JP4576342B2 - Adaptive image rotation encoding apparatus and decoding apparatus - Google Patents

Description

  The present invention relates to image signal processing, and in particular, when performing orthogonal transform typified by DCT (Discrete Cosine Transform) in image compression coding / decoding technology, the frequency components in the vertical and horizontal directions of the image are efficiently used. In general, the present invention relates to an encoding device and a decoding device that improve the efficiency of the entire encoding.

  MPEG-2 (see Non-Patent Document 1) standardized by ISO / IEC JTC1 / SC29 / WG11 (MPEG) and H.264 standardized by MPEG and ITU-T. In image compression coding standards such as H.264 / MPEG-4 AVC (see Non-Patent Document 2), the image signal is subjected to orthogonal transform processing represented by DCT, thereby reducing the spatial redundancy of the image signal. Removed. This orthogonal transformation process is performed in units of blocks on an image divided into rectangular blocks of, for example, horizontal 8 pixels × vertical 8 pixels, horizontal 4 pixels × vertical 4 pixels, etc. Separated for each component (conversion coefficient) up to the band frequency.

  In general, in image encoding, quantization processing is performed after the orthogonal transformation processing described above. Since human visual characteristics are insensitive to high frequency components, the conversion coefficients of high frequency components are coarse, and the conversion coefficients of low frequency components are finely quantized. Then, the quantized transform coefficients are scanned as shown in FIG. 6 in order from the low frequency to the high frequency. FIG. 6 is a diagram illustrating an example in which the zigzag scanning method is used as the scanning process, and illustrates that the scanning process is performed in numerical order. By such a scanning process, the quantized two-dimensional transform coefficient is rearranged into a one-dimensional signal. In the entropy encoding process that is performed after this, it is not necessary to encode the zero coefficient, so that the zero coefficient is processed by the scan process to increase the efficiency of the entropy encoding process. Can do. As the scan processing method, only several methods are defined in the standards of Non-Patent Document 1 and Non-Patent Document 2, and therefore the scan order cannot be changed in an arbitrary order. Actually, in addition to the zigzag scanning method shown in FIG. 6, there is a vertical priority scanning method according to the interlaced signal as shown in FIG. 7, and a total of two scanning methods are defined.

  By the orthogonal transformation process described above, the properties of the image signal can be analyzed for each frequency component. FIG. 8 is a diagram illustrating a first example of quantized transform coefficients and illustrates an example of a general image signal. According to FIG. 8, since a general image signal has many low frequency components, that is, flat portions of an image, and high frequency components, that is, fine stripes, edge components, or checkered patterns, there are relatively few conversion coefficients. It is concentrated in the region and not so distributed in the high region. On the other hand, when one of the vertical pattern and the horizontal pattern becomes dominant in the image signal, the conversion coefficients are distributed unevenly. FIG. 9 is a diagram illustrating a second example of quantized transform coefficients, and illustrates an example of an image signal in which a vertical pattern is dominant. Conversely, when the direction of the edge component of the image is varied, the deviation of the distribution of the transform coefficient is reduced.

ISO / IEC 13818-2 ITU-T Rec. H. H.264 | ISO / IEC 14496-10 Advanced Video Coding

  As described above, the characteristics of the image signal can be analyzed for each frequency component including the distribution direction by the orthogonal transform process performed in the image encoding. Then, quantization, scanning, and entropy coding are performed as subsequent processing. However, in the existing coding standard described above, the scan order is limited to only two, and the scan order cannot be changed in consideration of the directionality of the frequency components in the image. For this reason, useless zero coefficients are also encoded in the entropy encoding process depending on the distribution status of the frequency components, and the efficiency of the entire encoding is reduced.

  Accordingly, the present invention has been made to solve such a problem, and an object thereof is to provide an encoding device and a decoding device capable of improving the efficiency of the overall image encoding. .

  In order to solve the above problems, the present invention is characterized in that the original image is rotated by a predetermined angle so that the frequency components of the image are concentrated in a certain direction, and the present invention is quantized. It is characterized in that the scan order of transform coefficients is changed in accordance with the deviation of frequency components.

That is, the encoding device according to claim 1 of the present invention performs orthogonal transform and quantization on a rectangular image for each small region constituting the image, and scans the quantized two-dimensional transform coefficient to obtain a one-dimensional transform. In the encoding device that generates a transform coefficient and encodes the one-dimensional coefficient, the rectangular image is rotated to generate a diagonal image, and the pixels in the region other than the diagonal image in the new rectangular image are uniformly distributed. When the pixel value is set, the uniform pixel value is either a value obtained by holding the pixel value around the external neighborhood in the oblique image as it is or a value obtained by averaging the pixel values around the external neighborhood in the oblique image. a rotating section to be set as a generation unit for generating a new rectangular image including an oblique image generated by the rotation, the center position of the rotation angle and the rotation of the rectangular image to be rotated by the rotating unit, reference numeral An angle center position setting unit that is set based on an evaluation value indicating efficiency, and performing orthogonal transformation, quantization, scanning, and encoding on a new rectangular image generated by the generation unit To do.

  According to a second aspect of the present invention, there is provided an encoding device according to the present invention, which performs orthogonal transform and quantization on a rectangular image for each small region constituting an image, scans the quantized two-dimensional transform coefficient, and performs one-dimensional conversion. In an encoding device that generates transform coefficients and encodes the one-dimensional coefficients, the scan order for generating the one-dimensional transform coefficients from the quantized two-dimensional transform coefficients is determined by encoding efficiency. A scan order setting unit that is set based on the evaluation value is provided, and a one-dimensional transform coefficient is generated in the scan order set by the scan order setting unit, and encoding is performed.

In addition, the decoding apparatus according to the present invention receives a bit stream of an image encoded by the encoding apparatus, performs decoding, and reverse-scans the decoded one-dimensional transform coefficient to perform two-dimensional conversion. A decoding device that generates a coefficient and performs inverse quantization and inverse orthogonal transform on the two-dimensional transform coefficient, and inputs the bit stream of the image generated by the encoding device according to claim 1 to perform decoding. A new rectangular image generated by the encoding device by obtaining a rotation angle and a center position set in the encoding device from the decoding unit to be applied and the bit stream, rotating the image in reverse by the rotation angle and the center position, and And a reverse rotation unit for obtaining the oblique image .

  In addition, the decoding apparatus according to the present invention receives a bit stream of an image encoded by the encoding apparatus, performs decoding, and reverse-scans the decoded one-dimensional transform coefficient to perform two-dimensional conversion. A decoding device that generates coefficients and performs inverse quantization and inverse orthogonal transform on the two-dimensional transform coefficients, and inputs the bitstream of the image generated by the encoding device according to claim 2 for decoding. A decoding unit that performs, obtains a scan order set in the encoding device from the bit stream, and reverse-scans the one-dimensional transform coefficient according to the scan order to generate a two-dimensional transform coefficient; It is provided with.

  According to the present invention, it is possible to concentrate frequency components of image conversion coefficients in a certain direction (for example, a vertical frequency direction or a horizontal frequency direction). Also, the transform coefficients can be rearranged one-dimensionally in a scan order that matches the frequency component concentration direction (bias). As a result, as one-dimensional transform coefficients, zero coefficients can be gathered in the rear part, so that the amount of processing information for entropy coding can be reduced. Therefore, it is possible to improve the overall efficiency of image encoding.

  Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings. In addition, although the encoding method of a moving image was quoted by the above-mentioned non-patent documents 1 and 2, this embodiment is not only a moving image but also a still image. In order to simplify the following description, description of motion estimation and motion compensation used in the moving image coding method will be omitted, and processing for still images will be described. The present invention can also be applied to moving images and still image encoding that perform motion compensation and intra-screen prediction.

[Encoder]
First, an encoding apparatus according to an embodiment of the present invention will be described. FIG. 1 is a block diagram illustrating a configuration of an encoding device. The encoding device 1 includes a rotation unit 11, an orthogonal transformation unit 12, a quantization unit 13, a scanning unit 14, an entropy encoding unit 15, a rotation control unit 16, a local decoding unit 17, and a scan control unit 18. The encoding device 1 inputs an original image to be encoded by the encoding device 1, rotates the image, performs orthogonal transformation represented by DCT, and performs quantization, scanning, and entropy encoding. To generate a bitstream. Hereinafter, each part which comprises the encoding apparatus 1 is demonstrated in detail.

  The rotation unit 11 inputs an original image and inputs a rotation angle and a center position from the rotation control unit 16. Then, with respect to a rectangular image for each small area constituting the original image (for example, a rectangular image of horizontal 8 pixels × vertical 8 pixels), the image is rotated by the input rotation angle with reference to the input center position, and the oblique image is converted. Generate. In this case, the orthogonal transform unit 12 cannot perform the orthogonal transform on the outer peripheral portion of the oblique image as it is. Therefore, the rotation unit 11 further generates a new rectangular image so as to surround the diagonal image with respect to the generated diagonal image.

  FIG. 2 is a diagram illustrating a process for generating a new rectangular image in the rotation unit 11. As illustrated in FIG. 2, the pixel value of the external peripheral portion indicated by the black portion is unknown only by the rotation unit 11 generating an oblique image of the white portion. Therefore, in consideration of the encoding process, the rotation unit 11 obtains a black pixel value, a zero pixel value, a value obtained by directly holding a pixel value around the external neighborhood in the oblique image, and a pixel value around the external neighborhood in the oblique image. An average value or the like is set to the pixel value of the external peripheral portion indicated by black, and a new rectangular image is generated. As a result, the new rectangular image will be larger than the original oblique image, and it is expected that the encoding efficiency will decrease. However, since the pixels in the black part are composed of uniform pixel values, the amount of encoded information is hardly generated by these pixels. Accordingly, the encoding efficiency does not extremely decrease due to the enlargement of the image.

  In addition, when the rotation unit 11 rotates the rectangular image to generate the oblique image, the rotation unit 11 performs an interpolation process so that the pixel exists at the original pixel position. This interpolation process can be performed by a method such as bilinear interpolation. Here, it is expected that the image quality is deteriorated by the interpolation process. However, considering the encoding process by lowering the bit rate, for example, the image quality deterioration of the encoding cannot be avoided. For this reason, comparing the degree of image quality degradation due to interpolation processing with the degree of improvement in coding efficiency due to the adaptive rotation processing described above, the latter effect is superior to the former problem, and the coding efficiency is improved as a whole. Can be considered.

  Further, the rotation unit 11 outputs the rotation angle and the center position used for the rotation process to the entropy coding unit 15. Since the information on the rotation angle and the center position is necessary for decoding processing by the decoding device 2 described later, the entropy encoding unit 15 performs entropy encoding on the information together with the transform coefficient and the like.

  The orthogonal transformation unit 12 receives a new rectangular image generated by rotation by the rotation unit 11 and performs orthogonal transformation on the image signal of the rectangular image. The quantization unit 13 receives the two-dimensional transform coefficient orthogonally transformed by the orthogonal transform unit 12 and performs quantization. Then, the quantization unit 13 outputs the quantized transform coefficient to the scan unit 14 and the local decoding unit 17. The local decoding unit 17 receives the quantized transform coefficient from the quantizing unit 13 and performs inverse quantization, inverse orthogonal transform, and inverse rotation to generate a reconstructed image. Then, the local decoding unit 17 outputs the generated reconstructed image to the rotation control unit 16 and the scan control unit 18.

  The rotation control unit 16 inputs the original image, the reconstructed image from the local decoding unit 17, and the information amount of the bitstream from the entropy encoding unit 15, and the rotation unit 11 has a (previous) rotation angle and An evaluation value indicating the efficiency of encoding when rotated at the center position is calculated, a new (next) rotation angle and center position are set based on the evaluation value, and output to the rotation unit 11. Here, the original image is an image before rotation. The reconstructed image is an image in which the local decoding unit 17 reconstructs a signal that has been subjected to orthogonal transformation and quantization by rotating it at a certain rotation angle and center position. The information amount of the bitstream is a signal amount that has been subjected to orthogonal transformation, quantization, scanning, and entropy coding by rotating at a certain rotation angle and center position. The center position means a reference point when the rectangular image is rotated, and may be the center of the rectangular image, or may be a vertex such as the upper left or an arbitrary point.

Specifically, the rotation control unit 16 calculates a difference between the original image and the reconstructed image, and calculates a linear sum of the difference and the information amount of the bit stream as an evaluation value. Then, the rotation control unit 16 sets a new rotation angle and center position so that the evaluation value is minimized (in order to improve the evaluation). For example, when the evaluation value when the previous rotation angle is set to the plus side increases, the new rotation angle is set to the minus side. The same applies to the center position. Such processing is repeated to finally identify the most suitable rotation angle and center position. The formula for calculating the evaluation value is shown in formula (1).
minE = D + λR (1)
Here, E is an evaluation value, D is a difference between the original image and the reconstructed image, λ is a coefficient, and R is an information amount of the bit stream.

  The rotation control unit 16 repeats the above-described process for a preset number of times, and specifies the rotation angle and the center position when the evaluation value is the smallest as the most suitable rotation angle and center position. It may be. The rotation control unit 16 calculates a coding error as an evaluation value based on the original image and the reconstructed image, and the rotation angle and the center position when the evaluation value becomes smaller than a preset coding error. May be specified as the most suitable rotation angle and center position.

  In addition, the rotation control unit 16 does not arbitrarily set the rotation angle and the center position as described above, but prepares several combinations of the rotation angle and the center position in advance. An evaluation value based on the center position may be calculated, and the rotation angle and center position when the evaluation becomes the best may be specified as the most suitable rotation angle and center position. Also, several combinations of rotation angles and center positions are prepared in advance, and for each combination, a coding error is calculated as an evaluation value based on the original image and the reconstructed image, and this evaluation value is set in advance. The rotation angle and the center position when the encoding error is smaller than the encoding error may be specified as the most suitable rotation angle and center position.

  Furthermore, the rotation control unit 16 calculates an evaluation value when no rotation processing is performed in the rotation unit 11, compares the evaluation value with an evaluation value when the rotation processing is performed, and performs the rotation processing. If it is determined that the evaluation when there is no rotation is better than the evaluation when the rotation process is performed, a command indicating that the rotation process is not performed is output to the rotation unit 11. In this case, the rotation unit 11 outputs the rectangular image as it is to the orthogonal transform unit 12 without performing the rotation process.

  On the other hand, the scanning unit 14 inputs the transform coefficient quantized by the quantization unit 13 and converts the two-dimensional transform coefficient into a one-dimensional transform so that the entropy coding unit 15 in the subsequent stage can perform encoding. Sort by coefficient. This rearrangement is performed in the scan order set by the scan control unit 18. The scan order includes those shown in FIGS. 3 and 4 in addition to the zigzag scan order shown in FIG. 6 and the vertical priority scan order shown in FIG. FIG. 3 shows a scan order with priority in the vertical direction defined in consideration of the concentration of conversion coefficients in the vertical frequency direction. FIG. 4 shows a scan order with priority in the horizontal direction defined in consideration of the concentration of conversion coefficients in the horizontal frequency direction. For example, the scan control unit 18 sets a scan order from these. Details of the scan control unit 18 will be described later.

  In addition, the scan unit 14 outputs information on the scan order used for the scan process (for example, an identifier indicating the scan order) to the entropy encoding unit 15. Since this scan order information is necessary for decoding processing by the decoding device 2 to be described later, the entropy encoding unit 15 performs entropy encoding on this information together with transform coefficients and the like. Note that the scanning unit 14 and the decoding device 2 can specify the scan order by an identifier indicating the scan order.

  The entropy encoding unit 15 inputs a one-dimensional conversion coefficient from the scanning unit 14, inputs a rotation angle and a center position from the rotating unit 11, and inputs a scanning order from the scanning unit 14. Then, the entropy encoding unit 15 performs entropy encoding on the transform coefficient, the rotation angle, the center position, and the scan order, and generates and outputs a bit stream. In addition, the entropy encoding unit 15 calculates the information amount of the bit stream and outputs it to the rotation control unit 16 and the scan control unit 18.

  The scan control unit 18 inputs the original image, the reconstructed image from the local decoding unit 17, and the bitstream information amount from the entropy encoding unit 15, and has a (previous) scan order for the scan unit 14. The evaluation value indicating the coding efficiency when scanning is performed is calculated, a new (next) scan order is set based on the evaluation value, and is output to the scanning unit 14. The evaluation value calculation process is the same as the process of the rotation control unit 16 described above, and the scan order setting process is the same as the rotation angle and center position setting process of the rotation control unit 16.

  That is, the scan control unit 18 calculates a difference between the original image and the reconstructed image, and calculates a linear sum of the difference and the information amount of the bitstream as an evaluation value. Then, the scan control unit 18 sets a new scan order so that the evaluation value is minimized (in order to improve the evaluation). Such a process is repeated to finally identify the most suitable scan order.

  As in the case of the rotation control unit 16, the scan control unit 18 repeats the above-described processing for a preset number of times, and specifies the scan order when the evaluation value is the smallest as the most suitable scan order. The coding error may be calculated as an evaluation value, and the scan order when the evaluation value becomes smaller than a preset coding error is specified as the most suitable scan order. You may do it. In addition, as shown in FIGS. 6, 7, 3, and 4, the scan control unit 18 prepares several scan orders in advance, and calculates evaluation values based on these several scan orders. The scan order when the evaluation becomes the best may be specified as the most suitable scan order. Also, several scan orders are prepared in advance, and for each of these scan orders, a code is provided. The calculation order may be calculated as an evaluation value, and the scan order when the evaluation value becomes smaller than a preset encoding error may be specified as the most suitable scan order.

[Decryption device]
Next, the decoding apparatus according to the embodiment of the present invention will be described. FIG. 5 is a block diagram illustrating a configuration of the decoding apparatus. The decoding device 2 includes an entropy decoding unit 21, an inverse scan unit 22, an inverse quantization unit 23, an inverse orthogonal transform unit 24, and an inverse rotation unit 25. The decoding device 2 receives the bit stream output from the encoding device 1 and performs entropy decoding, inverse scanning, inverse quantization, inverse orthogonal transform, and inverse rotation to generate a decoded image that is an original image. . Hereinafter, each part which comprises the decoding apparatus 2 is demonstrated in detail.

  The entropy decoding unit 21 receives a bit stream from the encoding device 1 and performs entropy decoding to obtain a one-dimensional quantized transform coefficient, rotation angle, center position, and scan order. Then, the entropy decoding unit 21 outputs the one-dimensional transform coefficient to the reverse scanning unit 22, the rotation angle and the center position to the reverse rotation unit 25, and the scan order to the reverse rotation unit 25.

  The inverse scan unit 22 receives the one-dimensional transform coefficients and the scan order from the entropy decoding unit 21, reverse scans the one-dimensional transform coefficients according to the scan order, and rearranges the transform coefficients into two-dimensional transform coefficients. The inverse quantization unit 23 receives a two-dimensional transform coefficient from the inverse scan unit 22 and performs inverse quantization. The inverse orthogonal transform unit 24 receives the transform coefficient subjected to the inverse quantization from the inverse quantization unit 23 and performs an inverse orthogonal transform to obtain an image signal. Then, the reverse rotation unit 25 performs reverse rotation processing on the image signal input from the inverse orthogonal transform unit 24 using the rotation angle and center position input from the entropy decoding unit 21, and a decoded image having the same angle as the original image is obtained. obtain.

  As described above, according to the encoding device 1 according to the embodiment of the present invention, the original image is rotated by the rotation angle and the center position set from the predetermined evaluation value, and the scan obtained from the predetermined evaluation value is obtained. The transform coefficients are rearranged according to the order, and the rotation angle, the center position, and the scan order with the highest encoding efficiency are specified. As a result, the frequency components of the transform coefficients of the image can be concentrated in a certain direction (for example, the vertical frequency direction or the horizontal frequency direction), and the scan order of the transform coefficients is matched with the concentration direction (bias) of the frequency components. It can be rearranged in one dimension. That is, as a one-dimensional transform coefficient, zeros can be collected in the rear part, so that the amount of processing information for entropy coding can be reduced. Therefore, it is possible to improve the overall efficiency of image encoding.

  The present invention has been described with reference to the embodiment. However, the present invention is not limited to the above embodiment, and various modifications can be made without departing from the technical idea thereof. For example, in the above embodiment, the rotation control unit 16 and the scan control unit 18 calculate the difference between the original image and the reconstructed image, and use the linear sum of the difference and the information amount of the bitstream as the evaluation value. The new rotation angle, the center position, and the scan order are set so that the evaluation value is minimized. When such a bit stream is used, the scan unit obtains the information amount of the bit stream. 14 and the entropy encoding unit 15 are necessary, which increases the processing amount and may not be realistic. Therefore, the rotation control unit 16 and the scan unit 14 may use the difference between the original image and the reconstructed image as an evaluation value, and do not include the information amount of the bitstream in the evaluation value. Thereby, the processing load for setting the rotation angle, the center position, and the scan order can be reduced.

  In the above embodiment, the rotation control unit 16 sets the rotation angle and the center position while fixing the scan order, and the scan control unit 18 sets the scan order while fixing the rotation angle and the center position. However, the setting of the rotation angle and center position by the rotation control unit 16 and the setting of the scan order by the scan control unit 18 are mutually related. Therefore, the rotation control unit 16 and the scan control unit 18 may identify the optimum rotation angle, center position, and scan order while evaluating the effect of the evaluation value by combining them without fixing them. .

It is a block diagram which shows the structure of the encoding apparatus by embodiment of this invention. It is a figure for demonstrating the process of a rotation part. It is a figure which shows the example of the scanning method giving priority to the vertical direction. It is a figure which shows the example of the scanning method which gives priority to a horizontal direction. It is a block diagram which shows the structure of the decoding apparatus by embodiment of this invention. It is a figure which shows the example of the scanning method by a zigzag scan. It is a figure which shows the example of the scanning method by a vertical priority scan. It is a figure which shows the 1st example of the quantized conversion factor. It is a figure which shows the 2nd example of the transform coefficient quantized.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Encoding apparatus 2 Decoding apparatus 11 Rotation part 12 Orthogonal transformation part 13 Quantization part 14 Scan part 15 Entropy encoding part 16 Rotation control part 17 Local decoding part 18 Scan control part 21 Entropy decoding part 22 Inverse scanning part 23 Inverse quantum Conversion unit 24 Inverse orthogonal transform unit 25 Reverse rotation unit

Claims (8)

Orthogonal transformation and quantization are performed on the rectangular image of each small area constituting the image, the quantized two-dimensional transform coefficient is scanned to generate a one-dimensional transform coefficient, and the one-dimensional coefficient is encoded. In the encoding device for applying
A rotation unit that rotates the rectangular image to generate an oblique image;
When a new rectangular image including the diagonal image generated by the rotation is generated and a uniform pixel value is set for pixels in a region other than the diagonal image in the new rectangular image , the periphery of the external neighborhood in the diagonal image A value that holds the pixel value as it is or a value obtained by averaging pixel values around the external vicinity in the oblique image as a uniform pixel value ;
An angle center position setting unit that sets a rotation angle of the rectangular image rotated by the rotation unit and a center position of the rotation based on an evaluation value indicating encoding efficiency;
An encoding apparatus that performs orthogonal transformation, quantization, scanning, and encoding on a new rectangular image generated by the generation unit. The encoding device according to claim 1, wherein
A scan order setting unit that sets a scan order for generating a one-dimensional transform coefficient from the quantized two-dimensional transform coefficient based on an evaluation value indicating encoding efficiency;
An encoding apparatus that generates and encodes a one-dimensional transform coefficient in a scan order set by the scan order setting unit. The encoding device according to claim 1, wherein
And a scan order setting unit configured to set a scan order for generating a one-dimensional transform coefficient from the quantized two-dimensional transform coefficient based on an evaluation value indicating encoding efficiency. Encoding device. The encoding device according to claim 1 or 3,
The angle center position setting is performed by selecting a rotation angle from a plurality of preset rotation angles. In the encoding device according to any one of claims 1 to 4,
An encoding apparatus, wherein the evaluation value indicating the encoding efficiency is a value based on a difference between the rectangular image before the orthogonal transformation and the rectangular image after quantization. The encoding device according to claim 5, wherein
An encoding apparatus, wherein the evaluation value indicating the encoding efficiency is a value based on the difference and the information amount of the encoded rectangular image. A bit stream of an image encoded by an encoding device is input, decoding is performed, the decoded one-dimensional transform coefficient is reverse-scanned to generate a two-dimensional transform coefficient, and the two-dimensional transform In a decoding device that performs inverse quantization and inverse orthogonal transform on coefficients,
A decoding unit that inputs and decodes a bitstream of an image generated by the encoding device according to claim 1;
The rotation angle and center position set in the encoding device are obtained from the bitstream, the image is reversely rotated based on the rotation angle and center position, and the oblique image is obtained from the new rectangular image generated by the encoding device. And a reverse rotation unit for obtaining the decoding device. The decoding device according to claim 7, wherein
A decoding unit that inputs and decodes a bitstream of an image generated by the encoding device according to claim 2;
A reverse scan unit that obtains a scan order set in the encoding device from the bitstream, and reverse-scans the one-dimensional transform coefficients according to the scan order to generate a two-dimensional transform coefficient; A decoding device characterized.

Images