JP2007300678A - Information processing apparatus and method, recording medium, and program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To improve a coding efficiency, a compression efficiency or a transmission efficiency when transmitting image data. <P>SOLUTION: Classification adaptive processing section 24-1 to 24-n calculate a prediction value corresponding to a target pixel of a predicted frame stored in a prediction frame memory 21 by using pixel data configuring the classification adaptive processing, based on a prediction tap at each position within a searching area having, as a center, a target pixel of a reference frame stored in a reference frame memory 22. A comparator 25 detects a prediction value minimizing a difference between the prediction value and a target pixel value. A quantizer 28 quantizes a motion vector specified by the position of the prediction value and the position of the target pixel and transmits the result. <P>COPYRIGHT: (C)2008,JPO&INPIT

Description

  The present invention relates to an information processing apparatus and method, a recording medium, and a program, and more particularly, to an information processing apparatus and method, a recording medium, and a program that enable efficient encoding.

  Recently, compression methods represented by MPEG (Moving Picture Experts Group) have become widespread, and video data and audio data are compressed by the MPEG method and then recorded on a recording medium such as a hard disk, magnetic tape, or optical disk, or network Or transmitted via satellite.

  In the MPEG system, for example, as shown in FIG. 1, when a previous frame and a subsequent frame exist before and after the current frame, the current frame is divided into, for example, blocks of 8 × 8 pixels. Then, a block of 8 × 8 pixels is extracted within a predetermined search range of the previous frame, and block 1 of the current frame and block 2 of the previous frame are subjected to absolute difference sum, square error sum, or representative point matching for each pixel. Compared by processing. The position of the block 2 of the previous frame is sequentially moved within the search range.

  Then, the relative coordinate position having the smallest absolute difference sum, square error sum, or representative point matching value in the search range is estimated as a motion vector.

  Further, when the class classification adaptive process is used, the pixel at the block position is predicted from the pixel of the block 2 corresponding to the motion vector based on the class classification adaptive process, and the motion vector and the predicted value are transmitted.

  However, such a conventional predictive coding method using block matching and class classification adaptive processing has a problem that transmission efficiency, compression efficiency, and coding efficiency are poor.

  The present invention has been made in view of such a situation, and is intended to improve transmission efficiency, compression efficiency, and encoding efficiency.

  An information processing apparatus according to an aspect of the present invention includes a motion vector extraction unit that extracts a motion vector of first content data to be generated, a holding unit that holds second content data serving as a reference, and the motion vector extraction unit Based on the motion vector extracted by the above, a prediction tap extraction means for extracting a prediction tap from the second content data, and a classification classification adaptation process is performed on the prediction tap extracted by the prediction tap extraction means. , Calculating means for calculating a predicted value, and combining means for combining the predicted value calculated by the calculating means as the first content data.

  From the acquisition means for acquiring data related to the first content data, the data related to the first content data acquired by the acquisition means, from the data related to the first content data, the first Flag extraction for extracting a flag representing a prediction residual corresponding to a difference between a predicted value corresponding to the first content data generated by performing class classification adaptation processing on the second content data and the first content data Means can further be provided.

  When the flag extraction unit extracts the flag representing the prediction residual, the flag extraction unit extracts the prediction residual from data related to the first content data, and the data extraction unit extracts the prediction residual. Decoding means for decoding the received data can be further provided.

  The first content data and the second content data are image data, the first content data is a first frame of the image data, and the second content data is the first content data. The second frame may be a temporally previous frame.

  The computing means can compute the predicted value on a pixel-by-pixel basis.

  Update means for updating the second content data can be further provided.

  An information processing method, a recording medium, and a program according to one aspect of the present invention are a method, a recording medium, and a program corresponding to the information processing apparatus according to one aspect of the present invention described above.

  In the information processing apparatus and method, the recording medium, and the program according to one aspect of the present invention, the motion vector of the first content data to be generated is extracted, the second content data serving as a reference is held, and the extracted motion Based on the vector, a prediction tap is extracted from the second content data, a class classification adaptive process is performed on the extracted prediction tap, a prediction value is calculated, and the calculated prediction value is used as the first content data. Synthesized.

  As described above, according to the present invention, transmission efficiency, compression efficiency, and encoding efficiency can be improved.

  Furthermore, according to the present invention, it is possible to easily and reliably receive and decode transmission data with improved transmission efficiency, compression efficiency, or encoding efficiency.

  FIG. 2 shows a configuration example of a transmission apparatus in an image processing system to which the present invention is applied. In this transmission device 11, the reference frame memory 22 stores image data of a reference frame that is a reference for prediction. The predicted frame memory 21 sequentially stores image data of predicted frames predicted based on the image of the reference frame.

  The address setting unit 23-1 has a first search range (a search range corresponding to the target image of the predicted frame stored in the frame memory 21) in the pixel data stored in the frame memory 22. An address of an arbitrary number of pixels constituting the prediction tap and the class tap of the position (position corresponding to the motion vector v0) is set, and pixel data corresponding to the address, that is, pixel data constituting the prediction tap and the class tap Is supplied to the class classification adaptive processing unit 24-1. The class classification adaptation processing unit 24-1 performs the class classification adaptation process based on the prediction tap and the pixel data of the class tap supplied from the address setting unit 23-1, and the prediction frame stored in the prediction frame memory 21. The predicted value of the corresponding pixel (target pixel) is calculated.

  The address setting unit 23-2 extracts the pixel data of the prediction tap and the class tap at the second position (position corresponding to the second motion vector v1) within the search range, and sends it to the class classification adaptation processing unit 24-2. Supply. The class classification adaptation processing unit 24-2 performs class classification adaptation processing based on the prediction tap and class tap pixel data supplied from the address setting unit 23-2, and calculates a prediction value for the pixel of interest.

  Similar configurations are provided for the number of motion vectors (n) obtained by searching the search range. That is, n combinations of address setting units and class classification adaptive processing units are prepared. For example, assuming that the search range is from minus 8 to plus 8 in both the horizontal and vertical directions, the number of n is 289 (= 17 × 17).

  The comparator 25 compares the n prediction values supplied from the class classification adaptive processing units 24-1 to 24-n with the target pixel supplied from the prediction frame memory 21, and detects the difference as a prediction residual. In addition, the smallest prediction residual among the prediction residuals is selected as the evaluation value. The threshold determination unit 26 compares the minimum prediction residual (evaluation value) supplied from the comparator 25 with the threshold. When the evaluation value is equal to or smaller than the threshold value, the threshold value determination unit 26 supplies the motion vector to the post-processing unit 27. The threshold determination unit 26 outputs the motion vector and the prediction residual to the post-processing unit 27 when the prediction residual is larger than the threshold.

  When the motion vector is supplied from the threshold determination unit 26, the post-processing unit 27 outputs it as it is. When the motion vector and the prediction residual are supplied from the threshold determination unit 26, the post-processing unit 27 adds a flag indicating that it is a prediction residual to the prediction residual.

  The motion vector output from the post-processing unit 27 or the motion vector and the prediction residual to which the flag is added are supplied to the quantizer 28. The quantizer 28 quantizes the prediction residual among these by using, for example, a code such as Lloyd Max and run length, and transmits the result to a predetermined transmission path. In addition to various communication paths, this transmission path includes a recording medium that transmits data by recording and reproduction.

  For example, the control unit 30 including a microprocessor controls each unit of the transmission device 11. A magnetic disk 41, an optical disk 42, a magneto-optical disk 43, a semiconductor memory 44, or the like is appropriately attached to the control unit 30 via an interface 31 as necessary.

  FIG. 3 shows the configuration of the class classification adaptive processing units 24-1 to 24-n (hereinafter, simply referred to as the class classification adaptive processing unit 24 when it is not necessary to distinguish them individually). The class tap extraction unit 61 is a pixel supplied from a corresponding one of the address setting units 23-1 to 23-n (hereinafter simply referred to as the address setting unit 23 when it is not necessary to distinguish them individually). Class taps are extracted from the data and supplied to an ADRC (Adaptive Dynamic Range Coding) unit 62. The ADRC unit 62 performs, for example, 1-bit ADRC processing on the pixel data supplied from the class tap extraction unit 61.

  That is, the maximum value and the minimum value of the plurality of pixel data constituting the class tap are detected, and the difference between the maximum value and the minimum value is further calculated as a dynamic range. Each pixel data is normalized by subtracting the minimum value and further dividing by the dynamic range.

  The normalized data is compared with a predetermined reference value (for example, 0.5), and is set to 1 when it is larger than the reference value, for example, 0 when it is smaller than the reference value.

  That is, for example, when the number of pixels constituting the class tap is nine, 9-bit data is supplied from the ADRC unit 62 to the class code determining unit 63. The class code determination unit 63 determines the class code of the class tap based on the 9-bit data, and outputs it to the prediction coefficient memory 64.

  A prediction coefficient corresponding to the class code is stored in the prediction coefficient memory 64 in advance, and the prediction coefficient corresponding to the class code supplied from the class code determination unit 63 is output to the prediction value calculation unit 66.

  The prediction tap extraction unit 65 extracts pixel data constituting the prediction tap from the pixel data supplied from the address setting unit 23 and outputs the pixel data to the prediction value calculation unit 66. The prediction value calculation unit 66 multiplies the pixel data constituting the prediction tap supplied from the prediction tap extraction unit 65 by the prediction coefficient supplied from the prediction coefficient memory 64, and from the sum (that is, from linear linear combination). ) To calculate a predicted value.

  Next, processing of the transmission device 11 will be described with reference to the flowcharts of FIGS. 4 and 5.

  First, in step S <b> 11, the control unit 30 sets a transmission determination threshold for the threshold determination unit 26. This transmission determination threshold is used in the process of step S26 described later.

  In step S12, processing for accumulating previous frame data is executed, and in step S13, processing for accumulating current frame data is executed. That is, for example, image data for one frame is stored in the prediction frame memory 21, read out from there again, transmitted to the reference frame memory 22, and stored. Then, the predicted frame memory 21 stores image data for a new frame following that. In this way, the pixel data of the previous frame is stored in the reference frame memory 22, and the image data of the current frame (a frame temporally later than the reference frame) is stored in the prediction frame memory 21.

  In step S <b> 14, the prediction frame memory 21 extracts the data of the pixel of interest from the stored image data for one frame based on the control of the control unit 30, and supplies it to the comparator 25.

  In step S <b> 15, the control unit 30 sets a comparison value in the comparator 25. This comparison value is updated to a smaller prediction residual value in step S24 in order to obtain the evaluation value as the minimum prediction residual, and the maximum value is set as the initial value. The This comparison value is used in the process of step S23.

  Next, in step S16, addresses corresponding to the target pixel extracted in the process of step S14 are set for the address setting units 23-1 to 23-n, respectively. As a result, the pixel data including the class tap and the prediction tap corresponding to each search position (motion vector) in the search range corresponding to the target pixel from the address setting units 23-1 to 23-n is converted into the corresponding class classification. The data can be taken into the adaptive processing units 24-1 to 24-n.

  Therefore, in step S17, the class tap extraction unit 61 extracts class taps from the supplied pixel data. For example, as shown in FIG. 6, the class tap is 3 × 3 pixels shown in black in the figure out of 7 × 7 pixels centered on the target pixel of the prediction frame.

  In step S18, the ADRC unit 62 performs 1-bit ADRC processing on the nine pieces of pixel data constituting the class tap. As a result, the nine pieces of pixel data are converted into values of 0 or 1, respectively, and 9-bit data is supplied to the class code determination unit 63. Based on the 9-bit data supplied from the ADRC unit 62, the class code determination unit 63 determines a class code corresponding to the class tap composed of the nine pixel data, and outputs the class code to the prediction coefficient memory 64. .

  In step S <b> 19, the prediction coefficient memory 64 reads the prediction coefficient for the class code supplied from the class code determination unit 63 and outputs the prediction coefficient to the prediction value calculation unit 66.

  In step S <b> 20, the prediction tap extraction unit 65 acquires pixel data constituting the class tap from the pixel data supplied from the address setting unit 23. FIG. 7 shows an example of a class tap. In this example, among the 7 × 7 pixels centered on the target pixel of the prediction frame, 13 pixels shown in black at the center are class taps.

  In step S21, the predicted value calculation unit 66 executes a predicted value calculation process. That is, the prediction value calculation unit 66 calculates a linear primary combination of 13 pixel data constituting the prediction tap supplied from the prediction tap extraction unit 65 and 13 prediction coefficients supplied from the prediction coefficient memory 64. Then, the predicted value is calculated and output to the comparator 25.

  In step S22, the comparator 25 executes a prediction residual calculation process. That is, the comparator 25 includes the predicted value supplied from the class classification adaptive processing unit 24 (in this case, the class classification adaptive processing unit 24-1) and the target pixel (true value) supplied from the predicted frame memory 21. By calculating the difference, the prediction residual is calculated.

  In step S23, the comparator 25 compares the prediction residual (evaluation value) obtained in the process of step S22 with the comparison value. In this case, this comparison value is set to the maximum value in the process of step S15. Therefore, it is determined that the evaluation value is smaller than the comparison value, and the process proceeds to step S24, where the comparator 25 predicts the value of the comparison value set to the maximum value in the process of step S15 and calculated in the process of step S22. Update to residual (evaluation value). That is, a smaller value is set as the comparison value.

  In step S25, the threshold determination unit 26 holds a motion vector corresponding to the evaluation value obtained in the process of step S22 in a built-in memory (updated when there is a motion vector already held). . In this case, since the output of the class classification adaptive processing unit 24-1 is processed, this motion vector is v0.

  In step S26, the threshold determination unit 26 compares the prediction residual (evaluation value) obtained in the process of step S22 with the transmission determination threshold set in the process of step S11. When it is determined that the prediction residual (evaluation value) is equal to or larger than the transmission determination threshold, in step S27, the threshold determination unit 26 holds the prediction residual calculated in the process of step S22 as an evaluation value. To do.

  When it is determined that the prediction residual (evaluation value) is smaller than the transmission determination threshold, the process of step S27 is skipped.

  If it is determined in step S23 that the prediction residual (evaluation value) is equal to or greater than the comparison value, the processes in steps S24 to S27 are skipped.

  In step S28, the threshold determination unit 26 determines whether or not the processing for all positions within the search range has been completed. That is, it is determined here whether or not the processing for all the outputs of the class classification adaptive processing units 24-1 to 24-n has been completed. If there is an unprocessed output among the outputs of the class classification adaptive processing units 24-1 to 24-n, the process returns to step S16, and the subsequent processing is repeatedly executed.

  As described above, the processes in steps S16 to S28 are executed for all n predicted values output from the class classification adaptive processing units 24-1 to 24-n.

  If it is determined in step S28 that the processing of all positions within the search range has been completed, the process proceeds to step S29, and the threshold determination unit 26 stores a prediction residual (evaluation value) smaller than the transmission determination threshold. Determine whether. That is, n prediction residuals (evaluation values) are obtained by processing all positions in the search range, and at least one of the n prediction residuals is smaller than the transmission determination threshold. If there is a motion vector, the motion vector corresponding to the smallest one is held in the process of step S25. Therefore, in that case, in step S <b> 30, the threshold determination unit 26 reads out the motion vector held in the process of step S <b> 25 and outputs it to the post-processing unit 27.

  If it is determined in step S29 that there is no evaluation value smaller than the transmission determination threshold value, the prediction residual (evaluation value) held by the process in step S27 in the threshold value determination unit 26 and step S31 is used as the post-processing unit 27. Execute the process to output to.

  That is, if all n prediction residuals (evaluation values) are equal to or larger than the transmission determination threshold, the prediction residual corresponding to the smallest value among them is held in the process of step S27. Therefore, the threshold determination unit 26 outputs the prediction residual to the post-processing unit 27 as an evaluation value (minimum prediction residual) of the search range currently targeted.

  After the processes of steps S30 and S31, in step S32, the threshold determination unit 26 determines whether or not the processing of all search ranges within one frame has been completed, and there is still a search range that has not been completed. In that case, the process returns to step S14, and the subsequent processing is repeatedly executed.

  If it is determined in step S32 that the processing for all the search ranges has been completed, in step S33, the post-processing unit 27, when only the motion vector is supplied from the threshold determination unit 26, displays the motion vector. If the motion vector and the prediction residual (evaluation value) are output as is and supplied from the threshold determination unit 26, a flag indicating that this is a prediction residual is added to the prediction residual. To do.

  In step S34, the quantizer 28 outputs the motion vector supplied from the post-processing unit 27 as it is. When the motion vector and the prediction residual (evaluation value) have been supplied, the quantizer 28 quantizes the prediction residual and transmits it to the transmission path.

  As described above, as shown in FIG. 8, a predetermined pixel is selected as the target pixel 91 in the prediction frame 90 stored in the prediction frame memory 21. Then, a pixel corresponding to the target pixel 91 of the reference frame 80 stored in the reference frame memory 22 is selected as the target corresponding pixel 83, and a predetermined range centered on the target corresponding pixel 83 is set as the search range 81. The

  Within the search range 81, pixels in a predetermined range are selected as the prediction taps 82, and the prediction values are calculated as described above based on the pixels constituting the prediction taps 82. Then, a difference between the predicted value and the target pixel 91 is calculated as a prediction residual.

  The prediction tap 82 is sequentially moved to n positions within the search range 81 as indicated by prediction taps 82-1 to 82-n. Then, the smallest one of the n prediction residuals obtained corresponding to each of the n prediction taps is selected as the evaluation value of the search range 81.

  The above process is executed by sequentially selecting all the pixels in the prediction frame 90 as the target pixel 91.

  As described above, the image data transmitted to the transmission path is received and decoded by a receiving apparatus as shown in FIG.

  In this receiving apparatus 111, the inverse quantizer 121 is encoded by the transmitting apparatus 11, acquires the data transmitted through the transmission path, and performs inverse quantization. The transmission determination unit 122 reads a flag from the data supplied from the inverse quantizer 121 and outputs a motion vector of the transmitted data to the address setting unit 123. The address setting unit 123 extracts pixel data in a range corresponding to the motion vector from the pixel data of the reference frame already demodulated stored in the reference frame memory 124, and the class classification adaptive processing unit 125. To supply.

  The class classification adaptation processing unit 125 performs class classification adaptation processing on the pixel data supplied from the address setting unit 123 to generate pixel data.

  The class classification adaptation processing unit 125 has basically the same configuration as the class classification adaptation processing unit 24 of the transmission device 11 of FIG. Therefore, FIG. 3 is also referred to as the class classification adaptation processing unit 125 in the following description.

  When the transmission determination unit 122 determines that the data input from the inverse quantizer 121 includes prediction residual data, the transmission determination unit 122 supplies the data to the decoding unit 126. The decoding unit 126 decodes this prediction residual.

  The synthesizing unit 127 appropriately synthesizes the pixel data supplied from the class classification adaptive processing unit 125 and the decoding unit 126 on the memory, and outputs the pixel data for one frame. Part of the pixel data output from the combining unit 127 is supplied to the reference frame memory 124 as necessary, and is stored as a reference frame for processing of subsequent frames.

  For example, the control unit 128 configured by a microprocessor or the like controls the operation of each unit of the reception device 111.

  The control unit 128 is equipped with a magnetic disk 141, an optical disk 142, a magneto-optical disk 143, or a semiconductor memory 144 as necessary via an interface 131, and programs and data stored therein are appropriately installed. The

  Next, the decoding process in the receiving apparatus 111 will be described with reference to the flowchart of FIG.

  In step S <b> 51, the inverse quantizer 121 receives the quantized image data transmitted via the transmission path, performs inverse quantization, and supplies the quantized image data to the transmission determination unit 122. In step S52, the transmission determination unit 122 determines whether or not the data that has been transmitted is motion vector data. When it is determined that the input data includes motion vector data, the transmission determination unit 122 outputs the motion vector data to the address setting unit 123 in step S53. The address setting unit 123 extracts pixel data in the search range from the pixel data stored in the reference frame memory 124 and outputs the pixel data to the class classification adaptation processing unit 125.

  In step S54, the class classification adaptation processing unit 125 executes class classification adaptation processing. That is, the class tap extraction unit 61 and the prediction tap extraction unit 65 extract the pixel data of the class tap and the prediction tap from the pixel data in the search range based on the motion vector supplied from the address setting unit 123, respectively. The ADRC unit 62 performs 1-bit ADRC processing on the class tap, and the class code determination unit 63 determines the class code based on the data obtained as a result. The prediction coefficient memory 64 outputs a prediction coefficient corresponding to the class code to the prediction value calculation unit 66. The predicted value calculation unit 66 calculates a predicted value from a linear primary combination of a prediction tap and a prediction coefficient.

  The generated predicted value is supplied to the synthesis unit 127. In step S57, the synthesizing unit 127 synthesizes the prediction value supplied from the class classification adaptation processing unit 125 as pixel data at a corresponding position in the frame.

  In step S <b> 52, data (prediction error data) determined not to be a motion vector among the transmitted data is output from the transmission determination unit 122 to the decoding unit 126. In step S56, the decoding unit 126 performs a prediction residual decoding process. That is, the decoding unit 126 decodes the prediction residual supplied from the transmission determination unit 122. The prediction residual decoded by the decoding unit 126 is supplied to the combining unit 127, and is combined with the pixel data at the corresponding position decoded by the class classification adaptive processing unit 125 in the process of step S57.

  The prediction coefficient stored in the prediction coefficient memory 64 can be acquired by learning. FIG. 11 shows a configuration example of a learning device that performs this learning. In the learning device 160, the student data generation unit 161 generates student data by appropriately changing the pixel value of the image data as the input teacher data. The class tap extraction unit 162 extracts class taps from the student data supplied from the student data generation unit 161 and outputs the class taps to the ADRC unit 163. The ADRC unit 163 performs 1-bit ADRC processing on the class tap pixel data supplied from the class tap extraction unit 162 and outputs the obtained result to the class code determination unit 164.

  The class code determination unit 164 determines a class code based on the data input from the ADRC unit 163, and outputs the class code to the normal equation generation unit 166.

  The prediction tap extraction unit 165 extracts a prediction tap from the student data supplied from the student data generation unit 161 and outputs the prediction tap to the normal equation generation unit 166. The normal equation generation unit 166 is also supplied with teacher data as parent data from which the student data generation unit 161 generates student data. The normal equation generation unit 166 generates a linear linear combination normal equation in which the sum of products of the prediction coefficient (unknown number) and the student data is equal to the value of the teacher data.

  The prediction coefficient determination unit 167 obtains a prediction coefficient that is an unknown number from the normal equation generated by the normal equation generation unit 166 by using a general matrix solving method such as a sweep-out method, and stores the prediction coefficient in the prediction coefficient memory 64. .

  Next, the operation of the learning device 160 will be described. The class tap extraction unit 162 extracts class tap pixel data (3 × 3 pixel data in the example of FIG. 6) from the student data supplied from the student data generation unit 161, and outputs it to the ADRC unit 163. The ADRC unit 163 performs 1-bit ADRC processing on the pixel data of the class tap, and outputs the obtained result to the class code determination unit 164. The class code determination unit 164 determines a class code based on the data input from the ADRC unit 163 and outputs the class code to the normal equation generation unit 166.

  From the student data generated by the student data generation unit 161, pixel data constituting the prediction tap is extracted by the prediction tap extraction unit 165 (13 pixel data in the example of FIG. 6), and is output to the normal equation generation unit 166. Supplied. The normal equation generation unit 166 generates a linear linear combination normal equation for each class code, assuming that the product sum of the prediction tap pixel data and the prediction coefficient is equal to the teacher data, and outputs the normal equation to the prediction coefficient determination unit 167. The prediction coefficient determination unit 167 determines a prediction coefficient as an unknown number based on the sweep-out method and stores the prediction coefficient in the prediction coefficient memory 64.

  In the above, when it is determined in step S29 in FIG. 4 that there is no evaluation value smaller than the transmission determination threshold, the motion vector and the prediction residual are transmitted in the process of step S31. Instead of the prediction residual, it is also possible to send the predicted value calculated in step S21 or send the pixel data value of the target pixel itself.

  It is also possible to perform processing in units of blocks instead of in units of pixels. In this case, an absolute error sum is obtained from the predicted value and input value in the block, and if the error sum is smaller than a predetermined threshold, a motion vector is transmitted. The pixel data in the block may be transmitted.

  In this case, since only one bit is added to the flag in the block, the amount of extra information is reduced, and compression can be performed efficiently.

  Furthermore, as shown in FIG. 12, the threshold value determination unit 26 (threshold value determination process) may be omitted and a motion vector may be always transmitted. By doing so, it is possible to further improve the compression efficiency, transmission efficiency, and encoding efficiency.

  Note that the quantization of the first frame data may or may not be performed. Further, the reference frame may be updated every previous frame or every several frames. Alternatively, it is also possible not to update the reference frame at all.

  When the reference frame is updated, image degradation is small, and not only the demodulation side can reproduce a higher quality image, but also the influence of noise does not propagate to the later stage, so it is more robust. A simple image transmission method can be realized.

  As described above, according to the present invention, the prediction error is the same as or smaller than the absolute difference sum or the square error sum. Therefore, compared with the method using the absolute difference sum or the square error sum, The compression efficiency can be further increased, and a high-quality image can be transmitted.

  In the above description, the transmission device 11 and the reception device 111 are configured to be independent from each other. However, they can be integrated in one device. In particular, when the transmission path is configured by a recording medium, in a device that records and reproduces data with respect to the recording medium, both the transmission device and the reception device are arranged in one device.

  In the above description, image data has been described as an example. However, the present invention can also be applied to the case of transmitting content data other than image data.

  The series of processes described above can be executed by hardware, but can also be executed by software. When a series of processing is executed by software, a program constituting the software executes various functions by installing a computer incorporated in dedicated hardware or various programs. For example, a general-purpose personal computer is installed from a network or a recording medium.

  As shown in FIGS. 2 and 9, the recording medium is distributed to provide a program to the user separately from the apparatus main body, and includes magnetic disks 41 and 141 (including floppy disks) on which the program is recorded. ), Optical disks 42 and 142 (including CD-ROM (Compact Disk-Read Only Memory), DVD (Digital Versatile Disk)), magneto-optical disks 43 and 143 (including MD (Mini-Disk)), or semiconductor memory 44 , 144, etc., as well as a ROM in which a program is recorded, a hard disk included in the storage unit, etc. provided to the user in a state of being preinstalled in the apparatus main body. .

  In the present specification, the step of describing the program recorded on the recording medium is not limited to the processing performed in chronological order according to the described order, but is not necessarily performed in chronological order. It also includes processes that are executed individually.

  Further, in this specification, the system represents the entire apparatus constituted by a plurality of apparatuses.

It is a figure explaining block matching. It is a block diagram which shows the structural example of the transmitter to which this invention is applied. It is a block diagram which shows the structural example of the class classification adaptive process part of FIG. 3 is a flowchart for explaining processing of the transmission device in FIG. 3 is a flowchart for explaining processing of the transmission device in FIG. 2. It is a figure which shows the example of a class tap. It is a figure which shows the example of a prediction tap. It is a figure explaining the prediction process in the class classification adaptive process part of FIG. It is a block diagram which shows the structural example of the receiver which applied this invention. 10 is a flowchart for explaining the operation of the receiving apparatus of FIG. 9. It is a block diagram which shows the structural example of the learning apparatus which acquires a prediction coefficient. It is a block diagram which shows the other structural example of the transmitter to which this invention is applied.

Explanation of symbols

  21 prediction frame memory, 22 reference frame memory, 23-1 to 23-n address setting unit, 24-1 to 24-n class classification adaptive processing unit, 25 comparator, 26 threshold value determination unit, 27 post-processing unit, 28 quantum Generator

Claims (9)

Motion vector extraction means for extracting a motion vector of the first content data to be generated;
Holding means for holding second content data as a reference;
Prediction tap extraction means for extracting a prediction tap from the second content data based on the motion vector extracted by the motion vector extraction means;
A calculation unit that performs a class classification adaptive process on the prediction tap extracted by the prediction tap extraction unit, and calculates a prediction value;
An information processing apparatus comprising: combining means for combining the predicted value calculated by the calculating means as the first content data. Obtaining means for obtaining data related to the first content data;
From the data related to the first content data acquired by the acquiring means, from the data related to the first content data, the second content data generated by subjecting the second content data to class classification adaptation processing 2. The information processing according to claim 1, further comprising: a flag extracting unit that extracts a flag representing a prediction residual corresponding to a difference between a predicted value corresponding to one content data and the first content data. apparatus. Data extracting means for extracting the prediction residual from data related to the first content data when the flag extracting means extracts the flag representing the prediction residual;
The information processing apparatus according to claim 2, further comprising: a decoding unit that decodes the data extracted by the data extraction unit. The first content data and the second content data are image data,
The first content data is a first frame of the image data;
The information processing apparatus according to claim 1, wherein the second content data is a second frame that is temporally prior to the first frame. The information processing apparatus according to claim 1, wherein the calculation unit calculates the predicted value in units of one pixel. The information processing apparatus according to claim 1, further comprising updating means for updating the second content data. A motion vector extraction step of extracting a motion vector of the first content data to be generated;
A holding control step for controlling holding of the second content data as a reference;
A prediction tap extraction step of extracting a prediction tap from the second content data based on the motion vector extracted by the processing of the motion vector extraction step;
A calculation step of performing a class classification adaptation process on the prediction tap extracted by the processing of the prediction tap extraction step and calculating a prediction value;
A synthesis step of synthesizing the predicted value calculated by the processing of the calculation step as the first content data. A motion vector extraction step of extracting a motion vector of the first content data to be generated;
A holding control step for controlling holding of the second content data as a reference;
A prediction tap extraction step of extracting a prediction tap from the second content data based on the motion vector extracted by the processing of the motion vector extraction step;
A calculation step of performing a class classification adaptation process on the prediction tap extracted by the processing of the prediction tap extraction step and calculating a prediction value;
And a synthesis step of synthesizing the predicted value calculated by the processing of the calculation step as the first content data. A recording medium on which a computer-readable program is recorded. A motion vector extraction step of extracting a motion vector of the first content data to be generated;
A holding control step for controlling holding of the second content data as a reference;
A prediction tap extraction step of extracting a prediction tap from the second content data based on the motion vector extracted by the processing of the motion vector extraction step;
A calculation step of performing a class classification adaptation process on the prediction tap extracted by the processing of the prediction tap extraction step and calculating a prediction value;
A program that causes a computer to execute a synthesis step of synthesizing the predicted value calculated by the processing of the calculation step as the first content data.

Images