share parameters required for prediction, so that the decoder may predict the block information. The decoder predicts the information of the neighboring block that has already been decoded and reproduced, obtains the sum of the error information transmitted from the encoder and the information of the predicted neighboring block, and generates and reproduces information of a desired current block to be decoded. In this case, if parameters necessary for prediction are also transmitted from the encoder, the decoder uses the parameter to predict information of neighboring blocks.

Meanwhile, the conventional spatial intra prediction based coding method uses only a square block mode. This is because two-dimensional square transformations such as 4x4 and 8x8 transforms are used to increase the conversion efficiency. However, when only the square block mode is used, the pixels in the right and lower parts of the block are predicted from pixels far away from space, so the accuracy of the block prediction is inferior.

In order to solve the above-described problems, the present invention, in encoding or decoding an image, is mainly used to improve the encoding efficiency of an image by increasing the accuracy of prediction by predicting by differently determining a method of determining a prediction direction for each block to be encoded. There is a purpose.

In order to achieve the above object, the present invention provides a device for encoding a current block of an image, wherein a rectangular encoded bitstream is encoded by stepwise encoding a plurality of rectangular current blocks generated by dividing a rectangular block by a unit when the current block is input. A rectangular encoder for outputting; A prediction direction switching unit for changing a prediction direction of each rectangular current block according to a coefficient value of a quantization frequency coefficient of a rectangular previous block coded before each rectangular current block coded stepwise; A square encoder for outputting a square encoded bitstream by encoding the current block in units of square blocks; And an encoding selector configured to calculate the encoding cost of the rectangular encoded bitstream and the encoding cost of the square encoded bitstream, and output a bitstream having a minimum encoding cost.

According to another object of the present invention, in the method of encoding a current block of an image, when the current block is input, a rectangular encoded bitstream is output by stepwise encoding a plurality of rectangular current blocks generated by dividing the rectangular block into units. A rectangular encoding step; A prediction direction switching step of changing a prediction direction of each rectangular current block according to a coefficient value of a quantization frequency coefficient of a rectangular previous block coded before each rectangular current block coded stepwise; A square encoding step of outputting a square encoded bitstream by encoding the current block in units of square blocks; And an encoding selection step of calculating the encoding cost of the rectangular encoded bitstream and the encoding cost of the square encoded bitstream to output a bitstream having the minimum encoding cost.

According to still another object of the present invention, there is provided an apparatus for decoding an image, comprising: a prediction mode extraction unit for extracting information about a prediction mode from a bitstream; A decoder which extracts a quantized frequency coefficient sequence by decoding the bitstream; A prediction direction switching unit which extracts a prediction direction identifier from the quantized frequency coefficient sequence and changes the prediction direction according to the prediction mode; A rectangular reconstruction unit for reconstructing and outputting the current block step by step according to the prediction direction changed in units of rectangular blocks using the quantized frequency coefficient sequence; A square restoring unit for restoring and outputting a current block according to a prediction direction according to a prediction mode in units of square blocks using a quantized frequency coefficient sequence; And a bitstream identifier extractor for extracting the bitstream identifier from the bitstream and controlling the decoder to output the quantized frequency coefficient sequence to one of a rectangular reconstructor and a square reconstructor according to the bitstream identifier. Provided is a decoding device.

According to still another object of the present invention, there is provided a method of decoding an image, comprising: a prediction mode extraction step of extracting information on a prediction mode from a bitstream; Decoding a bitstream to extract a quantized frequency coefficient sequence; A prediction direction switching step of extracting a prediction direction identifier from the quantized frequency coefficient sequence and changing the prediction direction according to the prediction mode; Extracting a bitstream identifier from the bitstream; A rectangular reconstruction step of reconstructing the current block in units of rectangular blocks by using the quantized frequency coefficient sequence according to the bitstream identifier, and reconstructing and outputting the current block step by step according to the changed prediction direction; And a square reconstruction step of reconstructing the current block in units of square blocks by using the quantized frequency coefficient sequence according to the bitstream identifier, and reconstructing and outputting the current block according to the prediction direction according to the prediction mode. Provide a method.

Further, according to another object of the present invention, when predicting the current block of the image for image encoding, in the apparatus for switching the prediction direction, if the quantization frequency coefficient of the previous block coded before the current block is input and stored, When the current block is input, when the quantization frequency coefficients of the previous block are all '0', the reference prediction direction is determined as the prediction direction of the current block, and when one or more coefficients of the quantization frequency coefficients of the previous block are not '0', Provided is a prediction direction switching device, characterized in that the prediction direction of the minimum block is determined as the prediction direction of the current block by calculating the encoding cost of the plurality of prediction direction candidates.

Further, according to another object of the present invention, when predicting the current block of the image for image decoding, in the apparatus for switching the prediction direction, if the quantization frequency coefficient of the previous block decoded before the current block is stored, When the quantization frequency coefficients of the current block are input, when the quantization frequency coefficients of the previous block are all '0', the reference prediction direction is determined as the prediction direction of the current block, and at least one coefficient of the quantization frequency coefficients of the previous block is '0'. Otherwise, the prediction direction switching apparatus is provided, wherein the prediction direction identifier is extracted from the quantization frequency coefficient of the current block to determine the prediction direction indicated by the prediction direction identifier as the prediction direction of the current block.

As described above, according to the present invention, by predicting by differently determining a prediction direction for each block to be encoded, it is possible to increase the accuracy of prediction and to improve the encoding efficiency of an image.

Hereinafter, some embodiments of the present invention will be described in detail with reference to exemplary drawings. It should be noted that, in adding reference numerals to the constituent elements of the drawings, the same constituent elements are denoted by the same reference symbols as possible even if they are shown in different drawings. In addition, in describing the present invention, when it is determined that a detailed description of a related known configuration or function may obscure the gist of the present invention, the detailed description thereof will be omitted.

In describing the components of the present invention, terms such as first, second, A, B, (a), and (b) may be used. These terms are intended to distinguish the constituent elements from other constituent elements, and the terms do not limit the nature, order or order of the constituent elements. When a component is described as being "connected" or "connected" to another component, the component may be directly connected to or connected to that other component, but another component is "connected" between each component. It will be understood that "or" may be connected.

1 is an exemplary diagram showing nine conventional 4x4 intra prediction modes.

Intra prediction includes intra 4x4 prediction, intra 16x16 prediction, and intra 8x8 prediction, and each intra prediction includes a plurality of prediction modes. 1 illustrates nine prediction modes in intra 4x4 prediction.

Referring to FIG. 1, intra 4x4 prediction includes vertical mode, horizontal mode, direct current mode, diagonal down-left mode, diagonal down-right, and vertical. There are nine prediction modes, including the right-right mode, horizontal-down mode, vertical-left and horizontal-up modes.

Although not shown, intra 8x8 prediction has the same prediction mode as intra 4x4 prediction, and intra 16x16 prediction includes vertical mode, horizontal mode, direct current mode, and plane mode. There are four prediction modes.

2 is a block diagram schematically illustrating an electronic configuration of a video encoding apparatus according to an embodiment of the present invention.

The image encoding apparatus 200 according to an embodiment of the present invention is an apparatus for encoding an image, and may include a square encoder 210, a rectangular encoder 220, and an encoding selector 230.

The image encoding apparatus 200 may be a personal computer (PC), a notebook computer, a personal digital assistant (PDA), a portable multimedia player (PMP), a PlayStation Portable ), A mobile communication terminal, and the like, and may be a communication device such as a communication modem for performing communication with various devices or wired / wireless communication networks, a memory for storing various programs for encoding an image and data, a program And a microprocessor for calculating and controlling the microprocessor.

When the current block of the input image is input, the square encoder 210 outputs a square encoded bitstream by encoding the data in a square block unit. That is, as in normal video encoding, a bitstream is output by predicting a current block in units of square blocks according to a predetermined block mode, generating a residual block for the current block, and transforming, quantizing, and encoding the residual block. A bitstream encoded and output in block units is called a 'square encoded bitstream' in the present invention.

When the current block of the input image is input, the rectangular encoder 220 outputs a rectangular encoded bitstream by dividing the data into rectangular blocks in steps and encoding them in stages. That is, unlike conventional encoding, the current block according to the predetermined block mode is divided into rectangular block units according to the prediction direction of the current block, each rectangular block is predicted step by step, the residual block is transformed, quantized, and then squared into a square block. The bitstream is output by combining, scanning, and encoding the bitstream. The bitstream, which is encoded and output step by step in rectangular block units, is referred to as a rectangular encoded bitstream in the present invention.

The encoding selector 230 calculates an encoding cost of the rectangular encoded bitstream and an encoding cost of the square encoded bitstream and outputs a bitstream having a minimum encoding cost. That is, the encoding selector 230 calculates the cost of encoding the square encoded bitstream output from the square encoder 210 and encodes the rectangular encoded bitstream output from the rectangular encoder 220. The bitstreams having the minimum encoding cost are output by calculating the cost accordingly and comparing the respective costs.

3 is a block diagram schematically illustrating an electronic configuration of a square encoding apparatus according to an embodiment of the present invention.

Since the square encoding apparatus according to an embodiment of the present invention may be implemented by the square encoder 210 in FIG. 2, hereinafter, the square encoder 210 is referred to as a square encoder 210. The square encoder 210 according to an embodiment of the present invention includes a square subtractor 310, a square converter 320, a square quantizer 330, a square scan unit 340, a square encoder 350, A square inverse quantizer 360, a square inverse transformer 370, a square adder 380, and a square predictor 390 are included.

The square subtractor 310 generates a residual block by subtracting the prediction block predicted by the square predictor 380 from the current block of the input image. That is, the subtractor 310 subtracts the original pixel value of each pixel of the current block and the predicted pixel value of each pixel of the prediction block predicted by the square predictor 380. A residual signal of the difference between pixel values is generated as a residual block.

The square converter 320 converts the residual block into a frequency domain to generate a residual block having a frequency coefficient. Here, the square transform unit 320 may use a Discrete Cosine Transform (DCT) -based transform or a Hadamard Transform (DCT). However, the square transform unit 320 is not necessarily limited thereto. The residual signal can be converted into the frequency domain by using.

The square quantizer 330 quantizes the residual block transformed by the square transformer 320 to generate a residual block having a quantization frequency coefficient. Dead zone uniform threshold quantization (DZUTQ) or quantization weighted matrix (DZUTQ) may be used as the quantization method, but various quantization methods such as quantization improved therefrom may be used.

The square scan unit 340 generates a quantized frequency coefficient sequence by scanning the quantized frequency coefficients of the residual block quantized by the square quantizer 330 using various scan methods such as zigzag scan.

The square encoder 350 encodes the quantized residual block into a bitstream. That is, the square encoder 350 encodes the quantized frequency coefficient string generated by the square scan unit 340 to generate a bitstream. In addition, the square encoder 350 may encode not only the quantized residual block but also information on the prediction mode or the prediction direction determined by the square predictor 390 together with the residual block in a bitstream. As such an encoding technique, entropy encoding technology may be used, but various other encoding techniques may be used without being limited thereto. The bit stream output from the square encoder 350 is called a square encoded bit stream in the present invention.

The square inverse quantizer 360 inverse quantizes the quantized residual block to generate an inverse quantized residual block. That is, the square inverse quantizer 360 inverse quantizes the quantized frequency coefficients of the quantized residual block to generate a residual block having inverse quantized frequency coefficients.

The square inverse transform unit 370 generates an inverse transformed residual block by performing an inverse transform on the inverse quantized residual block. That is, the square inverse transform unit 370 inversely transforms the inverse quantization frequency coefficients of the inverse quantized residual block into the time domain to generate an inverse transformed residual block having pixel values.

The square adder 380 reconstructs the current block by adding the prediction block predicted by the square predictor 390 and the residual block inversely transformed by the square inverse transform unit 370, and restores the restored current block to the square predictor. Output to (390).

The square predictor 390 generates a prediction block by predicting the current block. That is, the square predictor 390 predicts a pixel value of each pixel of a current block to be encoded in an image according to a predetermined block mode and a prediction mode, thereby predicting a pixel value of each pixel. Generate a predictive block with a value. In this case, each pixel value of a previous block that has been previously encoded, decoded, and reconstructed is used to predict each pixel value of the current block, and a reconstructed block received from the square adder 380 before the current block is used.

4 is a block diagram schematically illustrating an electronic configuration of a rectangular encoding apparatus according to an embodiment of the present invention.

Since the rectangular encoding apparatus according to an embodiment of the present invention may be implemented by the rectangular encoder 220 in FIG. 2, the rectangular encoder 220 will be referred to below. The rectangular encoder 220 according to an embodiment of the present invention includes a block divider 410, a rectangular subtractor 420, a rectangular transform unit 430, a rectangular quantizer 440, a rectangular scan unit 450, The rectangular encoder 460 includes a rectangular inverse quantizer 470, a rectangular inverse transform unit 480, a rectangular adder 490, and a rectangular predictor 492.

The block dividing unit 410 outputs a plurality of rectangular current blocks by dividing the current block of the input image by rectangular blocks. Here, the current block is determined to be a square block according to a block mode and an encoding mode. The block dividing unit 410 predicts a pixel of the current block from a neighboring pixel closer in intra prediction to increase the efficiency of prediction. Is divided into rectangular blocks, and each divided rectangular current block can be predicted step by step.

Also, in dividing a square current block into a plurality of rectangular blocks, the block dividing unit 410 may determine a rectangular block size according to a prediction direction according to a prediction mode of the current block. That is, the rectangular current blocks divided by the block divider 410 may be Nx1 blocks, Nx2 blocks N / 2, 1xN blocks N, 2xN blocks N / 2, and the like. For example, assuming that the block mode of the current block is an NxN block, if the prediction direction according to the prediction mode is vertical, the current block is divided into Nx1 blocks, N, etc., and when the prediction direction is horizontal, N 1xN blocks The prediction mode may be divided into N / M blocks, such as 1x1 blocks and 4x4 blocks, when the prediction mode is a DC mode.

The rectangular subtractor 420 subtracts a plurality of rectangular prediction blocks predicted by the rectangular predictor 492 from the plurality of rectangular current blocks divided by the block divider 410 to generate a plurality of rectangular residual blocks.

The rectangular transform unit 430 converts the plurality of rectangular residual blocks generated by the subtractor 420 into the frequency domain. That is, the rectangular transform unit 430 generates a residual block having a frequency coefficient by converting each pixel value of the rectangular residual block into a frequency domain by using a DCT-based transform or the like. In this case, the frequency conversion of each pixel value of the rectangular residual block may be performed by matrix transformation or the like.

The rectangular quantization unit 440 quantizes the plurality of rectangular residual blocks transformed by the rectangular transform unit 430. That is, the rectangular quantization unit 440 quantizes frequency coefficients of the plurality of rectangular residual blocks to generate a residual block having quantized frequency coefficients.

The rectangular scan unit 450 generates a quantized square residual block by combining the plurality of residual blocks quantized by the rectangular quantization unit 440, and generates a quantized frequency coefficient sequence by scanning a quantized frequency coefficient. That is, the rectangular scan unit 450 combines the plurality of quantized residual blocks according to the prediction direction of the block mode and the prediction mode of the current block to generate a square residual block, and each quantization frequency of the square residual block. The coefficients are scanned according to a scanning method to generate a quantized frequency coefficient sequence.

The rectangular encoder 460 generates a bitstream by encoding the quantized frequency coefficient sequence generated by the rectangular scan unit 450. That is, the rectangular encoder 460 generates a bitstream by encoding the quantized frequency coefficient sequences by using various encoding methods such as entropy encoding. In this case, the rectangular encoder 460 may further encode a prediction mode of the current block.

As described above, when the rectangular encoding apparatus described above encodes the current block of the input image, the rectangular encoding apparatus outputs a plurality of rectangular current blocks by dividing the current block of the input image into rectangular block units. Predicting a plurality of rectangular current blocks stepwise to output a plurality of rectangular prediction blocks, subtracting a plurality of rectangular prediction blocks from the plurality of rectangular current blocks to generate a plurality of rectangular residual blocks, and generating a plurality of rectangular residual blocks Convert to the frequency domain and quantize the transformed rectangular residual blocks, combine the quantized rectangular residual blocks to generate a quantized square residual block, and generate quantized frequency coefficient sequences by scanning the quantized frequency coefficients of the square residual block Quantization frequency meter Encoding the heat to generate a bitstream.

5 is a block diagram schematically illustrating an electronic configuration of an encoding selection apparatus according to an embodiment of the present invention.

Since the encoding selecting apparatus according to the embodiment of the present invention may be implemented by the encoding selecting unit 230 in FIG. 2, the encoding selecting unit 230 will be referred to below. The encoding selector 230 according to an embodiment of the present invention includes a bitstream selector 510 and a bitstream identifier inserter 520.

When the bitstream selector 510 receives the rectangular encoded bitstream from the rectangular encoder 460 and receives the square encoded bitstream from the square encoder 350, the bitstream selector 510 calculates the rectangular encoding efficiency and the square encoding efficiency to obtain a larger encoding. Output a bitstream according to encoding with efficiency. In this case, the encoding efficiency may be calculated as an encoding cost, and an encoding having a minimum encoding cost may be determined as an encoding having a higher encoding efficiency.

The bitstream identifier inserter 520 generates an identifier for identifying the bitstream output by the bitstream selector 510 and inserts the identifier into the bitstream output from the bitstream selector 510.

6 is a flowchart illustrating an image encoding method according to an embodiment of the present invention.

The image encoding apparatus 200 generates a square encoded bitstream by encoding the current block of the input image to be encoded in a square block unit as it is (S610), and divides the current block into rectangular block units to encode the rectangular encoded bit stepwise. A stream is generated (S620).

When the square bitstream and the rectangular bitstream are generated, the image encoding apparatus 200 calculates the encoding efficiency of the square encoded bitstream and calculates the encoding efficiency of the rectangular encoded bitstream to output an encoded bitstream having high encoding efficiency (S630). ). Here, the encoding efficiency may be an encoding cost for encoding each encoded bitstream, and when the encoding cost is small, the encoding efficiency may be determined to be large. In addition, the image encoding apparatus 200 identifies whether the corresponding bitstream is a bitstream encoded using square encoding or a bitstream encoded using rectangular encoding while outputting a bitstream having high encoding efficiency. The bitstream identifier may be inserted into the bitstream and output.

7 is a flowchart illustrating a bitstream selection process according to an embodiment of the present invention.

In the image encoding method according to the embodiment of the present invention described above with reference to FIG. 6, the step of outputting a bitstream having high encoding efficiency as in step S630 may be performed through the process as illustrated in FIG. 7.

The encoding selector 230 of the image encoding apparatus 200 calculates the encoding cost of the square encoded bitstream and the encoding cost of the rectangular encoded bitstream in order to calculate the encoding efficiency of the square encoded bitstream and the rectangular encoded bitstream ( S710). Here, the coding cost may be a rate-distortion cost (RDcost), but other coding costs may be used.

The video encoding apparatus 200 having calculated the encoding cost compares the encoding cost of the rectangular encoding cost and the encoding cost of the square encoding bitstream (S720), so that the encoding cost of the rectangular encoding bitstream is smaller than that of the square encoding bitstream. In the case where the rectangular encoded bitstream is output (S730).

In addition, the image encoding apparatus 200 determines whether all of the quantization frequency coefficients of the quantized frequency coefficient sequence encoded by the rectangular encoded bitstream are all '0' (S740), so that at least one or more of the quantization frequency coefficients is not '0'. In this case, the bitstream identifier may be generated as '1' and the bitstream identifier is output together with the rectangular encoded bitstream (S750). If the quantization frequency coefficients are all '0', the bitstream identifier may not be generated and output. . That is, when the quantization frequency coefficients of the corresponding bitstream are all '0', since the video decoding apparatus can decode the rectangular coded bitstream and the square coded bitstream by square block units through normal decoding, the bitstream identifier can be decoded. Does not output, thereby saving bits for the bitstream identifier.

In addition, when the encoding cost of the rectangular encoded bit stream is greater than or equal to the encoding cost of the square encoded bit stream, the image encoding apparatus 200 outputs the square encoded bit stream (S760). In addition, it is checked whether the quantization frequency coefficients of the quantized frequency coefficient strings encoded with the square encoded bitstream are all '0' (S770). When at least one or more of the quantization frequency coefficients is not '0', the bitstream identifier is' It generates as 0 'and outputs the bitstream identifier together with the square encoded bitstream (S780). If the quantization frequency coefficients are all' 0 ', the bitstream identifier may not be generated and output. That is, when the quantization frequency coefficients of the corresponding bitstream are all '0', since the video decoding apparatus can decode the rectangular coded bitstream and the square coded bitstream by square block units through normal decoding, the bitstream identifier can be decoded. Does not output, thereby saving bits for the bitstream identifier.

8 is an exemplary view showing a rectangular current block according to an embodiment of the present invention.

In FIG. 8, it is assumed that the block mode of the current block is a 4x4 block mode, and the rectangular current block generated by dividing the current block into rectangular block units according to the prediction mode is exemplarily illustrated.

When the prediction mode of the current block is the horizontal mode, the rectangular current block is divided into a rectangular current block of size 1x4 as shown, and when the prediction mode of the current block is the vertical mode, the rectangular current block is divided into a rectangular current block of size 4x1. When the prediction mode of the block is the diagonal right mode, the block is divided into rectangular current blocks having a size of 4x1 similar to the vertical mode.

In FIG. 8, P ₀ ^* (m) (m = 0, 1, 2, 3) in each rectangular current block represents a prediction pixel predicted by the rectangular prediction unit 492, and P _n (m) (m = 0, 1, 2, 3, n = 0, 1, ..., N) represent pixels encoded before the rectangular current block according to the prediction direction.

When the rectangular prediction unit 492 generates a rectangular prediction block by predicting the rectangular current block, the rectangular prediction unit 492 may calculate a predicted pixel value of each pixel of the rectangular prediction block as shown in Equation (1).

Where _n is the pixel weight that can be adjusted by the user.

9 is an exemplary diagram for describing a process of performing a stepped intra prediction by the rectangular prediction unit according to an embodiment of the present invention.

In FIG. 9, when N = 1, α ₁ = 1, and the prediction mode is a horizontal mode, a process of predicting stepwise by dividing a 4 × 4 current block into a plurality of rectangular current blocks is shown.

As shown in FIG. 9, in the first step, pixels P11, P21, and P31 of the first rectangular current block of the current block using pixels A ', B', C ', and D' of the previously encoded rectangular reconstructed block. , P41 is predicted, and a rectangular reconstruction block is generated by encoding and reconstructing a rectangular residual block having the residual signal. The rectangular recovery block has pixels of P11 ', P21', P31 ', and P41'.

In the second step, pixels P12, P22, P32, and P42, which are the second rectangular current blocks of the current block, are predicted using the pixels P11 ', P21', P31 ', and P41' of the previously encoded rectangular block. A rectangular reconstruction block is generated by encoding and reconstructing a rectangular residual block having the residual signal. The rectangular recovery block has pixels of P12 ', P22', P32 ', and P42'.

In the third step, pixels P13, P23, P33, and P43, which are the third rectangular current block of the current block, are predicted using the pixels P12 ', P22', P32 ', and P42 of the previously encoded rectangular block. A rectangular reconstruction block is generated by encoding and reconstructing a rectangular residual block having the residual signal. The rectangular recovery block has pixels of P13 ', P23', P33 ', and P43'.

In the fourth step, pixels P14, P24, P34, and P44, which are the fourth rectangular current block of the current block, are predicted using the pixels P13 ', P23', P33 ', and P43' of the previously encoded rectangular block. A rectangular reconstruction block is generated by encoding and reconstructing a rectangular residual block having the residual signal. The rectangular recovery block has pixels of P14 ', P24', P34 ', and P44'.

As described above, the rectangular prediction unit 492 predicts a plurality of rectangular current blocks of the current block in stages, wherein stepwise prediction does not predict all rectangular current blocks at the same time but instead encodes them when predicting each rectangular current block. Is predicted using the pixels of the previous rectangular block.

10 is a block diagram schematically illustrating an electronic configuration of a rectangular scan apparatus according to an embodiment of the present invention.

The rectangular scan device according to an embodiment of the present invention may be implemented as a rectangular scan unit 450 in FIG. 4. Hereinafter, the rectangular scan device according to an embodiment of the present invention is called a rectangular scan unit 450.

The rectangular scan unit 450 according to an embodiment of the present invention may include a block combiner 1010, an initial scanning pattern determiner 1020, a scanning unit 1030, and a scanning pattern updater 1040. have.

The block combiner 1010 combines the plurality of quantized rectangular residual blocks transmitted from the rectangular quantization unit 440 according to a prediction direction according to the prediction mode of the current block to generate a square quantized square residual block. For example, the block combiner 1010 may combine the plurality of quantized residual blocks to match a block size according to a corresponding block mode and according to a prediction direction of a prediction mode, to generate a residual block having a square shape. That is, if the plurality of quantized residual blocks are four 4x1 blocks, the four 4x1 blocks may be sequentially combined, but may be combined in a vertical direction, which is a prediction direction, to generate a 4x4 block sized quantized residual block.

The initial scanning pattern determiner 1020 determines an initial scanning pattern for scanning the quantized frequency coefficients of the quantized square residual blocks combined by the block combiner 1010. In this case, the initial scanning pattern determiner 1020 determines the initial scanning pattern according to the size of the image. For example, when the size of the image is greater than or equal to the preset size, the initial scanning pattern determiner 1020 gives priority to the coefficient of the DC component among the quantization frequency coefficients of the square residual block and the lowest coefficient of the square residual block. In other words, the initial scanning pattern may be determined by giving a higher priority to the coefficient of the DC component. In addition, when the size of the image is smaller than the preset size, the initial scanning pattern determiner 1020 gives the same priority to the DC component coefficient and the lowest coefficient of the square residual block among the quantization frequency coefficients of the square residual block. Can be determined.

The scanning unit 1030 is a quantization frequency of the residual block quantized according to the determined scanning pattern (initial scanning pattern determined by the initial scanning pattern determiner 1020 or updated scanning pattern determined by the scanning pattern updater 104). The coefficients are scanned to produce a quantized frequency coefficient sequence.

The scanning pattern updater 1040 may adaptively update the scanning pattern according to a probability that a quantization frequency coefficient other than '0' occurs at each position of the square residual block. For example, each time the image encoding apparatus 200 encodes a current block, the scanning pattern updater 1040 calculates a probability that a quantization frequency coefficient other than '0' occurs at each position of the square residual block and has a high probability. By determining the scanning order in order, it is possible to adaptively update the scanning pattern.

11 is a flowchart illustrating a method of determining an initial scanning pattern according to an embodiment of the present invention.

The rectangular scan unit 450 generates a square residual block by combining a plurality of quantized rectangular residual blocks in a block mode unit (S1110) and checks whether the size of the image to which the current block belongs is larger than a preset size (S1120). ), If the size of the image is larger than the preset size, the DC component and the lowest coefficient of the square residual block are given priority among the quantization frequency coefficients of the square residual block, but the higher priority is given to the coefficient of the DC component. When the scanning pattern is determined (S1130), and the image size is smaller than the preset size, the initial scanning pattern is applied by giving the same priority to the DC component coefficient and the lowest coefficient of the square residual block among the quantization frequency coefficients of the square residual block. Determine (S1140).

12 is an exemplary view illustrating a plurality of rectangular residual blocks combined into square residual blocks according to a preferred embodiment of the present invention.

12 illustrates quantization frequency coefficients of a square residual block generated by combining a plurality of rectangular residual blocks in units of a 4x4 block mode or an 8x8 block mode when the prediction mode of the current block is the vertical mode.

A square residual block combined in 4x4 block mode is a rectangular residual block of 4x1 size combined in a vertical direction, and a square residual block combined in 8x8 block mode is an 8x1 rectangular residual block combined in a vertical direction.

Since the current block is divided into a plurality of rectangular current blocks to transform and quantize the residual blocks of each rectangular current block, the quantized frequency coefficient of the DC component is positioned to the left of each quantized rectangular residual block. Therefore, the coefficient of the DC component is driven to the left of the quantized square residual block combined in the square block unit.

13 is an exemplary view for explaining a process of determining an initial scanning pattern according to an embodiment of the present invention.

In a square residual block, a non-zero value frequently occurs in the coefficients of the DC component and the coefficients of the lowest part of the residual block. Accordingly, if scanning is performed first, the coefficients of the DC component and the coefficients of the lowermost portion having a high probability of occurrence of a value other than '0' may be improved.

Meanwhile, the probability that a non-zero value occurs in the quantization frequency coefficients of a square residual block varies according to the size of the image. When the size of the image increases, the probability that a non-zero value occurs in the coefficient of the DC component is lowest. Is higher than the coefficients of. Therefore, in the present invention, the initial scanning pattern is determined according to the size of the image.

For example, if a threshold size for comparing the image size is set in advance, and the image size is larger than or equal to the preset size, the DC component coefficient and the lowest coefficient of the square residual block have priority over the quantization frequency coefficients of the square residual block. An initial scanning pattern can be determined by ranking, but giving a higher priority to the coefficients of the DC component. In addition, when the size of the image is smaller than the preset size, the initial scanning pattern may be determined by giving the same priority to the DC component coefficient and the lowest coefficient of the square residual block among the quantization frequency coefficients of the square residual block.

FIG. 13 exemplarily illustrates an initial scanning pattern determined when the preset size for comparing the sizes of images is set to 720p. When the size of the image is QCIF or CIF, it may be determined that the size of the image is smaller than the preset size, and the initial scanning pattern may be determined by giving the same priority to the coefficient of the DC component and the lowest coefficient of the square residual block. In this case, starting from DC 3, AC ₁ (3), DC (2), DC (1), AC ₁ (2), AC ₁ (3),... , The initial scanning pattern is determined in the order of AC ₃ (0).

In addition, when the size of the image is 720p or 1080p, it may be determined that the size of the image is greater than or equal to the preset size, and the initial scanning pattern may be determined by giving a higher priority to the coefficient of the DC component. In this case, as illustrated arrow order, DC (3), DC ( 2), DC (1), DC (0), AC 1 (3), AC 1 (2), ... , The initial scanning pattern is determined in the order of AC ₃ (0).

14 is an exemplary view for explaining a process of updating a scanning pattern according to an embodiment of the present invention.

For example, as described above with reference to FIG. 13, when the initial scanning pattern is determined, the rectangular scan unit 450 scans the quantized frequency coefficients of the quantized square residual block according to the initial scanning pattern to generate a quantized frequency coefficient sequence and is rectangular. The encoder 460 encodes the quantized frequency coefficient sequence to generate a rectangular encoded bitstream. Then, when the next block after the current block is input, a plurality of rectangular next blocks are divided again, the residual blocks are transformed and quantized, the square residual blocks are combined into a square residual block, and then scanned. Instead of scanning by using the initial scanning pattern determined in Fig. 2, the scanning pattern is updated and scanned adaptively.

In this case, a criterion for updating the scanning pattern may be a probability that a quantization frequency coefficient other than '0' occurs for each position of the square residual block. For example, each time a plurality of blocks of an image are encoded, a scanning pattern is calculated by calculating a probability that a non-zero quantization frequency coefficient occurs at each position of a square residual block and determining the scanning order in the order of the greatest probability. Can be adaptively updated.

As shown in FIG. 14, when the initial scanning pattern is determined and the encoding is completed by scanning the corresponding square residual block, the process of encoding the next block is repeated, and each quantization frequency coefficient of each quantization frequency coefficient is repeated at least once. Calculate the probability of generating a non-zero quantized frequency coefficient for each position. For example, as illustrated, a probability distribution in which quantization frequency coefficients other than '0' may be generated at each position may be calculated. In this case, the scanning patterns may be sequentially determined and updated in order of high probability according to the probability distribution. The scanning pattern can be determined.

As described above, the image encoded in the bitstream by the video encoding apparatus 200 is real-time or non-real-time through the wired or wireless communication network such as the Internet, local area wireless communication network, wireless LAN network, WiBro network, mobile communication network or the like, The image decoding apparatus may be transmitted to a video decoding apparatus to be described later through a communication interface such as a universal serial bus (USB), decoded by the video decoding apparatus, and restored and reproduced.

15 is a block diagram schematically illustrating an electronic configuration of an image decoding apparatus according to an embodiment of the present invention.

The image decoding apparatus 1500 according to an embodiment of the present invention may include a prediction mode extractor 1510, a decoder 1520, a bitstream identifier extractor 1530, a rectangular decompressor 1540, and a square decompressor 1550. It may be configured to include).

The video decoding apparatus 1500 may be a personal computer (PC), a notebook computer, a personal digital assistant (PDA), a portable multimedia player (PMP), or a PlayStation Portable (PSP). ), A communication device such as a communication modem for communicating with various devices or a wired / wireless communication network, a memory for storing various programs and data for decoding an image, and executing a program. Means a variety of devices including a microprocessor for operation and control.

The prediction mode extractor 1510 extracts information about the prediction mode from the bitstream. Here, the information about the prediction mode is information for identifying the prediction mode of the current block determined by the image encoding apparatus 200 and includes information on whether the prediction mode of the current block is a horizontal mode, a vertical mode, a DC mode, or the like. do.

The decoder 1520 extracts a quantized frequency coefficient string by decoding the bitstream. When the decoder 1520 extracts and outputs the quantized frequency coefficient sequence, the quantized frequency coefficient sequence is output to one of the rectangular reconstructor 1540 and the square reconstructor 1550 under the control of the bitstream identifier extractor 1530. do.

The bitstream identifier extractor 1530 extracts the bitstream identifier from the bitstream using the quantized frequency coefficient sequence, and causes the decoder 1520 to decompose the quantized frequency coefficient sequence from the rectangular recovery unit 1540 according to the bitstream identifier. Control to output to one of the square recovery unit 1550.

For example, the bitstream identifier extractor 1530 extracts the bitstream identifier from the bitstream when at least one of the quantization frequency coefficients of the quantization frequency coefficient sequence transmitted from the decoder 1520 is not '0'. If the extracted bitstream identifier is '1', the bitstream identifier extractor 1530 restores the rectangular quantization frequency coefficient sequence from which the decoder 1520 decodes the bitstream since the bitstream is a rectangular encoded bitstream. If the bitstream identifier is '0' and the bitstream is a square encoded bitstream, the decoder 1520 outputs the quantized frequency coefficient sequence from which the bitstream is decoded to the square reconstructor 1550. To control. In addition, the bitstream identifier extractor 1530 does not extract the bitstream identifier from the bitstream when the quantization frequency coefficients of the quantization frequency coefficient sequence output from the decoder 1520 are all '0'. Control to output the quantized frequency coefficient sequence to the square recovery unit 1550.

The rectangular reconstructor 1540 reconstructs and outputs the current block of the image in steps of a quantized frequency coefficient sequence output from the decoder 1520 in the rectangular block unit according to the prediction mode. That is, the rectangular reconstruction unit 1540 generates a quantized residual block by inverse scanning when the extracted quantized frequency coefficient is input by decoding the bitstream, and inversely quantizes the plurality of quantized rectangular residual blocks by dividing into rectangular block units. Inverse transforming generates a plurality of rectangular prediction blocks by estimating a plurality of rectangular current blocks corresponding to a plurality of inversely transformed rectangular residual blocks, and adds a plurality of inversely transformed rectangular residual blocks to the plurality of rectangular prediction blocks. Then, a plurality of rectangular current blocks are restored, and each rectangular current block is combined to restore the current block of the image.

The square reconstructor 1550 reconstructs and outputs the current block according to the prediction mode in units of square blocks by using the quantized frequency coefficient sequence output from the decoder 1520.

16 is a block diagram schematically illustrating an electronic configuration of a rectangular decoding apparatus according to an embodiment of the present invention.

The rectangular decoding apparatus according to an embodiment of the present invention may be implemented by the rectangular reconstruction unit 1540 in FIG. 15. Hereinafter, the rectangular decoding apparatus according to an embodiment of the present invention is called a rectangular reconstruction unit 1540.

According to an exemplary embodiment, the rectangular reconstructor 1540 includes a rectangular inverse scan unit 1610, a rectangular inverse quantizer 1620, a rectangular inverse transform unit 1630, a rectangular predictor 1640, and a rectangular adder ( 1650).

When the quantized frequency coefficient extracted by decoding the bitstream is input, the rectangular inverse scan unit 1610 generates quantized residual blocks by inverse scanning, divides them into rectangular block units, and outputs a plurality of quantized rectangular residual blocks.

Here, the rectangular inverse scan unit 1610 may scan an quantization frequency coefficient by determining an initial inverse scanning pattern according to the size of an image. For example, if the size of the image is greater than or equal to the preset size, the DC component and the lowest coefficient of the square residual block are given priority among the quantized frequency coefficients of the quantized residual block, but are larger than the coefficient of the DC component. The initial inverse scanning pattern can be determined by giving priority to the image. If the image size is smaller than the preset size, the same priority is applied to the DC component coefficient and the lowest coefficient of the square residual block among the quantized frequency coefficients of the quantized residual block. The initial reverse scanning pattern can be determined.

Also, the rectangular inverse scan unit 1610 may adaptively update the inverse scanning pattern according to a probability that a quantization frequency coefficient other than '0' occurs at each position of the quantized residual block. Each time decoding, the inverse scanning pattern can be adaptively updated by calculating the probability of occurrence of a non-zero quantization frequency coefficient at each position of the quantized residual block and determining the inverse scanning order in the order of the highest probability. .

The rectangular inverse quantizer 1620 inverse quantizes the plurality of quantized rectangular residual blocks. The rectangular inverse transform unit 1630 inversely converts the plurality of inverse quantized rectangular residual blocks into the time domain.

The rectangular prediction unit 1640 generates a plurality of rectangular prediction blocks by estimating a plurality of rectangular current blocks corresponding to the plurality of inversely transformed rectangular residual blocks in steps. Here, the rectangular prediction unit 1640 predicts the plurality of rectangular current blocks step by step according to the prediction direction by the prediction mode extracted from the bitstream.

The rectangular adder 1650 reconstructs a plurality of rectangular current blocks by adding a plurality of inversely transformed rectangular residual blocks to the plurality of rectangular prediction blocks, combines each rectangular current block, and reconstructs an image and outputs a reconstructed image.

When the quantized frequency coefficient extracted by decoding the bitstream is input, the rectangular reconstructor 1540 generates quantized residual blocks by inverse scanning, divides them into rectangular block units, and outputs a plurality of quantized rectangular residual blocks. Inversely quantizes two quantized rectangular residual blocks, inversely transforms a plurality of inverse quantized rectangular residual blocks into a time domain, and predicts a plurality of rectangular current blocks corresponding to the plurality of inversely transformed rectangular residual blocks stepwise A prediction block is generated, and a plurality of inverse transformed rectangular residual blocks are added to the plurality of rectangular prediction blocks to restore the plurality of rectangular current blocks, and the respective rectangular current blocks are combined to reconstruct an image.

17 is a block diagram schematically illustrating an electronic configuration of a square decoding apparatus according to an embodiment of the present invention.

The square decoding apparatus according to the embodiment of the present invention may be implemented by the square reconstruction unit 1550 in FIG. 15. Hereinafter, the square decoding apparatus according to an embodiment of the present invention is called a square reconstruction unit 1550.

The square reconstructor 1550 may include a square inverse scan unit 1710, a square inverse quantizer 1720, a square inverse transform unit 1730, a square predictor 1740, and a square adder 1750. have.

The square inverse scan unit 1710 inversely scans the quantized frequency coefficient sequence output from the decoder 1520 to generate a quantized square residual block. In this case, the square inverse scan unit 1710 may adaptively update the scanning pattern in the encoding process.

The square inverse quantizer 1720 inverse quantizes the quantized square residual block. The square inverse transformer 1730 inversely transforms an inverse quantized square residual block into a time domain.

The square predictor 1740 generates a square prediction block by predicting a square current block corresponding to the inversely transformed square residual block. Here, the square predictor 1740 predicts the square current block according to the prediction direction by the prediction mode extracted from the bitstream.

The square adder 1750 reconstructs an image by adding an inverse transformed square residual block to the square prediction block to reconstruct the square current block, and outputs a reconstructed image.

18 is a flowchart illustrating an image decoding method according to an embodiment of the present invention.

The image decoding apparatus 1500 that receives and stores a bitstream of an image through a wired / wireless communication network or a cable, decodes and restores the image to reproduce the image according to a user's selection or an algorithm of another program being executed.

To this end, the image decoding apparatus 1500 extracts information about a prediction mode from the bitstream (S1810), and extracts a quantized frequency coefficient sequence by decoding the bitstream (S1820). The image decoding apparatus 1500 extracting the quantized frequency coefficient string checks whether all of the quantized frequency coefficients of the quantized frequency coefficient string are '0' (S1830). When the quantization frequency coefficients are all '0' as a result of checking in operation S1830, the image decoding apparatus 1500 restores and outputs the current block in the prediction mode in units of square blocks by using the quantization frequency coefficient sequence (S1840). If the quantization frequency coefficients are all '0' as a result of checking in step S1830, a bitstream identifier is extracted from the bitstream (S1850). The image decoding apparatus 1500 extracting the bitstream identifier checks whether the bitstream identifier is '1' (S1860), and if the bitstream identifier is '1', identifies the corresponding bitstream as a rectangular bitstream and quantizes it. The current block of the image is reconstructed and output step by step in the prediction mode in units of rectangular blocks using the frequency coefficient sequence (S1870). Also, if the bitstream identifier is not '1', the image decoding apparatus 1500 identifies that the bitstream is a square bitstream and proceeds to step S1840, and uses the quantization frequency coefficient sequence to predict the prediction mode in square block units. Restore and output the current block of the image.

19 is an exemplary diagram for describing a process of reverse scanning according to an initial reverse scanning pattern according to an embodiment of the present invention.

The image decoding apparatus 1500 determines an initial inverse scanning pattern according to the size of an image, and inversely scans a quantized frequency coefficient sequence according to the initial inverse scanning pattern to configure a quantized residual block. In FIG. 19, the quantization frequency coefficient sequence is "0, -1, 2, 0, 4, 0, 0, 0, ...", the image size is CIF, the block mode is 4x4 size, and the prediction mode is the vertical mode. The process of constructing the quantized residual block by reverse scanning according to the determined initial scanning pattern is shown.

Determining the initial reverse scanning pattern according to the size of the image is similar to determining the initial scanning pattern described above with reference to FIG. 13. That is, the probability that a non-zero value occurs in the quantization frequency coefficients of a square residual block depends on the size of the image, and when the image size increases, the probability that a non-zero value occurs in the coefficient of the DC component is the lowest. By using a characteristic that is higher than the coefficients of the part, the initial scanning pattern is determined according to the size of the image.

For example, if a threshold size for comparing the size of the image is set in advance, and the size of the image is larger than or equal to the preset size, the quantization frequency coefficients sequentially read from the quantization frequency coefficient sequence are converted into quantization frequency coefficients of the quantized residual block. Arrange the quantized frequency coefficients of the positions of the coefficients of the DC component and the positions of the coefficients of the lowest coefficient of the quantized residual block, and give the higher priority to the positions of the coefficients of the DC component. The initial inverse scanning pattern can be determined. In addition, when the size of the image is smaller than the preset size, the same priority is given to the position corresponding to the coefficient of the DC component among the positions of the quantization frequency coefficients of the quantized residual block and the position corresponding to the lowest coefficient of the square residual block. An initial reverse scanning pattern can be determined.

In the example of FIG. 19, when the preset size is 720p and the image size is CIF, since the size of the image is smaller than the preset size, the same priority is given to the position corresponding to the coefficient of the DC component and the position corresponding to the lowest coefficient. By inverse scanning, a quantized residual block as shown can be obtained.

In addition, when the quantized residual block is inversely scanned, a plurality of rectangular residual blocks are generated by dividing the quantized residual blocks into a plurality of vertical blocks according to the vertical direction, which is a prediction direction of the prediction mode of the current block.

20 is an exemplary diagram for explaining a process of updating a reverse scanning pattern according to an embodiment of the present invention.

As the image decoding apparatus 200 updates the scanning pattern, the image decoding apparatus 1500 may update the reverse scanning pattern. In other words, each time the block is decoded, the reverse scanning pattern is adaptively updated and scanned instead of using the initially determined initial reverse scanning pattern.

At this time, the criteria for updating the reverse scanning pattern is similar to the criteria for updating the scanning pattern. That is, it may be a probability that a quantization frequency coefficient other than '0' occurs for each position of the quantized residual block. For example, each time a plurality of blocks of an image are decoded, the probability of occurrence of a non-zero quantization frequency coefficient at each position of the quantized residual block is calculated and the reverse scanning order is determined in the order of the greatest probability. The inverse scanning pattern can be updated adaptively.

21 is a block diagram schematically illustrating an electronic configuration of a video encoding apparatus according to another embodiment of the present invention.

The image encoding apparatus 2100 according to another embodiment of the present invention may include a square encoder 210, an encoding selector 230, a prediction direction changer 2110, and a rectangular encoder 2120. .

The video encoding apparatus 2100 may be a personal computer (PC), a notebook computer, a personal digital assistant (PDA), a portable multimedia player (PMP), or a PlayStation Portable (PSP). ), A communication device such as a communication modem for communicating with various devices or a wired / wireless communication network, a memory for storing various programs and data for encoding an image, and executing a program. Means a variety of devices including a microprocessor for operation and control.

Since the square encoder 210 and the encoding selector 230 perform the same or similar roles as described above with reference to FIG. 2, a detailed description thereof will be omitted.

The prediction direction switching unit 2110 changes the prediction direction of each rectangular current block according to the coefficient value of the quantization frequency coefficient of the rectangular previous block coded before the rectangular current block coded stepwise by the rectangular encoder 2120.

To this end, the prediction direction switcher 2110 is connected to the rectangular encoder 2120 to receive and store the quantized frequency coefficients of the previous rectangular block encoded before each rectangular current block from the rectangular encoder 2120.

When the current block is input, the rectangular encoder 2120 sequentially outputs a rectangular encoded bitstream by sequentially encoding a plurality of rectangular current blocks generated by dividing the rectangular blocks into units. In this case, the rectangular encoder 2120 may predict the plurality of rectangular current blocks according to the prediction direction changed by the prediction direction switcher 2110.

The rectangular encoder 2120 has the same structure as the rectangular encoder 220 described above with reference to FIG. 4. However, the rectangular encoder 460 and the rectangular predictor 492 are connected to the prediction direction switching unit 2110, and the rectangular encoder 460 is a quantization frequency of each rectangular current block transmitted from the rectangular scan unit 450. The coefficient is transmitted to the prediction direction switching unit 2110, and the rectangular prediction unit 492 predicts each rectangular current block according to the prediction direction transmitted from the prediction direction switching unit 2110. Here, the prediction direction received by the rectangular prediction unit 492 from the prediction direction switching unit 2110 may be a prediction direction of the prediction mode of the current block, but may also be a prediction direction changed according to the quantization frequency coefficient of the previous rectangular block.

FIG. 22 is a block diagram schematically illustrating an electronic configuration of an apparatus for predicting direction change according to another embodiment of the present invention.

The prediction direction switching device according to another embodiment of the present invention may be implemented by the prediction direction switching unit 2110 in FIG. 21. Hereinafter, the prediction direction switching device according to another embodiment of the present invention is called a prediction direction switching unit 2110.

According to another exemplary embodiment of the present invention, the prediction direction switching unit 2110 may include a previous block coefficient storage unit 2210, a prediction direction determination unit 2220, an encoding cost calculator 2230, and a prediction direction identifier output unit 2240. It is configured to include.

The previous block coefficient storage unit 2210 stores the quantized frequency coefficients of the rectangular previous block. That is, each time the previous block coefficient storage unit 2210 encodes each rectangular current block by the rectangular encoder 2120. A quantization frequency coefficient of each rectangular current block is received and stored, and the quantization frequency coefficient becomes a quantization frequency coefficient of a previous rectangular block that is a reference for determining a prediction direction for predicting a next rectangular current block.

The prediction direction determiner 2220 changes the prediction direction for each rectangular current block when stepwise encoding a plurality of rectangular current blocks generated by dividing the current block into rectangular block units. The prediction direction determiner 2220 checks the coefficient values of the quantization frequency coefficients of the rectangular rectangular block stored in the previous block coefficient storage unit 2210, and when the coefficient values of the quantized frequency coefficients of the rectangular rectangular block are all '0'. The reference prediction direction is determined as the prediction direction of each rectangular current block. In addition, when one or more coefficient values of the quantization frequency coefficients of the rectangular previous block are not '0', the prediction direction determiner 2220 may predict that the encoding cost is the minimum (or the maximum coding efficiency) among the plurality of prediction direction candidates. The direction is determined as the prediction direction of each rectangular current block.

That is, if the coefficient values of the quantization frequency coefficients of the rectangular block previously encoded before encoding the rectangular current block to be currently predicted are all '0', this means that the prediction is almost complete and the encoding efficiency is very excellent. The reference prediction direction of the previous rectangular block is determined as the prediction direction of each rectangular current block. Here, the reference prediction direction refers to the prediction direction of the current block in which the rectangular current block is divided or the prediction direction of the previous rectangular block.

In addition, if at least one or more of the coefficient values of the quantization frequency coefficients of the previous rectangular block previously encoded before encoding the rectangular current block to be predicted are not '0', the prediction accuracy is low and the encoding efficiency is very high. In order to change the prediction direction without using the reference prediction direction as it is, the prediction direction having the smallest encoding cost among the plurality of prediction direction candidates is determined as the prediction direction of each rectangular current block. Here, the prediction direction candidate group is a plurality of prediction directions to be determined as the prediction directions of the rectangular current block, and refers to prediction directions other than the reference prediction direction. The prediction direction candidate group may be eight prediction directions except for the reference prediction direction among a total of nine prediction directions in the case of intra 4x4 prediction. However, as the number of prediction directions increases, the computational complexity increases and the coding efficiency may decrease, so that the number of prediction directions may be limited to a smaller number. For example, the prediction directions on both sides of the reference prediction direction may be selected as the prediction direction candidate group.

The encoding cost calculator 2230 calculates an encoding cost according to each prediction direction of the plurality of prediction direction candidates. Here, the encoding cost may be a rate-distortion cost, but is not necessarily limited thereto, and may be a cost for measuring encoding efficiency without being limited to the encoding cost. In addition, the encoding cost according to the prediction direction means a cost required when encoding in the prediction direction.

The prediction direction identifier output unit 2240 outputs a prediction direction identifier for identifying a prediction direction of each rectangular current block.

23 is a flowchart illustrating a prediction direction switching method according to an embodiment of the present invention.

The prediction direction switching unit 2110 stores the quantization frequency coefficients of the rectangular previous block transmitted from the rectangular encoder 2120 (S2310), and when determining the prediction direction of the rectangular current block, determines the quantization frequency coefficients of the previous rectangular block. When all of the quantization frequency coefficients are '0', the prediction direction of the current block is determined as the reference prediction direction (S2330). When at least one or more of the quantization frequency coefficients is not '0', the remaining prediction except for the reference prediction direction is determined. The prediction cost, that is, the encoding cost according to each prediction direction of the prediction direction candidate group is calculated (S2340), and the prediction direction having the minimum encoding cost is determined as the prediction direction of the current block by comparing each encoding cost (S2350). That is, the prediction direction is changed in the reference prediction direction.

24 and 25 are exemplary diagrams for describing a prediction direction candidate group according to another embodiment of the present invention.

Referring to FIG. 24, when the prediction mode of the current block is the horizontal mode, the prediction direction of the current block is a horizontal direction, and the prediction direction candidates are diagonally upper rightwards on both sides with respect to the horizontal direction that is the prediction direction of the current block. It can be in the direction of the bottom right and diagonally.

In this case, assuming that the reference prediction direction is a horizontal direction indicated by a dotted line, when the quantization frequency coefficients of the rectangular previous blocks (a, b, c, d) are all '0', the horizontal direction, which is the reference prediction direction, is the rectangular current as it is. It is the prediction direction of the block. However, when at least one or more of the quantization frequency coefficients of the previous rectangular block (a, b, c, d) is not '0', the diagonal upper right direction is a side direction except for the horizontal direction that is the reference prediction direction among the prediction direction candidates. The direction in which the encoding cost becomes the minimum among the right and bottom diagonal directions is the prediction direction of the rectangular current block.

Referring to FIG. 25, when the prediction mode of the current block is a diagonal lower right mode, the prediction direction of the current block is a diagonal right lower direction, and the prediction direction candidates are located on both sides of the diagonal right lower direction, which is the prediction direction of the current block. It may be a vertical direction and a horizontal direction of.

In this case, assuming that the reference prediction direction is a diagonal right lower direction indicated by a dotted line, when the quantization frequency coefficients of the previous rectangular blocks (a, b, c, d) are all '0', the diagonal right lower direction, which is the reference prediction direction, This is the prediction direction of the rectangular current block. However, when at least one or more of the quantization frequency coefficients of the previous rectangular block (a, b, c, d) is not '0', the horizontal direction is a side direction except for the lower right diagonal direction that is the reference prediction direction among the prediction direction candidates. The direction in which the encoding cost is the minimum among the and vertical directions is the prediction direction of the rectangular current block.

FIG. 26 is a block diagram schematically illustrating an electronic configuration of an image decoding apparatus 2600 according to another exemplary embodiment.

The image decoding apparatus 2600 according to another embodiment of the present invention may include a prediction mode extractor 1510, a decoder 1520, a bitstream identifier extractor 1530, a prediction direction identifier extractor 2610, and a rectangular reconstructor. 2620 and the square recovery unit 1550 may be configured.

The video decoding apparatus 1500 may be a personal computer (PC), a notebook computer, a personal digital assistant (PDA), a portable multimedia player (PMP), or a PlayStation Portable (PSP). ), A communication device such as a communication modem for communicating with various devices or a wired / wireless communication network, a memory for storing various programs and data for decoding an image, and executing a program. Means a variety of devices including a microprocessor for operation and control.

Since the prediction mode extractor 1510, the decoder 1520, the bitstream identifier extractor 1530, and the square reconstructor 1550 perform the same or similar functions as described above with reference to FIG. 15, a detailed description thereof will be omitted.

The prediction direction switching unit 2610 extracts the prediction direction identifier from the quantized frequency coefficient sequence delivered from the decoder 1520 and changes the prediction direction for each rectangular current block in the prediction direction indicated by the prediction direction identifier.

The rectangular reconstruction unit 2620 reconstructs and outputs the current block step by step according to the prediction direction changed by the prediction direction switching unit 2610 in units of rectangular blocks using the quantized frequency coefficient sequence. Here, the rectangular restoration unit 2620 has the same structure as the rectangular restoration unit 1540 described above with reference to FIG. 16. However, in the rectangular reconstruction unit 2620, the rectangular prediction unit 1640 is connected to the prediction direction switching unit 2610, and the rectangular prediction unit 1640 is not the prediction direction of the prediction mode of the current block but the prediction direction switching unit 2610. Predicts a rectangular current block according to the prediction direction delivered from

The image decoding apparatus 2600 according to another embodiment of the present invention is a method of decoding an image, extracts information about a prediction mode from a bitstream, decodes the bitstream to extract a quantization frequency coefficient sequence, and Extract the prediction direction identifier from the coefficient sequence to change the prediction direction according to the prediction mode, extract the bitstream identifier from the bitstream, and restore the current block in rectangular block units using the quantized frequency coefficient sequence according to the bitstream identifier. Restoring and outputting the current block step by step according to the changed prediction direction, and restoring the current block in square block units using the quantization frequency coefficient sequence according to the bitstream identifier, but restoring the current block according to the prediction direction according to the prediction mode. To print.

27 is a block diagram schematically illustrating an electronic configuration of an apparatus for predicting direction change according to another embodiment of the present invention.

The prediction direction switching device according to another embodiment of the present invention may be implemented as the prediction direction switching unit 2610 in FIG. 26. Hereinafter, the prediction direction switching device according to another embodiment of the present invention is called a prediction direction switching unit 2610.

The prediction direction switching unit 2610 according to another embodiment of the present invention may include a previous block coefficient storage unit 2710, a prediction direction determination unit 2720, and a prediction direction identifier extractor 2730.

The previous block coefficient storage unit 2710 stores the quantized frequency coefficients of the rectangular previous block decoded before each rectangular current block. That is, the previous block coefficient storage unit 2710 receives and stores the quantization frequency coefficients of the rectangular current block in the quantization frequency coefficient sequence transmitted from the decoder 1520, when the stored quantization frequency coefficients encode the next rectangular current block. It becomes the quantization frequency coefficient of the rectangular block before use.

The prediction direction determiner 2720 determines the reference prediction direction as the prediction direction of each rectangular current block when the coefficient values of the quantization frequency coefficients of the rectangular previous block stored in the previous block coefficient storage unit 2710 are all '0'. If one or more coefficient values of the quantization frequency coefficients of the quantization frequency coefficients of the previous rectangular block are not '0', the prediction direction indicated by the prediction direction identifier extracted from the quantization frequency coefficient sequence is used as the prediction direction of each rectangular current block. Decide

The prediction direction identifier extractor 2730 extracts the prediction direction identifier from the quantized frequency coefficient string transmitted from the decoder 1520 under the control of the prediction direction determiner 2720.

Here, the reference prediction direction may be one of the prediction direction of the current block and the prediction direction of the previous rectangular block, and the prediction direction candidate may be determined based on the reference prediction direction.

28 is a flowchart illustrating a prediction direction switching method according to another embodiment of the present invention.

According to another exemplary embodiment of the present invention, the prediction direction switching device, that is, the prediction direction switching unit 2610 stores the quantization frequency coefficients of the previous rectangular block (S2810) and determines whether all of the quantization frequency coefficients of the previous rectangular block are '0'. In operation S2820, when the quantization frequency coefficients of the rectangular previous block are all '0', the reference prediction direction is determined as the prediction direction of the rectangular current block (S2830), and at least one of the quantization frequency coefficients of the rectangular rectangular block is If it is not '0', the prediction direction identifier is extracted from the quantized frequency coefficient sequence (S2840), and the prediction direction indicated by the prediction direction identifier is determined as the prediction direction of the rectangular current block (S2850).

29 and 30 are exemplary diagrams for describing a prediction direction candidate group according to another embodiment of the present invention.

The reference prediction direction is determined as the prediction direction of the rectangular current block according to the coefficient value of the quantization frequency coefficient of the previous rectangular block, or among the prediction directions on both sides based on the prediction direction of the prediction mode of the current block indicated by the prediction direction identifier. One direction is determined as the prediction direction of the rectangular current block.

Through such embodiments, the following embodiments may be applied. In other words, when predicting, transforming, and quantizing a square block unit, each square block is predicted according to the prediction direction of the prediction mode of the corresponding block, and the quantization frequency coefficient of the previous square block is checked to determine the quantization frequency coefficient. By predicting by changing the prediction direction according to the coefficient value, the coding efficiency can be increased by increasing the accuracy of the prediction.

Therefore, when the prediction direction switching device according to another embodiment of the present invention predicts a current block of an image for image encoding, the prediction direction shifting device stores the quantized frequency coefficient of the previous block encoded before the current block and stores the quantized frequency coefficient of the current block. When the frequency coefficient is input, if the quantization frequency coefficients of the previous block are all '0', the reference prediction direction is determined as the prediction direction of the current block, and if one or more coefficients of the quantization frequency coefficients of the previous block are not '0', The encoding cost of the plurality of prediction direction candidates may be calculated to determine the prediction direction having the smallest encoding cost as the prediction direction of the current block. In addition, the prediction direction switching device may output a prediction direction identifier for identifying the prediction direction of the current block.

Also, when the prediction direction change device according to another embodiment of the present invention predicts a current block of an image for image decoding, the quantization frequency coefficient of the previous block decoded before the current block is stored, and the quantization of the current block is performed. When the frequency coefficient is input, if the quantization frequency coefficients of the previous block are all '0', the reference prediction direction is determined as the prediction direction of the current block, and if one or more coefficients of the quantization frequency coefficients of the previous block are not '0', A prediction direction identifier may be extracted from the quantization frequency coefficient of the current block to determine the prediction direction indicated by the prediction direction identifier as the prediction direction of the current block.

Here, the reference prediction direction may be one of the prediction direction of the current block and the prediction direction of the previous block, and the prediction direction candidate may be determined based on the reference prediction direction.

In the above description, it is described that all the components constituting the embodiments of the present invention operate in combination, but the present invention is not necessarily limited to these embodiments. That is, within the scope of the present invention, all of the components may be selectively coupled to one or more of them. In addition, although all of the components may be implemented as one independent hardware, some or all of the components may be selectively combined to perform a part or all of the functions in one or a plurality of hardware. As shown in FIG. The codes and code segments constituting the computer program may be easily deduced by those skilled in the art. Such a computer program can be stored in a computer-readable storage medium, readable and executed by a computer, thereby realizing an embodiment of the present invention. As the storage medium of the computer program, a magnetic recording medium, an optical recording medium, a carrier wave medium, or the like may be included.

Furthermore, the terms "comprises", "comprising", or "having" described above mean that a component can be implanted unless otherwise specifically stated, But should be construed as including other elements. All terms, including technical and scientific terms, have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs, unless otherwise defined. Commonly used terms, such as predefined terms, should be interpreted to be consistent with the contextual meanings of the related art, and are not to be construed as ideal or overly formal, unless expressly defined to the contrary.

The foregoing description is merely illustrative of the technical idea of the present invention, and various changes and modifications may be made by those skilled in the art without departing from the essential characteristics of the present invention. Therefore, the embodiments disclosed in the present invention are intended to illustrate rather than limit the scope of the present invention, and the scope of the technical idea of the present invention is not limited by these embodiments. The protection scope of the present invention should be interpreted by the following claims, and all technical ideas within the equivalent scope should be interpreted as being included in the scope of the present invention.

As described above, the present invention is applied to a field related to an apparatus and a method for encoding or decoding an image, so that the accuracy of prediction can be improved when coding or decoding an image, Is a very useful invention.

1 is an exemplary diagram showing nine conventional 4x4 intra prediction modes,

2 is a block diagram schematically illustrating an electronic configuration of an image encoding apparatus according to an embodiment of the present invention;

3 is a block diagram schematically illustrating an electronic configuration of a square-angle encoding apparatus according to an embodiment of the present invention.

FIG. 4 is a block diagram schematically illustrating an electronic configuration of a rectangular encoder according to an embodiment of the present invention. FIG.

5 is a block diagram schematically illustrating an electronic configuration of an encoding selection apparatus according to an embodiment of the present invention.

6 is a flowchart illustrating an image encoding method according to an embodiment of the present invention.

FIG. 7 is a flowchart illustrating a bitstream selection process according to an embodiment of the present invention. FIG.

8 is a diagram illustrating a rectangular current block according to an embodiment of the present invention.

9 is a diagram illustrating a process of performing a stepwise intra prediction by a rectangular prediction unit according to an embodiment of the present invention.

10 is a block diagram schematically showing an electronic configuration of a rectangular scanning device according to an embodiment of the present invention.

11 is a flowchart illustrating a method for determining an initial scanning pattern according to an exemplary embodiment of the present invention.

FIG. 12 is a diagram illustrating an example in which a plurality of rectangular residual blocks are combined into a square residual block according to a preferred embodiment of the present invention;

13 is an exemplary diagram for explaining a process of determining an initial scanning pattern according to an embodiment of the present invention;

FIG. 14 is an exemplary diagram for explaining a process of updating a scanning pattern according to an embodiment of the present invention;

15 is a block diagram schematically illustrating an electronic configuration of an image decoding apparatus according to an embodiment of the present invention.

16 is a block diagram schematically illustrating an electronic configuration of a rectangular decoding apparatus according to an embodiment of the present invention;

17 is a block diagram schematically illustrating an electronic configuration of a square decoding apparatus according to an embodiment of the present invention;

18 is a flowchart for explaining a video decoding method according to an embodiment of the present invention.

19 is an exemplary view for explaining a process of inverse scanning according to an initial reverse scanning pattern according to an embodiment of the present invention;

20 is an exemplary diagram for explaining a process of updating an inverse scanning pattern according to an embodiment of the present invention.

FIG. 21 is a block diagram schematically showing an electronic configuration of an image encoding apparatus according to another embodiment of the present invention; FIG.

22 is a block diagram schematically showing an electronic configuration of a prediction direction switching apparatus according to another embodiment of the present invention;

23 is a flowchart for explaining a prediction direction switching method according to an embodiment of the present invention;

24 and 25 are diagrams for explaining a prediction direction candidate group according to another embodiment of the present invention,

FIG. 26 is a block diagram schematically illustrating an electronic configuration of an image decoding apparatus 2600 according to another embodiment of the present invention.

27 is a block diagram schematically showing an electronic configuration of a prediction direction switching apparatus according to another embodiment of the present invention;

28 is a flowchart for explaining a prediction direction switching method according to another embodiment of the present invention;

29 and 30 are exemplary diagrams for describing a prediction direction candidate group according to another embodiment of the present invention.

Claims (18)

In the device for encoding the current block of the video, A rectangular encoder for outputting a rectangular encoded bitstream by stepwise encoding a plurality of rectangular current blocks generated by dividing the rectangular blocks in rectangular units when the current block is input; A prediction direction switching unit for changing a prediction direction of each rectangular current block according to a coefficient value of a quantization frequency coefficient of a rectangular previous block coded before each rectangular current block coded stepwise; A square encoder configured to output a square encoded bitstream by encoding the current block in units of square blocks; And An encoding selector configured to calculate a coding cost of the rectangular encoded bitstream and an encoding cost of the square encoded bitstream and output a bitstream having a minimum encoding cost An image encoding apparatus comprising a. The method of claim 1, wherein the prediction direction switching unit, A previous block coefficient storage unit for storing quantized frequency coefficients of the rectangular previous block; When the coefficient values of the quantization frequency coefficients are all '0', the reference prediction direction is determined as the prediction direction of each rectangular current block, and when one or more of the coefficient values of the quantization frequency coefficients are not '0', A prediction direction determiner which determines a prediction direction having the least encoding cost among the prediction direction candidates as prediction directions of the rectangular current blocks; An encoding cost calculator configured to calculate an encoding cost according to each prediction direction of the plurality of prediction direction candidates; And Prediction direction identifier output unit for outputting a prediction direction identifier for identifying the prediction direction of each rectangular current block An image encoding apparatus comprising a. The method of claim 2, wherein the reference prediction direction is And one of a prediction direction of the current block and a prediction direction of the rectangular previous block. The method of claim 2, wherein the prediction direction candidate is: And the image encoding apparatus is determined based on the reference prediction direction. In the method of encoding the current block of the video, A rectangular encoding step of outputting a rectangular encoded bitstream by gradually encoding a plurality of rectangular current blocks generated by dividing the rectangular blocks in units of rectangular blocks when the current block is input; A prediction direction changing step of changing a prediction direction of each rectangular current block according to a coefficient value of a quantization frequency coefficient of a rectangular previous block coded before each rectangular current block coded stepwise; A square encoding step of outputting a square encoded bitstream by encoding the square block unit when the current block is input; And An encoding selection step of calculating a coding cost of the rectangular encoded bitstream and an encoding cost of the square encoded bitstream and outputting a bitstream having a minimum encoding cost; Image encoding method comprising a. The method of claim 5, wherein the prediction direction switching step, A previous block coefficient storage step of storing the quantized frequency coefficients of the rectangular previous block; A prediction direction maintaining step of determining a reference prediction direction as a prediction direction of each rectangular current block when all coefficient values of the quantized frequency coefficients are '0'; A coding cost calculating step of calculating a coding cost according to each prediction direction of a plurality of prediction direction candidates when at least one of coefficient values of the quantization frequency coefficients is not '0'; A prediction direction changing step of determining a prediction direction having a minimum encoding cost according to each prediction direction as a prediction direction of each rectangular current block; And A prediction direction identifier output step of outputting a prediction direction identifier for identifying a prediction direction of each rectangular current block; Image encoding method comprising a. The method of claim 6, wherein the reference prediction direction, And one of a prediction direction of the current block and a prediction direction of the rectangular previous block. The method of claim 7, wherein the prediction direction candidate is: And the image encoding method is determined based on the reference prediction direction. In the apparatus for decoding an image, A prediction mode extraction unit for extracting information about the prediction mode from the bitstream; A decoder which extracts a quantized frequency coefficient sequence by decoding the bitstream; A prediction direction switching unit which extracts a prediction direction identifier from the quantized frequency coefficient sequence and changes the prediction direction according to the prediction mode; A rectangular reconstruction unit for reconstructing and outputting the current block step by step according to the changed prediction direction in units of rectangular blocks using the quantized frequency coefficient sequence; A square restoring unit for restoring and outputting the current block according to a prediction direction according to the prediction mode in units of square blocks using the quantized frequency coefficient sequence; And A bitstream identifier extractor which extracts a bitstream identifier from the bitstream and outputs the quantized frequency coefficient sequence to one of the rectangular decompressor and the square decompressor according to the bitstream identifier Video decoding apparatus comprising a. The method of claim 9, wherein the prediction direction switching unit, A previous block coefficient storage unit for storing the quantized frequency coefficients of the rectangular previous block decoded before each rectangular current block reconstructed step by step in the rectangular decompression unit; When the coefficient values of the quantization frequency coefficients are all '0', the reference prediction direction is determined as the prediction direction of each rectangular current block, and when at least one of the coefficient values of the quantization frequency coefficients is not '0', A prediction direction determiner for determining a prediction direction indicated by a prediction direction identifier extracted from a quantized frequency coefficient sequence as a prediction direction of each rectangular current block; And A prediction direction identifier extractor for extracting the prediction direction identifier from the quantized frequency coefficient sequence Video decoding apparatus comprising a. The method of claim 10, wherein the reference prediction direction, And a prediction direction of the current block and a prediction direction of the rectangular previous block. The method of claim 10, wherein the prediction direction determiner, And a prediction direction is determined according to the prediction direction identifier among prediction direction candidates determined based on the reference prediction direction. In the method of decoding an image, A prediction mode extraction step of extracting information on the prediction mode from the bitstream; Decoding the bitstream to extract a quantized frequency coefficient sequence; A prediction direction switching step of extracting a prediction direction identifier from the quantized frequency coefficient sequence and changing a prediction direction according to the prediction mode; Extracting a bitstream identifier from the bitstream; A rectangular reconstruction step of reconstructing a current block in units of rectangular blocks using the quantized frequency coefficient sequence according to the bitstream identifier, and reconstructing and outputting the current block in accordance with the changed prediction direction; And A square reconstruction step of reconstructing a current block in units of square blocks using the quantized frequency coefficient sequence according to the bitstream identifier, and reconstructing and outputting the current block according to a prediction direction according to the prediction mode. Image decoding method comprising a. An apparatus for switching a prediction direction when predicting a current block of an image for image encoding, If the quantization frequency coefficients of the previous block encoded before the current block are input, the quantization frequency coefficients are stored. If the quantization frequency coefficients of the previous block are all zero, the reference prediction direction is used as the prediction direction of the current block. If the one or more coefficients of the quantization frequency coefficients of the previous block is not '0', the encoding cost of the plurality of prediction direction candidates is calculated to determine the prediction direction having the minimum encoding cost as the prediction direction of the current block. Predictive direction switching device, characterized in that. The apparatus of claim 14, wherein the prediction direction switching device is And a prediction direction identifier for identifying the prediction direction of the current block. An apparatus for switching a prediction direction when predicting a current block of an image for image decoding, If the quantization frequency coefficients of the previous block decoded before the current block are input, the quantization frequency coefficients of the current block are input. And a prediction direction indicated by the prediction direction identifier by extracting a prediction direction identifier from the quantization frequency coefficients of the current block when one or more coefficients of the quantization frequency coefficients of the previous block are not '0'. Predicting direction as the prediction direction of the current block. The method of claim 14 or 16, wherein the reference prediction direction, And a prediction direction of the current block and a prediction direction of the previous block. The method of claim 14, wherein the prediction direction candidate is: Prediction direction switching device characterized in that determined based on the reference prediction direction.

Images