TW200948092A - Dynamic image encoding/decoding method and device - Google Patents

Abstract

Provided is a dynamic image encoding method which uses an encoded image as a reference image for predicting the next image to be encoded. The method includes: a step for applying a filter to a locally decoded image of an encoded image so as to generate a restored image; a step for setting filter coefficient information on the filter; a step for encoding the filter coefficient information; a step for encoding the locally decoded image to be used as a reference image or specific information indicating the restored image; and a step for storing the locally decoded image or the restored image as a reference image in a memory according to the specific information.

Description

200948092 VI. Description of the Invention: [Technical Field] The present invention relates to an animation encoding/decoding method and apparatus, and more particularly to the 'filter coefficient information of a loop filter set on the encoding side and transmitted An animation encoding/decoding method and apparatus for obtaining an image quality improvement effect by using the decoding side. Lu [Prior Art] In the animation encoding/decoding method of orthogonally converting an image in units of pixel blocks and performing quantization processing of a conversion coefficient, image quality deterioration called block distortion occurs in the decoded image. In response to this, G. Bjontegaard, "Deblocking filter for 4x4 based coding", ITU-T Q.15/SG16 VCEG document, Q 1 5 - J-27, May 2000 (hereinafter referred to as "Deblocking filter for 4x4 based coding" The deblocking filter described in the method is to make the upper block distortion visually less conspicuous by applying a low-pass filter at the junction of the block, and obtain a subjectively good image. The block diagram of the decoding/decoding device with deblocking filter in "Deblocking filter for 4x4 based coding" is shown in Figure 34. The block filter is recorded, which is the deblocking filter processing shown in Figure 34. Like section 901, it is used in the loop of the encoding and decoding device, and is therefore also referred to as a loop filter. By designing a loop filter, the block distortion of the reference image used in the prediction can be reduced, especially in a high-compression bit rate band which is prone to block distortion, which can improve the coding efficiency. . -5- 200948092 However, the above-mentioned deblocking filter is a process that reduces the visually conspicuous deterioration by blurring the block boundary. It is not necessarily the case that the error with the input image is small, and sometimes it may cause loss. Fine texture, etc., resulting in reduced image quality. Further, since the image processed by the filter is used as a reference image and used for prediction of the next image to be encoded, there is a problem that the image quality due to the filter is reduced and the image is propagated to the predicted image. On the other hand, a filter that is different from the loop filter and acts only on the image output by the decoder is called a back-end filter. The latter filter is used because the image after the filter is not used as a reference image, so that the effect of the filter does not propagate to the predicted image. Japanese Laid-Open Patent Publication No. 2001-2751 1 0 provides an animation decoding method for dynamically switching whether the deblocking filter is used as a loop filter or as a back-end filter. A block diagram of an encoding/decoding apparatus having a switching method of loop/rear filter in the Japanese Patent Publication No. 2001-2751-01 is shown in Fig. 35. In the animation decoding method of the Japanese Laid-Open Patent Publication No. 2001-275110, the deblocking filter processing unit 902 included in the animation decoding device of FIG. 35 generates a decoded image to which the deblocking filter is applied, and The image is output and output. On the other hand, in the encoding parameter extracting unit 904, the quantization parameter is extracted from the encoded data; in the switching unit 903, based on the 前 of the pre-quantized parameter, whether or not the image after the filter processing is to be regarded as the reference image is controlled. And use. By means of the action in the switching unit 903, the deblocking filter is used as a loop filter on the high compression bit rate band with a high deblocking filter effect, on the low compression bit rate band. This is controlled by using the deblocking filter as a back-end filter. However, in the Japanese Patent Publication No. 6-1-200948092, the same problem is not found on the encoding side, so that there is a problem of mismatch between the encoding side and the decoding side. Moreover, since * is not a technique for improving the image quality of the reference image on the encoding side, it is impossible to obtain the encoding efficiency improvement effect. On the other hand, S. Wittmann and T. Wdi, “Post-filter SEI message for 4:4:4 coding", JVT of ISO/IEC MPEG & ITU-T VCEG, JVT-S030, April 2006 (hereinafter referred to as "Post-filter 0 SEI message for 4:4:4 coding" describes an animation encoding/decoding method which encodes the filter coefficient information of the back-end filter on the encoding side and uses it on the decoding side. The decoded filter coefficient information is used for the back-end filter processing. The block diagram of the animation encoding/decoding apparatus in "Post-filter SEI message for 4:4:4 coding" is shown in Fig. 36. • The animation of Fig. 36 The rear filter setting unit 905 included in the encoding device sets the predetermined filter coefficient information and outputs the mirror coefficient information 90. The filter coefficient information 90 is encoded, decoded on the decoding side, and Q is decoded in the animation. The rear-end filter processing unit 906 included in the device is used to perform the rear-end filter processing. In the animation encoding/decoding method of "Post-filter SEI message for 4:4:4 coding", the decoded side is reduced on the encoding side. Error with the input image The filter coefficient information is set, so that the image quality of the output image to which the back-end filter is applied on the decoding side is improved. However, the method of "Postfilter SEI message for 4:4:4 coding" is not to draw The image that has been upgraded is used as a reference image, and the encoding efficiency improvement effect cannot be obtained on the encoding side. 200948092 [Summary of the Invention] As described above, the method disclosed by Deblocking filter for 4x4 based coding does not necessarily result in image quality. Ascension, there is a problem that the deterioration of the image quality caused by the filter propagates to the predicted image. Further, in the method disclosed in Japanese Laid-Open Patent Publication No. 2001-275 1 10, the switching of the loop/post-segment filtering is performed only on the decoding side, so that there is a problem of mismatch between the encoding side and the decoding side. Moreover, the method disclosed in "Post-filter SEI message for 4:4:4 coding" is to improve the image quality of the output image on the decoding side, and it is impossible to promote the image quality of the reference image used for prediction. Get the coding efficiency improvement effect. It is an object of the present invention to provide an animation encoding/decoding method and apparatus for encoding the filter coefficient information set on the encoding side and decoding the filter coefficient information on the decoding side for use in animation encoding/decoding. By switching the loop filter processing in the same way by the encoding side and the decoding side, it is possible to suppress the propagation of image quality deterioration while promoting the image quality of the reference image used for prediction. Upgrade. The same state of the present invention provides an animation coding method, which belongs to an animation coding method for using a coded image as a reference image for prediction of the next coded image, which is characterized by: having an encoding a partial decoded image of the image 'applicable filter' to generate a reconstructed image; and a step of setting the filter coefficient information of the pre-filter; and a step of encoding the pre-filter filter information; and The step of encoding the local decoded image or the pre-reset information as the reference image is to be used as the reference image, and the recording of the local decoded image or the pre-recovered image based on the pre-recorded specific information* The steps in the memory. According to another aspect of the present invention, there is provided an animation decoding method for an animation decoding method to use a decoded image as a reference image for prediction of a next incoming image, which is characterized in that: ❹ image application filter is generated to generate a step of restoring the image; and a step of decoding the filter number information; and decoding the specific information to be used as the decoded image or the restored image to be used as a reference; and decoding the image or restoring based on the specific information The image is stored as a step in the memory. [Embodiment] Hereinafter, embodiments of the present invention will be described with reference to the drawings.第 (First Embodiment) An animation composition according to the first embodiment will be described with reference to Fig. 1 . In the following, the animation coding apparatus 1000 shown in FIG. 1 has a pre-forming unit 101, a subtractor 102, a conversion quantization unit 103, an entropy coding, an inverse conversion, and an inverse quantization unit. 105, the adder 106, the filter 107, the reference image buffer 108, and is controlled by the encoding control. The image is characterized by the pre-image, which is decoded by the decoding method: the step of decoding the image of the filter and the code is the tear of the reference image device. Measured signal generation unit 10 Mirror processing unit 109 -9- 200948092 The prediction signal generation unit 101 acquires the encoded reference video signal 19 stored in the reference video buffer 108 and performs predetermined prediction processing. , the predicted image signal 1 1 is output. The prediction processing may be, for example, prediction of the temporal direction of motion prediction and motion compensation, or prediction of the spatial direction from the encoded image in the picture. The subtractor 102 calculates the difference between the obtained input image signal 10 and the predicted image signal 11, and outputs the predicted error image signal 12. The conversion/quantization unit 103 first obtains the prediction error video signal 12 and performs conversion processing. Here, the conversion/quantization unit 103 performs orthogonal conversion of the prediction error video signal 12 using, for example, DCT (Discrete Cosine Transform), and generates a conversion coefficient. As another embodiment, a conversion method may be generated by a method such as wavelet conversion or independent component analysis. Next, the conversion/quantization unit 103 performs quantization processing of the generated conversion coefficient based on the quantization parameter set by the encoding control unit 109, and outputs the quantized conversion coefficient 1 3 . The quantized conversion coefficient 13 is input to the entropy coding unit 104, which will be described later, and is also input to the inverse conversion/inverse quantization unit 105. The inverse transform/inverse quantization unit 105 inversely quantizes the quantized transform coefficients 13 in accordance with the quantization parameters set by the encoding control unit 109, and inversely converts the obtained transform coefficients (for example, inverse discrete cosine transform). Etc.), the prediction error image signal 15 is output. The adder 016 adds the prediction error video signal 15 obtained from the inverse conversion/inverse quantization unit 156 and the predicted video signal 11 generated in the prediction signal generation unit 1 〇1. The local decoded video signal 16 is output.

-10- 200948092 The loop filter processing unit 107 obtains the local decoded video signal I6 and the input video signal 10, and uses the reference video signal 19 and the filter coefficient information 17 to indicate that the local decoded image or the restored image is to be used. The specific information used as the reference image is specifically used, and specifically, the switching information 18 for switching to the local decoded image or the restored image is output. The detailed description of the loop filter processing unit 107 will be described later. The reference video buffer 1 〇 8 is used to temporarily store the reference video signal Q 19 that has been acquired from the loop filter processing unit 107. The reference video signal 19 stored in the reference video buffer 108 is referred to when the predicted video signal generation unit 101 generates the predicted video signal 11. On the other hand, the entropy coding unit 104 obtains the filter coefficient information 丨7 and the switching information 1 in addition to the quantized conversion coefficient 1 3, and also obtains the prediction mode information, the block size switching information, and the motion vector. Encoding parameters such as quantization parameters are subjected to entropy coding (for example, Huffman coding or arithmetic coding, etc.) and then output to the coded material 14. The coding control unit 109 performs control of generating the coding amount feedback control, quantization control, mode control, and the like, and performs overall coding processing. Next, the loop filter processing unit 107 in the animation encoding apparatus according to the first embodiment will be described in detail with reference to Figs. 2 and 3 . Hereinafter, the components of FIGS. 2 and 3 will be described separately. The loop filter processing unit 107 shown in FIG. 2 includes a filter setting unit 110, a switching filter processing unit U1, and a switching information generating unit 112. Further, the "switching mirror processing unit 1 1 1 is as shown in FIG. As shown, the filter processing unit 113 and the return mirror switching unit 114 are provided. The switch SW of Fig. 3 switches the connection between terminal -11 - 200948092 sub A and terminal B. The filter setting unit 110 obtains the local decoded video signal 16 and the input video signal 1〇, and sets the determined filter coefficient information 17. Details of the setting method of the filter coefficient information 17 will be described later. The filter coefficient information 17 that has been set is input to the switching filter processing unit 111 and the entropy encoding unit 104 which will be described later. The switching filter processing unit 111 has a filter processing unit 113 and a loop filter switching unit 114 therein, and acquires the local decoded image 16, the filter coefficient information 17 and the switching information 18, and then outputs the reference image signal 19. . The switching information generating unit 112 acquires the reference video signal 19 from the switching filter processing unit ill, acquires the input video signal 1〇, and generates the switching information 18 in accordance with the predetermined switching determination method. The generated switching information 18 is input to the switching filter processing unit 111 and also input to the entropy encoding unit 104. Details of the handover determination method will be described later. The filter processing unit 113 obtains the local decoded video signal 16 and the filter coefficient information 17, and filters the local decoded video signal 16 according to the filter coefficient information 17 to generate the restored video signal 20. The restored image signal 20 that has been generated is input to the loop filter switching unit 114 which will be described later. The loop filter switching unit 114 obtains the switching information 18, and switches the connection of the terminal A and the terminal B with the switch SW provided therein according to the switching information 18, and uses the local decoded video signal 16 or the restored video signal 20 as a reference. The image signal 19 is output. The above is the configuration of the animation encoding device according to the first embodiment. Next, the operation of the loop filter in the animation encoding device according to the first embodiment will be described in detail with reference to Figs. 1, 2, 3 and 4. -12- 200948092 FIG. 4 is a flowchart showing the operation of the loop filter in the animation encoding device 10000 according to the first embodiment. First, once the input* video signal 10 is input to the animation encoding process 1000 of FIG. 1, the subtractor 102 performs subtraction processing of the input video signal 10 and the predicted video signal 11 obtained from the prediction signal generating unit 101 to generate a prediction error. Image signal 12. The generated prediction error video signal 12 is converted and quantized by the conversion/quantization unit 103, and is output as a conversion coefficient 12 quantized by φ, and is encoded by the entropy coding unit 104. On the other hand, the quantized conversion coefficient 13 is inversely converted and inversely quantized by the inverse conversion/inverse quantization unit 105 included in the animation coding apparatus 1 000, and is output as the prediction error video signal 15. The prediction error video signal 15 is added to the predicted video signal 11 outputted by the prediction signal generating unit 101, and is added to the adder 106 to generate a local decoded video signal 16. The series of processing described above is a common encoding process in the so-called hybrid coding animation coding for predictive processing and conversion processing. The characteristic processing in the animation encoding device 1000 according to the first embodiment, that is, the operation of the loop filter, will be described in detail with reference to FIGS. 2, 3, and 4. First, the filter setting unit 110 included in the feedback filter processing unit 107 of Fig. 2 receives the local decoded video signal 16 and the input video signal 10, and sets the filter coefficient information 17 (step S1100). Here, the filter setting unit 110 uses a two-dimensional Wiener filter generally used in image restoration to design a filter coefficient such that a filter-processed image is applied to the locally decoded image signal 16 and the input image signal 10 is used. Level-13- 200948092 The squared error is minimized, and the designed filter coefficient and the 表示 indicating the filter size are set as the filter coefficient information 17. The filter setting unit 110 outputs the set filter coefficient information 17 to the filter processing unit 113 of Fig. 3 and also outputs it to the entropy encoding unit 104. Next, the switching filter processing unit 111 of FIG. 2 receives the local decoded video signal 16, the filter coefficient information 17 and the switching information 18, and based on the switching information 18, the local decoded video signal 16 or has been in the filter processing unit 113. The generated restored video signal 20 is output as the reference video signal 19 (steps S1101 to S1109). Here, first, in the case where the local decoded video signal 16 is regarded as the reference video signal 19 and the restored video signal 20 is referred to as the reference video signal 19, the switching information generating unit 112 performs the switching determination process. The switching information 18 required to determine which of the locally decoded video signal 16 or the restored video signal 20 is to be used as the reference video signal 19 is generated. Based on the generated switching information 18, the switch SW inside the loop filter switching unit 114 is switched to output the reference image signal 19. The detailed description of the operations from step S1101 to step S1109 in the loop filter processing unit 1 to 7 is as follows. First, the switch SW in the loop filter switching unit 114 of Fig. 3 is connected to the terminal A, and the local decoded video signal 16 is input as the reference video signal 19 to the switching information generating unit 112 (step S1101). Then, the switch SW is connected to the terminal B, and the restored video signal 20 is input to the switching information generating unit 112 as the reference video signal 19 (step S1102). Here, once the switch SW is connected to the terminal B, the filter The processing unit 113 performs a filter processing on the local decoded video signal 16 based on the filter coefficient information 17 to generate a restored video signal 20. As an example, if the pixel of the position (X, y) on the locally decoded image is F(x, y), the width of the 2-dimensional filter is W, the height is Η, and the filter coefficient is h(i, j). (i - ^W' w = W/2' h = H/2), the restored image G(x, Υ ) can be expressed by the following equation. [Number 1] G(x»y)= i=-wj=-h

G, the switching information generating unit 112 of FIG. 2 calculates the residual squared sum SSDA of the locally decoded video signal 16 and the input video signal 10, and the residual squared sum SSDB of the restored video signal 20 and the input video signal 10 (step S1103). ). Here, it is assumed that the switching determination process is performed on each partial area of the image, so that the pixel position in the local area is i, the total number of pixels N, the local decoded video signal 16 is Fi, the restored video signal 20 is Gi, and the input video signal is input. 10 is Ii', then SSDA and SSDB can be expressed by the following formula. [Number 2] ssda /=1

SSDB i = l Next, the following handover determination processing is performed based on ssda and ssdb (step S1104). If SSDAo is below SSDB, 0 is set to the handover information 18, i.e., l〇〇p_filter_flag (step s 1 105). Otherwise ‘If SSDA〇 is greater than SSDb, set -15-200948092 1 to loop_filter flag (step S110). Here, an example of a reference pixel in which the video is divided into 16x16 pixel units, which is called a macroblock, and the above-described switching determination processing is performed in this unit is shown in Fig. 33. When the local area is a macro block, the N system of the above [number 2] is 256, and the switching information 18 is outputted in a macro block unit. In another embodiment, the switching determination process may determine the image signal according to the frame unit or the slice unit or the block unit different from the size of the macro block. In this case, the switching information 18 also corresponds to the determination. The unit of results is output. Next, the loop filter switching unit 114 of Fig. 3 receives the loop information_flag_flag, which is the generated switching information 18, and switches the switch SW provided inside based on the l〇〇p_filter_flag (step S1107). When the loop_filter_flag is 0, the loop filter switching unit 114 connects the switch SW to the terminal A, and temporarily stores the locally decoded video signal 16 as the reference video signal 19 in the reference video buffer 1〇8 (step) 81108). On the other hand, when 1〇〇卩_^1161>_^& Lu is 1, the loop filter switching unit 114 connects the switch SW to the terminal B', and restores the restored image signal 20 as the reference image signal 19 It is temporarily stored in the reference image buffer 108 (step S1109). The above is the operation from step S1101 to step s 1 1 0 9 in the loop filter processing unit 107. Finally, the filter coefficient information 1 7 generated in the filter setting unit 11 及 and the switching information 18 ′ generated in the switching information generating unit 112 are encoded by the entropy encoding unit 104 and converted to be quantized. The coefficient 13, the prediction mode information, the block size switching information, the motion vector, the quantization parameter, and the like are multiplexed into a bit stream and then transmitted to the animation decoding device 2000 to be described later (step SI 1 10). Here, the syntax of the filter coefficient information 17 and the switching information 18 method will be described in detail with reference to FIG. 31. In the following example, the mirror coefficient information 17 is set in slice units, and the switching information 18 is set in macroblock units. The φ grammar system is mainly composed of three parts. In the high-level grammar (1 900 ), the grammar information of the upper layer above the slice is inserted. In the slice level grammar (1 903), the information necessary for each slice is clearly recorded; in the macro block level grammar (1 907), the conversion coefficient data necessary for each macro block is clearly recorded or Forecast mode information, motion vectors, etc. Each grammar is further composed of a more detailed grammar, a high-level grammar (1 900 ), which is composed of sequence parameter set grammar (1901) and image parameter set grammar (1902). The grammar is composed. The slice level syntax (1 903), 〇 is composed of a slice header syntax (1 904), a slice data syntax (1 905), and a loop filter data syntax (1906). Then, the macro block level grammar (1907) is composed of a macro block grammar (1908) and a macro block prediction grammar (1 909). In the loop filter data syntax (1 906), as shown in FIG. 32(a), the parameters related to the loop filter of the present embodiment, that is, the filter coefficient information 17 and the switching information 1 8 are described. . Here, the filter coefficient information 1 7 is the filter_coeff[cy][cx] of Fig. 32 (a) is the coefficient of the 2-dimensional filter, and the filter_size_y and filter_size_x are the parameters for determining the filter size. In the -17-200948092, the 値' indicating the filter size is described in the grammar. However, as another embodiment, a predetermined fixed 値 may be used instead of the grammar. However, if the filter size is set to be fixed, it is necessary to use the same flaw in the animation encoding apparatus 100 0 and the animation decoding apparatus 200 described later. Further, in Fig. 32 (〇's l〇op_filter_flag is the switching information 18, the total number of macroblocks in the slice, that is, the NumOfMacroblock loop filter flag > is transmitted. The above is about the animation coding apparatus 1000. Description of the operation of the loop filter. Next, the animation decoding device for the animation encoding device 100 will be described. The configuration of the animation decoding device according to the first embodiment will be described with reference to Fig. 5. The components of the animation decoding apparatus shown in FIG. 5 include an entropy decoding unit 2〇1, an inverse conversion/inverse quantization unit 202, a prediction signal generation unit 203, an adder 204, and a switching. The filter processing unit 205 and the reference video buffer 206 are controlled by the decoding control unit 207. The entropy decoding unit 2 0 1 ' is in accordance with the syntax structure shown in Fig. 31 for high-level syntax, slice level syntax, and giant Each of the block level grammars sequentially decodes the code columns of the grammars of the coded data 14, and restores the quantized conversion coefficients 1 3 , the filter coefficient information 17 , the switching information 1 8 , and the like. The inverse quantization unit 202 obtains the quantized conversion coefficient 13 and performs inverse quantization and inverse orthogonal conversion (for example, inverse discrete cosine conversion) to output a prediction error video signal 15. Here, although it is for inverse orthogonality In the case of performing the wavelet conversion or the like in the animation encoding device 1000, the inverse conversion/inverse quantization unit 202 performs the inverse quantization and inverse wavelet conversion corresponding thereto. The prediction signal generation unit 203 acquires the decoded reference video signal 19 stored in the reference video buffer 206 and performs predetermined prediction processing to output the predicted video signal 11. The prediction processing uses, for example, motion compensation. The prediction of the time direction or the prediction of the spatial direction from the image of the Q-coded image in the picture may be performed, but it is necessary to perform the same prediction process as the animation coding apparatus 1 ,, which must be noted. The unit 204 adds the obtained prediction error image signal 15 and the predicted image signal 11 to generate a decoded image signal 21. Switching filter processing The unit 205 obtains the decoded video signal 21, the filter 'coefficient information 17 and the switching information 18, and outputs the reference video signal 19. The detailed description of the switching filter processing unit 205 will be described later. The reference video signal 19 that has been acquired from the switching filter processing unit G 205 is temporarily stored. The reference video signal 19 stored in the reference video buffer 20 6 is generated by the predicted signal generating unit 203. When the video signal 11 is predicted, the decoding control unit 207 is referred to, and decoding timing control or the like is performed to control the entire decoding. Next, the switching filter processing unit 205 in the animation decoding device according to the first embodiment will be described in detail with reference to Fig. 6 . Hereinafter, the components of Fig. 6 will be described separately. -19- 200948092 The switching filter processing unit 205A shown in Fig. 6 includes a filter processing unit 208 and a return filter switching unit 209. The switch SW switches the connection between the terminal A and the terminal B. The filter processing unit 208 receives the decoded video signal 21 and the filter coefficient information 17 that has been restored by the entropy decoding unit 201, and filters the decoded video signal 21 according to the filter coefficient information 17 to generate a restored image. Image signal 20. The restored image 20 is input to the loop filter switching unit 209, which will be described later, and is also outputted as the timing at which the output of the video signal 22 is managed by the decoding control unit 207. The loop filter switching unit 209 receives the switching information 18 that has been restored by the entropy decoding unit 201, and switches the connection of the terminal A and the terminal B with the switch SW provided therein according to the switching information 18, and decodes the video signal. 21 or the restored video signal 20 is output as the reference video signal 19. The above is the configuration of the animation decoding device according to the first embodiment. Next, the operation of the loop filter in the animation decoding device according to the first embodiment will be described in detail with reference to Figs. 5, 6 and 7'. Further, Fig. 7 is a flowchart showing the operation of the loop filter in the animation decoding device 2000 according to the first embodiment. First, 'once the encoded data 14' is input to the animation decoding device 2 000 of FIG. 5, the entropy decoding unit 201' is divided by the conversion coefficient 13, the filter coefficient information 17, the switching information 18', the prediction mode information, and the region. Block size switching information, motion vectors, quantization parameters, etc., are decoded in accordance with the syntax structure of FIG. Then, the conversion coefficient 13 decoded by the entropy decoding unit 201 is input to the inverse conversion/inverse quantization unit 202, and inverse quantization is performed in accordance with the quantization parameter set by the 200948092 code control unit 207. The inverse orthogonal transform (e.g., discrete cosine transform *, etc.) is performed on the transform coefficients that have been inversely quantized, and the prediction error video signal 15 is restored. The prediction error video signal - 15 is added to the predicted video signal 1 1 outputted by the prediction signal generating unit 203 in the adder 204 to generate a decoded video signal 21. The above-mentioned series of processing is a decoding process which is generally common in animation coding of a so-called hybrid coding which performs prediction processing and conversion processing. Here, the characteristic processing in the animation decoding device 2000 according to the first embodiment, that is, the operation of the loop filter, will be described in detail with reference to FIGS. 6 and 7 . First, the entropy decoding unit 201 entropy decodes the filter coefficient information 17 and the switching information 18 in accordance with the syntax structure of Fig. 31 (step S2100). In the loop 泸 mirror data grammar (1906) to which the slice level grammar (19〇3) in the grammatical structure of FIG. 31 belongs, as shown in FIG. 32(a), the loop filter of the present embodiment is described. The relevant parameters are the filter coefficient 〇 information 17 and the switching information 18. Here, the filter coefficient information 17, that is, filter_coeff[cy][Cx] of Fig. 32 (a) is a coefficient of a two-dimensional filter, and filter_size_y and fiHer_size_x are 値 which determine the size of the filter. In this case, the 尺寸 of the filter size is described in the grammar. However, as another embodiment, the predetermined fixed 値 may be used as the filter size without being described in the grammar. However, if the filter size is set to a fixed size, the same animation must be used in the aforementioned animation encoding apparatus 1000 and animation decoding apparatus 2000. This must be noted. Moreover, the loop_filter_flag of FIG. 3 2 (a) is the switching information 18, and the total number of blocks in the slice 200948092, that is, the total number of blocks of NumOfMacroblock, l〇〇p_filter_flag, is decoded. Next, the filter processing unit 208 of Fig. 6 receives the decoded filter coefficient information 17 (step S2101). The loop filter switching unit 209 receives the decoded switching information 18, that is, the l〇〇p_filter_flag (step S2102), and switches the switch SW (which is provided therein) based on the switch of the l〇〇p_filter_flag ( Step S2103). When the loop_filter_flag is 〇, the loop filter switching unit 209 connects the switch SW to the terminal A, and temporarily stores the decoded video signal 2 1 as the reference video signal 19 in the reference image buffer 206 (step S2104). ). On the other hand, when l〇〇p_filter_flag is 1, the loop filter switching unit 209 connects the switch SW to the terminal B, and temporarily restores the restored video signal 20 as the reference video signal 19 to the reference image buffer. 206 (step S2105). Here, once the switch SW is connected to the terminal B, the filter processing unit 1 1 3 performs filter processing on the decoded video signal 2 1 based on the filter coefficient information 17 to generate the restored video signal 20. As an example, if the pixel at the position (x, y) on the decoded image is F(x, y), the width of the 2-dimensional filter is W, the height is H, and the filter coefficient is h(i, j) ( w^ w , -hS j S h, w = W/2, h = H/2 ), the restored image G ( x, y ) can be represented by [number 1]. Here, the loop filter switching unit 209 accepts the switching information 18 in the macro block unit in accordance with the syntax of Fig. 32 (a), and switches the switch SWT. As another embodiment, when the animation coding apparatus 1 is in a frame unit or a slice unit or a block unit different from the size of the macro block, the unit 20-2240092 and the measured portion 20 When the code information is encoded in the code to encode the switching information 18, 'the animation decoding device 2000 performs the decoding of the switching information 18 in the same unit' to switch the switch SW of the switching unit 209. • The above is an explanation of the operation of the loop filter in the animation decoding device 2000. As described above, the animation encoding apparatus according to the first embodiment can perform the filter processing by setting the filter coefficient information of the loop filter so that the error between the input image and the pre-φ signal is minimized. The picture quality is improved. Moreover, by switching the filter area for each field by the loop filter switching unit 114, which of the locally decoded video signal 16 and the restored video signal is to be used as a reference image, the image quality is caused by the filter. In the low field, the restored video signal 20 is not used as a reference image, which prevents the degradation of the image quality, and the improvement of the image quality is to use the restored image signal 20 as a reference image. It can improve the prediction accuracy. Further, according to the animation decoding apparatus 2000 according to the first embodiment, by performing filter processing and switching processing using the filter coefficient information switching information similar to the animation encoding apparatus 1000, the animation editing apparatus 1000 can be secured. The reference image is synchronized with the reference image in the animation decoding device 2000. Further, in the switching filter processing unit 205A of the animation decoding device 2000 according to the first embodiment, the reference video signal 19 is output as the output video signal 22, but the switching filter processing as shown in FIG. 8 is also possible. Similarly, the decoded video signal 21 is output as the output video signal -23-200948092 22, or the restored video signal 20 can be regarded as the output video signal as in the switching filter processing unit 205C of FIG. 22 and output. In this case, switching information indicating that the decoded image or the restored image is used as the output image is generated. Further, as in the switching filter processing unit 205D of Fig. 10, a rear filter switching unit 2 1 0 is newly provided, and the switch SW2 is switched by switching the information 1 to switch the output video signal 22. In this case, as the switching information 18 for switching the output video signal 22, for example, as shown in Fig. 32(d), the post_filter_flag is described in the syntax in the slice unit, thereby switching the switch SW2. Similarly to l〇〇p_filter_flag, p〇st_filter_flag can also be described by a tile unit or a macro block unit or a block unit different from the size of the macro block. Further, in the animation encoding device 1000 and the animation decoding device 2000 according to the first embodiment, although the local decoded video signal 16 is subjected to filter processing, the local decoded video signal 16 may be subjected to the previous execution of the past region. Image after block filter processing. Further, the animation encoding device 10 00 and the animation decoding device 20 00 can be realized, for example, by using, for example, a general-purpose computer device as a basic hardware. That is, the prediction signal generation unit 101, the subtractor 102' conversion/quantization unit 103, the entropy coding unit 104, the inverse conversion, the inverse quantization unit 105, the adder 1〇6, the loop filter processing unit 107, and the reference image buffer The area 108, the encoding control unit 109, the filter setting unit 110, the switching filter processing unit 111, the switching information generating unit 112, the filter processing unit 113, the loop filter switching unit 114, the entropy decoding unit 201, and the inverse conversion. The quantization unit 202, the prediction signal generation unit 200948092 203, the adder 204, the loop filter processing unit 205, the reference video buffer 206, the decoding control unit 207, the filter processing unit 208, the loop filter switching unit 209, and The rear-end filter switching unit 210 can be realized by executing a program by a processor mounted in the computer device*. In this case, the animation encoding device 1000 and the animation decoding device 2000 are implemented by attaching the above-mentioned program to the computer device in advance, or by storing it in a memory medium such as a CD-ROM or distributing the writing program through the network. This program is implemented by suitably mounting the φ to a computer device. Further, the reference video buffer 108 and the reference video buffer 206 can be suitably used in an expansion memory such as a built-in computer device, a hard disk, or a CD-R, a CD_RW, a DVD-RAM, or a DVD-R. The media is waiting to be realized. (Second Embodiment) A configuration of an animation encoding apparatus according to a second embodiment will be described with reference to Fig. 11 . Hereinafter, the components of Fig. 11 will be described separately. The animation encoding device 3000 shown in FIG. 11 includes a switching information generation prediction unit 301A, a loop filter processing unit 302, a local decoded image buffer 303, a restored image buffer 304, and a subtractor 1〇2. The conversion unit 103, the entropy coding unit 104, the inverse conversion/inverse quantization unit 1〇5, and the adder 1 〇6 are controlled by the coding control unit 109. Here, the 'reduction unit 102, the conversion/quantization unit 1〇3, the entropy coding unit ι4, the inverse conversion/inverse quantization unit 105, the adder 106, and the coding control unit 1〇9' are the same as the first embodiment. The constituent elements of the same number in the animation encoding apparatus 1000 of Fig. 1 perform the same operation, and therefore, the explanation is omitted here: 3 - 25 - 200948092. As shown in Fig. 12, the switching information generation predicting unit 301A has a reference switching predicting unit 305A and a switching information generating unit 112 in its internal portion. In this case, the switching information generating unit 1 1 2 performs the same operation as the constituent elements of the same number in the animation encoding device according to the first embodiment. Therefore, the description thereof is omitted here. The switching information generation prediction unit 301A obtains the local decoded video signal 16, the restored video signal 20, and the input video signal 10, and outputs the predicted video signal 11 and the switching information 18. The loop mirror processing unit 312 has a filter setting unit 110 and a filter processing unit 113 as shown in Fig. 14 . Since the filter setting unit 110 and the filter processing unit 113 perform the same operations as the components of the same number in the animation encoding device according to the first embodiment, the description thereof will be omitted. The loop filter processing unit 302 obtains the local decoded video signal 16 and the input video signal 1〇, and outputs the restored video signal 20 and the filter coefficient information 17 for output. The local decoded video buffer 303 is obtained by acquiring the local decoded video signal 16 generated by the adder 1 〇 6 and temporarily storing it. The local decoded video signal 16 stored in the local decoded video buffer 303 is input to the switching information generation predicting unit 301A. The restored video buffer 3 04 acquires the restored video signal 20 generated by the loop filter processing unit 302 and temporarily stores it. The restored video signal 20' stored in the restored image buffer 304 is input to the switching information generation predicting unit 301A. The reference switching prediction unit 305A has a prediction unit 200948092 generation unit 101 and a loop filter switching unit 114 as shown in Fig. 13 . Here, the prediction signal generation unit 101 and the loop filter switching unit 114 perform the same operations as the constituent elements* of the same number in the animation encoding apparatus 1000 described in the first embodiment. Referring to the switching prediction unit 305A, the local decoded video signal 16, the restored video signal 20, and the switching information 18 are obtained, and based on the acquired switching information 18, the local decoded video signal 16 or the restored video signal 20 is used as the As a reference image, the prediction process is performed in the e-line, and the predicted image signal 1 1 is output. The prediction processing system may use, for example, prediction of the temporal direction of motion prediction and motion compensation, or prediction of the spatial direction from the decoded image in the picture. The above is the configuration of the animation encoding device according to the second embodiment. Next, the operation of the animation encoding apparatus according to the second embodiment will be described in detail with reference to Figs. 11, 12, 13, 14, and 15. Further, Fig. 15 is a flowchart showing the operation of the loop filter in the animation encoding device 3 000 according to the second embodiment. First, similarly to the conventional hybrid coding or the animation coding apparatus 1000 according to the first embodiment, prediction, conversion, quantization, and entropy coding are performed, and local decoding is performed inside the coding apparatus to generate a locally decoded video signal 16 Next, the generated partial decoded video signal 16 is temporarily stored in the local decoded video buffer 310 (step S3100). The filter setting unit 11 in the loop filter processing unit 312 obtains the local decoded video signal 16 and the input video signal 10' to set the subtraction coefficient information 17 (step S3101). Here, the 2-dimensional Wiener filter generally used in image restoration is used, and the filter coefficient is designed to be -27-200948092, which causes the image of the filter image processed by the local decoded image signal 16 to be averaged with the input image signal ίο. The squared error becomes the minimum 'The designed filter coefficient and the 表示' indicating the filter size are set as the filter coefficient information 17. The filter coefficient information 17' that has been set is similarly outputted to the filter processing unit 113' inside the loop filter processing unit 302, and is also output to the entropy coding unit 104. The filter processing unit 113 in the loop filter processing unit 302 performs filter processing on the locally decoded video signal 16 using the filter coefficient information 17 obtained by the filter setting unit 11 to generate a restored image. Signal 20 (step S3102). The generated restored video signal 20 is temporarily stored in the restored video buffer 310 (step S3103). The filter coefficient information 17 generated by the filter setting unit 110 is encoded by the entropy encoding unit 104, and is multiplexed together with the quantized conversion coefficient 13, prediction mode information, block size switching information, motion vector, quantization parameter, and the like. The bit stream is converted into a stream stream and transmitted to the animation decoding device 4000 (to be described later) (step S3104). At this time, the filter coefficient information 17 is described in the loop filter data syntax (1 906) in the slice level syntax (1903) in the syntax structure of FIG. 31, and is described as FIG. 3 2 (b) ) shown. Here, the filter coefficient information 17, that is, filter_coeff[cy][cx] of Fig. 32(b) is a coefficient of a two-dimensional filter, and filter_size_y and filter_size_x are 値 which determine the size of the galvanometer. Here, the 表示 indicating the size of the filter is described in the grammar. However, as another embodiment, the pre-defined fixed 値 may be used as the filter size without being described in the grammar. However, if the size of the subtraction mirror is set to be fixed, it is necessary to use the same flaw in the animation encoding apparatus 3000 and the animation decoding apparatus 4000 described later in -28-200948092, which must be noted. Then, the switching information generation predicting unit 301A performs the predetermined prediction processing by using the local decoded video-image signal 16 or the restored video signal 20 as the reference video, and outputs the predicted video signal 1 1 (step S3 1 05 to 3 1 1). 3). Here, first, the switching prediction unit 305A acquires the predicted video when the partial decoded video is the reference video and the predicted video when the restored video is the φ reference video, and the switching information generating unit 112 performs the based on these. The switching determination process generates switching information 1 8 for determining which of the locally decoded video or the restored video is to be used as the reference video. The switching prediction unit 3 05A switches the switch SW inside the loop filter switching unit 114 of Fig. 13 based on the generated switching information 18, and then outputs the predicted video signal 11. The detailed description of the actions of steps 3105 to 3113 is as follows. First, referring to the switching prediction unit 3 0 5 A, the switch SW in the loop filter switching unit 114 of FIG. 13 is connected to the terminal A, and the local decoded video signal 16 is regarded as the localized decoded video signal 16 in the prediction signal generating unit 101. The predicted video signal 11 is obtained by referring to the video, and is input to the switching information generating unit 112 of Fig. 12 (step S3 105). Next, referring to the switching prediction unit 305A, the switch SW is connected to the terminal B, and the predicted video signal 20 is used as the reference image in the prediction signal generating unit 1〇1 to obtain the predicted video signal 11, and the switch is input to FIG. The information generating unit 112 (step S3 106). Next, the switching information generating unit 112 of FIG. 2 calculates the residual squared sum SSDA of the predicted image and the input video signal 1 取得 obtained from the locally decoded video signal 16 and the reconstructed video signal 20 The obtained residual image summation SSDB of the predicted image and the input image signal 10 (step S3107) aSSDA and SSDB, if the predicted image obtained from the locally decoded video signal 16 is Fi, and the reconstructed video signal 20 is If the obtained predicted image is Gi, it can be expressed by the same expression as [2]. The switching information generating unit 112 performs the following switching determination processing based on the SSDAR SSDB (step S3108). If SSDA〇 is below SSDB, 切换 is set to the switching information 18, i.e., l〇oP_filter_flag (step S3109). On the other hand, if SSDA 〇 is 値 larger than SSDB, 1 is set for loop_filter_flag (step S3110). Here, the above-mentioned macroblock is subjected to the above-described switching determination processing, and the switching information 18 is outputted in the macroblock unit. In another embodiment, the switching determination process may be determined according to a frame unit or a slice unit or a block unit different from the size of the macro block. In this case, the switching information 18 also corresponds to the determination result. Unit to output. Further, when the prediction processing in the prediction signal generating unit 1 0 1 is motion prediction, the switching determination processing is described above, and a general method may be used as the motion prediction processing system, for example, for the residual square sum D, The code amount R of the parameter information such as the switching information, the reference image index, and the motion vector can be added, and the cost J can be calculated for each of the local decoded video and the restored video as follows, and the cost J can be used for the determination. [Number 3]

J = D+ λ xR

The λ system of [Number 3] is given by a fixed number and is determined based on the quantization width or the quantization parameter -30-200948092. It is possible to encode the switching information, the reference image index, and the motion vector, which are obtained at such a cost j as the minimum 値. Further, in the present embodiment, although the sum of squares of residuals is used, as other embodiments, absolute residual sums may be used, or these may be converted into Hadamards, or approximates, etc. . Alternatively, the activity of the input image can be used to create a cost' or a quantization function can be made using a quantization width or quantization parameter. Further, the prediction process is not limited to the motion error. As long as it is a prediction that requires any parameter as the prediction q process, the code J of the parameter can be used as the R and the cost J can be calculated using the above formula. Then, the loop filter switching unit 114 of FIG. 13 receives the generated switching information 18, that is, loop_filter_flag, and switches the switch SW provided inside based on l〇〇p_filter_flag (step S3111). . When the loop_filter_flag is 0, the loop filter switching unit 114 connects the switch SW to the terminal A, and outputs the local decoded video signal 16 as the reference video signal 19, and generates the 解码 in the prediction signal generating unit 1〇1. The video signal 11 is predicted (step S3112). On the other hand, when the loop_filter_flag is 1, the loop filter switching unit 14 connects the switch SW to the terminal B, and the restored video signal 20 is output as the reference video signal 19 to the prediction signal generating unit 1〇1. The predicted video signal 1 1 is generated (step S3 1 1 3 ). The above is the operation from step 3105 to step 3 1 1 3 in the switching information generation prediction unit 301A. Finally, the entropy coding unit 104 encodes the switching information 18, and together with the quantized conversion coefficient 13, the prediction mode information, the block size switching resource, the motion vector, the quantization parameter, etc. The stream is sent to an animation decoding device 4 000, which will be described later (step S3 114). At this time, the switching information 18, that is, l〇〇p_filter_flag' is in the macroblock layer syntax (1908) belonging to the macroblock level syntax (1907) in the syntax structure of Fig. 31, and is described as 32(c). The above is an explanation of the operation of the loop filter in the animation encoding device 3000. Next, an animation decoding device with respect to the animation encoding device 3000 will be described. The configuration of the animation decoding apparatus according to the second embodiment will be described with reference to Fig. 16 . Hereinafter, the components of Fig. 16 will be described separately. The animation decoding apparatus 400 of the present invention has a reference switching prediction unit 401A, a decoded video buffer 502, a restored video buffer 403, an entropy decoding unit 201, an inverse conversion/inverse quantization unit 202, and an addition. The filter 204 and the filter processing unit 208 are controlled by the decoding control unit 207. Here, the entropy decoding unit 201, inverse conversion. The inverse quantization unit 202, the adder 204, the filter processing unit 208, and the decoding control unit 207 perform the same operations as the components of the same number in the animation decoding device 2000 according to the first embodiment. The description is omitted. The reference prediction unit 4 0 1 A has a prediction signal generation unit 203 and a loop filter switching unit 209 as shown in Fig. 17 . Since the prediction signal generation unit 2〇3 and the loop filter switching unit 209 perform the same operations as the components of the same number in the animation decoding device 2000 according to the first embodiment, the description thereof is omitted here. The reference switching prediction unit 40 1 A ' receives the decoded video signal 21, the restored video signal 20, and the switching information 18 that has been decoded in the entropy solution -32-200948092 code portion 20 1 , and is decoded based on the received switching information 18 Any one of the video signal 21 or the restored video signal 20 outputs a predicted video signal 1 1 as a predetermined prediction process for the reference image. • The prediction processing may use, for example, prediction of the temporal direction of motion compensation, or prediction of the spatial direction from the decoded image in the picture, but it is necessary to perform the same prediction processing as the animation coding apparatus 3000, which must be performed. Keep an eye out. The decoded video buffer 402 is obtained by acquiring the decoded video signal 21 generated by the adder 204 and temporarily storing it. The decoded video signal 21 stored in the decoded video buffer 602 is input to the reference switching prediction unit 40 1A. The restored image buffer 403 acquires the restored video signal 20 generated by the filter processing unit 208 and temporarily stores it. The restored video signal 20 stored in the restored video buffer 403 is input to the reference switching prediction unit 40 1A. The above is the configuration of the animation decoding device according to the second embodiment. 〇 Next, the operation of the animation decoding apparatus according to the second embodiment will be described in detail using Figs. 16, 17, and 18'. Further, Fig. 18 is a flowchart showing the operation of the look-ahead filter in the animation decoding device 4000 according to the second embodiment. First, once the encoded material 14 is input to the animation decoding device 4000, the entropy decoding unit 2 0 1, in addition to the conversion coefficient 1 3, the filter coefficient information 17, the switching information 18, and the prediction mode information, the block size. The switching information, the motion vector, the quantization parameter, and the like are decoded in accordance with the syntax structure of FIG. 31 (step S4100). -33- 200948092 The loop filter data syntax (1 906) belonging to the slice level syntax (1903) in the syntax structure of Fig. 31 is shown in Fig. 32 (b), and the filter coefficient information 17 is described. Here, the filter coefficient information 17, that is, filter_coeff[cy][cx] of Fig. 32(b) is a coefficient of the two-dimensional filter, 'filter_size_y and filter_size_x are 値 which determine the size of the subtraction mirror. In this case, the 滤 of the filter size is described in the grammar. However, as another embodiment, the predetermined fixed 値 may be used as the filter size without being described in the grammar. However, if the filter size is set to fixed 値, the same enthalpy must be used in the animation encoding device 3000 and the animation decoding device 4000, which must be noted. Further, in the macroblock block syntax (1 908) to which the macroblock level syntax (1907) in the syntax structure of Fig. 31 belongs, as shown in Fig. 32 (c), l〇〇p_filter_flag is described as Switch information 18. The conversion coefficient 13 decoded by the entropy decoding unit 201 is subjected to entropy decoding, inverse quantization, inverse conversion, and reference, similarly to the previous hybrid coding or the animation decoding apparatus 2000 according to the first embodiment. The predicted video signal 11 outputted by the switching prediction unit 401A is added to be decoded video signal 21 and output. The decoded video signal 21 is output to the filter processing unit 208 and temporarily stored in the decoded video buffer 402 (step S4101). The filter processing unit 208' acquires the filter coefficient information 17 restored by the entropy decoding unit 201 (step S4102). Further, the filter processing unit 2 08' receives the decoded video signal 21, performs filter processing on the decoded video using the filter coefficient information 17, and generates a restored video signal 20 (step -34 - 200948092, step S4103). The filter processing unit 208 outputs the generated restored video signal 20 as the output video signal 22, and temporarily stores it in the restored video buffer 403 (step S41〇4). Next, referring to the switching prediction unit 401A, the decoded video signal 21 or the restored video signal 20 is used as a reference video to generate a predicted video signal 11 (steps 4105 to 4108). Hereinafter, the operation of steps 4105 to 4108 will be described. Q First, the reference switching prediction unit 401A of Fig. 17 acquires the decoded switching information 18, i.e., l〇op_filter_flag (step S4105). Next, referring to the loop filter switching unit 209 inside the switching prediction unit 401A, the switch SW is switched based on the acquired l〇〇p_filter_flag (step S4106). When loop_filter_flag is 0, the loop-filter switching unit 209 connects the switch SW to the terminal A, acquires the decoded video signal 21 as the reference video signal 19, and sends it to the prediction signal generating unit 203. The prediction signal generation unit 203 generates the predicted video signal 1 1 based on the reference video signal 119 corresponding to the decoded video signal 21 (step S4107). On the other hand, when lo〇p_filter_flag is 1, the loop filter switching unit 209 connects the switch SW to the terminal B', and takes the restored video signal 20 as the reference video signal 19, and then sends it to the prediction signal generation. Part 203. The prediction signal generation unit 203 generates the predicted video signal 1 1 based on the reference video signal 1 9 ' corresponding to the restored video signal 20 (step S4108). The above is the operation of steps S4105 to S4 108 in the switching prediction unit 401A. -35- 200948092 Here, referring to the switching prediction unit 401A, the switching information 18 is obtained in units of macro blocks in accordance with the syntax of Fig. 32(c), and the switch SW is switched. In another embodiment, when the animation encoding device 3000 encodes the switching information 18 in a tile unit or a slice unit or a block unit different from the size of the macro block, the animation decoding device 4000 The decoding of the switching information 18 and the switching of the switch SW of the loop filter switching unit 209 are performed in the same unit. The above is an explanation of the operation of the loop filter in the animation decoding device 4000. As described above, according to the animation encoding apparatus according to the second embodiment, the filter processing is performed by setting the filter coefficient information of the loop filter that minimizes the error between the input image and the prediction signal, thereby promoting the reference. The image quality of the image is improved. Further, by the loop filter switching unit 114 included in the switching information generation and prediction unit 301A, it is switched whether or not the local decoded video signal 16 and the restored video signal 20 are to be used as reference images for prediction processing. In the field where the image quality is lowered due to the filter, the restored image signal 20 is not used as the reference image, thereby preventing the deterioration of the image quality, and the image is improved in the field where the image quality is improved. The signal 20 is used as a reference image, thereby promoting prediction accuracy. Further, according to the animation decoding device 40 00 according to the second embodiment, the filter processing and the reference image switching can be performed by using the same filter coefficient information and switching information as the animation encoding device 3000, thereby ensuring the animation. The reference image in the encoding device 3000 is synchronized with the reference image in the animation decoding device 4000. In the animation decoding apparatus 4000 according to the second embodiment, the restored video signal 20 generated by the filter processing unit 208 is output as the output video signal 22, but as another embodiment, The decoded video signal 21 can also be output as the output video signal 22. Further, in the animation encoding apparatus 3 and the animation decoding apparatus 4000 according to the second embodiment, although the local decoded video signal 16 is subjected to filter processing, the local decoded video signal 16 can be previously executed. Image processed by the past block filter. Further, in the animation encoding/decoding apparatus according to the second embodiment, in the above-described embodiment, the restored video signal 20 temporarily stored in the restored video buffer is acquired and used by the reference switching predicting unit 401A. In addition, in the other embodiment, the restored image buffer may be provided, and the filter processing unit may be provided in the reference switching prediction unit, and the restored image may be generated inside the reference switching prediction unit. The animation coding apparatus and the animation decoding apparatus of the embodiment are shown in FIG. 19 to FIG. 23, and the animation coding apparatus 3001 of FIG. 19 includes a switching information generation prediction unit 301B, and the switching information generation prediction unit 301B of FIG. There is a reference switching prediction unit 305B. The reference switching prediction unit 305B of Fig. 21 is provided with a filter processing unit 113 therein, whereby the restored video signal 20 can be generated by acquiring the locally decoded image signal 16 and the filter coefficient information 17. Further, the animation decoding device 4001 of Fig. 22 has a reference switching prediction unit 401B. The reference switching prediction unit 401B of Fig. 23 includes a filter processing unit 208 therein, whereby the restored video signal 20 can be generated by acquiring the decoded video signal 21 and the filter -37-200948092 coefficient information 17. By using the animation encoding device 301 and the animation decoding device 4,000 as described above, the animation encoding device 3000 and the animation decoding device can be realized while reducing the amount of memory required to restore the image buffer. The same action of 4000. Further, the animation encoding devices 3000 and 3001 and the animation decoding devices 4000 and 400 1 ' can be realized by, for example, using a general-purpose computer device as a basic hardware. In other words, the switching information generation prediction unit 301, the loop filter processing unit 312, the local decoded video buffer 303, the restored video buffer 304, the reference switching prediction unit 305, the prediction signal generation unit 101, and the subtraction The unit 102, the conversion/quantization unit 103, the entropy coding unit 104, the inverse conversion/inverse quantization unit 105, the adder 106, the encoding control unit 109, the filter setting unit 110, the switching information generating unit 112, the filter processing unit 113, and the back The circle filter switching unit 114, the reference switching prediction unit 40 1 , the decoded video buffer 40 2 , the restored video buffer 403 , the entropy decoding unit 201 , the inverse conversion • inverse quantization unit 202 , the prediction signal generation unit 203 , and the adder 204. The decoding control unit 207, the filter processing unit 208, and the loop filter switching unit 209 can be realized by executing a program by a processor mounted on the computer device. In this case, the animation encoding devices 3000 and 3001 and the animation decoding devices 4000 and 4001 are implemented by attaching the above-mentioned program to the computer device in advance, or may be stored in a memory medium such as a CD-ROM or distributed through a network. The above program is implemented by appropriately installing the program to a computer device. Further, the local decoded video buffer 303, the restored video buffer 304, the decoded video buffer 402, and the restored video buffer 403' 200948092 can be suitably used for the extended memory built in the upper computer device. , hard disk or CD-R, CD-RW, DVD-RAM, DVD-R, etc. to achieve the media. (Third Embodiment) A configuration of an animation encoding apparatus according to a third embodiment will be described with reference to Fig. 24'. Hereinafter, the components of Fig. 24 will be described separately. φ The animation coding apparatus 5000 shown in Fig. 24 includes a loop filter processing unit 501, a prediction signal generation unit 1〇1, a subtractor i〇2, and a conversion. The quantization unit 103, the entropy coding unit 104, the inverse conversion/inverse quantization unit 105, the adder 106, and the reference video buffer 108 are controlled by the coding control unit 109, and the prediction signal generation unit 101 and the subtractor 102 are controlled. Conversion. The quantization unit 103, the entropy coding unit 104, and the inverse conversion. The inverse quantization unit 105, the adder 106, the reference video buffer 108, and the encoding control unit 109 are configured by the same numbered components in the animation encoding device 1000 of Fig. 1 described in the first embodiment. The same operation, so the description is omitted here. The loop filter processing unit 501 receives the local decoded video signal 16 and the input video signal 1 〇, and outputs the reference video signal 19 and the filter coefficient information 17. The detailed description of the loop filter processing unit 501 will be described later. Next, the loop filter processing unit 501 in the animation encoding apparatus according to the third embodiment will be described in detail with reference to Figs. 25 and 26 . Hereinafter, the components of FIGS. 25 and 26 will be described separately. -39- 200948092 The loop filter processing unit 501 shown in Fig. 25 includes a filter setting unit 110 and a switching information generation filter processing unit 502. Further, the switching information generation filter processing unit 502 is as shown in Fig. 26. The switching filter processing unit 111 and the switching information generating unit 503 are provided. Here, the filter setting unit 110 and the switching filter processing unit 111 perform the same operations as the components of the same number in the loop filter processing unit 107 of FIG. 2 described in the first embodiment. Description is omitted here. The switching information generating unit 503 is controlled by the encoding control unit 109, receives the locally decoded video signal 16, and generates the switching information 18 in accordance with the predetermined switching determination method. The generated switching information 18 is input to the switching filter processing unit 111. Details of the handover determination method will be described later. The above is the configuration of the animation encoding device according to the third embodiment. Next, the operation of the loop filter in the animation encoding device according to the third embodiment will be described in detail with reference to Figs. 25, 26 and 27. Fig. 27 is a flowchart showing the operation of the loop filter in the animation encoding device 5000 according to the third embodiment. First, the filter setting unit 110 shown in Fig. 25 acquires the local decoded video signal 16 and the input video signal 10, and sets the filter coefficient information 17 (step S5100). Here, the filter setting unit U0 uses a two-dimensional Wiener filter generally used in image restoration, and the filter coefficient is designed such that the image after the filter processing is applied to the locally decoded image signal 16 and the input image signal is 1〇. The average squared error is minimized, and the designed filter coefficient and the 表示 indicating the filter size are set as the filter coefficient information 17. The set filter coefficient information 17 is output to the switching information generation filter processing unit 502 shown in Figs. 25-40-200948092, and is also output to the entropy encoding unit 104. * The switching information generation unit 503 provided in the switching information generation filter processing unit 052 shown in Fig. 26 acquires the local decoded video signal 16 (step S5101). The switching information generating unit 503 calculates the difference between the target pixel in the partial decoded video signal 16 and the peripheral pixel and the SAD in the pixel unit (step S5102). 〇 SAD, if the coordinates of the pixels in the locally decoded image are x, y and the locally decoded image is F(x, y), it can be expressed by the following equation. [Equation 4] SAD = Σ Σ \F{x, y) - F(x + i, y + ;)| - Next, the following switching determination processing is performed using the SAD and the threshold 値T set in advance to the encoding control unit 109. (Step S5 103). If SAD is T or less, 0 Q is set to the switching information 18, i.e., l00p_filter_flag (step S5104). On the other hand, if SAD is 値 greater than T, 1 is set to l〇op_filter_flag (step S5105). Here, although the switching information 18 is obtained in units of pixels, as another embodiment, the switching determination processing may be performed in accordance with a frame unit or a slice unit or a macro block unit or a size different from a macro block. The block unit is used for the determination. In this case, the switching information 18 is also output in units corresponding thereto. In addition, although the absolute difference between the target pixel and the peripheral pixel is used here, as another embodiment, the difference square sum may be used as long as it is an index that can be calculated from the locally decoded image, and activity or space may be used. Frequency, edge strength, edge direction and other indicators. Again, here, although -41-.  In 200948092, the above-mentioned index is calculated from the locally decoded image. However, the restored image obtained by filtering the locally decoded image may be acquired. The above-mentioned index is calculated from the restored image. Further, as another embodiment, the switching determination process may be performed based on a quantization parameter, a block size, a prediction mode, a motion vector, a conversion coefficient, and the like belonging to a part of the encoded information. Next, the switching filter processing unit 111 shown in FIG. 26 acquires the locally decoded video signal 16 and the filter coefficient information 17 in addition to the generated switching information 18, and performs the processing according to the first embodiment. The same operation of steps S1107 to S1109 of FIG. 4 in the animation encoding apparatus 1000 outputs the reference video signal 19. In other words, the loop filter switching unit 1 of FIG. 3 acquires the generated switching information 18, that is, l〇op_filter_flag', based on l〇〇P_filter_fUg, switches the switch SW provided therein (step S5106). . When l〇〇p_filter_flag is 0, the loop filter switching unit 114 connects the switch SW to the terminal A, and temporarily stores the local decoded video signal 16 as the reference video signal 19 in the reference image buffer 108 ( Step S5107). On the other hand, when 1〇(^_^^1"_0&§ is 1, the loop filter switching unit 114 connects the switch SW to the terminal B, and the restored video signal 20 is regarded as the reference image signal 19. Temporarily stored in the reference image buffer 108 (step S5108). Finally, the filter coefficient information 17 generated by the filter setting unit 1 1 is encoded by the entropy encoding unit 104, together with the quantized conversion coefficient 13 The prediction mode information, the block size switching information, the motion vector, the quantization parameter, and the like - 42 - 200948092 are multiplexed into a bit stream, and then transmitted to the animation decoding device 6000 described later (step S5109). The mirror coefficient information 17, which is included in the slice filter syntax (1906) in the slice level syntax (1903) in the syntax structure of Fig. 31, is described as shown in Fig. 32(b). The coefficient information 17 is also the filter_coeff [cy][cx] of Fig. 32 (b) is the coefficient of the 2D filter, and the filter_size_y and filter_siZe_x are the parameters determining the size of the filter. Here, the size of the filter φ mirror will be indicated. 'Described in grammar, but as other implementations The state may be used in the grammar instead of the predetermined fixed 値 as the filter size. However, if the filter size is set to fixed 値, the animation encoding device 5000 and the animation decoding device described later are used. The same enthalpy must be used in 6000. This must be noted. The above is a description of the actions involved in the loop mirror in the animation encoding device 5000. Next, the animation decoding device Q with respect to the animation encoding device 6000 will be described. The configuration of the animation decoding device according to the third embodiment will be described with reference to Fig. 28'. Hereinafter, the components of Fig. 28 will be described. The animation decoding device 6000 shown in Fig. 28 has switching information generation. The filter processing unit 601, the entropy decoding unit 201, the inverse conversion/inverse quantization unit 202, the prediction signal generation unit 203, the adder 204, and the reference video buffer 206 are controlled by the decoding control unit 207. The 'entropy decoding unit 201, the inverse conversion/inverse quantization unit 202, the prediction signal generation unit 203, the adder 204, the reference video buffer 206, and the decoding control unit 2 0 7 ' are In the same manner as the components of the same number in the animation decoding apparatus 2000 of the embodiment of the present invention, the same operation is omitted here. The switching information generation filter processing unit 601 obtains the decoded video signal 21 and The filter coefficient information 17 outputs the reference video signal 19 and the output video signal 22. The detailed description of the switching information generation filter processing unit 601 will be described later. Next, the switching information generation filter processing unit 601 in the animation decoding device according to the third embodiment will be described in detail with reference to Fig. 29 . Hereinafter, the components of Fig. 29 will be described separately. The switching information generation filter processing unit 601 shown in Fig. 29 includes a switching information generation unit 602 and a switching filter processing unit 205. Here, the switching filter processing unit 205 performs the same operation as the components of the same number in the animation decoding device 2000 of Fig. 5 described in the first embodiment, and thus the description thereof will be omitted. The switching information generating unit 602 is controlled by the decoding control unit 207 to obtain the decoded video signal 21, and generates the switching information 18 in accordance with the predetermined switching determination method. The generated switching information 18 is input to the switching filter processing unit 205. Details of the handover determination method will be described later. The above is the configuration of the animation decoding device according to the third embodiment. Next, the operation of the animation decoding apparatus according to the third embodiment will be described in detail with reference to Figs. 28, 29 and 30. Fig. 30 is a flowchart showing the operation of the loop filter in the animation decoding device 6000 according to the third embodiment. First, the entropy decoding unit 201 decodes the filter coefficient information 17 in accordance with the syntax structure of Fig. 31 (step S6100). The loop filter data syntax '(1 906) in the slice level syntax (1 903) in the syntax structure of Fig. 31 is as shown in Fig. 32 (b), and the filter coefficient information 17 is described. Here, the filter coefficient information 17 is that the filter_coeff[cy][cx] of Fig. 32 (b) is the coefficient of the 2-dimensional filter, and the filter_size_y& filter_size_x is the 决定 which determines the size of the filter. Here, the 表示 indicating the size of the filter is described in the grammar. However, as another embodiment, the predetermined 値 may be used instead of the grammar. However, if the filter size is set to fixed 値, the same 値 must be used in the animation encoding device 5000 and the animation decoding device 6000, which must be noted. The switching information generation filter processing unit 601 obtains the decoded filter 'coefficient information 17 (step S6101). Further, the switching information generation filter processing unit 601' obtains the decoded video signal 21 obtained from the adder 024 (step S6102). The switching information generating unit 602 included in the switching information generation filter processing unit 601 shown in FIG. 29 sets the absolute difference between the pixel of interest in the decoded video signal 21 and its peripheral pixels, and SAD, in units of pixels. And it is calculated (step S6103). SAD, if the coordinates of the pixels in the decoded image are x, y, and the decoded image is ', can be expressed as [4]. The following switching determination processing is performed using the SAD and the threshold 値T set in advance to the decoding control unit 207 (step S 6 1 〇 4 ). Here, the threshold 値T is the same as the threshold 値τ set in the animation coding apparatus 5〇〇〇, and the forced point must be noticed. If s AD is Τ or less, the switching information 18 is also -45- 200948092, that is, loop_filter_flag is set to 0 (step S6105). On the other hand, if SAD is 値 greater than T, 1 is set for loop_filter_flag (step S6106). Here, although the switching information 18' is obtained in units of pixels, as another embodiment, the switching determination process may be in accordance with the frame unit or the slice unit or the macro block unit or the size of the macro block. The block unit is used for the determination. In this case, the switching information 18 is also decoded in units corresponding thereto. Further, although the absolute difference between the target pixel and the peripheral pixel is used here, as another embodiment, a difference square sum may be used, and any index that can be calculated from the decoded image may be used, and an active or spatial frequency may be used. , edge strength, edge direction and other indicators. Here, although the above-mentioned index is calculated from the decoded image, the restored image obtained by performing the filter processing on the decoded image may be acquired, and the above-mentioned index may be calculated from the restored image. Further, as another embodiment, the switching determination process may be performed based on a quantization parameter, a block size, a prediction mode, a motion vector, a conversion coefficient, and the like belonging to a part of the encoded information. In any case, the switching information generating unit 602 in the animation decoding device 6 000 must perform the same switching determination processing as the switching information generating unit 503 in the animation encoding processing 5,000, which must be noted. The switching filter processing unit 205 shown in FIG. 29 acquires the decoded switching information 18 and acquires the decoded video signal 21 and the filter coefficient information 17' by performing the animation decoding device 2 according to the first embodiment. In the same operation as step S2103 to step S2105 of FIG. 7, the reference video signal 19 is output. That is, the loop filter switching unit 205A of Fig. 6 obtains the generated switching information 18, i.e., l〇〇P_filter_flag, and switches the switch SW provided therein based on the l〇〇p_filter_flag.  (Step S6107). When l〇〇p_filter_flag is 0, the loop filter switching unit 205A connects the switch SW to the terminal A, and temporarily stores the decoded video signal 21 as the reference video signal 19 in the reference image buffer 206 ( Step S6108). On the other hand, when loop_filter_flag is 1, the loop filter switching unit 205A connects the switch SW to the terminal φ sub B, and temporarily restores the restored video signal 20 as the reference video signal 19 to the reference image buffer. 206 (step S6109). As described above, the switching information generation filter processing unit 601 uses the acquired decoded video signal 21, and is similar to the switching information generation filter processing unit 502 of FIG. 26 in the animation encoding apparatus according to the third embodiment. 'Create, to generate the switching information 18, and output the reference image signal 19. The above is an explanation of the operation of the loop mirror in the animation decoding device 6000. As described above, according to the animation encoding apparatus according to the third embodiment, by performing filter processing by setting filter coefficient information of the loop filter that minimizes the error between the input image and the predicted image, the filter processing can be promoted. The image quality of the reference image is improved. Further, the loop filter switching unit 114 switches between the local decoded video signal 16 and the restored video signal 18 as a reference video for each local area switching, and is calculated from the local decoded video 16 The indicator is used to switch between each partial field, thereby preventing the spread of the image quality caused by the filter, which can promote the coding efficiency. Further, according to the animation decoding apparatus according to the third embodiment, the filter processing is performed by using the same filter coefficient information as that of the animation encoding apparatus, and the same switching determination processing is performed by the -47-200948092. The reference image in the animation encoding device is synchronized with the reference image in the animation decoding device. Further, according to the animation encoding apparatus and the animation/decoding apparatus according to the third embodiment, the encoding side generates switching information based on an index that can be calculated from the locally decoded video image, thereby enabling decoding from the decoding side. The image calculates the same switching information. Therefore, the amount of coding when the switching information is encoded can be reduced. In addition, in the animation decoding device 600 0 according to the third embodiment, the switching filter processing unit 205 may be designed as shown in FIG. 8 or in the same manner as the animation decoding device 2000 according to the first embodiment. The configuration shown in FIG. 9 is to output the decoded video 21 or the reference video 19 as the output video signal 2 2 . Further, the switching filter processing unit 205 may be configured as shown in FIG. 10, and the switching information generating unit newly generates the switching information for the rear-end filter, thereby providing the rear-end filter switching unit 210 of FIG. The switch SW2 Q is switched to switch the output image signal 22. Further, in the animation encoding apparatus 5000 and the animation decoding apparatus 6000 according to the third embodiment, although the local decoded video signal 16 is subjected to filter processing, the local decoded video signal 16 may be subjected to the previous execution of the past block filtering. Mirrored image. Further, the animation encoding device 5000 and the animation decoding device 6000 can be realized by, for example, using a general-purpose computer device as a basic hardware. In other words, the loop filter processing unit 501, the switching information generation filter processing unit -48-200948092 5 02, the switching information generation unit 503, the prediction signal generation unit 1〇1, 102, the conversion/quantization unit 103, and the entropy coding unit 104, inverse conversion.  The section 105, the adder 106, the reference video buffer 1〇8, the code•1〇9, the filter setting unit 110, the switching filter processing unit 111, the switching to the filter processing unit 601, the switching information generating unit 602, The entropy decoding inverse transforming unit 202, the prediction signal generating unit 203, the dancing, switching filter processing unit 205, the reference video buffer 206, and the buffer unit 207 can be mounted on the computer device. It is implemented by executing the program. In this case, the animation encoding device 5 000 decoding device 6000 is implemented by pre-installing the above-mentioned program to a computer, or may be stored in a memory medium such as a CD-ROM, and the network distributes the writing program to appropriately program the program. It is implemented by installing it to electricity. Further, the reference video buffer 108 and the reference buffer 206 can be suitably used for expansion, hard disk or CD-R, CD-RW, DVD-RAM, DVD-R, etc. built in a computer device, etc. To achieve it. (Fourth Embodiment) In the first embodiment, the example in which the macro block unit of the information 16 pixels is switched is set. On the other hand, the setting unit of the switching information in the form is not limited to the upper block, but may be a sequence, a frame, or a division into a pixel block or the like. In the present embodiment, the first embodiment of the inverse quantizer control unit information generation 201, the 204 decoding controller and the animation device, or the brain device image storage medium is used. 6 X The slice mode of the macro area is described in the encoding and decoding method of -49-200948092, which is a method for setting the switching information to each pixel block and adaptively switching the size of the pixel block. . The switching filter processing unit 111 in the encoding processing and decoding apparatus according to the fourth embodiment will be described with reference to Figs. 37 and 38. The switching filter processing unit 111 of FIGS. 37 and 38 is a modification of the switching filter processing unit 111 of FIGS. 3 and 6, and is a case where only the restored video signal 20 is regarded as a reference image by the switch SW. The composition of the filter processing is performed. That is, in FIG. 37, the switch SW is provided in the front stage of the filter processing unit 113, and the terminal A of the switch SW is directly derived to the reference image buffer 108, and the local decoded video signal 16 is directly used as the reference video signal. Stored in buffer 108. The terminal B of the switch SW is transmitted through the filter processing unit 113 to the reference video buffer 1〇8, and the local decoded video signal 16 is filtered in the filter processing unit 113 in accordance with the filter coefficient information 17. The mirror processing is then stored in the buffer 1〇8 as a reference image signal. Further, in Fig. 38, a switch SW is provided in the front stage of the filter processing unit 208, and the terminal A of the switch SW is directly led to the output line, and the decoded video signal 21 is directly output to the output line. The terminal B of the switch SW is passed through the filter processing unit 208 and is derived to the output line. The decoded video signal is filtered by the filter processing unit 208 in response to the filter coefficient information 17, and then treated as a filter. The reference image signal 19 and the output image signal 22 are output to the output line. According to this configuration, only when the loop information_flag_flag is 1 for the switching information 18, the filter processing unit 113 or 208 can perform the filter processing. Therefore, the processing cost can be further reduced as compared with the configuration of FIGS. 3 and 6. . The modifications of Figs. 37 and 38 can of course also be applied to the first

-50- 200948092 1 to the third embodiment. The switching filter processing unit & HI of Figs. 37 and 38 of the present embodiment has the filter processing unit 113 and the switching filter processing unit 111 in the first embodiment. The loop filter switching unit 114 obtains the local decoded video signal 16 or the decoded video signal 21, the filter coefficient information 17, and the switching information 18, and outputs the reference video signal 19. Further, in the present embodiment, the field setting information 23 is acquired at the same time, and is input to the loop filter switching unit 114. Here, the field setting information 23 is information for controlling the timing of the switch SW in accordance with the block size, and indicates the size of one block when the screen is divided into rectangular blocks. The field setting information 23 is a 可 which can directly use the block size, and can also be an index in the 'domain size specification table for the predetermined block size. An example of the above-mentioned field size specification table is shown in Fig. 39 (a). In the field size specification table of Fig. 39 (a), the smallest block size, that is, the index of the square block of the vertical and horizontal 4 © pixels is set to 〇, and the number of pixels of one side is doubled each time, and the number of pixels is 512x5. 8 block sizes up to 12 pixels. By maintaining the same field size specification table on the encoding and decoding side, the code determined by the field size specification table of FIG. 39(a) is encoded as block information on the encoding side, and is used on the decoding side. The same field size specification table determines the block size based on the decoded block information. Further, the field size specification table (a) is provided in the encoding control unit and the decoding control unit. Next, the syntax of the field-containing setting information 23 - 51 - 200948092 described in the present embodiment will be described using FIG. 42 . The loop filter data syntax (1 9 03 )' in the syntax structure of Fig. 31 described in the present embodiment is described as shown in Fig. 42. The filter_block_size of Fig. 42(a) indicates the area setting information 23, and the NumOfBlock is the total number of blocks in one slice determined by the block size as indicated by filter_block_size. For example, when the field size specification table of Fig. 39 (a) is used, once the field setting information, that is, the index is set to 0, the slice is divided into 4x4 pixel blocks, and for every 4x4 pixel block, l〇 〇p_filter_flag is encoded. Thereby, in the slicing, both the encoding side and the decoding side can switch the switch SW in the switching filter processing unit 111 every 4×4 pixel block, and as another embodiment, the pair may be used. The basic method of dividing the block size is used as the field setting information. For example, as shown in the field size specification table of Fig. 39 (b), four types of block division methods of "no division", "horizontal division", "longitudinal division", and "vertical division 2 division" are prepared, and Give the index number separately. Thereby, when the basic block size is set to a 16x16 pixel block and an 8x8 pixel block, the block shape as shown in Fig. 40 can be taken separately. As for the grammar of these, as shown in FIG. 42(b), the field setting information, that is, the filter block size, is described in the loop of the basic block size. The number of sub-blocks determined by filter_block_size, that is, the number of NumOfSubblocks. The loop_filter_flag is encoded. As an example, an example of a reference image when the basic block size is set to 16x16 and 8x8 is shown in Fig. 41. Further, the field size specification table (b) is provided in the encoding control unit and the decoding control unit. 200948092 In this way, according to the fourth embodiment, by using the image to be set and encoded, the diversity can be achieved by drawing a picture or drawing on the divided blocks. The switching timing of each local area within the grid. Here, when the block size is adapted to φ to encode the amount of coding necessary, the finer the block is, the more it will be increased, and thus, a plurality of the above-mentioned loop filters can be prepared. Switch the table by the amount of image coding and the information obtained by the party. For example, as shown in Fig. 39 (c), for example, the encoding control unit and the resolution size and image type can be determined, and the quantization size field size specification table can be determined. The larger the image size, the larger the size is prepared, and the image type is a B image that is encoded by the I image >P image > The field size specification table for the time block. Further, regarding the quantization parameter, the coding amount of the conversion coefficient is smaller, so that the domain size specification table of the block is made. The field setting required for the animation coding and decoding domain segmentation described above has the switching information set on the block division of the screen, and the information can be switched to the size when the filter processing switching is adaptively controlled. The smaller, that is, the picture segmentation can lead to a reduction in coding efficiency. The domain size specification table is improved in a plurality of domain size specification code control sections based on different domain size specifications on the encoding side and the decoding side, and can be used in accordance with the quantization parameter for the shadow size. For example, if the image size is relative to the field size of the block, the table is specified. When it is set as a general user, due to the tendency of the image, the larger the size of the quantization parameter is prepared, the larger the size is used. -53-200948092 By switching a plurality of field size specification tables having different block sizes as above, a limited index number can be used to more adaptively select the block size. Moreover, the balance between the coding amount of the conversion coefficient or the coding parameter and the amount of coding necessary for switching the information can be efficiently controlled. In still another embodiment, the block information system can also be used, and the area of the motion compensation block size and the conversion block size that have been used in encoding and decoding at the time of generating the locally decoded image or the decoded image can be used. The block size. In this case, in the loop filter data syntax (1906), it is not necessary to describe the block information, and the switching timing of the filter processing can be changed, and the amount of coding involved in the block information can be reduced. Further, although the case where the intra-screen is divided into rectangular blocks has been described here, the segmentation method in which the same field division can be realized in both the encoding device and the decoding device is not limited to the rectangular block. (Fifth Embodiment) Next, an animation coding method according to a fifth embodiment will be described with reference to Figs. 43 and 44. The loop filter processing unit 1A of FIG. 43 includes the same components as the loop filter processing unit 1〇7 of FIG. 2 described in the first embodiment, and has a filter setting unit 110 therein. The switching filter processing unit 111 and the switching information generating unit 112 obtain the local decoded video signal 16 and the input video signal 10, and output the filter coefficient information 17, the switching information 18, and the reference video signal 19. On the other hand, in the present embodiment, it is designed to set the switching information 18 for each partial field of the intra-screen division -54-200948092, and input the switching information 18 to the filter setting unit 110, thereby switching based on Information 18, Selectively 'Get the field used for filter settings. * The operation related to the reticle mirror in the animation encoding device according to the fifth embodiment will be described in detail using the flowchart of Fig. 44. In the flowchart of Fig. 44, R is the maximum number of times the filter coefficient information is set, and N is the total number of partial fields into which the screen is divided. First, Q sets the filter coefficient information setting frequency r to 0 (step S7100), and sets the switching information, i.e., l〇〇p_filter_flag, to 1 in all fields (step S7101). Next, r is incremented by 1 (step S7102). Thereafter, the filter setting unit 110 of Fig. 43 sets the filter coefficient information (step S1100). Here, similarly to the first embodiment, a two-dimensional Wiener filter generally used in image restoration is used, and the filter coefficient is designed to be an image obtained by applying a filter to the locally decoded video signal 16 (refer to the image). The average squared error of the signal 19) and the input image signal 〇ίο is minimized, and the designed filter coefficient and the 表示 indicating the filter size are set as the filter coefficient information 17. Here, in the filter setting unit 11 of FIG. 43 described in the fifth embodiment, it is designed to use only the switching information that has been input, that is, the field in which the lo〇P-filter-flag is set to 1. To calculate the average squared error. Here, when the initial filter coefficient information of r = 1 is set, the l〇〇P_filter^_flag of all the fields is set to 1 in step S7101. Therefore, similarly to the first embodiment, it has been designed. A filter coefficient information that minimizes the average squared error of the entire picture will be generated. -55- 200948092 Next, the switching filter processing unit 111 of Fig. 43 sets the switching information 18 for each partial field in which the screen is divided. That is, if the field number is assumed to be η, η is first set to 0 (step S7103), and η is incremented by 1 (step S7104). Next, the processing of steps S1 101 to S1 109 which are the same as the first embodiment is performed on the nth field. The above process is repeated until η reaches the total number Ν (step S 7 1 0 5 ). After setting l〇〇P_filter_flag for all fields in the screen, the series of processing is repeated until the filter coefficient information setting number r reaches the maximum number R of predetermined filter coefficient information (step S71 06). As described above, according to the animation coding method according to the fifth embodiment, the setting of the filter coefficient information after the second time or later can be limited to the field in which l〇〇p_filter_flag has been set to 1 and the average square error is minimized. Filter. For example, when the reference image obtained by the switching information set for the first time is as shown in FIG. 3, the setting of the second filter coefficient information may be limited to "applicable back". A macroblock of the circle filter is set to set a filter that minimizes the average squared error. Therefore, for the "Macro block to which the loop filter is applied", it is possible to set a filter with a larger image quality improvement effect. In addition, the present invention is not limited to the above-described first embodiment to the third embodiment, and the constituent elements may be modified and embodied in the implementation stage without departing from the spirit and scope of the invention. As it is, in the implementation stage, the constituent elements can be deformed and embodied without departing from the scope of the object. Further, the plural constituent elements disclosed in the above embodiments can be combined as appropriate to form various inventions. For example, it is also possible to delete a plurality of constituent elements from all the constituents shown in the embodiment. It is even possible to combine constituent elements across different embodiments. (Sixth embodiment) Next, an animation coding apparatus and an animation decoding apparatus according to a sixth embodiment will be described. 45 is a view showing a filter processing unit existing in both the animation encoding processing and the animation decoding apparatus, and the filter processing unit 1 1 3 and the third implementation corresponding to the first and second embodiments are illustrated. The filter processing unit of the filter processing unit 208 of the fourth embodiment. The local decoded video filter processing unit 701 of FIG. 45 inputs a locally decoded video signal, switching information, filter information, and encoding information if it is an animation encoding device, and inputs decoding if it is an animation decoding device. -_ Image signal, switching information, filter information, encoding information, and generating reference image signals. The filter processing unit 701 is composed of a filter boundary determination processing unit 702, a switch 703, a deblocking filter processing unit 704, a switch 705, and a video restoration Q filter processing unit 706. The filter boundary determination processing unit 7〇2′ determines whether the pixel to be converted or the motion compensation unit, that is, the pixel boundary portion, should be deblocked using the coded information from the locally decoded video signal or the decoded video signal. The pixels are mirrored and the switch 703 and switch 705 are controlled. At this time, the 'filter boundary determination processing unit 702' connects the switch 703 and the switch 705 to the terminal A' when the input image belongs to the block boundary pixel, and the deblocking filter processing unit 704 goes to the pixel signal. Block filter processing; when it is determined that the input image is not a block boundary pixel, the switch -57-200948092 07 and the switch 70 5 are connected to the terminal B', and the image restoration filter processing unit 706 performs image restoration on the input image. Mirror processing. The deblocking filter processing unit 704 uses the filter coefficient prepared in advance or the filter coefficient given from the outside of the local decoded image filter processing unit for the pixel determined by the block boundary determination processing unit 702 as the block boundary. Information to perform conversion processing or filter processing to offset block distortion caused by motion compensated prediction (eg, averaging of pixel signals). The image restoration filter processing unit 706' determines that the pixel boundary determination processing unit 702 determines that the pixel is not the block boundary, and performs local decoding by the filter coefficient given from the outside of the local decoded image filter processing unit 701. Image restoration processing. The video restoration filter processing unit 706 assumes that the switching filter processing units 111, 205A, 205B, 205C, and 205D and the second embodiment of FIGS. 3 and 6 and FIGS. 8 and 9 and FIG. 10 of the first embodiment are assumed. The filter processing units 1 13 and 208 of FIGS. 14 and 16 of the embodiment, the reference switching prediction units 305B and 401B of FIGS. 21 and 23 of the second embodiment, and the switching information of FIGS. 26 and 29 of the third embodiment. The filter processing units 502 and 601 and the switching filter processing units 1 1 and 205A of Figs. 37 and 38 of the fourth embodiment are directly replaced. According to the sixth embodiment, the block boundary portion can be subjected to filter processing in consideration of the difference in the nature of the decoded video signal caused by the block distortion caused by the conversion processing or the motion compensation processing, thereby improving the overall image. Restore performance. (Seventh embodiment)

-58- 200948092 Next, an animation coding apparatus and a decoding apparatus according to a seventh embodiment will be described with reference to Figs. 46 to 49. First, the configuration of the animation encoding apparatus according to the seventh embodiment will be described with reference to Fig. 46'. The following description of the constituent elements of Fig. 46 will be respectively described. The animation encoding device 70 00 shown in Fig. 46 includes a subtractor 102, a conversion and quantization unit 103, an entropy encoding unit 104, an inverse conversion, an inverse quantization unit 105, an adder 106, a Deblocking filter unit 801, and a filter setting. The switching information generating unit 802, the decoded video buffer 803, the predicted video creating unit 804, and the motion vector generating unit 805 are controlled by the encoding control unit 1〇9. Here, the subtractor 102, the conversion/quantization unit 103, the entropy coding unit 104, the inverse conversion/inverse quantization unit 105, the adder 106, and the coding control unit 1〇9 are shown in Fig. 11 according to the second embodiment. The components of the same number in the animation coding device 3,000 perform the same operations, and thus the description thereof is omitted here. In addition, the filter setting unit 802 of the filter setting unit 802 performs the same operation as the switching information generating unit 112 in the loop filter processing unit 107 of Fig. 2 described in the first embodiment. The specific operation is that, similar to the conventional hybrid coding or the animation coding apparatus 1000 according to the first embodiment or the animation coding apparatus 3000 according to the second embodiment, the animation coding apparatus 700 is corrective for prediction errors. Perform conversion, quantization, entropy coding, and perform local decoding by inverse quantization and inverse conversion. For the local decoded signal, after the filter processing is performed by the Deblocking filter section to remove the distortion of the block boundary, the local decoded signal after the filter processing -59-200948092 is temporarily stored in the decoded image buffer 8 03. The dropper setting/switching information generating unit 8 02 receives the decoded image signal and the input image signal of the localized portion processed by the filter, generates filter coefficients, and switches the information. The filter setting/switching information generating unit 902 outputs the generated filter coefficient information to the decoded image buffer 803 and also outputs it to the entropy encoding unit 104. The entropy coding unit 1〇4 encodes the filter coefficient and the switching information generated by the filter setting/switching information generating unit 802, and further converts the quantized conversion coefficient with the prediction mode information and the block size switching information. The motion vector, the quantization parameter, and the like are multiplexed into a bit stream and transmitted to the animation decoding device 8000, which will be described later. At this time, the filter coefficient information and the switching information are transmitted as the encoded information of the input image, and are transmitted in accordance with the syntax of Fig. 32 and the like. In the decoded video buffer 803, the local decoded video to be referred to by the predicted video generating unit 804 and the filter coefficients and switching information corresponding to the locally decoded video are stored. The predicted image creating unit 804 uses the locally decoded video, the filter coefficient, the switching information, and the motion vector information generated by the motion vector generating unit 805 managed by the decoded video buffer 803 to create a motion compensated prediction. image. The subtractor 1 0 2 generates a prediction error signal between the predicted image and the input image that has been created. On the other hand, the motion vector information is encoded by the entropy coding unit 104 and multiplexed with other information. The configuration of the animation decoding apparatus according to the seventh embodiment will be described with reference to Fig. 47'. Hereinafter, the components of Fig. 47 will be described separately. The animation decoding apparatus 8000 shown in Fig.-60-200948092 47 includes an entropy decoding unit 20.1, an inverse conversion/inverse quantization unit 202, an adder 204, a Deblocking filter unit 81 1 , a decoded image buffer 813, and a predicted image. The preparation unit 814 is controlled by the solution control unit 207. The entropy decoding unit 201, the inverse transform, the inverse quantization unit 202, the adder 204, and the decoding control unit 207 perform the same operations as the components of the same number in the animation decoding device 2000 according to the first embodiment. Therefore, the description is omitted here. The specific operation is performed by the entropy decoding unit 201 in the same manner as the decoding of the previous hybrid code or the animation decoding device 2000 according to the first embodiment or the animation encoding device 4000 according to the second embodiment. The decoded signal is subjected to inverse quantization and inverse conversion, and a prediction error signal is generated, and is added to the predicted image in the adder 204 to generate a decoded image. The deblocking filter unit 811 performs filter processing for removing the block boundary distortion, and outputs the decoded image and stores it in the decoded image buffer 813. In the decoded video buffer 813, the filter coefficients and the switching information corresponding to the decoded video referred to by the predicted video generating unit 814 and the decoded video decoded by the entropy decoding unit 201 are stored. The predicted image creating unit 813 creates the restored filter processing and motion based on the motion vector information decoded by the entropy decoding unit 201 and the decoded video, filter coefficient, and switching information from the decoded video buffer 402. The predicted image of the compensation. The adaptive image restoration processing is applied to the motion compensation prediction by the composition of the animation encoding processing and the animation decoding -61 - 200948092 as shown in Figs. 46 and 47 to achieve an improvement in encoding efficiency. Fig. 48 is a view showing an embodiment of the specific embodiment of the predicted image creating unit 814 of Figs. 46 and 47. Hereinafter, the components of Fig. 48 will be described separately. The predicted image creating unit 804 shown in Fig. 4 is composed of a switch 8 2 1 , a restoration image creating unit 822, and an interpolation image creating unit 823. The switch 821' is a switch for switching whether or not to perform the restoration filter processing on the decoded video referred to based on the motion vector information, and is switched based on the switching information generated by the filter setting information switching unit 802. When the switch 821 is switched to the terminal A, the restored image creating unit 822 performs the decoded image referred to by the motion vector information using the filter coefficient set by the filter setting/switching information generating unit 802. Restore filter processing. When the switch 821 is switched to the terminal B, the decoded video is directly input to the interpolated video creating unit 823. The interpolated image creating unit 8 23 generates an interpolated image 'of a decimal point pixel position based on the motion vector information and uses it as a predicted image. With this configuration, a combination of adaptive image restoration processing and motion compensation prediction can be realized. Fig. 49 is a view showing another embodiment of the specific embodiment of the predicted image forming unit of Figs. 46 and 47. The following description of the components of Fig. 49 will be given. The predicted image creating unit 804 shown in FIG. 49 is composed of a switch 831, an integer pixel filter unit 832, a bit expansion unit 833, an interpolation image creating unit 843, a switch 835, and a weighted prediction image creating unit 836. The bit reduction unit 837 is composed of. The switch 831 is based on the decoded video of the 200948092 based on the motion vector information, and the switching switch that performs the restoration filter processing in integer pixel units is generated based on the filter setting/switching information generating unit 802 of FIG. Switching information while being switched. When the switch 831 is 'switched to the terminal A, the integer pixel filter unit 832 uses the filter coefficient set by the filter setting/switching information generating unit 802 to decode the decoded image based on the motion vector information. , perform restoration filter processing. In this case, when the pixel bit length of the decoded video image is N-bit, the pixel bit length of the output of the integer pixel filter unit 832 becomes the M-bit of M2 N. When the switch 83 1 is switched to the terminal B, the decoded image is input to the bit expansion unit 83 3 , and the N-bit decoded image is expanded to become the M-bit of the M2N. Specifically, the bit expansion unit 83 3 performs arithmetic shift processing of (M-N) bits to the left of the pixel 値V. The interpolation image creating unit 834 generates an interpolated image of the decimal point pixel position based on the motion vector information. In this case, the length of the pixel bit of the input M-bit is such that the output becomes the L-bit pixel bit length of L2N. The switch 83 5 is switched by the encoding control unit 109 or the switch controlled by the decoding control unit 207 based on the encoding information, and switches whether or not to perform weighted prediction. When the switch 835 is switched to the terminal A, the weighted prediction image creating unit 836 creates a predicted image based on the weighted prediction equation performed in H.264/AVC or the like. At this time, the input of the L-bit pixel bit length is processed to be an output of the N-bit pixel bit length. When the switch 835 is switched to the terminal B, the bit reduction unit 847 performs rounding processing so that the input of the L-bit becomes the N-bit of L2N. -63- 200948092 By constructing the inside of the predicted image creating unit 804 by making the pixel bit length larger than the pixel bit length of the decoded image, the calculation of the restoration processing, the interpolation processing, and the weighting prediction processing can be reduced. Into the error, it is possible to achieve predictive image creation with better coding efficiency. (Eighth Embodiment) In the fourth embodiment, the unit for setting switching information is set to each pixel block, and the field setting information indicating the size of the pixel block or the dividing method is used to adaptively switch pixels. The method of block size. In the present embodiment, as a method of adaptively switching the pixel block size in the screen, a method of hierarchically dividing the inside of the parent block of a predetermined size into small-sized sub-blocks is used, and FIG. 50 to FIG. 55 are used. To explain. Here, a method of hierarchically dividing a parent block into blocks is explained using a 4-point tree structure. Fig. 50 is a diagram showing hierarchical block division of a 4-point tree structure. The parent block Bq of level 0 in Fig. 50. It is divided into sub-blocks Βι, 〇~Βι, 3 in the hierarchy 1, and is divided into sub-blocks Β2, 〇~B2, 1 5 in the hierarchy 2. Next, a description will be given of a method of expressing a segmentation tree of a 4-point tree structure. Figure 51 shows the split tree represented by level 3 using the block split information represented by 0 or 1. When the block division information is 1, it indicates that the sub-block is to be divided into the next layer, and when it is 0, it indicates that it is not divided. The block represented by the block splitting information set as shown in FIG. 51 is divided into four sub-blocks by the parent block as shown in FIG. 52, and the sub-block is further divided into four sub-blocks. The child (sun) block is again divided into 4 sub-children. -64- 200948092 Switching information is set for each of the blocks thus divided into different sizes. That is, it is equivalent to setting the switching information for the block corresponding to the block division 'in the split tree of FIG. 51'. • The size of the sub-blocks within the parent block is determined by the size of the parent block and the depth of the hierarchy. Fig. 53 is a view showing the sizes of the sub-blocks of the hierarchy 〇 ~ 4 for the parent block size when the 4-point tree structure is used. In Fig. 53, for example, when the parent block size is 32x3 2, the sub-dimension of the level 4 is 2x2. The local area within the picture, that is, the inside of the parent block, is divided into sub-blocks that are sequentially smaller in accordance with the depth of the hierarchy. Next, the syntax including the field setting information described in the present embodiment will be described with reference to Fig. 54. The loop filter data syntax (1 903 ) in the syntax structure of Fig. 31 described in the present embodiment is described as shown in Fig. 54. The filter_bl 〇 Ck_siZe of Fig. 54 is the 决定 which determines the block size as in the fourth embodiment, and here represents the parent block size. Next, max_layer_level is the maximum value of the depth 〇 of the obtained hierarchy. In the present embodiment, maxjayerjevel is encoded in slice units as shown in the syntax of Fig. 54, but as another embodiment, encoding may be performed in a sequence, a frame or a parent block unit. Fig. 55 is a diagram showing the syntax at the time of encoding in parent block units. Further, if both the encoding device and the decoding device use 値 which indicates the maximum depth of the same level, the max_layer_level may not be included in the grammar, but a predetermined 値 may be used. Further, in the present embodiment, although the parent block size is represented by filter_block_size, it may be a parameter indicating the minimum block size. In this case, -65-200948092, max_layer_level and filter_block_size are used, by the predetermined calculation formula. To derive the parent block size. For example, in the case of a 4-point tree structure, if filter_block_size is B and max_layer_level is L, the parent block size P can be represented by P = Bx2L. Next, NumOfParentBlock is the total number of parent blocks in one slice, and NumOfChildBlock[丨ayer] is the total number of child blocks in the hierarchy indicated by the layer in one parent block. For example, if it is a 4-point tree construct, then NumOfChildBlock[0] = l , NumOfChildBlock[l] = 4 ,

NumOfChi 1 dB 1 o ck [2] = 1 6 , NumOfChi 1 dB 1 ock[3 ] = 64 ,

NumOfChildBlock[layer] can be represented by 4lay. Next, explain valid_block_flag and block partitioning flag. After valid_block_flag is 0 or 1, the initial value of valid_block_flag is set to 〇. block_partitioning_flag is block partitioning information, as described above. When there is a division, it is set to 1, and when there is no division, it is set to 〇. Thus, according to the animation coding and decoding method according to the eighth embodiment, the hierarchical division information is used as the domain division information. , the primary block of the reference can be hierarchically divided into sub-blocks, and by setting the switching information for each divided sub-block, the filter can be adaptively controlled according to each local area in the frame. The switching timing of the processing. Further, by setting the unit for setting the switching information so as to be variable in the screen, the loop filter performs fine-grained domain segmentation in the field where the image quality is greatly affected, and the loop filtering In the field where the mirror has less influence on the image quality, the rougher domain segmentation is performed, whereby the switching information can be set efficiently. -66 - 200948092 In addition, although the division method as the parent block is a description of the division method by the 4-point tree structure, the division method that can be divided in the same area as both the encoding device and the decoding device can be performed. The present invention is not limited to the four-point tree structure. Here, although the case where the hierarchical structure in the parent block is divided into rectangular blocks has been described, the segmentation method that can be divided in the same domain as both the encoding device and the decoding device can be performed. It is not limited to a rectangular block. Q According to the present invention, the filter coefficient information set on the encoding side is encoded, and the filter coefficient information is decoded on the decoding side to be used in the animation encoding/decoding. The encoding side and the decoding side switch to the 圏 filter processing in the same process, and it is possible to suppress the propagation of the image quality deterioration while promoting the image quality improvement of the reference image used in the prediction, so that the coding efficiency is improved. [Effect of industrial use] The image encoding and decoding method and apparatus of the present invention can be used in communication media and storage media. [Fig. 1] Fig. 1 is a block diagram of the animation coding apparatus according to the first embodiment. [Fig. 2] Fig. 2 is a first embodiment. A block diagram of the loop filter processing unit in the animation coding apparatus described above - 67 - 200948092 [Fig. 3] Fig. 3 is a block diagram of the switching filter processing unit in the animation encoding apparatus according to the first embodiment Fig. 4 is a flowchart showing the operation of the animation encoding apparatus according to the first embodiment. Fig. 5 is a block diagram of the animation decoding apparatus according to the first embodiment. [Fig. 6] Fig. 6 is a block diagram of a first switching filter processing unit in the animation decoding device according to the first embodiment. [ Fig. 7] Fig. 7 is a flowchart showing the operation of the animation decoding apparatus according to the first embodiment. [Fig. 8] Fig. 8 is a block diagram showing a second switching filter processing unit in the animation decoding device according to the first embodiment. [Fig. 9] Fig. 9 is a block diagram of a third switching filter processing unit in the animation decoding device according to the first embodiment. [ Fig. 10] Fig. 10 is a block diagram showing a fourth switching filter processing unit in the animation decoding device according to the first embodiment. Fig. 11 is a block diagram of a first animation encoding apparatus according to a second embodiment. [Fig. 12] Fig. 12 is a block diagram of a switching information generation prediction unit in the first animation encoding device according to the second embodiment. [Fig. 13] Fig. 1 is a block diagram of a reference switching prediction unit in the first animation encoding device according to the second embodiment. Fig. 14 is a block diagram of a loop filter processing unit in the first animation encoding device according to the second embodiment. -68- 200948092 [Fig. 15] Fig. 15 is a flowchart showing the operation of the first animation encoding apparatus according to the second embodiment. Fig. 16 is a block diagram of the first animation decoding device according to the second embodiment. [Fig. 17] Fig. 1 is a block diagram of a reference switching prediction unit in the first animation decoding device according to the second embodiment. [Fig. 18] Fig. 18 is a flowchart showing the operation of the animation decoding device φ according to the second embodiment. Fig. 19 is a block diagram showing a second animation encoding apparatus according to the second embodiment. [Fig. 20] Fig. 20 is a block diagram showing a switching information generation prediction unit in the second animation encoding device according to the second embodiment. [Fig. 21] Fig. 21 is a block diagram of a reference switching prediction unit in the second animation encoding device according to the second embodiment. Fig. 22 is a block diagram showing a second animation decoding apparatus according to the second embodiment. [Fig. 23] Fig. 23 is a block diagram of a reference switching prediction unit in the second animation decoding device according to the second embodiment. Fig. 24 is a block diagram showing an animation encoding apparatus according to a third embodiment. [Fig. 25] Fig. 25 is a block diagram showing a loop filter processing unit in the animation encoding device according to the third embodiment. [ Fig. 26] Fig. 26 is a block diagram of a switching information generation filter processing unit in the animation encoding apparatus according to the third embodiment. - 69 - 200948092 [Fig. 27] Fig. 27 is an animation described in the third embodiment. A flow chart of the operation of the encoding device. [Fig. 28] Fig. 28 is a block diagram of the animation decoding apparatus according to the third embodiment. [Fig. 29] Fig. 29 is a block diagram showing a switching information generation filter processing unit in the animation decoding device according to the third embodiment. Fig. 30 is a flow chart showing the operation of the animation decoding apparatus according to the third embodiment. Fig. 31 is a view showing the syntax structure of the first, second, and third embodiments. Fig. 32 is a view showing the syntax of the loop filter data described in the first, second, and third embodiments. [Fig. 33] Fig. 33 is an example of a reference image when a loop filter is switched for each macroblock. [Fig. 34] Fig. 34 is a block diagram of the encoding/decoding apparatus in Non-Patent Document 1. [Fig. 35] Fig. 35 is a block diagram of an encoding/decoding apparatus in Patent Document 1. [Fig. 36] Fig. 36 is a block diagram of the encoding/decoding apparatus in Non-Patent Document 2. [Fig. 37] Fig. 37 is a block diagram showing a switching filter processing unit in the animation encoding device according to the fourth embodiment. [Fig. 38] Fig. 38 is a block diagram showing a switching filter processing unit in the animation decoding apparatus according to the embodiment of Table 4. [0039] Fig. 39 is a diagram showing a reference table for determining a block size and a block dividing method according to the fourth embodiment. [Fig. 40] Fig. 40 is a diagram showing an example of block division according to the fourth embodiment. [Fig. 41] Fig. 41 is a diagram showing an example of a reference image when switching is performed in the block division method according to the fourth embodiment. [Fig. 42] Fig. 42 is a diagram showing the syntax of the loop filter data φ according to the fourth embodiment. [Fig. 43] Fig. 43 is a block diagram showing a loop filter processing unit in the animation encoding device according to the fifth embodiment. Fig. 44 is a flow chart showing the operation of the animation encoding apparatus according to the fifth embodiment. Fig. 45 is a block diagram showing a partial decoded video filter processing unit according to the sixth embodiment. [ Fig. 46] Fig. 46 is a block diagram of the animation encoding apparatus Q according to the seventh embodiment. Fig. 47 is a block diagram showing an animation decoding apparatus according to an eighth embodiment. Fig. 4 is a block diagram showing an example of a predicted image creating unit according to the seventh embodiment. [Fig. 49] Fig. 49 is a block diagram showing another example of the predicted image creating unit according to the seventh embodiment. Fig. 50 is a diagram showing an example of hierarchical block division described in the ninth embodiment. [71] Fig. 51 is a diagram showing an example of a segmentation tree and a divided block according to the ninth embodiment. [Fig. 52] Fig. 52 is a view showing an example of a divided block in the ninth embodiment. Fig. 53 is a view showing an example of the block size of each level described in the ninth embodiment. Fig. 54 is a diagram showing the syntax of the block division information according to the ninth embodiment. @ [Fig. 5 5] Fig. 5 is a diagram showing another syntax of the block division information described in the ninth embodiment. [Main component symbol description] 1 〇: Input video signal 1 1 : Predicted video signal 1 2 : Prediction error video signal 1 3 : Quantized conversion coefficient 0 14 : Coded data 1 5 : Prediction error video signal 16: Local decoded image Signal 1 7 : Filter coefficient information 1 8 : Switching information 1 9 : Reference video signal 2 〇: Restoring video signal 21 : Decoding video signal - 72 - 200948092 22 : Output video signal 23 : Field setting information 101 : • 102 : 103 : 104 : 105 : © 106 : 107 : 108 ·_ 109 : 110: 111: 112: 113: 〇114: 201 : 202 : 203 : 204 :

205 : 20 5 A 205B 20 5 C Predictive signal generation unit reducer conversion/quantization unit entropy coding unit inverse conversion • inverse quantization unit adder loop filter processing unit reference image buffer coding control unit filter setting unit switching filter Mirror processing unit switching information generation unit, hob mirror processing unit, loop filter switching unit, entropy decoding unit, inverse conversion, inverse quantization unit, prediction signal generation unit, addition unit switching filter processing unit: switching filter processing unit: switching filter processing unit: Switching filter processing unit-73 200948092 205D : switching filter processing unit 206 : reference video buffer 2 0 7 : decoding control unit 208 : filter processing unit 209 : loop filter switching unit 2 1 0 : rear filter Switching unit 3 0 1 : switching information generation prediction unit 3 0 1 A : switching information generation prediction unit 3 0 1 B : switching information generation prediction unit 302: loop filter processing unit 3 03 : local decoded video buffer 304: The restored video buffer 3 05 : Reference switching prediction unit 3 05A : Reference switching prediction unit 3 0 5 B : Reference switching prediction unit 401 : Reference switching prediction unit 4 0 1 A : Reference switching prediction unit 4 0 1 B : Reference switching Prediction section 4 0 2: Decoded video buffer 403: restored video buffer 501: loop filter processing unit 5 02: switching information generation filter processing unit 03: switching information generation unit 601: switching information generation filter processing unit - 74 - 200948092 602 : switching information generation unit 701 : local decoded video filter processing unit '702 : filter boundary determination processing unit ' 7 〇 3 : switch 704 : deblocking furnace mirror processing unit 7 0 5 : switch 706 : image restoration Filter processing unit 801 : Deblocking filter unit 8 02 : Filter setting/switching information generating unit 803 : Decoded video buffer 804 : Predicted video creating unit 805 : Motion vector generating unit 811 : Deblocking filter unit 813 : Decoding Video buffer 8 1 4 : Predicted video preparation unit Q 821 : Switch 822 : Restored video creation unit 823 : Interpolated video creation unit 8 3 1 : Switch 8 3 2 : Integer pixel filter unit 8 3 3 : Bit expansion Part 834: Interpolation image creation unit 8 3 5 : Switch 83 6 : Weighted prediction image creation unit -75 - 200948092 8 3 7 : Bit reduction unit 847 : Bit reduction unit 901 : Deblocking filter processing unit 902: Deblocking filter processing unit 903: switching unit 904: encoding parameters Output section 905: Rear section filter setting section 906: Rear section filter processing section 1 000: Animation encoding apparatus 1900: Cartridge level syntax 1901: Sequence parameter set syntax 1 902: Image parameter set syntax 1 903: Slice level syntax 1 904: Slice header syntax 1 905 : Slice data syntax 1 906 : Loop filter data syntax 1 907 : Macro block level syntax 1908: Macro block layer syntax 1909 : Macro block prediction syntax 2000 : Animation decoding device 3 000 : animation encoding device 3 00 1 : animation encoding device 4000 : animation decoding device 400 1 : animation decoding device - 76 - 200948092 5000: animation encoding device 6000 : animation decoding device 7000 : animation encoding device 8000 : animation decoding device SW : switch A, B: terminal.

-77-

Claims (1)

200948092 VII. Patent application scope: 1. An animation coding method belongs to an animation coding method for predicting an image to be encoded by using the encoded image as a reference image, which is characterized by: a step of locally decoding the image of the encoded image 'applicable filter' to generate a reconstructed image; and a step of setting filter coefficient information of the predescriptor filter; and encoding the preamble filter coefficient information; and a step of encoding the specific information of the pre-recorded local decoded image or the pre-recovered image as a reference image, and encoding the pre-recorded local decoded image or the pre-recovered image as a reference image based on the pre-recorded specific information. The steps in the memory. 2. An animation decoding method, which is an animation decoding method for predicting an image to be decoded by using a decoded image as a reference image, and is characterized in that: a filter is applied to the decoded image to generate a step of restoring the image; and a step of decoding the filter coefficient information of the pre-filter; and decoding the specific information to be used as the reference image to use the previously decoded image or the pre-recovered image; and The step of storing the pre-decoded image or the pre-recovered image as a reference image in the memory based on the pre-recorded specific information. 3. An animation coding method, which is an animation coding method for predicting an image to be encoded by using a coded image as a reference image, and is characterized in that it has: -78- 200948092 a locally decoded image of the image, a filter for generating a reconstructed image; and a step of setting a filter coefficient information of the front filter; and * a step of encoding the front filter coefficient information; and The step of encoding the specific information of the pre-recorded local decoded image or the pre-recovered image as the reference image, and encoding the pre-recorded local decoded image or the pre-recorded original image as the reference image based on the pre-recorded specific information The steps to use when predicting. 4. An animation decoding method, which is an animation decoding method for predicting an image to be decoded by using a decoded image as a reference image, and is characterized in that: a filter is applied to the decoded image to generate a restoration a step of decoding; and a step of decoding the filter coefficient information of the pre-filter; and decoding the specific information for indicating that the pre-recorded local decoded image or the pre-recovered image is used as the reference image; And 〇 based on the pre-recorded specific information, the pre-recorded decoded image or the pre-recorded restored image is used as a reference image and used in the prediction. 5. An animation coding method, which is an animation coding method for predicting an image to be encoded by using the encoded image as a reference image, and is characterized by: having: local decoding of the encoded image Image, applying a filter to generate a reconstructed image; and setting the filter coefficient information of the pre-filter; and encoding the pre-filter coefficient information; and -79-200948092 based on The pre-recorded partial decoded image or the pre-recovered image is used as a reference image to be used as a reference image, and the pre-recorded partial decoded image or the pre-recorded restored image is stored as a reference image in the memory. 6. An animation decoding method, which is an animation decoding method for predicting an image to be decoded by using a decoded image as a reference image, and is characterized in that: a filter is applied to the decoded image to generate a step of restoring the image; and decoding the filter coefficient information of the pre-filter; and decoding the pre-record based on the specific information used to indicate that the pre-recorded local decoded image or the pre-recovered image is used as the reference image The image or the pre-recovered image is stored as a reference image in the memory. 7. The animation encoding method according to the first or third aspect of the patent application, comprising: specific information for indicating that the pre-decoded video or the pre-reconstructed video is used as an output image on the decoding side. , the step of encoding on the encoding side. 8. The animation decoding method according to the second or fourth aspect of the patent application, wherein: the specific information for indicating that the pre-decoded video or the pre-reconstructed video is used as an output video is used for decoding And the step of outputting the pre-decoded image or the pre-recovered image as an output image based on the specific information of the object ij. 9. In the animation decoding method described in the sixth paragraph of the patent application, the -80-200948092 includes: a specific information for indicating that the pre-decoded video or the pre-reconstructed video is used as an output image. The step of generating; and the step of outputting the pre-decoded image or the pre-recovered image as an output image based on the pre-recorded specific information. 1 〇 · An animation coding method as described in any of items 1, 3, and 7 of the patent application, wherein the pre-recorded specific information is coded according to each partial field in the frame or the frame 。. 1 1 The animation decoding method according to any one of claims 2, 4, and 8, wherein the pre-recording specific information is decoded in accordance with each partial field in the frame or the frame. 12. The method of encoding an animation according to any one of claims 1, 3, and 7, wherein: the step of: calculating a difference between a pre-recorded local decoded image and a pre-recovered image and an input original image; And 〇 generates a step for indicating the pre-recorded specific information of the image having the smaller error as the reference image. The animation coding method described in claim 5, wherein the method includes: using an index that can be obtained from a previously recorded partial decoded image or a pre-recorded decoded image, that is, an absolute sum of difference from a peripheral pixel, and a difference squared Any of the above, activity, spatial frequency, edge strength, and edge direction to generate the pre-recorded specific information. 14. The animation decoding method according to claim 6 or claim 9, wherein the method includes: using an indicator that can be obtained from a previously decoded partial decoded image or a pre-recorded -81 - 200948092 decoded image, that is, a peripheral pixel The steps of generating the pre-recorded specific information by using either absolute difference sum, difference sum of squares, activity, spatial frequency, edge intensity, or edge direction 〇 15. The animation code described in item 5 of the patent application scope The method s includes the step of generating a pre-recorded specific information using any of a part of the encoded information, that is, a quantization parameter, a block size, a prediction mode, a motion vector, and a conversion coefficient information. 16. The animation decoding method as described in claim 6 or claim 9, wherein the method includes: using a part of the encoded information, that is, a quantization parameter, a block size, a prediction mode, a motion vector, and a conversion coefficient information. Above, to generate the steps of the pre-recorded specific information. 17. An animation coding apparatus, which is an animation coding apparatus that uses a coded video as a reference image and is used for prediction of a next coded image, and includes: a restoration video generation unit; For the locally decoded image of the encoded image, a filter is applied to generate the restored image; and a setting unit is used to set the filter coefficient information of the front filter; and a decoding unit is used to add the information of the front mirror coefficient The encoding unit and the encoding unit are configured to encode the specific information used to indicate that the pre-recorded local decoded image or the pre-recorded restored image is used as a reference image; and the storage unit is based on the pre-recorded specific information. The locally decoded image or the pre-recovered image is stored as a reference image in the memory. -82- 200948092 18--Animated decoding device is an animation decoding device belonging to the prediction that the decoded image is used as a reference image for the next decoded image, and is characterized by: 'Reconstructed image generation a portion for applying a filter to the decoded image to generate a restored image; and a first decoding unit for decoding filter coefficient information of the pre-filter; and a second decoding unit for use in the second decoding unit The specific information for the purpose of using the pre-recorded partial decoded image or the pre-recovered restored image as a reference image is decoded, and the storage unit uses the pre-recorded decoded image or the pre-reposted restored image as the reference image based on the pre-recorded specific information. Save in memory. _ 19. An animation coding method is an animation coding method for predicting an image to be encoded by using the encoded image as a reference image, and is characterized in that: 〇 a specific information encoding step is used To encode specific information, which is used to indicate whether the restored image generated by applying the filter to the locally decoded image of the encoded image is used as a pre-referenced image, or whether the pre-recorded locally decoded image is used as a reference image. The use and storage steps are based on the pre-recorded specific information, and the pre-recorded local decoded image or the pre-recovered restored image is stored as a reference image in the cell. The animation coding method described in claim 19, wherein the pre-recording specific information coding step is to set and encode the pre-recorded specific information according to each partial field in the frame. -83- 200948092 The procedure for encoding the field setting information for the size of the local area beforehand is determined. 21. The animation encoding method as recited in claim 19, wherein the pre-recording specific information encoding step is configured to encode the pre-recorded specific information according to each partial field in the frame, and includes setting based on the pre-recorded specific information. The steps for the filter coefficient information of the pre-filter. 22. The animation encoding method as recited in claim 19, wherein the pre-recording specific information encoding step encodes the pre-recorded field information according to each partial field in the frame or the frame. 23. The method of encoding an animation as described in claim 20 or 22, wherein the pre-recorded field setting information is used to indicate the size of the field of the predecessor. 24. The method of encoding an animation as recited in claim 20 or 22, wherein the pre-recorded field setting information is used to indicate a segmentation method required to divide a local region of a predetermined size. 25. The method of encoding an animation according to claim 24, wherein the pre-recording field setting information is used to indicate a size of a pre-recorded partial field determined according to a pre-prepared domain size specification table or a pre-recording method. Index. 26. The animation coding method according to claim 25, wherein a plurality of pre-registration field size specification tables are prepared, based on one or more of an image size, an image type, and a quantization parameter for specifying a quantization thickness. Switch it. 27. An animation decoding method is an animation decoding method for predicting an image to be decoded by using the decoded image as a -84-200948092 reference image, which is characterized by: * specific information decoding The step is to decode the specific information, which is used to indicate whether the restored image generated by applying the filter to the decoded image is used as a reference image or whether the pre-recorded decoded image is used as a reference image; And the saving step is based on the pre-recorded specific information, and the pre-decoded image 0 or the pre-recovered image is stored as a reference image in the memory. 28. The animation decoding method according to claim 27, wherein the pre-recording specific information decoding step is to decode the pre-recorded specific information and set according to each partial field in the frame, and includes the domain setting. The information decoding step is used to decode the field setting information for determining the size of the pre-recorded local area. 29. The animation decoding method as recited in claim 27, wherein the pre-recording field setting information decoding step is to set the pre-recording field information to be decoded according to each partial field in the frame or the frame. 30. An animation decoding method as described in claim 28 or claim 29, wherein the pre-recording field setting information is used to indicate the size of the field of the preceding office. 3 1. The animation decoding method as described in claim 28 or claim 29, wherein the 'previous field setting information' is used to indicate a division method required to divide the internal area of the predetermined size. 32. If the method of decoding the mobile document described in the 28th or 29th paragraph of the patent application is 'the 'Bug field setting information', the pre-recorded field determined according to the pre-prepared field size specification table of -85- 200948092 Size or an index that represents the pre-segmentation method. 33. The animation decoding method according to claim 32, wherein a plurality of pre-registration field size specification tables are prepared based on one or more of image size, image type, and quantization parameter for specifying quantization thickness. Switch it. 34. An animation coding method, characterized in that the encoded local decoded image described in any one of the applications i, 3, 5, 13, 15, 19, 20, 21, 22, and 24 is applicable. The step of generating a reconstructed image by the filter includes: a step of determining whether the locally decoded image is a block boundary; and a filter for removing the block distortion for the pixel that has determined that the pre-recorded locally decoded image is a block boundary The steps of processing. 3 5. An animation decoding method, characterized in that the step of applying a filter to generate a reconstructed image for the decoded image described in any one of claims 2, 4, 6, 9, 27, 28, and 29 is And including: a step of determining whether the locally decoded image is a block boundary; and a step of performing filter processing for removing the block distortion for the pixel that has determined that the pre-recorded locally decoded image is a block boundary. 36. An animation coding method is an animation coding method used for motion compensation prediction using a coded image as a reference image. The feature is: a method for: locally decoding an image of the encoded image, applying a filter 'To generate a reconstructed image; and -86- 200948092 to set the filter coefficient information of the pre-filter; and to encode the pre-filter coefficient information; and 'will be used to represent the pre-recorded local decoded image Or the pre-recovery image is used as a specific information for the purpose of the reference image, and the step of encoding; and based on the pre-recorded specific information, the pre-recorded local decoded image or the pre-recovered image is used as the reference image for motion compensation prediction. Step 0 ❹ 37. The animation decoding method belongs to an animation decoding method for using a decoded image as a reference image for motion compensation prediction, and is characterized in that: a filter is applied to the decoded image to generate a restoration. The steps of the image; and the step of decoding the filter coefficient information of the pre-filter; and a step of decoding the specific information used to use the pre-recorded local decoded image or the pre-recovered image as a reference image, and decoding the pre-decoded image or the pre-recovered image as a reference image based on the pre-recorded specific information. The steps to be used in motion compensation prediction. 38. The animation coding method described in claim 20 or 22, wherein the pre-recorded field setting information is used to divide the pre-recorded local area into a smaller class according to the deeper the hierarchy. Split the information. 39. The animation coding method according to claim 38, wherein the pre-hierarchical segmentation information is field segmentation information indicating whether or not each domain in each hierarchy is to be divided. 4. The animation coding method described in claim 38, -87- 200948092 wherein the pre-hierarchical segmentation information ' contains the largest class information' which is used to indicate the maximum depth of the hierarchy. 4 1. The animation decoding method described in claim 28 or 29, wherein the pre-recording field setting information is used to divide the pre-recorded local area into the smaller required segments according to the deeper level of the hierarchy. Split the information. 4 2. The animation decoding method as described in claim 41, wherein the pre-hierarchical segmentation information is field segmentation information indicating whether or not each domain in each hierarchy is to be divided. 4 3. The animation decoding method as recited in claim 41, wherein the pre-hierarchical segmentation information' contains the largest class information' and the system uses $ to indicate the maximum depth of the hierarchy. -88-