JP4140662B2 - Coefficient seed data generation apparatus and coefficient data generation apparatus - Google Patents

Description

  The present invention relates to a coefficient seed data generation apparatus and a coefficient data generation apparatus. Specifically, parameter input means for inputting a plurality of types of parameter values corresponding to parameters included in the generation formula and for stepwise determining the quality of output information obtained by the student signal corresponding to the first information signal By obtaining coefficient seed data corresponding to the plurality of types of parameters, only the specific quality of the output obtained by the second information signal can be adjusted step by step by changing the value of the input parameter Is.

  As one of image signal compression encoding methods, an MPEG2 (Moving Picture Expert Group phase 2) encoding method is used. In the transmission / reception or recording / reproducing system based on MPEG2, the image signal is subjected to compression coding processing according to MPEG2 to be transmitted or recorded, and the received or reproduced image signal is expanded corresponding to the compression coding processing according to MPEG2. By decoding, the original image signal is restored.

  In the encoding process based on MPEG2, additional information for decoding process is transmitted together with the encoded image data in order to make the encoding process versatile and improve the compression efficiency by encoding. . The additional information is inserted into the header of the MPEG2 stream and transmitted to the decoding apparatus.

  The characteristics of an image signal obtained by decoding, not limited to MPEG, vary greatly depending on the encoding / decoding method applied. For example, the physical characteristics (frequency characteristics, etc.) differ greatly depending on the signal type such as a luminance signal, color difference signal, and three primary color signal. This difference also remains in the decoded signal that has undergone the encoding / decoding process. Also, in general, in image coding / decoding processing, the number of pixels to be encoded is often reduced by introducing spatiotemporal thinning processing. The spatio-temporal resolution characteristics of images differ greatly depending on the thinning method. Furthermore, even when the difference in spatio-temporal resolution characteristics is small, image quality characteristics such as S / N and coding distortion amount greatly differ depending on the compression rate (transmission rate) conditions in encoding.

  The present applicant has previously proposed a classification adaptation process (see Patent Document 1). This is because, in the learning process, the actual image signal (teacher signal and student signal) is used to obtain and accumulate prediction coefficients for each class in advance, and in the actual image conversion process, the class is determined from the input image signal. The output pixel value is obtained by the prediction calculation of the prediction coefficient corresponding to the class and the plurality of pixel values of the input image signal. The class is determined according to the distribution and waveform of pixel values in the spatial and temporal vicinity of the pixel to be created. By calculating a prediction coefficient using an actual image signal and calculating a prediction coefficient for each class, it is possible to create a resolution higher than that of an input image signal as compared with a simple interpolation filter process.

  In the above-described class classification adaptive processing, the resolution of the image obtained by the output image signal is fixed, and it is set to the user's preferred resolution according to the image content, etc., as in the conventional adjustment of contrast, sharpness, etc. I could not. Therefore, the present applicant has further proposed that the user can arbitrarily adjust the resolution by changing the input parameter value (see Patent Document 2 and Patent Document 3).

JP-A-8-51599 JP 2001-238185 A Japanese Patent Application No. 2000-348730

  For example, a digital image signal obtained by decoding an MPEG2 stream is subjected to class classification application processing as an input image signal, and resolution, block distortion, mosquito distortion, noise removal degree, etc. are arbitrarily changed by changing input parameter values. It is possible to adjust. In this case, it is convenient if only a specific quality among the resolution, block distortion, mosquito distortion, noise removal degree, etc. can be adjusted step by step by changing the value of the input parameter.

  An object of the present invention is to convert a first information signal, which is an encoded digital information signal or a digital information signal obtained by decoding the encoded digital information signal, into a second information signal. An object of the present invention is to provide a coefficient seed data generation device and a coefficient data generation device for adjusting only a specific quality of the obtained output in stages by changing the value of an input parameter.

  The coefficient seed data generation device according to the present invention generates a first information signal made up of a plurality of information data, which is generated by decoding an encoded digital information signal, from the first information signal. Coefficient type that is coefficient data in a generation formula for generating coefficient data used in an estimation formula used when converting to an information signal to be converted into a second information signal composed of a plurality of information data with improved quality An apparatus for generating data, which corresponds to the parameters included in the generation formula, and includes a plurality of types of parameters for determining in stages the quality of output information obtained by the student signal corresponding to the first information signal. A parameter input means for inputting a value and a digital information signal obtained by encoding a teacher signal corresponding to the second information signal; and decoding the digital information signal in the decoding process. The signal is changed in accordance with the parameter value input to the parameter input means, the decoding means for obtaining the student signal corresponding to the parameter value, and the student signal output from the decoding means, the teacher signal First data selection means for selecting a plurality of first information data located in the vicinity of the target position, and level distribution patterns of the plurality of first information data selected by the first data selection means are detected. And a class detection unit that detects a class to which the information data of the target position belongs based on the level distribution pattern, and a plurality of positions positioned around the target position in the teacher signal from the student signal output from the decoding unit. The second data selection means for selecting the second information data, the class detected by the class detection means, and the second data selection means. By using the information data of the target position in the plurality of second information data and the teacher signal, for each class, are those comprising a calculating means for obtaining the coefficient seed data.

  In the present invention, a parameter value that determines the quality of the output obtained by the student signal corresponding to the first information signal is input. The digital information signal obtained by decoding the teacher signal corresponding to the second information signal is encoded to obtain the student signal. The signal is determined according to the parameter value input in the decoding process. Is changed to obtain a student signal corresponding to the value of this parameter.

  Based on the student signal, a plurality of first information data located around the attention position in the teacher signal is selected, and based on the plurality of first information data, a class to which the information data of the attention position belongs is selected. Detected. Here, when additional information used when decoding the digital information signal is added to the encoded digital information signal, the class is based on the additional information together with the plurality of first information data described above. By detecting, more detailed classification can be performed.

  Further, based on this student signal, a plurality of second information data located around the attention position in the teacher signal is selected. Then, the value of the input parameter is changed step by step, and the class to which the attention position information data in the teacher signal belongs, the selected second information data, and the attention position information data in the teacher signal are used. Thus, coefficient seed data is obtained for each class.

  Here, the coefficient seed data is coefficient data in a generation formula including the above parameters for generating coefficient data used in an estimation formula used when converting from the first information signal to the second information signal. By using this coefficient seed data, it is possible to obtain coefficient data corresponding to the arbitrarily changed parameter value by the generation formula. Thus, when the user converts the first information signal into the second information signal using the estimation formula, the user can change the parameter value to improve the quality of the output obtained by the second information signal. Can be adjusted arbitrarily.

  As described above, the student signal is obtained by decoding the digital information signal obtained by encoding the teacher signal, and the parameter is changed by changing the signal according to the parameter value in the decoding process. A student signal corresponding to the value is obtained.

  For example, when the encoding is encoding with orthogonal transformation (discrete cosine transformation, warblet transformation, discrete sine transformation, etc.), when performing inverse orthogonal transformation according to the parameter value in the decoding process At least one of the frequency coefficient and the inverse transform base used in the above is changed.

  A plurality of student signals can be obtained by setting the gains of all AC (alternating current) coefficients among the frequency coefficients to the same value and changing the gains according to the parameter values. In this case, coefficient seed data is obtained in which the resolution at the output obtained by the second information signal described above can be adjusted stepwise according to the value of the parameter.

  When the frequency coefficient is quantized, a plurality of student signals can be obtained by changing the quantization step of the DC (direct current) coefficient among the frequency coefficients in accordance with the parameter value. In this case, coefficient seed data that can adjust the block distortion in the output obtained by the above-described second information signal in a stepwise manner according to the value of the parameter is obtained.

  A plurality of student signals can be obtained by decreasing the gain of the AC coefficient of the frequency coefficient as the frequency increases and changing the degree of decrease in the gain of the AC coefficient according to the parameter value. In this case, coefficient seed data that can adjust the mosquito noise in the output obtained by the above-described second information signal stepwise according to the value of the parameter is obtained.

  Moreover, a plurality of student signals can be obtained by adding random noise to at least the frequency coefficient or the inverse transform base and changing the level of the random noise according to the parameter value. In this case, coefficient seed data that can adjust the noise suppression degree in the output obtained by the above-described second information signal stepwise according to the parameter value is obtained.

  The coefficient data generation apparatus according to the present invention provides a first information signal composed of a plurality of pieces of information data, which is generated by decoding an encoded digital information signal, with a higher quality of information than the first information signal. Is a device for generating coefficient data of an estimation formula used when converting to a second information signal composed of a plurality of information data with improved quality, obtained by a student signal corresponding to the first information signal Parameter input means for inputting a plurality of types of parameter values for determining the quality of output information in stages, and decoding a digital information signal obtained by encoding a teacher signal corresponding to the second information signal And decoding means for changing a signal in accordance with the parameter value input to the parameter input means in the decoding process to obtain a student signal corresponding to the parameter value; Selected by the first data selecting means for selecting a plurality of first information data located around the position of interest in the teacher signal, and the first data selecting means. Class detection means for detecting a level distribution pattern of the plurality of first information data, and detecting a class to which the information data of the target position belongs based on the level distribution pattern, and a student signal output from the decoding means From the second data selection means for selecting a plurality of second information data located around the position of interest in the teacher signal, the class detected by the class detection means, and the selection by the second data selection means A combination of the class and the parameter values using the plurality of second information data and the information data of the position of interest in the teacher signal To, those comprising calculating means for obtaining the coefficient data.

  In the present invention, a parameter value that determines the quality of the output obtained by the student signal corresponding to the first information signal is input. The digital information signal obtained by encoding the teacher signal corresponding to the second information signal is decoded to obtain the student signal. The signal is determined according to the parameter value input in the decoding process. Is changed to obtain a student signal corresponding to the value of this parameter.

  Based on the student signal, a plurality of first information data located around the attention position in the teacher signal is selected, and based on the plurality of first information data, a class to which the information data of the attention position belongs is selected. Detected. Here, when additional information used when decoding the digital information signal is added to the encoded digital information signal, the class is based on the additional information together with the plurality of first information data described above. By detecting, more detailed classification can be performed.

  Further, based on this student signal, a plurality of second information data located around the attention position in the teacher signal is selected. Then, the value of the input parameter is changed step by step, and the class to which the attention position information data in the teacher signal belongs, the selected second information data, and the attention position information data in the teacher signal are used. Thus, coefficient data is obtained for each combination of class and input parameter value.

  As described above, coefficient data of the estimation formula used when converting the first information signal to the second information signal is generated. When converting from the first information signal to the second information signal, In the second information signal, the class data to which the information data of the target position belongs and the coefficient data corresponding to the changed parameter value are selectively used, and the information data of the target position is calculated by the estimation formula. . Thereby, when converting from the first information signal to the second information signal using the estimation formula, the quality of the output obtained by the second information signal is arbitrarily adjusted by changing the parameter value. it can.

  As described above, the student signal is obtained by decoding the digital information signal obtained by encoding the teacher signal, and the parameter is changed by changing the signal according to the parameter value in the decoding process. A student signal corresponding to the value is obtained.

  For example, when the encoding is encoding with orthogonal transformation (discrete cosine transformation, warblet transformation, discrete sine transformation, etc.), when performing inverse orthogonal transformation according to the parameter value in the decoding process At least one of the frequency coefficient and the inverse transform base used in the above is changed. As a result, as in the above-described coefficient seed data generation apparatus, it is possible to obtain coefficient data that can adjust the resolution, block distortion, mosquito noise, and noise suppression degree in a stepwise manner in accordance with the parameter values.

  According to this invention, the values of a plurality of types of parameters are input corresponding to the parameters included in the generation formula and for stepwise determining the quality of the output information obtained by the student signal corresponding to the first information signal. By providing parameter input means and obtaining coefficient seed data corresponding to the plurality of types of parameters, only the specific quality of the output obtained by the second information signal can be adjusted step by step by changing the value of the input parameter .

  Hereinafter, embodiments of the present invention will be described with reference to the drawings. FIG. 1 shows a configuration of a digital broadcast receiver 100 as an embodiment.

  The digital broadcast receiver 100 includes a microcomputer, and includes a system controller 101 for controlling the operation of the entire system, and a remote control signal receiving circuit 102 for receiving a remote control signal. The remote control signal receiving circuit 102 is connected to the system controller 101, receives a remote control signal RM output from the remote control transmitter 200 according to a user operation, and supplies an operation signal corresponding to the signal RM to the system controller 101. Is configured to do.

  Also, the digital broadcast receiver 100 is supplied with a receiving antenna 105 and a broadcast signal (RF modulated signal) captured by the receiving antenna 105, performs channel selection processing, demodulation processing, error correction processing, etc. And a tuner unit 106 that obtains an MPEG2 stream as an encoded image signal.

  The digital broadcast receiver 100 also temporarily decodes the MPEG2 decoder 107 that decodes the MPEG2 stream output from the tuner unit 106 to obtain the image signal Va, and the image signal output from the MPEG2 decoder 107. And a buffer memory 108 for storing.

  Further, the digital broadcast receiver 100 converts the image signal Va stored in the buffer memory 108 into an image signal Vb that presents a user-desired image quality, and outputs from the image signal processing unit 110. And a display unit 111 for displaying an image based on the image signal. The display unit 111 includes a display such as a CRT (cathode-ray tube) display or an LCD (liquid crystal display).

The operation of the digital broadcast receiver 100 shown in FIG. 1 will be described.
The MPEG2 stream output from the tuner unit 106 is supplied to the MPEG2 decoder 107 and decoded. The image signal Va output from the decoder 107 is supplied to the buffer memory 108 and temporarily stored.

  Thus, the image signal Va stored in the buffer memory 108 is supplied to the image signal processing unit 110 and converted into an image signal Vb that presents a user-desired image quality. In the image signal processing unit 110, pixel data constituting the image signal Vb is obtained from the pixel data constituting the image signal Va.

  The image signal Vb output from the image signal processing unit 110 is supplied to the display unit 111, and an image based on the image signal Vb is displayed on the screen of the display unit 111.

  The user can adjust the resolution, block distortion, mosquito noise, or noise suppression degree in the image displayed on the screen of the display unit 111 as described above by operating the remote control transmitter 200. In the image signal processing unit 110, pixel data of the image signal Vb is calculated by an estimation formula, as will be described later. As the coefficient data of this estimation formula, the data corresponding to the value of the parameter p changed by the user's operation of the remote control transmitter 200 is generated and used by the generation formula including this parameter p. Thereby, the resolution, block distortion, mosquito noise, or noise suppression degree in the image by the image signal Vb output from the image signal processing unit 110 corresponds to the changed value of the parameter p.

Next, details of the image signal processing unit 110 will be described.
The image signal processing unit 110 selectively extracts and outputs a plurality of pixel data located around the target position in the image signal Vb from the image signal Va stored in the buffer memory 108 and outputs the first and second taps. Selection circuits 121 and 122 are provided.

  The first tap selection circuit 121 selectively extracts a plurality of pixel data of prediction taps used for prediction. The second tap selection circuit 122 selectively extracts a plurality of pixel data of class taps used for class classification. The pixel data of the prediction tap and the class tap are pixel data located around the target position in the image signal Vb.

  Here, the periphery of the target position means a position that is spatially (horizontal and vertical) and temporal (frame direction) close to the target position.

  FIG. 2A shows an example of the arrangement of prediction taps, and FIG. 2B shows an example of the arrangement of class taps. 2A and 2B, “◯” indicates the position of the tap, and “X” indicates the position of interest. In the present embodiment, seven pixel data in the current field (shown with a solid line) and the previous field (shown with a broken line) are extracted as pixel data of a prediction tap, and the current field (shown with a solid line) and the previous field (shown with a broken line). 13 pixel data are extracted as pixel data of the class tap.

  Returning to FIG. 1, the image signal processing unit 110 also includes a feature amount extraction unit 123 that extracts a feature amount from the pixel data of the class tap selectively extracted by the second tap selection circuit 122. The feature amount extraction unit 123 generates an ADRC code as first class information indicating a space class by performing processing such as 1-bit ADRC (Adaptive Dynamic Range Coding) on the pixel data of the class tap.

  ADRC calculates the maximum value and minimum value of a plurality of pixel data of a class tap, calculates a dynamic range that is the difference between the maximum value and the minimum value, and requantizes each pixel value in accordance with the dynamic range. . In the case of 1-bit ADRC, the pixel value is converted to 1 bit depending on whether it is larger or smaller than the average value of the plurality of pixel values of the class tap. The ADRC process is a process for making the number of classes representing the level distribution of pixel values relatively small. Therefore, not only ADRC but encoding which compresses the bit number of pixel values, such as VQ (vector quantization), may be used.

  The image signal processing unit 110 includes an additional information class generation unit 124. The additional information class generation unit 124 generates second class information indicating the additional information class based on the additional information used when the MPEG2 stream is decoded. In the present embodiment, this additional information is encoding control information (picture information PI, motion vector information MI) extracted from the MPEG2 stream by the MPEG2 decoder 107.

  Further, the image signal processing unit 110 performs class classification based on the first class information (ADRC code) generated by the feature amount extraction unit 123 and the second class information generated by the additional information class generation unit 124. A class code generation unit 125 that generates a class code CL indicating the result is included.

  The image signal processing unit 110 has a coefficient memory 126. The coefficient memory 126 stores, for each class, a plurality of coefficient data Wi (i = 1 to n) used in an estimation formula used in an estimated prediction calculation circuit 127 described later. The coefficient data Wi is information for converting the image signal Va into the image signal Vb. The class code CL output from the class code generation unit 125 described above is supplied to the coefficient memory 126 as read address information, and the coefficient data Wi of the estimation formula corresponding to the class code CL is read from the coefficient memory 126. , And supplied to the estimated prediction calculation circuit 127.

  The image signal processing unit 110 has an information memory bank 128. In the estimated prediction calculation circuit 127, the pixel data y of the target position in the image signal Vb to be generated is calculated from the pixel data xi of the prediction tap and the coefficient data Wi read from the coefficient memory 126 by the estimation formula (1). Calculated. In the equation (1), n represents the number of prediction taps selected by the first tap selection circuit 121.

The coefficient data Wi of this estimation formula is generated by a generation formula including the parameter p as shown in the formula (2). Information in the memory bank 128, the coefficient seed data w _i0 to w _i3 is the coefficient data in the production equation is, for each class, stored.
_{Wi = w i0 + w i1 p} + w i2 p 2 + w i3 p 3 ··· (2)

Further, the image signal processing unit 110 uses the coefficient seed data w _{i0 to} w _{i3 of} each class and the value of the parameter p, and the coefficient of the estimation formula corresponding to the value of the parameter p for each class according to the equation (2). A coefficient generation circuit 129 that generates data Wi (i = 1 to n) is included. The coefficient generation circuit 129 is loaded with the above-described class seed coefficient data w _{i0 to} w _i3 from the information memory bank 128. The coefficient generation circuit 129 is supplied with the value of the parameter p from the system controller 101.

  The coefficient data Wi (i = 1 to n) of each class generated by the coefficient generation circuit 129 is stored in the coefficient memory 126 described above. The generation of coefficient data Wi for each class in the coefficient generation circuit 129 is performed, for example, in each vertical blanking period. Thereby, even if the value of the parameter p is changed by the user's operation of the remote control transmitter 200, the coefficient data Wi of each class stored in the coefficient memory 126 is immediately changed to the one corresponding to the value of the parameter p. The user can smoothly adjust the resolution, block distortion, mosquito noise, noise suppression degree, and the like.

Here, the coefficient seed data w _{i0 to} w _i3 stored in the information memory bank 128 as described above will be further described. The coefficient seed data w _{i0 to} w _i3 are signals obtained by processing the MPEG2 stream encoded in the same manner as the MPEG2 stream input to the MPEG2 decoder 107 described above according to the value of the parameter p. Is generated in advance.

That is, the coefficient seed data w _{i0 to} w _i3 are respectively generated according to a plurality of values of the parameter p, and using a plurality of student signals corresponding to the image signal Va and a teacher signal corresponding to the image signal Vb. Pre-generated. In this case, the student signal is obtained by decoding the MPEG2 stream obtained by MPEG2 encoding the teacher signal, and the signal is changed according to the value of the parameter p in the process of decoding, so that the parameter A student signal corresponding to the value of p is obtained. In the present embodiment, at least one of the DCT coefficient (frequency coefficient) or the inverse transform base used when performing the inverse discrete cosine transform (inverse DCT) is changed according to the value of the parameter p.

For example, it is possible to obtain a plurality of student signals by changing the gain according to the value of the parameter p while setting the gain of all AC coefficients of the DCT coefficients to the same value. In the information memory bank 128, when this way the coefficient seed data w _i0 to w _i3 generated in advance using a plurality of student signal obtained is stored is obtained by the image signal Vb above image Can be adjusted stepwise according to the value of the parameter p.

Further, for example, a plurality of student signals can be obtained by changing the quantization step of the DC coefficient among the DCT coefficients according to the value of the parameter p. In the information memory bank 128, when this way the coefficient seed data w _i0 to w _i3 generated in advance using a plurality of student signal obtained is stored is obtained by the image signal Vb above image The block distortion can be adjusted stepwise according to the value of the parameter p.

In addition, for example, a plurality of student signals can be obtained by reducing the gain of the AC coefficient of the DCT coefficient as the frequency increases and changing the degree of decrease in the gain of the AC coefficient according to the parameter value. Can do. In the information memory bank 128, when this way the coefficient seed data w _i0 to w _i3 generated in advance using a plurality of student signal obtained is stored is obtained by the image signal Vb above image The mosquito noise can be adjusted stepwise according to the value of the parameter p.

Also, for example, a plurality of student signals can be obtained by adding random noise to at least the DCT coefficient or the inverse transform base and changing the level of the random noise according to the value of the parameter p. In the information memory bank 128, when this way the coefficient seed data w _i0 to w _i3 generated in advance using a plurality of student signal obtained is stored is obtained by the image signal Vb above image The noise suppression degree can be adjusted stepwise according to the value of the parameter p.

  In addition, the image signal processing unit 110 uses the estimation formula (1) from the prediction tap pixel data xi selectively extracted by the first tap selection circuit 121 and the coefficient data Wi read from the coefficient memory 126. And an estimated prediction calculation circuit 127 for calculating pixel data y of the target position in the image signal Vb to be created.

Next, the operation of the image signal processing unit 110 will be described.
From the image signal Va stored in the buffer memory 108, the second tap selection circuit 122 selectively extracts pixel data of class taps located around the target position in the image signal Vb to be created. The pixel data of this class tap is supplied to the feature amount extraction unit 123. In the feature amount extraction unit 123, processing such as 1-bit ADRC is performed on the pixel data of the class tap, and an ADRC code as first class information indicating a space class is generated.

  Encoding control information (picture information PI, motion vector information MI) extracted from the MPEG2 stream by the MPEG2 decoder 107 is supplied to the additional information class generation unit 124. The additional information class generation unit 124 generates second class information indicating the additional information class based on the encoding control information.

  Then, the first class information (ADRC code) generated by the feature amount extraction unit 123 and the second class information generated by the additional information class generation unit 124 are supplied to the class code generation unit 125. The class code generation unit 125 generates a class code CL indicating the result of class classification based on the first class information and the second class information.

The class code CL represents the detection result of the class to which the pixel data at the target position in the image signal Vb belongs. The class code CL is supplied to the coefficient memory 126 as read address information. In the coefficient memory 126, for example, in each vertical blanking period, the coefficient generation circuit 129 estimates for each class using the coefficient seed data w _{i0 to} w _i3 corresponding to the value of the parameter p adjusted by the user. Coefficient data Wi (i = 1 to n) of the equation is obtained and stored (see equation (2)).

  As described above, the class code CL is supplied to the coefficient memory 126 as read address information, whereby the coefficient data Wi corresponding to the class code CL is read from the coefficient memory 126 and supplied to the estimated prediction calculation circuit 127. The Further, the pixel data of the prediction tap located around the target position in the image signal Vb is selectively extracted by the first tap selection circuit 121 from the image signal Va stored in the buffer memory 108.

  The estimated prediction calculation circuit 127 uses the pixel data xi of the prediction tap and the coefficient data Wi read from the coefficient memory 126, and based on the estimation formula (see formula (1)), attention is paid to the image signal Vb to be generated. Pixel data y at the position is obtained.

  In this manner, the image signal processing unit 110 calculates the pixel data y using the coefficient data Wi (i = 1 to n) of the estimation formula corresponding to the value of the parameter p. Therefore, the user can arbitrarily adjust the resolution, block distortion, mosquito noise, or noise suppression degree of the image by the image signal Vb by changing the value of the parameter p.

The coefficient seed data w _{i0 to} w _i3 for generating the coefficient data Wi is an MPEG2 stream encoded in the same manner as the MPEG2 stream input to the MPEG2 decoder 107 described above, according to the value of the parameter p. Are generated in advance based on the signal obtained by the processing, and only the specific quality of the image by the image signal Vb, for example, only the resolution, only the block distortion, only the mosquito noise, or only the noise suppression degree is stepwise. Can be adjusted.

Next, an example of a method for generating the coefficient seed data w will be described. In this example, an example in which coefficient seed data w _{i0 to} w _i3 which are coefficient data in the generation formula of the above-described formula (2) is obtained is shown.

Here, for the following explanation, t _j (j = 0 to 3) is defined as shown in equation (3).
t ₀ = 1, t ₁ = p, t ₂ = p ² , t ₃ = p ³ (3)
Using this equation (3), equation (2) can be rewritten as equation (4).

Finally, the undetermined coefficient w _ij is obtained by learning. That is, for each class, the coefficient value that minimizes the square error is determined using the pixel data of the student signal corresponding to the image signal Va and the pixel data of the teacher signal corresponding to the image signal Vb. This is a so-called least square method. If the learning number is m, the residual in the kth (1 ≦ k ≦ m) learning data is e _k , and the sum of the squared errors is E, using Eqs. (1) and (2), E is (5 ) Expression. Here, x _ik represents k-th pixel data at the i-th predicted tap position of the student signal, and y _k represents pixel data of the corresponding k-th teacher signal.

The solution according to the minimum square method, determine the w _ij such that 0 is partial differentiation by (5) of w _ij. This is shown by equation (6).

Hereinafter, when X _ipjq and Y _ip are defined as in equations (7) and (8), equation (6) can be rewritten using a matrix as in equation (9).

This equation is generally called a normal equation. This normal equation is solved for w _ij by using a sweep-out method (Gauss-Jordan elimination method) or the like, and coefficient seed data is calculated.

  FIG. 3 shows an example of the concept of the above-described coefficient seed data generation method. A plurality of student signals corresponding to the image signal Va are generated from the teacher signal corresponding to the image signal Vb. For example, the parameter p is varied in nine steps to generate nine types of student signals. The coefficient seed data is generated by performing learning between the plurality of student signals and the teacher signal generated as described above.

FIG. 4 shows the configuration of the coefficient seed data generation device 150 for generating the coefficient seed data w _{i0 to} w _i3 stored in the information memory bank 128 of the digital broadcast receiver 100 described above.

  The coefficient seed data generation device 150 includes an input terminal 151 to which a teacher signal ST corresponding to the image signal Vb is input, an MPEG2 encoder 152 that encodes the teacher signal to obtain an MPEG2 stream, An MPEG2 decoder 153 that decodes the MPEG2 stream to obtain the student signal SS. The value of the parameter p is input to the MPEG2 decoder 153.

  In the decoder 153, in the decoding process, the signal is changed according to the value of the parameter p, and a student signal corresponding to the value of the parameter p is obtained. In this embodiment, at least one of the DCT coefficient (frequency coefficient) or the inverse transform base used when performing inverse discrete cosine transform (inverse DCT) is changed.

  Here, in DCT, a DCT coefficient (frequency coefficient) F (u, v) for a pixel value f (i, j) is given by equation (10). On the other hand, in inverse DCT, a pixel value f (i, j) with respect to a DCT coefficient (frequency coefficient) F (u, v) is given by equation (11). Here, C (w) = 1 / √2 (w = 0) and C (w) = 1 (w ≠ 0). Thus, in inverse DCT, the pixel value f (i, j) is obtained by a product-sum operation of the DCT coefficient (frequency coefficient) and the inverse transform base.

For example, it is possible to obtain a plurality of student signals by changing the gain according to the value of the parameter p while setting the gain of all AC coefficients of the DCT coefficients to the same value. Here, if the original AC coefficient is AC [m] and the gain is gain, the new AC coefficient newAC [m] is newAC [m] = AC [m] * gain. In this case, since the correlation between the AC coefficients can be maintained, the coefficient seed data w _{i0 to} w _i3 generated using the plurality of student signals obtained in this way is obtained by the image signal Vb described above. The resolution of the image can be adjusted stepwise according to the value of the parameter p.

Further, for example, a plurality of student signals can be obtained by changing the quantization step of the DC coefficient among the DCT coefficients according to the value of the parameter p. Here, assuming that the original DC coefficient is DC [m], the original quantization step is Qst, and the new quantization step is newQst, the new DC coefficient newDC [m] is newDC [m] = {DC [ m] * Qst / newQst}. {} Represents an integer operation. In this case, it is possible to manipulate the variation of the DC coefficient, and the coefficient seed data w _{i0 to} w _i3 generated using the plurality of student signals obtained in this way is the block distortion of the image obtained by the image signal Vb described above. Can be adjusted stepwise according to the value of the parameter p.

Further, for example, by reducing the gain of the AC coefficient of the DCT coefficient as the high frequency region is changed, and the degree of decrease in the high frequency gain of the AC coefficient is changed according to the value of the parameter p, a plurality of student signals are obtained. be able to. Here, if the original AC coefficient is AC [m] and the gain is gain [m], the new AC coefficient newAC [m] is newAC [m] = AC [m] * gain [m]. In this case, the correlation between the AC coefficients is broken stepwise, and the coefficient seed data w _{i0 to} w _i3 generated using the plurality of student signals obtained in this way are the above-described image signals. The mosquito noise of the image obtained by Vb can be adjusted stepwise according to the value of the parameter p.

Also, for example, a plurality of student signals can be obtained by adding random noise to at least the DCT coefficient or the inverse transform base and changing the level of the random noise according to the value of the parameter p. When the original AC coefficient is AC [m] and the random noise is rdmN, the new AC coefficient newAC [m] is newAC [m] = AC [m] + rdmN. Similarly, if the original DC coefficient is DC [m] and the random noise is rdmN, the new DC coefficient newDC [m] is newDC [m] = DC [m] + rdmN. Similarly, when the original base is REF [m] and the random noise is rdmN, the new base newREF [m] is newREF = REF [m] + rdmN. In this case, coefficient level data w _{i0 to} w _i3 generated using a plurality of student signals obtained by switching the level of random noise step by step is the noise of the image obtained by the image signal Vb described above. The degree of suppression can be adjusted stepwise according to the value of the parameter p.

FIG. 6 shows the configuration of the MPEG2 decoder 153.
The decoder 153 has an input terminal 181 to which an MPEG2 stream is input, and a stream buffer 182 that temporarily stores the MPEG2 stream input to the input terminal 181.

  The decoder 153 extracts the DCT (Discrete Cosine Transform) coefficient as a frequency coefficient from the MPEG2 stream stored in the stream buffer 182 and the extraction circuit 183 extracts the DCT coefficient. A variable length decoding circuit 184 that performs variable length decoding on DCT coefficients that have been subjected to variable length coding, for example, Huffman coding.

  The decoder 153 also extracts an extraction circuit 185 for extracting quantization characteristic designation information from the MPEG2 stream stored in the stream buffer 182, and the quantization characteristic designation information extracted by the extraction circuit 185. An inverse quantization circuit 186 that performs inverse quantization on the quantized DCT coefficient output from the variable length decoding circuit 184, and an inverse DCT circuit that performs inverse DCT on the DCT coefficient output from the inverse quantization circuit 186 187.

  Further, the decoder 153 stores the image signals of the I picture (Intra-Picture) and the P picture (Predictive-Picture) in a memory (not shown) and uses the image signals from the inverse DCT circuit 187. When an image signal of a P picture or a B picture (Bidirectionally predictive-Picture) is output, a prediction memory circuit 188 that generates and outputs a corresponding reference image signal Vref is provided.

  In addition, when the inverse DCT circuit 187 outputs a P picture or B picture image signal, the decoder 153 adds an adder circuit 189 that adds the reference image signal Vref generated by the prediction memory circuit 188 to the image signal. Have. When the image signal of the I picture is output from the inverse DCT circuit 187, the reference image signal Vref is not supplied from the prediction memory circuit 188 to the adder circuit 189, and accordingly, the adder circuit 189 outputs from the inverse DCT circuit 187. The image signal of the I picture is output as it is.

  In addition, the decoder 153 supplies the I-picture and P-picture image signals output from the adder circuit 189 to the prediction memory circuit 188 and stores them in the memory, and at the same time, the image of each picture output from the adder circuit 189. It has a picture selection circuit 190 that rearranges and outputs signals in the correct order, and an output terminal 191 that outputs an image signal output from the picture selection circuit 190.

  The decoder 153 also extracts an extraction circuit 192 that extracts encoding control information, that is, picture information PI and motion vector information MI, from the MPEG2 stream stored in the stream buffer 182, and an output that outputs the encoding control information. Terminal 193. The motion vector information MI extracted by the extraction circuit 192 is supplied to the prediction memory circuit 188, and the prediction memory circuit 188 performs motion compensation when generating the reference image signal Vref using the motion vector information MI. The picture information PI extracted by the extraction circuit 192 is supplied to the prediction memory circuit 188 and the picture selection circuit 190, and the prediction memory circuit 188 and the picture selection circuit 190 identify pictures based on the picture information PI. .

The operation of the MPEG2 decoder 153 shown in FIG. 6 will be described.
The MPEG2 stream stored in the stream buffer 182 is supplied to the extraction circuit 183, and DCT coefficients as frequency coefficients are extracted. This DCT coefficient is variable-length encoded, and this DCT coefficient is supplied to the variable-length decoding circuit 184 and decoded. Then, the quantized DCT coefficient output from the variable length decoding circuit 184 is supplied to the inverse quantization circuit 186 and subjected to inverse quantization.

  The DCT coefficients output from the inverse quantization circuit 186 are subjected to inverse DCT by the inverse DCT circuit 187, and an image signal of each picture is obtained. The picture signal of each picture is supplied to the picture selection circuit 190 via the addition circuit 189. In this case, the reference image signal Vref output from the prediction memory circuit 188 is added by the addition circuit 189 to the image signals of the P picture and the B picture. The picture signals of each picture are rearranged in the correct order by the picture selection circuit 190 and output to the output terminal 191.

  Returning to FIG. 4, the coefficient seed data generation apparatus 150 selectively extracts a plurality of pixel data located around the target position in the teacher signal ST from the student signal SS output from the MPEG2 decoder 153. A first tap selection circuit 154 and a second tap selection circuit 155 that output the first tap. The first and second tap selection circuits 154 and 155 are configured similarly to the first and second tap selection circuits 121 and 122 of the image signal processing unit 110 described above, respectively.

  In addition, the coefficient seed data generation device 150 includes a feature amount extraction unit 156 that extracts a feature amount from pixel data of class taps that are selectively extracted by the second tap selection circuit 155. The feature amount extraction unit 156 is configured in the same manner as the feature amount extraction unit 123 of the image signal processing unit 110 described above, and generates an ADRC code as first class information indicating a space class.

  Also, the coefficient seed data generation device 150 generates second class information indicating the additional information class based on the encoded information (picture information PI, motion vector information MI) extracted from the MPEG2 stream by the MPEG2 decoder 153. An additional information class generation unit 157 is generated. The additional information class generation unit 157 is configured in the same manner as the additional information class generation unit 124 of the image signal processing unit 110 described above.

  In addition, the coefficient seed data generation device 150 performs class classification based on the first class information (ADRC code) generated by the feature amount extraction unit 156 and the second class information generated by the additional information class generation unit 157. A class code generation unit 158 that generates a class code CL indicating the result of the above. The class code generation unit 158 is configured in the same manner as the class code generation unit 125 of the image signal processing unit 110 described above. The class code CL indicates a class to which the pixel data y at the target position in the teacher signal ST belongs.

The coefficient seed data generation device 150 also includes a delay circuit 159 for adjusting the time of the teacher signal ST supplied to the input terminal 151, and each attention position obtained from the teacher signal ST adjusted in time by the delay circuit 159. Pixel data y, prediction tap pixel data xi selectively extracted by the first tap selection circuit 154 in correspondence with the pixel data y of each target position, the value of the parameter p, and each target position A normal equation (see equation (9)) for obtaining coefficient seed data w _{i0 to} w _i3 is generated for each class from the class code CL generated by the class code generation unit 158 corresponding to each pixel data y. A normal equation generating unit 160 for performing

In this case, one piece of learning data is generated by a combination of one piece of pixel data y and pixel data xi of n prediction taps corresponding to the piece of pixel data y. A plurality of student signals SS are sequentially generated, and learning data is generated between the teacher signal and each student signal. As a result, the normal equation generation unit 160 generates a normal equation in which many pieces of learning data having different values of the parameter p are registered, and the coefficient seed data w _{i0 to} w _i3 can be obtained.

In addition, the coefficient seed data generation device 150 is supplied with the data of the normal equation generated for each class by the normal equation generation unit 160, solves the normal equation, and generates coefficient seed data w _{i0 to} w _{i3 for} each class. Is provided with a coefficient seed data determining unit 161 and a coefficient seed memory 162 for storing the obtained coefficient seed data. In the coefficient seed data determination unit 161, the normal equation is solved, for example, by a sweeping method, and coefficient coefficient data is obtained.

The operation of the coefficient seed data generation device 150 shown in FIG. 4 will be described.
A teacher signal ST corresponding to the image signal Vb is supplied to the input terminal 151, and the teacher signal ST is encoded by the MPEG2 encoder 152 to generate an MPEG2 stream. This MPEG2 stream is supplied to the MPEG2 decoder 153.

  The value of the parameter p is input to the MPEG2 decoder 153 so that the signal changes in accordance with the value of the parameter p in the decoding process. That is, in the present embodiment, as described above, at least one of the DCT coefficient (frequency coefficient) or the inverse transform base used when performing the inverse discrete cosine transform (inverse DCT) according to the value of the parameter p. Is changed. Thereby, the student signal SS corresponding to the value of the parameter p is obtained.

  Further, from the student signal SS output from the MPEG2 decoder 153, the second tap selection circuit 155 selectively extracts pixel data of class taps located around the target position in the teacher signal ST. The pixel data of this class tap is supplied to the feature amount extraction unit 156. The feature amount extraction unit 156 generates an ADRC code as first class information indicating a space class based on the pixel data of the class tap.

  Also, encoded information (picture information PI, motion vector information MI) extracted from the MPEG2 stream by the MPEG2 decoder 153 is supplied to the additional information class generation unit 157. The additional information class generation unit 157 generates second class information indicating the additional information class based on the encoded information.

  Then, the first class information (ADRC code) generated by the feature amount extraction unit 156 and the second class information generated by the additional information class generation unit 157 are supplied to the class code generation unit 158, respectively. In this class code generation unit 158, based on the first class information and the second class information, a class code CL indicating the class (class classification result) to which the pixel data y at the target position in the teacher signal ST belongs is generated. The

  Also, from the student signal SS output from the MPEG2 decoder 153, the first tap selection circuit 154 selectively extracts pixel data of the prediction tap located around the target position in the teacher signal ST.

Then, the pixel data y of each target position obtained from the teacher signal ST time-adjusted by the delay circuit 159 and the first tap selection circuit 154 selectively corresponding to the pixel data y of each target position. The normal equation generation unit 160 uses the predicted tap pixel data xi, the value of the parameter p, and the class code CL generated by the class code generation unit 158 corresponding to the pixel data y of each target position. For each class, a normal equation (see equation (9)) for generating coefficient seed data w _{i0 to} w _i3 is generated.

Then, each normal equation is solved by the coefficient seed data decision section 161, coefficient seed data w _i0 to w _i3 of each class is determined, their coefficient seed data w _i0 to w _i3 is stored in the coefficient seed memory 162 .

As described above, the coefficient seed data generation device 150 shown in FIG. 4 uses the estimation formula (see formula (1)) for each class stored in the information memory bank 128 of the image signal processing unit 110 in FIG. It is possible to generate coefficient seed data w _{i0 to} w _i3 which are coefficient data in a generation formula (see formula (2)) for determining the coefficient data Wi to be obtained.

Next, another example of the coefficient seed data generation method will be described. Also in this example, an example in which coefficient seed data w _{i0 to} w _i3 which are coefficient data in the generation formula of the above-described formula (2) is obtained is shown.

FIG. 7 shows the concept of this example. A plurality of student signals are generated from the teacher signal. For example, the value of the parameter p is varied in 9 steps to generate 9 types of student signals. Learning is performed between each student signal and the teacher signal generated in this way, and coefficient data Wi of the estimation formula (1) is generated. Then, using the coefficient data Wi generated corresponding to each student signal, coefficient seed data w _{i0 to} w _i3 , which is coefficient data in the generation formula of formula (2), is generated.

  First, how to obtain coefficient data of the estimation formula will be described. Here, an example is shown in which the coefficient data Wi (i = 1 to n) of the estimation formula (1) is obtained by the least square method.

In the above equation (1), before learning, coefficient data W ₁ , W ₂ ,..., W _n are undetermined coefficients. Learning is performed on a plurality of signal data for each class. When the number of learning data is m, the following equation (12) is set according to equation (1). n indicates the number of prediction taps.
y _k = W ₁ × x _k1 + W ₂ × x _k2 +... + W _n × x _kn (12)
(K = 1, 2,..., M)
When m> n, the coefficient data W ₁ , W ₂ ,..., W _n are not uniquely determined, so the element e _k of the error vector e is defined by the following equation (13), and (14) Coefficient data that minimizes e ² in the equation is obtained. Coefficient data is uniquely determined by a so-called least square method.
e _k = y _k − {w ₁ × x _k1 + w ₂ × x _k2 +... + w _n × x _kn } (13)
(K = 1, 2, ... m)

As a practical calculation method for obtaining coefficient data that minimizes e ² in equation (14), first, as shown in equation (15), e ² is expressed as coefficient data Wi (i = 1, 2,. .., N), and the coefficient data Wi may be obtained so that the partial differential value becomes 0 for each value of i.

  A specific procedure for obtaining the coefficient data Wi from the equation (15) will be described. If Xji and Yi are defined as in equations (16) and (17), equation (15) can be written in the form of a determinant of equation (18).

  Equation (18) is generally called a normal equation. Coefficient data Wi (i = 1, 2,..., N) can be obtained by solving this normal equation by a general solution method such as a sweep-out method (Gauss-Jordan elimination method).

  Next, how to obtain coefficient seed data using coefficient data Wi generated corresponding to each student signal will be described.

It is assumed that coefficient data of a certain class obtained by learning using a student signal corresponding to the value of the parameter p becomes k _pi . Here, i is the number of the prediction tap. The coefficient seed data of this class is obtained from this k _pi .

The coefficient data Wi (i = 1 to n) is expressed by the above-described equation (2) using the coefficient seed data w _{i0 to} w _i3 . Here, considering that the least square method is used for the coefficient data Wi, the residual is expressed by equation (19).

Here, t _j is shown in the above equation (3). When the least square method is applied to equation (19), equation (20) is obtained.

Here, if X _jk and Y _j are defined as in the equations (21) and (22), respectively, the equation (20) is rewritten as in the equation (23). The equation (23) is also a normal equation, and the coefficient seed data w _{i0 to} w _i3 can be calculated by solving this equation by a general solution such as a sweep-out method.

  FIG. 8 shows a configuration of a coefficient seed data generation device 150A that generates coefficient seed data based on the concept shown in FIG. In FIG. 8, parts corresponding to those in FIG. 4 are denoted by the same reference numerals, and detailed description thereof is omitted.

  The coefficient seed data generation device 150A includes pixel data y at each position of interest obtained from the teacher signal ST time-adjusted by the delay circuit 159, and a first tap selection circuit corresponding to the pixel data y at each position of interest. From the pixel data xi of the prediction tap selectively extracted in 154 and the class code CL generated by the class code generation unit 158 corresponding to the pixel data y of each target position, coefficient data Wi ( A normal equation generation unit 163 that generates a normal equation (see equation (18)) for obtaining i = 1 to n) is provided.

  In this case, one piece of learning data is generated by a combination of one piece of pixel data y and pixel data xi of n prediction taps corresponding to the piece of pixel data y. A plurality of student signals SS are sequentially generated, and learning data is generated between the teacher signal and each student signal. As a result, the normal equation generation unit 163 generates a normal equation for obtaining coefficient data Wi (i = 1 to n) for each class corresponding to each student signal.

The coefficient seed data generation device 150A is supplied with the data of the normal equation generated by the normal equation generation unit 163, solves the normal equation, and obtains the coefficient data Wi of each class corresponding to each student signal. A normal equation ((23)) for obtaining coefficient seed data w _{i0 to} w _i3 for each class using the coefficient data determination unit 164 and coefficient data Wi corresponding to the value of the parameter p and each student signal. And a normal equation generation unit 165 that generates (see Equation).

The coefficient seed data generation device 150A is supplied with the data of the normal equation generated for each class by the normal equation generation unit 165, solves the normal equation for each class, and generates coefficient seed data w _{i0 to} w _{i3 for} each class. A coefficient seed data determination unit 166 and a coefficient seed memory 162 for storing the determined coefficient seed data w _{i0 to} w _i3 .

  The rest of the coefficient seed data generation device 150A shown in FIG. 8 is configured in the same manner as the coefficient seed data generation device 150 shown in FIG.

Next, the operation of the coefficient seed data generation device 150A shown in FIG. 8 will be described.
A teacher signal ST corresponding to the image signal Vb is supplied to the input terminal 151, and the teacher signal ST is encoded by the MPEG2 encoder 152 to generate an MPEG2 stream. This MPEG2 stream is supplied to the MPEG2 decoder 153.

  The value of the parameter p is input to the MPEG2 decoder 153 so that the signal changes in accordance with the value of the parameter p in the decoding process. That is, in the present embodiment, as described above, at least one of the DCT coefficient (frequency coefficient) or the inverse transform base used when performing the inverse discrete cosine transform (inverse DCT) according to the value of the parameter p. Is changed. Thereby, the student signal SS corresponding to the value of the parameter p is obtained.

  Further, from the student signal SS output from the MPEG2 decoder 153, the second tap selection circuit 155 selectively extracts pixel data of class taps located around the target position in the teacher signal ST. The pixel data of this class tap is supplied to the feature amount extraction unit 156. The feature amount extraction unit 156 generates an ADRC code as first class information indicating a space class based on the pixel data of the class tap.

  Also, encoded information (picture information PI, motion vector information MI) extracted from the MPEG2 stream by the MPEG2 decoder 153 is supplied to the additional information class generation unit 157. The additional information class generation unit 157 generates second class information indicating the additional information class based on the encoded information.

  Then, the first class information (ADRC code) generated by the feature amount extraction unit 156 and the second class information generated by the additional information class generation unit 157 are supplied to the class code generation unit 158, respectively. In this class code generation unit 158, based on the first class information and the second class information, a class code CL indicating the class (class classification result) to which the pixel data y at the target position in the teacher signal ST belongs is generated. The

  Also, from the student signal SS output from the MPEG2 decoder 153, the first tap selection circuit 154 selectively extracts pixel data of the prediction tap located around the target position in the teacher signal ST.

  Then, the pixel data y of each target position obtained from the teacher signal ST time-adjusted by the delay circuit 159 and the first tap selection circuit 154 selectively corresponding to the pixel data y of each target position. The normal equation generation unit 163 uses the pixel data xi of the predicted tap, the value of the parameter p, and the class code CL generated by the class code generation unit 158 corresponding to the pixel data y of each target position. , A normal equation (((n = 1 to n)) for obtaining coefficient data Wi (i = 1 to n) for each class for a plurality of student signals SS respectively corresponding to a plurality of values of the parameter p output from the MPEG2 decoder 153. 18) is generated.

Then, the coefficient data determining unit 164 solves the normal equation, and the coefficient data Wi of each class corresponding to each of the plurality of student signals SS is obtained. The normal equation generation unit 165 uses the coefficient data Wi of each class corresponding to each of the plurality of student signals SS to obtain a normal equation (equation (23)) for obtaining coefficient seed data w _{i0 to} w _i3 for each class. Reference) is generated.

Then, the normal equation is solved by the coefficient seed data determination section 166, the coefficient seed data w _i0 to w _i3 of each class is determined, the coefficient seed data w _i0 to w _i3 is stored in the coefficient seed memory 162.

As described above, the coefficient seed data generation device 150A shown in FIG. 8 also generates the coefficient seed data w _{i0 to} w _i3 of each class stored in the information memory bank 128 of the image signal processing unit 110 in FIG. Can do.

  In the image signal processing unit 110 of FIG. 1, the generation formula of the formula (2) is used to generate the coefficient data Wi (i = 1 to n), but it is expressed by a polynomial having a different order or another function. It is also possible to use this formula.

  Further, in the image signal processing unit 110 of FIG. 1, in addition to the first class information (ADRC code) indicating the space class output from the feature amount extraction unit 123, the additional information generated by the additional information class generation unit 124. Based on the second class information indicating the class, the class code generation unit 125 generates the class code CL. However, the class code CL may be generated using only the first class information. However, by generating the class code CL in consideration of the second class information, finer classification can be performed, and the generation accuracy of the pixel data of the image signal Vb can be improved.

  Note that the processing in the image signal processing unit 110 in FIG. 1 can be realized by software, for example, by an image signal processing apparatus 300 as shown in FIG.

  First, the image signal processing apparatus 300 shown in FIG. 9 will be described. The image signal processing apparatus 300 includes a CPU 301 that controls the operation of the entire apparatus, a ROM (read only memory) 302 that stores an operation program of the CPU 301, coefficient seed data, and the like, and a RAM ( random access memory) 303. These CPU 301, ROM 302, and RAM 303 are each connected to a bus 304.

  The image signal processing apparatus 300 also includes a hard disk drive (HDD) 305 as an external storage device and a drive (FDD) 307 that drives a floppy (registered trademark) disk 306. These drives 305 and 307 are each connected to a bus 304.

  In addition, the image signal processing apparatus 300 includes a communication unit 308 that connects to a communication network 400 such as the Internet by wire or wirelessly. The communication unit 308 is connected to the bus 304 via the interface 309.

  In addition, the image signal processing device 300 includes a user interface unit. The user interface unit includes a remote control signal receiving circuit 310 that receives a remote control signal RM from the remote control transmitter 200, and a display 311 that includes an LCD (liquid crystal display) or the like. The receiving circuit 310 is connected to the bus 304 via the interface 312, and similarly the display 311 is connected to the bus 304 via the interface 313.

  Further, the image signal processing apparatus 300 has an input terminal 314 for inputting the image signal Va and an output terminal 315 for outputting the image signal Vb. The input terminal 314 is connected to the bus 304 via the interface 316, and similarly, the output terminal 315 is connected to the bus 304 via the interface 317.

  Here, instead of storing the processing program, coefficient seed data, and the like in the ROM 302 in advance as described above, for example, they are downloaded from the communication network 400 such as the Internet via the communication unit 308 and stored in the hard disk or RAM 303 for use. You can also Further, these processing programs, coefficient seed data, and the like may be provided on a floppy (registered trademark) disk 306.

  Further, instead of inputting the image signal Va to be processed from the input terminal 314, it may be recorded in advance on a hard disk or downloaded from the communication network 400 such as the Internet via the communication unit 308. Further, instead of outputting the processed image signal Vb to the output terminal 315 or in parallel therewith, it is supplied to the display 311 to display an image, further stored in a hard disk, the Internet via the communication unit 308, etc. It may be sent to the communication network 400.

  A processing procedure for obtaining the image signal Vb from the image signal Va in the image signal processing apparatus 300 shown in FIG. 9 will be described with reference to the flowchart of FIG.

  First, in step ST1, processing is started, and in step ST2, for example, an image signal Va for one frame or one field is input into the apparatus from the input terminal 314. Thus, the pixel data constituting the image signal Va input from the input terminal 314 is temporarily stored in the RAM 303. When the image signal Va is recorded in advance in the hard disk drive 307 in the apparatus, the image signal Va is read from the drive 307, and the pixel data constituting the image signal Va is temporarily stored in the RAM 303. To do.

  In step ST3, it is determined whether or not processing of all frames or all fields of the image signal Va has been completed. When the process is finished, the process ends in step ST4. On the other hand, when the process is not finished, the process proceeds to step ST5.

  In step ST5, the value of the parameter p, which is the image quality designation value input by the user's operation of the remote control transmitter 200, is acquired from the RAM 303, for example. In step ST6, using the value of the parameter P and the coefficient seed data of each class, the coefficient data Wi of the estimation formula (see formula (1)) of each class is generated by a generation formula (for example, formula (2)). To do.

  Next, in step ST7, pixel data of class taps and prediction taps are acquired from the image signal Va input in step ST2 corresponding to the target position in the image signal Vb. In step ST8, it is determined whether or not the processing for obtaining the pixel data of the image signal Vb has been completed in the entire region of the pixel data of the image signal Va for one frame or one field input in step ST2. If completed, the process returns to step ST2, and the process proceeds to input processing of the image signal Va for the next one frame or one field. On the other hand, when the process has not ended, the process proceeds to step ST9.

  In step ST9, a class code CL is generated from the pixel data of the class tap acquired in step ST7. In step ST10, using the coefficient data Wi corresponding to the class code CL and the pixel data of the prediction tap, the pixel data of the target position in the image signal Vb is generated by the estimation formula, and then the process returns to step ST7. Then, the same processing as described above is repeated.

  As described above, by performing processing according to the flowchart shown in FIG. 10, the pixel data of the input image signal Va can be processed to obtain the pixel data of the image signal Vb. As described above, the image signal Vb obtained by such processing is output to the output terminal 315, supplied to the display 311 to display an image, and further supplied to the hard disk drive 305 to be supplied to the hard disk. Or is recorded.

  In step ST9, the class code CL is generated from the pixel data of the class tap. However, in addition to the image signal Va, encoding control information (picture information PI, motion vector information MI) related to the image signal Va is also input from the input terminal 314, and the image signal processing unit 110 in FIG. As in, the class code CL may be generated using this encoding control information.

  Further, although illustration of the processing device is omitted, the processing in the coefficient seed data generation device 150 in FIG. 4 can also be realized by software.

A processing procedure for generating coefficient seed data will be described with reference to the flowchart of FIG.
First, in step ST21, processing is started, and in step ST22, an image quality pattern (specified by the value of parameter p) used for learning is selected. In step ST23, it is determined whether learning has been completed for all image quality patterns. If learning has not been completed for all image quality patterns, the process proceeds to step ST24.

  In this step ST24, the teacher signal is inputted for one frame or one field. In step ST25, it is determined whether or not the processing of all the frames or fields of the teacher signal has been completed. When the process is completed, the process returns to step ST22, the next image quality pattern is selected, and the same process as described above is repeated. On the other hand, when not completed, the process proceeds to step ST26.

  In step ST26, a student signal is generated based on the image quality pattern selected in step ST22 from the teacher signal input in step ST24. In step ST27, the pixel data of the class tap and the prediction tap are acquired from the student signal generated in step ST26 corresponding to the attention position in the teacher signal input in step ST24. In step ST28, it is determined whether or not the learning process has been completed in all regions of the teacher signal input in step ST24. When the learning process is completed, the process returns to step ST24, the teacher signal for the next one frame or one field is input, and the same process as described above is repeated. On the other hand, when the process has not ended, the process proceeds to step ST29.

  In step ST29, a class code CL is generated from the pixel data of the class tap acquired in step ST27. In step ST30, a normal equation (see equation (9)) is generated. Thereafter, the process returns to step ST27.

  If learning for all image quality patterns is completed in step ST23, the process proceeds to step ST31. In this step ST31, the coefficient seed data of each class is calculated by solving the normal equation by a sweeping method or the like. In step ST32, the coefficient seed data is stored in the memory, and then the process is terminated in step ST33. .

  In this way, by performing the processing according to the flowchart shown in FIG. 11, coefficient seed data can be obtained by the same method as the coefficient seed data generating apparatus 150 shown in FIG.

  In step ST29, the class code CL is generated from the pixel data of the class tap. However, similar to the coefficient seed data generation device 150 of FIG. 4, the class code CL may be generated using also the encoding control information (picture information PI, motion vector information MI) related to the student signal.

  Although illustration of the processing device is omitted, the processing in the coefficient seed data generation device 150A in FIG. 8 can also be realized by software.

A processing procedure for generating coefficient seed data will be described with reference to the flowchart of FIG.
First, in step ST41, processing is started, and in step ST42, an image quality pattern (specified by the value of parameter p) used for learning is selected. In step ST43, it is determined whether or not learning has been completed for all image quality patterns. If learning has not been completed for all image quality patterns, the process proceeds to step ST44.

  In this step ST44, the teacher signal is inputted for one frame or one field. In step ST45, it is determined whether or not the processing of all the frames or fields of the teacher signal has been completed. If not completed, a student signal is generated in step ST46 based on the image quality pattern selected in step ST42 from the teacher signal input in step ST44.

  In step ST47, the pixel data of the class tap and the prediction tap are acquired from the student signal generated in step ST46 corresponding to the attention position in the teacher signal input in step ST44. In step ST48, it is determined whether or not the learning process has been completed in all regions of the teacher signal input in step ST44. When the learning process is finished, the process returns to step ST44, the teacher signal for the next one frame or one field is input, and the same process as described above is repeated. On the other hand, when the learning process is not ended, the process proceeds to step ST49.

  In step ST49, a class code CL is generated from the pixel data of the class tap acquired in step ST47. In step ST50, a normal equation for obtaining coefficient data (see equation (18)) is generated. Thereafter, the process returns to step ST47.

  When the processing is completed in step ST45 described above, in step ST51, the normal equation generated in step ST50 is solved by a sweeping method or the like to calculate coefficient data for each class. Thereafter, the process returns to step ST42, the next image quality pattern is selected, the same processing as described above is repeated, and the coefficient data of each class corresponding to the next image quality pattern is obtained.

  If the coefficient data calculation process for all image quality patterns is completed in step ST43, the process proceeds to step ST52. In step ST52, a normal equation (see equation (23)) for obtaining coefficient seed data is generated from the coefficient data for all image quality patterns.

  Then, in step ST53, the coefficient equation data of each class is calculated by solving the normal equation generated in step ST52 by the sweep method or the like, and in step ST54, the coefficient species data is stored in the memory, and then the step In ST55, the process ends.

  In this way, by performing processing according to the flowchart shown in FIG. 12, coefficient seed data can be obtained by the same method as the coefficient seed data generating apparatus 150A shown in FIG.

  In step ST49, the class code CL is generated from the pixel data of the class tap. However, similar to the coefficient seed data generation device 150A of FIG. 8, the class code CL may be generated using also the encoding control information (picture information PI, motion vector information MI) related to the student signal.

  Next, another embodiment of the present invention will be described.

  FIG. 13 shows a configuration of a digital broadcast receiver 100A as another embodiment. In FIG. 13, portions corresponding to those in FIG.

  The digital broadcast receiver 100A is obtained by replacing the image signal processing unit 110 of the digital broadcast receiver 100 shown in FIG. 1 with the image signal processing unit 110A, and operates in the same manner as the digital broadcast receiver 100.

  Details of the image signal processing unit 110A will be described. In this image signal processing unit 110A, parts corresponding to those of the image signal processing unit 110 shown in FIG.

  The image signal processing unit 110A has an information memory bank 128A. In this information memory bank 128A, coefficient data Wi (i = 1 to n) is stored in advance for each combination of class and parameter p values. A method for generating the coefficient data Wi will be described later.

In this way, the coefficient data Wi stored in the information memory bank 128A is also encoded in the MPEG2 format in the same manner as the coefficient seed data w _{i0 to} w _i3 stored in the information memory bank 128 of the image signal processing unit 110 in FIG. The MPEG2 stream encoded in the same manner as the MPEG2 stream input to the decoder 107 is generated in advance based on a signal obtained by processing according to the value of the parameter p.

  That is, the coefficient data Wi is generated according to a plurality of values of the parameter p, and is generated in advance using a plurality of student signals corresponding to the image signal Va and a teacher signal corresponding to the image signal Vb.

  In this case, the student signal is obtained by decoding the MPEG2 stream obtained by MPEG2 encoding the teacher signal, and the signal is changed according to the value of the parameter p in the process of decoding, so that the parameter A student signal corresponding to the value of p is obtained. In this case, at least one of the DCT coefficient (frequency coefficient) or the inverse transform base used when performing inverse discrete cosine transform (inverse DCT) is changed according to the value of the parameter p.

  For example, it is possible to obtain a plurality of student signals by changing the gain according to the value of the parameter p while setting the gain of all AC coefficients of the DCT coefficients to the same value. When the coefficient data Wi generated in advance using a plurality of student signals obtained in this way is stored in the information memory bank 128A, the resolution of the image obtained by the image signal Vb is set to the parameter p. It can be adjusted in stages according to the value.

  Further, for example, a plurality of student signals can be obtained by changing the quantization step of the DC coefficient among the DCT coefficients according to the value of the parameter p. When the coefficient data Wi generated in advance using a plurality of student signals obtained in this way is stored in the information memory bank 128A, the block distortion of the image obtained by the image signal Vb is set as the parameter p. It becomes possible to adjust in steps according to the value of.

  In addition, for example, a plurality of student signals can be obtained by reducing the gain of the AC coefficient of the DCT coefficient as the frequency increases and changing the degree of decrease in the gain of the AC coefficient according to the parameter value. Can do. When coefficient data Wi generated in advance using a plurality of student signals obtained in this way is stored in the information memory bank 128A, the mosquito noise of the image obtained by the image signal Vb is set as the parameter p. It becomes possible to adjust in steps according to the value of.

  Also, for example, a plurality of student signals can be obtained by adding random noise to at least the DCT coefficient or the inverse transform base and changing the level of the random noise according to the value of the parameter p. When the coefficient seed data Wi generated in advance using a plurality of student signals obtained in this way is stored in the information memory bank 128A, the noise suppression degree of the image obtained by the image signal Vb described above is set. The adjustment can be made in stages according to the value of the parameter p.

The operation of the image signal processing unit 110A will be described.
From the image signal Va stored in the buffer memory 108, the second tap selection circuit 122 selectively extracts pixel data of class taps located around the target position in the image signal Vb to be created. The pixel data of this class tap is supplied to the feature amount extraction unit 123. In the feature amount extraction unit 123, processing such as 1-bit ADRC is performed on the pixel data of the class tap, and an ADRC code as first class information indicating a space class is generated.

  Encoding control information (picture information PI, motion vector information MI) extracted from the MPEG2 stream by the MPEG2 decoder 107 is supplied to the additional information class generation unit 124. The additional information class generation unit 124 generates second class information indicating the additional information class based on the encoding control information.

  Then, the first class information (ADRC code) generated by the feature amount extraction unit 123 and the second class information generated by the additional information class generation unit 124 are supplied to the class code generation unit 125. The class code generation unit 125 generates a class code CL indicating the result of class classification based on the first class information and the second class information.

  The class code CL represents the detection result of the class to which the pixel data at the target position in the image signal Vb belongs. The class code CL is supplied to the coefficient memory 126 as read address information.

  In the coefficient memory 126, for example, coefficient data Wi of each class corresponding to the value of the parameter p adjusted by the user during the vertical blanking period is loaded from the information memory bank 128A and stored under the control of the system controller 101. Is done.

  When the coefficient data Wi corresponding to the adjusted parameter p value is not stored in the information memory bank 128A, the coefficient data Wi corresponding to the values before and after the adjusted parameter p value is stored in the information memory bank 128A. The coefficient data corresponding to the adjusted value of the parameter p may be obtained by reading from the memory bank 128A and performing an interpolation calculation process using them.

  As described above, the class code CL is supplied to the coefficient memory 126 as read address information, whereby the coefficient data Wi corresponding to the class code CL is read from the coefficient memory 126 and supplied to the estimated prediction calculation circuit 127. The Further, the pixel data of the prediction tap located around the target position in the image signal Vb is selectively extracted by the first tap selection circuit 121 from the image signal Va stored in the buffer memory 108.

  The estimated prediction calculation circuit 127 uses the pixel data xi of the prediction tap and the coefficient data Wi read from the coefficient memory 126, and based on the estimation formula (see formula (1)), attention is paid to the image signal Vb to be generated. Pixel data y at the position is obtained.

  Thus, in the image signal processing unit 110A, the pixel data y is calculated using the coefficient data Wi (i = 1 to n) of the estimation formula corresponding to the value of the parameter p. Therefore, the user can arbitrarily adjust the resolution, block distortion, mosquito noise, or noise suppression degree of the image by the image signal Vb by changing the value of the parameter p.

  The coefficient data Wi is based on a signal obtained by processing an MPEG2 stream encoded in the same manner as the MPEG2 stream input to the MPEG2 decoder 107 described above according to the value of the parameter p. It is generated in advance, and only the specific quality of the image based on the image signal Vb, for example, only the resolution, only the block distortion, only the mosquito noise, or only the noise suppression degree can be adjusted in stages.

Next, a method for generating the coefficient data Wi (i = 1 to n) will be described.
In the above, as another example of the coefficient seed data generation method, for each student signal obtained by changing the value of the parameter p stepwise, the coefficient data Wi of each class is generated by learning using the signal, then use the coefficient data Wi for each student signal, have been described herein what obtaining coefficient seed data w ₁₀ to w _n9 for each class. The coefficient data Wi for each combination of class and parameter p values stored in advance in the information memory bank 128A can be generated in the same manner as in the first half of this coefficient seed data generation method.

  FIG. 14 shows the coefficient data generation device 170. In this coefficient data generation device 170, parts corresponding to those of the coefficient seed data generation device 150A shown in FIG.

  This coefficient data generation device 170 has a coefficient memory 167. The coefficient memory 167 stores coefficient data Wi of each class corresponding to each value of the parameter p determined by the coefficient data determination unit 164. The rest of the coefficient data generation device 170 is configured in the same manner as the coefficient seed data generation device 150A shown in FIG. 8 and operates in the same manner.

  14 generates coefficient data Wi for each combination of class and parameter p values stored in the information memory bank 128A of the image signal processing unit 110A of FIG. be able to.

  Also, the processing in the image signal processing unit 110A in FIG. 13 can be realized by software, for example, by the image signal processing device 300 shown in FIG. In this case, coefficient data is stored in advance in the ROM 302 or the like and used.

  A processing procedure for obtaining the image signal Vb from the image signal Va in the image signal processing apparatus 300 shown in FIG. 9 will be described with reference to the flowchart of FIG.

  First, in step ST61, the processing is started, and in step S62, for example, an image signal Va for one frame or one field is input into the apparatus from the input terminal 314. Thus, the pixel data constituting the image signal Va input from the input terminal 314 is temporarily stored in the RAM 303. When the image signal Va is recorded in advance in the hard disk drive 307 in the apparatus, the image signal Va is read from the drive 307, and the pixel data constituting the image signal Va is temporarily stored in the RAM 303. To do.

  In step ST63, it is determined whether or not the processing of all frames or all fields of the image signal Va has been completed. When the process is finished, the process ends at step ST64. On the other hand, when the process is not finished, the process proceeds to step ST65.

  In step ST65, the value of the parameter p, which is an image quality designation value input by the user's operation of the remote control transmitter 200, is acquired from the RAM 303, for example. In step ST66, based on the acquired value of the parameter p, the coefficient data Wi of each class corresponding to the value of the parameter p is read from the ROM 302 or the like and temporarily stored in the RAM 303.

  Next, in step ST67, pixel data of class taps and prediction taps are acquired from the image signal Va input in step ST62, corresponding to the target position in the image signal Vb. In step ST68, it is determined whether or not the processing for obtaining the pixel data of the image signal Vb has been completed in the entire area of the pixel data of the image signal Va for one frame or one field input in step ST62. If completed, the process returns to step ST62, and the process proceeds to input processing of the image signal Va for the next one frame or one field. On the other hand, when the process has not ended, the process proceeds to step ST69.

  In step ST69, the class code CL is generated from the pixel data of the class tap acquired in step ST67. In step ST70, the coefficient data Wi corresponding to the class code CL and the pixel data of the prediction tap are used to generate pixel data of the target position in the image signal Vb by the estimation formula, and then the process returns to step ST67. Then, the same processing as described above is repeated.

  As described above, by performing processing according to the flowchart shown in FIG. 15, the pixel data of the input image signal Va can be processed to obtain the pixel data of the image signal Vb. As described above, the image signal Vb obtained by such processing is output to the output terminal 315, supplied to the display 311 to display an image, and further supplied to the hard disk drive 305 to be supplied to the hard disk. Or is recorded.

  In step ST69, the class code CL is generated from the pixel data of the class tap. However, in addition to the image signal Va, encoding control information (picture information PI, motion vector information MI) related to the image signal Va is also input from the input terminal 314, and the image signal processing unit 110A in FIG. As in, the class code CL may be generated using this encoding control information.

  Although illustration of the processing device is omitted, the processing in the coefficient data generation device 170 in FIG. 14 can also be realized by software.

A processing procedure for generating coefficient data will be described with reference to the flowchart of FIG.
First, in step ST81, processing is started, and in step ST82, an image quality pattern (specified by the value of parameter p) used for learning is selected. In step ST83, it is determined whether learning has been completed for all image quality patterns. If learning has not been completed for all image quality patterns, the process proceeds to step ST84.

  In this step ST84, the teacher signal is inputted for one frame or one field. In step ST85, it is determined whether or not the processing of all the frames or fields of the teacher signal has been completed. If not completed, a student signal is generated in step ST86 based on the image quality pattern selected in step ST82 from the teacher signal input in step ST84.

  Then, in step ST87, pixel data of class taps and prediction taps are acquired from the student signal generated in step ST86, corresponding to the target position in the teacher signal input in step ST84. In step ST88, it is determined whether or not the learning process has been completed in all regions of the teacher signal input in step ST84. When the learning process is finished, the process returns to step ST84, the teacher signal for the next one frame or one field is input, and the same process as described above is repeated. On the other hand, when the learning process is not ended, the process proceeds to step ST89.

  In step ST89, a class code CL is generated from the pixel data of the class tap acquired in step ST87. In step ST90, a normal equation for obtaining coefficient data (see equation (18)) is generated. Thereafter, the process returns to step ST87.

  When the processing is completed in step ST85 described above, in step ST91, the normal equation generated in step ST90 is solved by a sweeping method or the like to calculate coefficient data of each class. Thereafter, the process returns to step ST82, the next image quality pattern is selected, the same processing as described above is repeated, and the coefficient data of each class corresponding to the next image quality pattern is obtained.

  When the coefficient data calculation processing for all image quality patterns is completed in step ST83, the class coefficient data for all image quality patterns is stored in the memory in step ST92, and then in step ST93. The process ends.

  In this way, by performing processing according to the flowchart shown in FIG. 16, coefficient data can be obtained by a method similar to that of the coefficient data generating apparatus 170 shown in FIG. In step ST89, the class code CL is generated from the pixel data of the class tap. However, similarly to the coefficient data generation device 170 of FIG. 14, the class code CL may be generated using also the encoding control information (picture information PI, motion vector information MI) related to the student signal.

Next, still another embodiment of the present invention will be described.
FIG. 17 shows a configuration of a digital broadcast receiver 100B as another embodiment. In FIG. 17, parts corresponding to those in FIG. 1 are given the same reference numerals.

  The digital broadcast receiver 100B is obtained by replacing the image signal processing unit 110 of the digital broadcast receiver 100 shown in FIG. 1 with the image signal processing unit 110B, and operates in the same manner as the digital broadcast receiver 100.

  Details of the image signal processing unit 110B will be described. In this image signal processing unit 110B, portions corresponding to those of the image signal processing unit 110 shown in FIG. 1 are denoted by the same reference numerals, and detailed description thereof is omitted.

  The image signal processing unit 110B has an accumulation table 131. In the accumulation table 131, difference data DF is stored in advance for each combination of class and parameter p values. The difference data DF is difference data of pixel values or DCT coefficients (frequency coefficients) obtained by DCT processing. The accumulation table 131 is supplied with the class code CL output from the class code generation unit 125 and the value of the parameter p output from the system controller 101 as read address information. The difference data DF corresponding to the value of the class code CL and the parameter p is read from the accumulation table 131 and supplied to the adding unit 134 described later.

The difference data DF stored in the accumulation table 131 is also an MPEG2 decoder, similarly to the coefficient seed data w _{i0 to} w _i3 stored in the information memory bank 128 of the image signal processing unit 110 in FIG. The MPEG2 stream encoded in the same manner as the MPEG2 stream input to 107 is generated in advance based on a signal obtained by processing according to the value of the parameter p.

  That is, the difference data DF is generated according to a plurality of values of the parameter p, and is generated in advance using a plurality of student signals corresponding to the image signal Va and a teacher signal corresponding to the image signal Vb.

  In this case, the student signal is obtained by decoding the MPEG2 stream obtained by MPEG2 encoding the teacher signal, and the signal is changed according to the value of the parameter p in the process of decoding, so that the parameter A student signal corresponding to the value of p is obtained. In this case, at least one of the DCT coefficient (frequency coefficient) or the inverse transform base used when performing inverse discrete cosine transform (inverse DCT) is changed according to the value of the parameter p.

  For example, it is possible to obtain a plurality of student signals by changing the gain according to the value of the parameter p while setting the gain of all AC coefficients of the DCT coefficients to the same value. When difference data DF previously generated using a plurality of student signals obtained in this way is stored in the accumulation table 131, the resolution of the image obtained by the image signal Vb is set to the value of the parameter p. It becomes possible to adjust in stages according to.

  Further, for example, a plurality of student signals can be obtained by changing the quantization step of the DC coefficient among the DCT coefficients according to the value of the parameter p. In the case where the difference data DF generated in advance using a plurality of student signals obtained in this way is stored in the accumulation table 131, the block distortion of the image obtained by the image signal Vb is set to the parameter p. It can be adjusted in stages according to the value.

  In addition, for example, a plurality of student signals can be obtained by reducing the gain of the AC coefficient of the DCT coefficient as the frequency increases and changing the degree of decrease in the gain of the AC coefficient according to the parameter value. Can do. In the case where the difference data DF generated in advance using a plurality of student signals obtained in this way is stored in the accumulation table 131, the mosquito noise of the image obtained by the image signal Vb is set to the parameter p. It can be adjusted in stages according to the value.

  Also, for example, a plurality of student signals can be obtained by adding random noise to at least the DCT coefficient or the inverse transform base and changing the level of the random noise according to the value of the parameter p. When difference data DF previously generated using a plurality of student signals obtained in this way is stored in the accumulation table 131, the noise suppression degree of the image obtained by the image signal Vb is set as the parameter p. It becomes possible to adjust in steps according to the value of.

  The digital broadcast receiver 100B also includes a DCT circuit 132 that obtains a DCT coefficient by performing DCT processing on the image signal Va stored in the buffer memory 108, and a DCT coefficient output from the DCT circuit 132 is on the a side. And a selector switch 133 to which the image signal Va output from the buffer memory 108 is input to the fixed terminal on the b side. The changeover switch 133 is connected to the b side when the difference data DF stored in the accumulation table 131 is the difference data of the pixel value, and the a side when the difference data DF is the difference data of the DCT coefficient obtained by the DCT processing. Connected to.

  Further, the image signal processing unit 110B outputs the difference data DF read from the accumulation table 131 to the pixel data (pixel value or DCT coefficient) x corresponding to the target position in the image signal Vb output from the movable terminal of the changeover switch 133. Are added to generate pixel data y of the target position in the image signal Vb.

  In addition, the image signal processing unit 110B includes an inverse DCT circuit 135 that performs inverse DCT processing on the output signal of the adder 134, and the output signal of the inverse DCT circuit 135 is input to a fixed terminal on the a side. and a selector switch 136 to which the output signal of the adder 134 is input to a fixed terminal on the b side. This changeover switch 136 is connected to the b side when the difference data DF stored in the accumulation table 131 is the difference data of the pixel value, and is the a side when it is the difference data of the DCT coefficient obtained by the DCT processing. Connected to. A signal output from the movable terminal of the changeover switch 136 is supplied to the display unit 111 as an image signal Vb.

  Further, the image signal processing unit 110B includes a tap selection circuit 122B corresponding to the second tap selection circuit 122 in the image signal processing unit 110 illustrated in FIG. The tap selection circuit 122B selectively extracts a plurality of pixel data of class taps used for class classification.

  Here, when the difference data DF stored in the accumulation table 131 is pixel value difference data, the tap selection circuit 122B stores a plurality of pixel data located around the target position in the image signal Vb as class taps. Are selectively extracted as pixel data. The class code generator 125 generates a class code CL indicating the class to which the pixel data (one piece) at the target position in the image signal Vb belongs. Then, based on the class code CL, one difference data DF corresponding to the target position in the image signal Vb is read from the accumulation table 131 and supplied to the adding unit 134.

  On the other hand, when the difference data DF stored in the accumulation table 131 is DCT coefficient difference data obtained by DCT processing, the tap selection circuit 122B uses the pixel block of interest (for example, the pixel block ( A plurality of pixel data corresponding to a DCT block as a unit of DCT processing) is selectively extracted as class tap pixel data. Then, the class code generator 125 generates a class code CL indicating the class to which the target pixel block in the image signal Vb belongs. Then, based on the class code CL, a plurality of difference data DF corresponding to the number of pixel data in the pixel block is read from the accumulation table 131 and supplied to the adding unit 134.

The operation of the image signal processing unit 110B will be described.
First, a case where the difference data DF stored in the accumulation table 131 is pixel value difference data will be described. In this case, the changeover switches 133 and 136 are respectively connected to the b side.

  From the image signal Va stored in the buffer memory 108, the tap selection circuit 122B selectively extracts pixel data of the class tap located around the target position in the image signal Vb to be created. The pixel data of this class tap is supplied to the feature amount extraction unit 123. In the feature amount extraction unit 123, processing such as 1-bit ADRC is performed on the pixel data of the class tap, and an ADRC code as first class information indicating a space class is generated.

  Encoding control information (picture information PI, motion vector information MI) extracted from the MPEG2 stream by the MPEG2 decoder 107 is supplied to the additional information class generation unit 124. The additional information class generation unit 124 generates second class information indicating the additional information class based on the encoding control information.

  Then, the first class information (ADRC code) generated by the feature amount extraction unit 123 and the second class information generated by the additional information class generation unit 124 are supplied to the class code generation unit 125. The class code generation unit 125 generates a class code CL indicating the result of class classification based on the first class information and the second class information. The class code CL indicates a class to which pixel data (one piece) at the target position in the image signal Vb belongs.

  The class code CL is supplied to the accumulation table 131 as read address information. Based on the class code CL, one difference data DF corresponding to the target position in the image signal Vb is read from the accumulation table 131 and supplied to the adding unit 134.

  Of the image signal Va stored in the buffer memory 108, pixel data x corresponding to the target position in the image signal Vb is supplied to the adding unit 134 via the b side of the changeover switch 133. In the adder 134, the difference data DF read from the accumulation table is added to the pixel data x, and the pixel data y at the target position in the image signal Vb is generated. The pixel data y is output as an output signal of the image signal processing unit 110B via the b side of the changeover switch 136.

  Next, a case where the difference data DF stored in the accumulation table 131 is DCT coefficient difference data obtained by DCT processing will be described. In this case, the changeover switches 133 and 136 are respectively connected to the a side.

  From the image signal Va stored in the buffer memory 108, the tap selection circuit 122B selectively extracts a plurality of pixel data corresponding to the target pixel block in the image signal Vb to be created as pixel data of the class tap. The pixel data of this class tap is supplied to the feature amount extraction unit 123. In the feature amount extraction unit 123, processing such as 1-bit ADRC is performed on the pixel data of the class tap, and an ADRC code as first class information indicating a space class is generated.

  Encoding control information (picture information PI, motion vector information MI) extracted from the MPEG2 stream by the MPEG2 decoder 107 is supplied to the additional information class generation unit 124. The additional information class generation unit 124 generates second class information indicating the additional information class based on the encoding control information.

  Then, the first class information (ADRC code) generated by the feature amount extraction unit 123 and the second class information generated by the additional information class generation unit 124 are supplied to the class code generation unit 125. The class code generation unit 125 generates a class code CL indicating the result of class classification based on the first class information and the second class information. The class code CL indicates the class to which the target pixel block in the image signal Vb belongs.

  The class code CL is supplied to the accumulation table 131 as read address information. Based on the class code CL, a plurality of difference data DF is read from the accumulation table 131 and supplied to the adding unit 134.

  Also, DCT is obtained for a plurality of pixel data of the image signal Va corresponding to the pixel block of interest in the image signal Vb (all pixel data in this block becomes pixel data of the target position) obtained from the DCT circuit 132. A plurality of DCT coefficients x constituting the DCT block obtained by performing the processing are supplied to the adding unit 134 via the a side of the changeover switch 133. In the adder 134, the plurality of difference data DF read from the accumulation table is added to the plurality of DCT coefficients x, respectively, and the DCT coefficient y of the DCT block corresponding to the target pixel block in the image signal Vb is generated. .

  A plurality of DCT coefficients y of each DCT block sequentially output from the adder 134 corresponding to each block of interest in the image signal Vb are supplied to the inverse DCT circuit 135. In the inverse DCT circuit 135, inverse DCT processing is performed on a plurality of DCT coefficients y of each DCT block to obtain pixel data (pixel value) of each pixel block. In this manner, the pixel data output from the inverse DCT circuit 135 is output as an output signal of the image signal processing unit 110B via the a side of the changeover switch 136.

  In this way, in the image signal processing unit 110B, the difference data DF corresponding to the value of the parameter p is used to calculate pixel data (DCT coefficient) y. Therefore, the user can arbitrarily adjust the resolution, block distortion, mosquito noise, or noise suppression degree of the image by the image signal Vb by changing the value of the parameter p.

  The difference data DF is preliminarily determined based on a signal obtained by processing an MPEG2 stream encoded in the same manner as the MPEG2 stream input to the MPEG2 decoder 107 described above according to the value of the parameter p. Only the specific quality of the image generated by the image signal Vb, for example, only the resolution, only the block distortion, only the mosquito noise, or only the noise suppression level can be adjusted stepwise.

  FIG. 18 shows a configuration of the difference data generation device 210 that generates the difference data DF to be stored in the accumulation table 131 of the image signal processing unit 110B of FIG. 18, parts corresponding to those in FIG. 4 are given the same reference numerals, and detailed descriptions thereof are omitted.

  The differential data generation device 210 performs DCT processing on the student signal SS output from the MPEG2 decoder 153 to obtain a DCT coefficient, and the DCT coefficient output from the DCT circuit 171 is a side. The changeover switch 172 to which the student signal SS output from the MPEG2 decoder 153 is input is input to the fixed terminal on the b side. The changeover switch 172 is connected to the b side when the difference data DF accumulated in the accumulation table 177 described later is the difference data of the pixel value, and the a side when the difference data DF is the difference data of the DCT coefficient obtained by the DCT processing. Connected to.

  Further, the differential data generation device 210 performs a DCT process on the teacher signal ST time-adjusted by the delay circuit 159 to obtain a frequency coefficient, and the frequency coefficient output from the DCT circuit 173 is the a side. And a changeover switch 174 to which the teacher signal ST adjusted in time by the delay circuit 159 is input to the b-side fixed terminal. The changeover switch 174 is connected to the b side when the difference data DF accumulated in the accumulation table 177 to be described later is the difference data of the pixel value, and the a side when the difference data DF is the difference data of the DCT coefficient obtained by the DCT processing. Connected to.

  Further, the difference data generation device 210 is output from the movable terminal of the changeover switch 174 from pixel data (pixel value or DCT coefficient) y at the target position of the teacher signal ST output from the movable terminal of the changeover switch 174. A subtracting unit 175 that subtracts pixel data (pixel value or DCT coefficient) x corresponding to the target position of the teacher signal ST to obtain difference data df is provided.

  Further, the difference data generation device 210 applies the class code CL output from the class code generation unit 158 and the value of the parameter p input to the MPEG2 decoder 153 to the difference data df sequentially output from the subtraction unit 175. Is stored in the accumulation table 177 as difference data DF, and an averaging process is performed for each combination of class and parameter p.

  Further, the difference data generation device 210 has a tap selection circuit 155A corresponding to the second tap selection circuit 155 in the coefficient seed data generation device 150 shown in FIG. The tap selection circuit 155A selectively extracts a plurality of pixel data of class taps used for class classification.

  Here, when the difference data DF stored in the accumulation table 177 is pixel value difference data, in the tap selection circuit 155A, a plurality of pixel data located around the target position in the teacher signal ST is a pixel of a class tap. It is selectively retrieved as data. Then, the class code generation unit 158 generates a class code CL indicating the class to which the pixel data (one piece) at the target position in the teacher signal ST belongs.

  On the other hand, when the difference data DF stored in the accumulation table 177 is DCT coefficient difference data obtained by DCT processing, the tap selection circuit 155A uses, for example, the target pixel block (DCT processing) in the teacher signal ST from the student signal SS. A plurality of pixel data corresponding to a DCT block which is a unit of () are selectively extracted as class tap pixel data. The class code generation unit 158 generates a class code CL indicating the class to which the pixel block of interest belongs in the teacher signal ST.

  The rest of the difference data generation device 210 shown in FIG. 18 is configured in the same manner as the coefficient seed data generation device 150 shown in FIG.

Next, the operation of the difference data generation device 210 shown in FIG. 18 will be described.
First, a case where the difference data DF stored in the accumulation table 177 is pixel value difference data will be described. In this case, the changeover switches 172 and 174 are respectively connected to the b side.

  A teacher signal ST corresponding to the image signal Vb is supplied to the input terminal 151, and the teacher signal ST is encoded by the MPEG2 encoder 152 to generate an MPEG2 stream. This MPEG2 stream is supplied to the MPEG2 decoder 153.

  The value of the parameter p is input to the MPEG2 decoder 153 so that the signal changes in accordance with the value of the parameter p in the decoding process. That is, in the present embodiment, as described above, at least one of the DCT coefficient (frequency coefficient) or the inverse transform base used when performing the inverse discrete cosine transform (inverse DCT) according to the value of the parameter p. Is changed. Thereby, the student signal SS corresponding to the value of the parameter p is obtained.

  Of the teacher signal ST time-adjusted by the delay circuit 159, the pixel data y at the target position is supplied to the subtraction unit 175 via the b side of the changeover switch 174. Of the student signal SS output from the MPEG2 decoder 153, pixel data x corresponding to the position of interest in the teacher signal ST is supplied to the subtracting unit 175 via the b side of the changeover switch 172. Then, in the subtraction unit 175, the pixel data x is subtracted from the pixel data y to generate difference data df. The difference data df corresponding to each position of interest in the teacher signal ST sequentially output from the subtraction unit 175 is supplied to the accumulation control unit 176.

  Also, from the student signal SS obtained by the MPEG2 decoder 153, the pixel data of the class tap located around the target position in the teacher signal ST is selectively extracted by the tap selection circuit 155A. The pixel data of this class tap is supplied to the feature amount extraction unit 156. In the feature amount extraction unit 156, processing such as 1-bit ADRC is performed on the pixel data of the class tap, and an ADRC code serving as first class information indicating a space class is generated.

  Encoding control information (picture information PI, motion vector information MI) extracted from the MPEG2 stream by the MPEG2 decoder 153 is supplied to the additional information class generation unit 157. The additional information class generation unit 157 generates second class information indicating the additional information class based on the encoding control information.

  Then, the first class information (ADRC code) generated by the feature amount extraction unit 156 and the second class information generated by the additional information class generation unit 157 are supplied to the class code generation unit 158. The class code generation unit 158 generates a class code CL indicating the result of class classification based on the first class information and the second class information. The class code CL indicates a class to which pixel data (one piece) at the target position in the teacher signal ST belongs.

  The class code CL is supplied to the accumulation control unit 176. The accumulation control unit 176 is also supplied with the value of the parameter p input to the MPEG2 decoder 153. The accumulation control unit 176 performs an averaging process on the difference data df sequentially output from the subtraction unit 175 based on the class code CL and the value of the parameter p for each combination of the class and the parameter p. Is stored in the accumulation table 177 as difference data DF.

  Next, a case where the difference data DF stored in the accumulation table 177 is DCT coefficient difference data obtained by DCT processing will be described. In this case, the changeover switches 172 and 174 are respectively connected to the a side.

  A teacher signal ST corresponding to the image signal Vb is supplied to the input terminal 151, and the teacher signal ST is encoded by the MPEG2 encoder 152 to generate an MPEG2 stream. This MPEG2 stream is supplied to the MPEG2 decoder 153.

  The value of the parameter p is input to the MPEG2 decoder 153 so that the signal changes in accordance with the value of the parameter p in the decoding process. That is, in the present embodiment, as described above, at least one of the DCT coefficient (frequency coefficient) or the inverse transform base used when performing the inverse discrete cosine transform (inverse DCT) according to the value of the parameter p. Is changed. Thereby, the student signal SS corresponding to the value of the parameter p is obtained.

  A DCT block obtained by performing DCT processing on a plurality of pixel data of the target pixel block (all pixel data in this block becomes pixel data at the target position) in the teacher signal ST obtained from the DCT circuit 173 The plurality of DCT coefficients y are supplied to the subtracting unit 175 via the a side of the changeover switch 174.

  The subtracting unit 175 also includes a plurality of DCT blocks obtained by performing DCT processing on a plurality of pixel data of the student signal SS corresponding to the target pixel block in the teacher signal ST obtained from the DCT circuit 171. DCT coefficients x are supplied through the a side of the changeover switch 172. The subtracting unit 175 then subtracts the plurality of DCT coefficients x from the plurality of DCT coefficients y to generate a plurality of difference data df. A plurality of difference data df corresponding to each target pixel block in the teacher signal ST sequentially output from the subtraction unit 175 is supplied to the accumulation control unit 176.

  Further, from the student signal SS obtained by the MPEG2 decoder 153, a plurality of pixel data corresponding to the pixel block of interest in the teacher signal ST is selectively extracted as class tap pixel data by the tap selection circuit 155A. The pixel data of this class tap is supplied to the feature amount extraction unit 156. In the feature amount extraction unit 156, processing such as 1-bit ADRC is performed on the pixel data of the class tap, and an ADRC code serving as first class information indicating a space class is generated.

  Encoding control information (picture information PI, motion vector information MI) extracted from the MPEG2 stream by the MPEG2 decoder 153 is supplied to the additional information class generation unit 157. The additional information class generation unit 157 generates second class information indicating the additional information class based on the encoding control information.

  Then, the first class information (ADRC code) generated by the feature amount extraction unit 156 and the second class information generated by the additional information class generation unit 157 are supplied to the class code generation unit 158. The class code generation unit 158 generates a class code CL indicating the result of class classification based on the first class information and the second class information. The class code CL indicates the class to which the target position pixel block in the teacher signal ST belongs.

  The class code CL is supplied to the accumulation control unit 176. The accumulation control unit 176 is also supplied with the value of the parameter p input to the MPEG2 decoder 153. The accumulation control unit 176 performs an averaging process on each of the plurality of difference data df sequentially output from the subtraction unit 175 for each combination of the class and the parameter p based on the value of the class code CL and the parameter p. The result is stored in the accumulation table 177 as a plurality of difference data DF.

  As described above, the difference data generation device 210 illustrated in FIG. 18 generates the difference data DF for each combination of the class and parameter p values stored in the accumulation table 131 of the image signal processing unit 110B in FIG. Can do.

  Note that the processing in the image signal processing unit 110B of FIG. 17 can also be realized by software, for example, by an image signal processing device 300 as shown in FIG. In this case, the difference data is stored in advance in the ROM 302 or the like and used.

  A processing procedure for obtaining the image signal Vb from the image signal Va in the image signal processing apparatus 300 shown in FIG. 9 will be described with reference to the flowchart of FIG.

  First, in step ST101, processing is started, and in step S102, for example, an image signal Va for one frame or one field is input into the apparatus from the input terminal 314. Thus, the pixel data constituting the image signal Va input from the input terminal 314 is temporarily stored in the RAM 303. When the image signal Va is recorded in advance in the hard disk drive 307 in the apparatus, the image signal Va is read from the drive 307, and the pixel data constituting the image signal Va is temporarily stored in the RAM 303. To do.

  In step ST103, it is determined whether or not processing of all frames or all fields of the image signal Va has been completed. If the process is finished, the process ends in step ST104. On the other hand, when the process is not finished, the process proceeds to step ST105.

  In step ST105, the value of the parameter p, which is an image quality designation value input by the user's operation of the remote control transmitter 200, is acquired from the RAM 303, for example. In step ST106, the pixel data of the class tap is acquired from the image signal Va input in step ST102 corresponding to the target position in the image signal Vb. In step ST107, a class code CL is generated from the pixel data of the class tap.

  Next, in step ST108, based on the value of the parameter p acquired in step ST105 and the class code CL generated in step ST107, the difference data DF corresponding to the value of the parameter p and the class code CL is obtained from the ROM 302 or the like. Read and temporarily store in RAM 303.

  Next, in step ST109, the difference data DF read in step ST108 is added to the pixel data x corresponding to the target position in the image signal Vb in the image signal Va, and the pixel at the target position in the image signal Vb. Data y is generated.

  Here, when the difference data DF stored in the ROM 302 or the like is the difference data of the DCT coefficient obtained by the DCT process, the pixel data y as the addition result is a DCT coefficient. Therefore, in step ST109, the inverse DCT is further performed. Process. In that case, DCT processing is performed on the image signal Va input in step ST102 described above, and the pixel data x corresponding to the target position in the image signal Vb is used as the DCT coefficient.

  Next, in step ST110, it is determined whether or not the processing for obtaining the pixel data of the image signal Vb in the entire region of the pixel data of the image signal Va for one frame or one field input in step ST102 is completed. If completed, the process returns to step ST102, and the process proceeds to input processing of the image signal Va for the next one frame or one field. On the other hand, when the process has not ended, the process returns to step ST106, and the same process as described above is repeated.

  In this way, by performing processing according to the flowchart shown in FIG. 19, the pixel data of the input image signal Va can be processed, and the pixel data of the image signal Vb can be obtained. As described above, the image signal Vb obtained by such processing is output to the output terminal 315, supplied to the display 311 to display an image, and further supplied to the hard disk drive 305 to be supplied to the hard disk. Or is recorded.

  In step ST107, the class code CL is generated from the pixel data of the class tap. However, in addition to the image signal Va, encoding control information (picture information PI, motion vector information MI) related to the image signal Va is also input from the input terminal 314, and the image signal processing unit 110B in FIG. As in, the class code CL may be generated using this encoding control information.

  Although illustration of the processing device is omitted, the processing in the differential data generation device 210 in FIG. 18 can also be realized by software.

A processing procedure for generating difference data will be described with reference to the flowchart of FIG.
First, in step ST121, processing is started, and in step ST122, an image quality pattern (specified by the value of parameter p) is selected. In step ST123, it is determined whether or not the difference data DF generation processing has been completed for all image quality patterns. If the generation process of the difference data DF has not been completed for all image quality patterns, the process proceeds to step ST124.

  In this step ST124, the teacher signal is inputted for one frame or one field. In step ST125, it is determined whether or not processing of all the frames or fields of the teacher signal has been completed. If not completed, a student signal is generated in step ST126 based on the image quality pattern selected in step ST122 from the teacher signal input in step ST124.

  In step ST127, class tap pixel data is acquired from the student signal generated in step ST126, corresponding to the target position in the teacher signal input in step ST124. In step ST128, a class code CL is generated from the pixel data of the class tap.

  Next, in step ST129, the difference data df is obtained by subtracting the pixel data x of the student signal corresponding to the target position of the teacher signal from the pixel data y of the target position of the teacher signal. Furthermore, in this step ST129, based on the class code CL generated in step ST128, an averaging process is performed for each class to generate difference data DF.

  Next, in step ST130, it is determined whether or not the generation process of the difference data DF has been completed in all regions of the teacher signal input in step ST124. When the generation process of the difference data DF has been completed, the process returns to step ST124, the next one frame or one field of teacher signals are input, and the same process as described above is repeated. On the other hand, when the difference data generation process has not ended, the process returns to step ST127 and the same process as described above is repeated.

  When the process is completed in step ST125 described above, the process returns to step ST122, the next image quality pattern is selected, the same process as described above is repeated, and the difference of each class corresponding to the next image quality pattern is determined. Find the data DF.

  Here, when generating the difference data of the DCT coefficient obtained by the DCT processing as the difference data DF, it is necessary to use the difference data df as the subtraction result as the DCT coefficient. In that case, DCT processing is performed on the teacher signal input in step ST124 described above, and pixel data y corresponding to the target position in the teacher signal is used as a DCT coefficient. Also, the student signal generated in step ST126 is subjected to DCT processing, and the pixel signal x of the student signal corresponding to the target position in the teacher signal is used as the DCT coefficient.

  When the difference data DF calculation processing for all image quality patterns is completed in step ST123, the class difference data DF for all image quality patterns is stored in the memory in step ST131, and then step ST132 is performed. Then, the process ends.

  As described above, by performing the processing according to the flowchart shown in FIG. 20, the difference data DF can be obtained by the same method as that of the difference data generation device 210 shown in FIG. In step ST128, the class code CL is generated from the pixel data of the class tap. However, similar to the difference data generation device 210 of FIG. 18, the class code CL may be generated using also the encoding control information (picture information PI, motion vector information MI) related to the student signal.

Next, another embodiment of the present invention will be described.
FIG. 21 shows the configuration of a digital broadcast receiver 100C as another embodiment. In FIG. 21, parts corresponding to those in FIG.

  The digital broadcast receiver 100C includes a microcomputer, and includes a system controller 101 for controlling the operation of the entire system, and a remote control signal receiving circuit 102 that receives a remote control signal. The remote control signal receiving circuit 102 is connected to the system controller 101, receives a remote control signal RM output from the remote control transmitter 200 according to a user operation, and supplies an operation signal corresponding to the signal RM to the system controller 101. Is configured to do.

  Also, the digital broadcast receiver 100C is supplied with a receiving antenna 105 and a broadcast signal (RF modulated signal) captured by the receiving antenna 105, performs channel selection processing, demodulation processing, error correction processing, etc. And a stream buffer 109 for temporarily storing the MPEG2 stream Vc output from the tuner unit 106. The tuner unit 106 obtains an MPEG2 stream as an encoded image signal.

  Also, the digital broadcast receiver 100C converts the MPEG2 stream Vc stored in the stream buffer 109 into a decoded image signal Vd that presents a user-desired image quality, and this image signal. And a display unit 111 that displays an image based on the image signal Vd output from the processing unit 110C.

The operation of the digital broadcast receiver 100C shown in FIG. 21 will be described.
The MPEG2 stream Vc output from the tuner unit 106 is supplied to the stream buffer 109 and temporarily stored.

  In this way, the MPEG2 stream Vc stored in the stream buffer 109 is supplied to the image signal processing unit 110C and converted into a decoded image signal Vd that presents a user-desired image quality. In this image signal processing unit 110C, pixel data constituting the image signal Vd is obtained from the DCT coefficients constituting the MPEG2 stream Vc.

  The image signal Vd output from the image signal processing unit 110C is supplied to the display unit 111, and an image based on the image signal Vd is displayed on the screen of the display unit 111.

  The user can adjust the resolution, block distortion, mosquito noise, or noise suppression degree in the image displayed on the screen of the display unit 111 as described above by operating the remote control transmitter 200. As will be described later, in the image signal processing unit 110C, pixel data of the image signal Vd is calculated by an estimation formula. As coefficient data of this estimation formula, the parameter p changed by the user's operation of the remote control transmitter 200 is used. The one corresponding to the value is generated and used by a generation formula including this parameter p. As a result, the resolution, block distortion, mosquito noise, or noise suppression degree in the image based on the image signal Vd output from the image signal processing unit 110C corresponds to the changed value of the parameter p.

  Details of the image signal processing unit 110C will be described. In this image signal processing unit 110C, parts corresponding to those of the image signal processing unit 110 shown in FIG.

  The image signal processing unit 110C includes a DCT coefficient extraction circuit 221 that extracts a DCT coefficient as a frequency coefficient from the MPEG2 stream Vc stored in the stream buffer 109, and variable-length encoding extracted by the extraction circuit 221 such as Huffman A variable-length decoding circuit 222 that performs variable-length decoding on the encoded DCT coefficients.

  The image signal processing unit 110C also includes a quantization characteristic designation information extraction circuit 223 that extracts quantization characteristic designation information from the MPEG2 stream Vc stored in the stream buffer 109, and a quantization characteristic extracted by the extraction circuit 223. And an inverse quantization circuit 224 that performs inverse quantization on the quantized DCT coefficient output from the variable length decoding circuit 222 based on the designation information.

  Further, the image signal processing unit 110C selectively extracts a DCT coefficient corresponding to the target position in the image signal Vd from the DCT coefficient output from the inverse quantization circuit 224, and the image signal from the reference image signal Vref described later. There are first and second tap selection circuits 225 and 226 for selectively extracting and outputting a plurality of pixel data located around the target position in Vd.

  The first tap selection circuit 225 selectively extracts a plurality of pixel data (including DCT coefficients) of prediction taps used for prediction. FIG. 2A shows an example of the arrangement of pixel data of prediction taps that are selectively extracted from the reference image signal Vref in the first tap selection circuit 225. In this figure, “◯” indicates the tap position of the prediction tap, and “X” indicates the position of interest. In the present embodiment, seven pieces of pixel data in the current field (shown by a solid line) and the previous field (shown by a broken line) are extracted as pixel data of a prediction tap.

  FIG. 22A shows an example of the DCT coefficient of the prediction tap that is selectively extracted from the DCT coefficient output from the inverse quantization circuit 224 in the first tap selection circuit 225. In the present embodiment, a pixel block to which the target position in the image signal Vd belongs (corresponding to an 8 × 8 small block serving as a unit of DCT processing) is a target pixel block, and a block of DCT coefficients corresponding to the target pixel block. All DCT coefficients (hereinafter referred to as “DCT blocks”), that is, 64 (8 × 8) DCT coefficients are extracted as DCT coefficients of the prediction tap.

  The second tap selection circuit 226 selectively extracts a plurality of pixel data (including DCT coefficients) of class taps used for class classification. FIG. 2B shows an example of the arrangement of pixel data of class taps selectively extracted from the reference image signal Vref in the second tap selection circuit 226. In this figure, “◯” indicates the tap position of the prediction tap, and “X” indicates the position of interest. In the present embodiment, 13 pieces of pixel data in the current field (shown with solid lines) and the previous field (shown with broken lines) are extracted as pixel data of class taps.

  FIG. 22B shows an example of class tap DCT coefficients that are selectively extracted from the DCT coefficients output from the inverse quantization circuit 224 in the second tap selection circuit 226. In the present embodiment, 320 (8 × 8 × 5) DCTs of a total of five DCT blocks including a DCT block corresponding to the pixel block of interest and four DCT blocks adjacent in the vertical and horizontal directions. The coefficients are extracted as class tap DCT coefficients.

  In addition, the image signal processing unit 110C includes a feature amount extraction unit 123 that extracts feature amounts from pixel data (including DCT coefficients) of class taps selectively extracted by the second tap selection circuit 226. The feature amount extraction unit 123 generates an ADRC code as first class information indicating a space class by performing processing such as 1-bit ADRC (Adaptive Dynamic Range Coding) on the pixel data of the class tap.

  Also, the image signal processing unit 110C includes an encoding control information extraction circuit 227 that extracts encoding control information (picture information PI, motion vector information MI) as additional information from the MPEG2 stream Vc stored in the stream buffer 109. And an additional information class generation unit 124 that generates second class information indicating the additional information class based on the encoding control information extracted by the extraction circuit 227.

  In addition, the image signal processing unit 110C performs class classification based on the first class information (ADRC code) generated by the feature amount extraction unit 123 and the second class information generated by the additional information class generation unit 124. A class code generation unit 125 that generates a class code CL indicating the result is included. The class code CL indicates the class to which the pixel data at the target position in the image signal Vd belongs.

  The image signal processing unit 110 </ b> C has a coefficient memory 126. The coefficient memory 126 stores, for each class, a plurality of coefficient data Wi (i = 1 to n) used in an estimation formula used in an estimated prediction calculation circuit 127 described later. The coefficient data Wi is information for converting the MPEG2 stream Vc into the image signal Vd. The class code CL output from the class code generation unit 125 described above is supplied to the coefficient memory 126 as read address information, and the coefficient data Wi of the estimation formula corresponding to the class code CL is read from the coefficient memory 126. , And supplied to the estimated prediction calculation circuit 127.

  The image signal processing unit 110 </ b> C has an information memory bank 128. In the estimated prediction calculation circuit 127 described later, the position of interest in the image signal Vd to be created is calculated from the pixel data xi of the prediction tap and the coefficient data Wi read from the coefficient memory 126 by the above-described estimation formula (1). Pixel data y is calculated. In Expression (1), n represents the number of prediction taps selected by the first tap selection circuit 225.

The coefficient data Wi of this estimation formula is generated by a generation formula including the parameter p, as shown in the above formula (2). Information in the memory bank 128, the coefficient seed data w _i0 to w _i3 is the coefficient data in the production equation is, for each class, stored.

Further, the image signal processing unit 110C uses the coefficient seed data w _{i0 to} w _{i3 of} each class and the value of the parameter p, and the coefficient of the estimation formula corresponding to the value of the parameter p for each class according to the equation (2). A coefficient generation circuit 129 that generates data Wi (i = 1 to n) is included. The coefficient generation circuit 129 is loaded with the above-described class seed coefficient data w _{i0 to} w _i3 from the information memory bank 128. The coefficient generation circuit 129 is supplied with the value of the parameter p from the system controller 101.

  The coefficient data Wi (i = 1 to n) of each class generated by the coefficient generation circuit 129 is stored in the coefficient memory 126 described above. The generation of coefficient data Wi for each class in the coefficient generation circuit 129 is performed, for example, in each vertical blanking period. Thereby, even if the value of the parameter p is changed by the user's operation of the remote control transmitter 200, the coefficient data Wi of each class stored in the coefficient memory 126 is immediately changed to the one corresponding to the value of the parameter p. The user can smoothly adjust the resolution, block distortion, mosquito noise, noise suppression degree, and the like.

Here, the coefficient seed data w _{i0 to} w _i3 stored in the information memory bank 128 as described above will be further described. The coefficient seed data w _{i0 to} w _i3 are converted into a student signal obtained by processing an MPEG2 stream encoded in the same manner as the MPEG2 stream Vc stored in the stream buffer 109 according to the value of the parameter p. Pre-generated based on.

That is, the coefficient seed data w _{i0 to} w _i3 are respectively generated according to a plurality of values of the parameter p, and using a plurality of student signals corresponding to the MPEG2 stream Vc and a teacher signal corresponding to the image signal Vd. Pre-generated. In this case, a student signal corresponding to the value of the parameter p is obtained by processing an MPEG2 stream obtained by MPEG2 encoding the teacher signal according to the value of the parameter p. In this case, the DCT coefficient (frequency coefficient) is changed according to the value of the parameter p.

For example, it is possible to obtain a plurality of student signals by changing the gain according to the value of the parameter p while setting the gain of all AC coefficients of the DCT coefficients to the same value. When the coefficient seed data w _{i0 to} w _i3 generated in advance using the student signal obtained in this way is stored in the information memory bank 128, the resolution of the image obtained by the image signal Vd described above is set. The adjustment can be made in stages according to the value of the parameter p.

Further, for example, a plurality of student signals can be obtained by changing the quantization step of the DC coefficient among the DCT coefficients according to the value of the parameter p. In the information memory bank 128, when this way the coefficient seed data w _i0 to w _i3 generated in advance using a plurality of student signal obtained is stored, the image obtained by the image signal Vd of the above The block distortion can be adjusted stepwise according to the value of the parameter p.

In addition, for example, a plurality of student signals can be obtained by reducing the gain of the AC coefficient of the DCT coefficient as the frequency increases and changing the degree of decrease in the gain of the AC coefficient according to the parameter value. Can do. In the information memory bank 128, when this way the coefficient seed data w _i0 to w _i3 generated in advance using a plurality of student signal obtained is stored, the image obtained by the image signal Vd of the above The mosquito noise can be adjusted stepwise according to the value of the parameter p.

Further, for example, a plurality of student signals can be obtained by adding random noise to the DCT coefficient and changing the level of the random noise in accordance with the value of the parameter p. The information memory bank 128, when this way the coefficient seed data w _i0 to w _i3 generated in advance using a plurality of student signal obtained is stored, the image obtained by the image signal Vd of the above The noise suppression degree can be adjusted stepwise according to the value of the parameter p.

  In addition, the image signal processing unit 110 </ b> C uses the pixel data (including DCT coefficients) xi of the prediction tap selectively extracted by the first tap selection circuit 225 and the coefficient data Wi read from the coefficient memory 126 as (1 The estimation prediction calculation circuit 127 that calculates pixel data y of the target position in the image signal Vd to be created is provided by the estimation formula (1).

  The image signal processing unit 110 </ b> C includes a prediction memory circuit 228. The prediction memory circuit 228 stores image signals of an I picture (Intra-Picture) and a P picture (Predictive-Picture) in a memory (not shown), and uses these image signals to use an inverse quantization circuit 224. When a DCT coefficient related to a P picture or a B picture (Bidirectionally predictive-Picture) is output, the corresponding reference image signal Vref is generated and supplied to the first and second tap selection circuits 225 and 226. Incidentally, when the DCT coefficient related to the I picture is output from the inverse quantization circuit 224, the reference image signal Vref is not supplied to the first and second tap selection circuits 225 and 226.

  In addition, the image signal processing unit 110C includes a picture selection circuit 229. The picture selection circuit 229 supplies the image signals of the I picture and P picture output from the estimated prediction calculation circuit 127 to the prediction memory circuit 228 and stores them in the memory, and also outputs each of the image signals output from the estimated prediction calculation circuit 127. The picture signals of pictures are rearranged in the correct order and output.

  The motion vector information MI extracted by the extraction circuit 227 described above is supplied to the prediction memory circuit 228, and the prediction memory circuit 228 performs motion compensation when generating the reference image signal using the motion vector information MI. The picture information PI extracted by the extraction circuit 227 is supplied to the prediction memory circuit 228 and the picture selection circuit 229, and the prediction memory circuit 228 and the picture selection circuit 229 identify pictures based on the picture information PI. .

Next, the operation of the image signal processing unit 110C will be described.
The MPEG2 stream Vc stored in the stream buffer 109 is supplied to the extraction circuit 221 to extract DCT coefficients as frequency coefficients. This DCT coefficient is variable length encoded, and this DCT coefficient is supplied to the variable length decoding circuit 222 and decoded. Then, the quantized DCT coefficient output from the variable length decoding circuit 222 is supplied to the inverse quantization circuit 224 and subjected to inverse quantization.

  In the second tap selection circuit 226, a plurality of class tap pixel data (including DCT coefficients) is selectively extracted. That is, the DCT coefficient corresponding to the target position in the image signal Vd is selectively extracted from the DCT coefficient output from the inverse quantization circuit 224, and the image signal is output from the reference image signal Vref supplied from the prediction memory circuit 228. A plurality of pixel data located around the target position in Vd are selectively extracted.

  As described above, the pixel data of the class tap extracted by the second tap selection circuit 226 is supplied to the feature amount extraction unit 123. In the feature amount extraction unit 123, processing such as 1-bit ADRC is performed on the pixel data of the class tap, and an ADRC code as first class information indicating a space class is generated.

  Also, the encoding control information (picture information PI, motion vector information MI) extracted from the MPEG2 stream Vc by the extraction circuit 227 is supplied to the additional information class generation unit 124. The additional information class generation unit 124 generates second class information indicating the additional information class based on the encoding control information.

  Then, the first class information (ADRC code) generated by the feature amount extraction unit 123 and the second class information generated by the additional information class generation unit 124 are supplied to the class code generation unit 125. The class code generation unit 125 generates a class code CL indicating the result of class classification based on the first class information and the second class information.

The class code CL represents the detection result of the class to which the pixel data at the target position in the image signal Vd belongs. The class code CL is supplied to the coefficient memory 126 as read address information. In the coefficient memory 126, for example, in each vertical blanking period, the coefficient generation circuit 129 estimates for each class using the coefficient seed data w _{i0 to} w _i3 corresponding to the value of the parameter p adjusted by the user. Coefficient data Wi (i = 1 to n) of the equation is obtained and stored (see equation (2)).

  As described above, the class code CL is supplied to the coefficient memory 126 as read address information, whereby the coefficient data Wi corresponding to the class code CL is read from the coefficient memory 126 and supplied to the estimated prediction calculation circuit 127. The

  The first tap selection circuit 225 selectively extracts a plurality of pixel data (including DCT coefficients) of the prediction tap. That is, the DCT coefficient corresponding to the target position in the image signal Vd is selectively extracted from the DCT coefficient output from the inverse quantization circuit 224, and the image signal is output from the reference image signal Vref supplied from the prediction memory circuit 228. A plurality of pixel data located around the target position in Vd are selectively extracted. In this way, the pixel data of the prediction tap extracted by the first tap selection circuit 265 is supplied to the estimated prediction calculation circuit 127.

  The estimated prediction calculation circuit 127 uses the pixel data xi of the prediction tap and the coefficient data Wi read from the coefficient memory 126 to draw attention in the image signal Vd to be created based on the estimation expression (see Expression (1)). Pixel data y at the position is obtained. The picture signals of the pictures output from the estimated prediction calculation circuit 127 are rearranged in the correct order by the picture selection circuit 229 and output as an image signal Vd.

  In this way, the image signal processing unit 110C calculates the pixel data y using the coefficient data Wi (i = 1 to n) of the estimation formula corresponding to the value of the parameter p. Therefore, the user can arbitrarily adjust the resolution, block distortion, mosquito noise, or noise suppression degree of the image by the image signal Vd by changing the value of the parameter p.

Further, the coefficient seed data w _{i0 to} w _i3 for generating the coefficient data Wi is an MPEG2 stream encoded in the same manner as the MPEG2 stream Vc stored in the stream buffer 109 described above according to the value of the parameter p. Is generated in advance based on the student signal obtained by processing, and only the specific quality of the image by the image signal Vd, for example, only the resolution, only the block distortion, only the mosquito noise, or only the noise suppression degree is stepwise. Can be adjusted.

The coefficient seed data w _{i0 to} w _i3 stored in the information memory bank 128 of the image signal processing unit 110C in FIG. 21 is also the coefficient seed data w _i0 to w _i0 stored in the information memory bank 128 of the image signal processing unit 110 in FIG. Similar to w _i3 , it is generated by learning between a plurality of student signals and teacher signals (see FIG. 3).

FIG. 23 shows a configuration of a coefficient seed data generation device 150C for generating the coefficient seed data w _{i0 to} w _i3 stored in the information memory bank 128 of the image signal processing unit 110C described above. In FIG. 23, portions corresponding to those in FIG.

  The coefficient seed data generation device 150C includes an input terminal 151 to which a teacher signal ST corresponding to the image signal Vd is input, and an MPEG2 code that encodes the teacher signal ST to obtain an MPEG2 stream that is a student signal SS. Generator 152.

  The coefficient seed data generation device 150C also includes a DCT coefficient extraction circuit 231 that extracts a DCT coefficient as a frequency coefficient from the student signal SS obtained by the MPEG2 encoder 152, and a variable length encoding extracted by the extraction circuit 231. For example, it has a variable length decoding circuit 232 that performs variable length decoding on DCT coefficients that are Huffman encoded.

  The coefficient seed data generation device 150C also includes a quantization characteristic designation information extraction circuit 233 that extracts quantization characteristic designation information from the student signal SS obtained by the MPEG2 encoder 152, and a quantization that is extracted by the extraction circuit 233. And an inverse quantization circuit 234 that performs inverse quantization on the quantized DCT coefficient output from the variable length decoding circuit 232 based on the characteristic designation information.

  The value of the parameter p is input to the inverse quantization circuit 234. In the inverse quantization circuit 234, the signal is changed in accordance with the value of the parameter p in the process of inverse quantization signal processing, and the DCT coefficient related to the student signal corresponding to the value of the parameter p is obtained. It is done. In the present embodiment, the DCT coefficient (frequency coefficient) is changed.

For example, DCT coefficients related to a plurality of student signals can be obtained by changing the gain according to the value of the parameter p while setting the gains of all the AC coefficients of the DCT coefficients to the same value. Here, if the original AC coefficient is AC [m] and the gain is gain, the new AC coefficient newAC [m] is newAC [m] = AC [m] * gain. In this case, since the correlation between the AC coefficients can be maintained, the coefficient seed data w _{i0 to} w _i3 generated using the DCT coefficients related to the plurality of student signals obtained in this way are the above-described image signals. The resolution of the image obtained by Vd can be adjusted stepwise according to the value of the parameter p.

Further, for example, by changing the quantization step of the DC coefficient among the DCT coefficients according to the value of the parameter p, it is possible to obtain DCT coefficients related to a plurality of student signals. Here, assuming that the original DC coefficient is DC [m], the original quantization step is Qst, and the new quantization step is newQst, the new DC coefficient newDC [m] is newDC [m] = {DC [ m] * Qst / newQst}. {} Represents an integer operation. In this case, the variation of the DC coefficient can be manipulated, and the coefficient seed data w _{i0 to} w _i3 generated using the DCT coefficients related to the plurality of student signals obtained in this way are obtained by the image signal Vd described above. The block distortion of the image can be adjusted stepwise according to the value of the parameter p.

Further, for example, by reducing the gain of the AC coefficient of the DCT coefficient as the high frequency region is changed, and the degree of decrease in the high frequency gain of the AC coefficient is changed according to the value of the parameter p, DCT coefficients can be obtained. Here, if the original AC coefficient is AC [m] and the gain is gain [m], the new AC coefficient newAC [m] is newAC [m] = AC [m] * gain [m]. In this case, the correlation between the AC coefficients is broken stepwise, and the coefficient seed data w _{i0 to} w _i3 generated using the DCT coefficients related to the plurality of student signals thus obtained are The mosquito noise of the image obtained by the image signal Vd described above can be adjusted stepwise according to the value of the parameter p.

Further, for example, by adding random noise to the DCT coefficient and changing the level of the random noise according to the value of the parameter p, DCT coefficients related to a plurality of student signals can be obtained. When the original AC coefficient is AC [m] and the random noise is rdmN, the new AC coefficient newAC [m] is newAC [m] = AC [m] + rdmN. Similarly, if the original DC coefficient is DC [m] and the random noise is rdmN, the new DC coefficient newDC [m] is newDC [m] = DC [m] + rdmN. In this case, by switching the level of random noise in stages, the coefficient seed data w _{i0 to} w _i3 generated using the DCT coefficients related to the plurality of student signals obtained thereby are obtained by the image signal Vd described above. The noise suppression degree of the image to be adjusted can be adjusted stepwise according to the value of the parameter p.

  Further, the coefficient seed data generation device 150C selectively extracts a DCT coefficient corresponding to the target position in the image signal Vd from the DCT coefficient output from the inverse quantization circuit 234, and also extracts an image from a reference image signal Vref described later. There are first and second tap selection circuits 235 and 236 that selectively extract and output a plurality of pixel data located around the target position in the signal Vd. The first and second tap selection circuits 235 and 236 are configured in the same manner as the first and second tap selection circuits 225 and 226 of the image signal processing unit 110C described above, respectively.

  The coefficient seed data generation device 150 </ b> C includes a feature amount extraction unit 156 that extracts a feature amount from the class tap pixel data that is selectively extracted by the second tap selection circuit 236. The feature amount extraction unit 156 is configured in the same manner as the feature amount extraction unit 123 of the image signal processing unit 110C described above, and generates an ADRC code as first class information indicating a space class.

  The coefficient seed data generation device 150C extracts encoding control information (picture information PI, motion vector information MI) as additional information from the student signal SS obtained by the MPEG2 encoder 152, and an encoding control information extraction circuit 237. And an additional information class generation unit 157 that generates second class information indicating the additional information class based on the encoding control information extracted by the extraction circuit 237. The additional information class generation unit 157 is configured in the same manner as the additional information class generation unit 124 of the image signal processing unit 110C described above.

  Further, the coefficient seed data generation device 150C performs class classification based on the first class information (ADRC code) generated by the feature amount extraction unit 156 and the second class information generated by the additional information class generation unit 157. A class code generation unit 158 that generates a class code CL indicating the result of the above. The class code generation unit 158 is configured in the same manner as the class code generation unit 125 of the image signal processing unit 110C described above.

Further, the coefficient seed data generation device 150C has a delay circuit 159 for adjusting the time of the teacher signal ST supplied to the input terminal 151, and each attention position obtained from the teacher signal ST adjusted in time by the delay circuit 159. Pixel data y, prediction tap pixel data xi selectively extracted by the first tap selection circuit 235 in correspondence with the pixel data y of each target position, the value of the parameter p, and each target position A normal equation (see equation (9)) for obtaining coefficient seed data w _{i0 to} w _i3 is generated for each class from the class code CL generated by the class code generation unit 158 corresponding to each pixel data y. A normal equation generating unit 160 for performing

In this case, one piece of learning data is generated by a combination of one piece of pixel data y and pixel data xi of n prediction taps corresponding to the piece of pixel data y, but the inverse quantization circuit 234 depends on the value of the parameter p. The DCT coefficients related to a plurality of student signals are sequentially obtained from the, and learning data is generated between the teacher signal and each student signal. As a result, the normal equation generation unit 160 generates a normal equation in which many pieces of learning data having different values of the parameter p are registered, and the coefficient seed data w _{i0 to} w _i3 can be obtained.

Further, the coefficient seed data generation device 150C is supplied with data of normal equations generated for each class by the normal equation generation unit 160, solves the normal equations, and generates coefficient seed data w _{i0 to} w _{i3 for} each class. Is provided with a coefficient seed data determining unit 161 and a coefficient seed memory 162 for storing the obtained coefficient seed data. In the coefficient seed data determination unit 161, the normal equation is solved, for example, by a sweeping method, and coefficient coefficient data is obtained.

  In addition, the coefficient seed data generation device 150 </ b> C includes a prediction memory circuit 238. The prediction memory circuit 228 stores in the memory (not shown) the image signals of the I picture and the P picture of the student signal SS among the teacher signals ST time-adjusted by the delay circuit 159, and also stores these image signals. When the DCT coefficient related to the P picture or B picture is output from the inverse quantization circuit 234, the corresponding reference image signal Vref is generated and supplied to the first and second tap selection circuits 235 and 236. Incidentally, when the DCT coefficient related to the I picture is output from the inverse quantization circuit 234, the reference image signal Vref is not supplied to the first and second tap selection circuits 235 and 236.

  The coefficient seed data generation device 150C has a picture selection circuit 239. The picture selection circuit 239 supplies the image signals of the I picture and the P picture among the teacher signals ST adjusted in time by the delay circuit 159 to the prediction memory circuit 238 and stores them in the memory.

  The motion vector information MI extracted by the extraction circuit 237 is supplied to the prediction memory circuit 238, and the prediction memory circuit 238 performs motion compensation when generating the reference image signal Vref using the motion vector information MI. The picture information PI extracted by the extraction circuit 237 is supplied to the prediction memory circuit 238 and the picture selection circuit 239, and the prediction memory circuit 238 and the picture selection circuit 239 identify pictures based on the picture information PI. .

The operation of the coefficient seed data generation device 150C shown in FIG. 23 will be described.
A teacher signal ST corresponding to the image signal Vd is supplied to the input terminal 151. The teacher signal ST is encoded by the MPEG2 encoder 152 to generate an MPEG2 stream as the student signal SS. .

  The student signal SS obtained by the MPEG2 encoder 152 is supplied to the extraction circuit 231 to extract DCT coefficients as frequency coefficients. This DCT coefficient is variable-length encoded, and this DCT coefficient is supplied to the variable-length decoding circuit 232 and decoded. Then, the quantized DCT coefficient output from the variable length decoding circuit 232 is supplied to the inverse quantization circuit 234 and subjected to inverse quantization.

  The value of the parameter p is input to the inverse quantization circuit 234. In the inverse quantization circuit 234, the signal changes in accordance with the value of the parameter p in the process of inverse quantization signal processing. That is, in the present embodiment, as described above, the DCT coefficient is changed according to the value of the parameter p. Thereby, the DCT coefficient concerning the student signal corresponding to the value of the parameter p is obtained.

  Further, the second tap selection circuit 236 selectively extracts a plurality of pixel data (including DCT coefficients) of the class tap. That is, the DCT coefficient corresponding to the target position in the teacher signal ST is selectively extracted from the DCT coefficient output from the inverse quantization circuit 234, and the teacher signal is received from the reference image signal Vref supplied from the prediction memory circuit 228. A plurality of pixel data located around the target position in ST are selectively extracted.

  Thus, the pixel data of the class tap extracted by the second tap selection circuit 236 is supplied to the feature amount extraction unit 156. In the feature amount extraction unit 156, processing such as 1-bit ADRC is performed on the pixel data of the class tap, and an ADRC code serving as first class information indicating a space class is generated.

  Also, the encoding control information (picture information PI, motion vector information MI) extracted from the student signal SS by the extraction circuit 237 is supplied to the additional information class generation unit 157. The additional information class generation unit 157 generates second class information indicating the additional information class based on the encoding control information.

  Then, the first class information (ADRC code) generated by the feature amount extraction unit 156 and the second class information generated by the additional information class generation unit 157 are supplied to the class code generation unit 158. The class code generation unit 158 generates a class code CL indicating the result of class classification based on the first class information and the second class information.

  The first tap selection circuit 235 selectively extracts a plurality of pixel data (including DCT coefficients) of the prediction tap. That is, the DCT coefficient corresponding to the position of interest in the teacher signal ST is selectively extracted from the DCT coefficient output from the inverse quantization circuit 234, and the teacher signal from the reference image signal Vref supplied from the prediction memory circuit 238. A plurality of pixel data located around the target position in ST are selectively extracted.

The pixel data y of each target position obtained from the teacher signal ST time-adjusted by the delay circuit 159 and the first tap selection circuit 235 selectively extract the pixel data y corresponding to each pixel data y of each target position. The normal equation generation unit 160 uses the predicted tap pixel data xi, the value of the parameter p, and the class code CL generated by the class code generation unit 158 corresponding to the pixel data y of each target position. For each class, a normal equation (see equation (9)) for generating coefficient seed data w _{i0 to} w _i3 is generated.

Then, each normal equation is solved by the coefficient seed data decision section 161, coefficient seed data w _i0 to w _i3 of each class is determined, their coefficient seed data w _i0 to w _i3 is stored in the coefficient seed memory 162 .

As described above, in the coefficient seed data generation device 150C shown in FIG. 23, it is used in the estimation formula (see formula (1)) for each class stored in the information memory bank 128 of the image signal processing unit 110C in FIG. It is possible to generate coefficient seed data w _{i0 to} w _i3 which are coefficient data in a generation formula (see formula (2)) for determining the coefficient data Wi to be obtained.

  In the coefficient seed data generation device 150C shown in FIG. 23, the reference image signal Vref is generated based on the time-adjusted teacher signal ST. However, the present invention is not limited to this. As shown in FIG. 24, an arithmetic circuit 239A may be provided, and the reference image signal Vref may be generated based on the image signal generated by the arithmetic circuit 239A.

The arithmetic circuit 239A is supplied with the pixel data xi of the prediction tap selectively extracted by the first tap selection circuit 235, the value of the parameter p, and the class code CL generated by the class code generation unit 158. The Although detailed description is omitted, the arithmetic circuit 239A is configured in the same manner as the estimated prediction arithmetic circuit 127, the coefficient memory 126, the coefficient generation circuit 129, and the information memory bank 128 in the image signal processing unit 110C shown in FIG. Yes. Note that the information memory bank 128 stores coefficient seed data w _{i0 to} w _i3 generated in advance in a system different from the coefficient seed data generating apparatus 150C shown in FIG.

  In the image signal processing unit 110C of FIG. 21, the generation formula of the formula (2) is used to generate the coefficient data Wi (i = 1 to n), but it is expressed by a polynomial having a different order or another function. It is also possible to use this formula.

  Further, in the image signal processing unit 110 </ b> C of FIG. 21, in addition to the first class information (ADRC code) indicating the space class output from the feature amount extraction unit 123, the additional information generated by the additional information class generation unit 124. Based on the second class information indicating the class, the class code generation unit 125 generates the class code CL. However, the class code CL may be generated using only the first class information. However, by generating the class code CL in consideration of the second class information, finer classification can be performed, and the generation accuracy of the pixel data of the image signal Vd can be increased.

  Although detailed description is omitted, the processing in the image signal processing unit 110C in FIG. 21 is realized by software, for example, by the image signal processing device 300 shown in FIG. 9, as in the processing in the image signal processing unit 110 in FIG. It is also possible to do this (see the flowchart in FIG. 10). Further, although detailed description is omitted, the processing in the coefficient seed data generation device 150C of FIGS. 23 and 24 can also be realized by software in the same manner as the processing in the coefficient seed data generation device 150 of FIG. (Refer to the flowchart of FIG. 11).

23 and 24 correspond to the coefficient seed data generation device 150 shown in FIG. 4, but correspond to the coefficient seed data generation device 150A shown in FIG. You can also That is, the coefficient data Wi of each class corresponding to each value of the parameter p is obtained, and thereafter, the coefficient data Wi of each class corresponding to each value of the parameter p is used to calculate the coefficient seed data w _{i0 to} w _{i3 of} each class. Is what you want.

Further, in the image signal processing unit 110C illustrated in FIG. 21, the coefficient generation circuit 129 uses the coefficient of each class corresponding to the value of the parameter p based on the coefficient seed data w _{i0 to} w _i3 stored in the information memory bank 128. Data Wi is generated, and the coefficient data Wi of each class is stored in the coefficient memory 126 and used. However, like the image signal processing unit 110A of FIG. 13, the coefficient data Wi of each combination of the class and the parameter p is stored in the information memory bank 128A, and the coefficient data of each class corresponding to the value of the parameter p. Wi may be configured to be loaded from the information memory bank 128A into the coefficient memory 126 and used.

Next, still another embodiment of the present invention will be described.
FIG. 25 shows a configuration of a digital broadcast receiver 100D as still another embodiment. In FIG. 25, portions corresponding to those in FIGS. 1 and 21 are denoted by the same reference numerals.

  The digital broadcast receiver 100D includes a microcomputer, and includes a system controller 101 for controlling the operation of the entire system, and a remote control signal receiving circuit 102 that receives a remote control signal. The remote control signal receiving circuit 102 is connected to the system controller 101, receives a remote control signal RM output from the remote control transmitter 200 according to a user operation, and supplies an operation signal corresponding to the signal RM to the system controller 101. Is configured to do.

  Also, the digital broadcast receiver 100D is supplied with a receiving antenna 105 and a broadcast signal (RF modulated signal) captured by the receiving antenna 105, and performs a channel selection process, a demodulation process, an error correction process, and the like, and a predetermined program And a stream buffer 109 for temporarily storing the MPEG2 stream Vc output from the tuner unit 106. The tuner unit 106 obtains an MPEG2 stream as an encoded image signal.

  The digital broadcast receiver 100D also converts the MPEG2 stream Vc stored in the stream buffer 109 into a decoded image signal Vd that presents the user-desired image quality, and this image signal. And a display unit 111 that displays an image based on the image signal Vd output from the processing unit 110D.

The operation of the digital broadcast receiver 100D shown in FIG. 25 will be described.
The MPEG2 stream Vc output from the tuner unit 106 is supplied to the stream buffer 109 and temporarily stored.

  The MPEG2 stream Vc stored in the stream buffer 109 in this way is supplied to the image signal processing unit 110D and converted into a decoded image signal Vd that presents the user-desired image quality. In this image signal processing unit 110D, a new DCT coefficient for obtaining the image signal Vd is obtained from the DCT coefficient constituting the MPEG2 stream Vc, and the inverse DCT is performed on the new DCT coefficient to obtain the image signal Vd. It is done.

  The image signal Vd output from the image signal processing unit 110D is supplied to the display unit 111, and an image based on the image signal Vd is displayed on the screen of the display unit 111.

  The user can adjust the resolution, block distortion, mosquito noise, or noise suppression degree in the image displayed on the screen of the display unit 111 as described above by operating the remote control transmitter 200. As will be described later, in the image signal processing unit 110D, the DCT coefficient for obtaining the image signal Vd is calculated by the estimation formula. The coefficient data of the estimation formula is changed by the user's operation of the remote control transmitter 200. A value corresponding to the value of the parameter p is generated and used by a generation formula including the parameter p. As a result, the resolution, block distortion, mosquito noise, or noise suppression degree in the image based on the image signal Vd output from the image signal processing unit 110D corresponds to the changed value of the parameter p.

  Details of the image signal processing unit 110D will be described. In this image signal processing unit 110D, portions corresponding to those of the image signal processing unit 110 shown in FIG. 1 and the image signal processing unit 110C shown in FIG. 21 are denoted by the same reference numerals, and detailed description thereof is omitted.

  The image signal processing unit 110D extracts an extraction circuit 221 that extracts a DCT coefficient as a frequency coefficient from the MPEG2 stream Vc stored in the stream buffer 109, and variable-length coding extracted by the extraction circuit 221 such as Huffman coding And a variable length decoding circuit 222 that performs variable length decoding on the DCT coefficients.

  The image signal processing unit 110D also includes a quantization characteristic designation information extraction circuit 223 that extracts quantization characteristic designation information from the MPEG2 stream Vc stored in the stream buffer 109, and a quantization characteristic extracted by the extraction circuit 223. And an inverse quantization circuit 224 that performs inverse quantization on the quantized DCT coefficient output from the variable length decoding circuit 222 based on the designation information.

  Further, the image signal processing unit 110D first and second tap selection circuits that selectively extract and output the DCT coefficient corresponding to the position of interest in the image signal Vd from the DCT coefficient output from the inverse quantization circuit 224. 241,242.

  The first tap selection circuit 241 selectively extracts a plurality of DCT coefficients of a prediction tap used for prediction. FIG. 22A shows an example of a DCT coefficient of a prediction tap that is selectively extracted from the DCT coefficient output from the inverse quantization circuit 224 in the first tap selection circuit 241. In the present embodiment, a DCT block for obtaining a pixel block of interest of the image signal Vd is a target DCT block, and all DCT coefficients of the DCT block corresponding to this target DCT block, that is, 64 (8 × 8) pieces of DCT blocks. The DCT coefficient is extracted as the DCT coefficient of the prediction tap.

  The second tap selection circuit 242 selectively extracts a plurality of DCT coefficients of class taps used for class classification. FIG. 22B shows an example of DCT coefficients of class taps that are selectively extracted from the DCT coefficients output from the inverse quantization circuit 224 in the second tap selection circuit 242. In the present embodiment, 320 (8 × 8 × 5) DCTs of a total of five DCT blocks including a DCT block corresponding to the target DCT block and four DCT blocks adjacent in the vertical and horizontal directions. The coefficients are extracted as class tap DCT coefficients.

  The image signal processing unit 110 </ b> D includes a feature amount extraction unit 123 that extracts a feature amount from the class tap DCT coefficients selectively extracted by the second tap selection circuit 242. The feature amount extraction unit 123 generates an ADRC code as the class code CL by performing processing such as 1-bit ADRC on the DCT coefficient of the class tap. This class code CL indicates the class to which the target DCT block belongs.

  The image signal processing unit 110D has a coefficient memory 126. The coefficient memory 126 stores, for each class, a plurality of coefficient data Wi (i = 1 to n) used in an estimation formula used in an estimated prediction calculation circuit 127 described later. The coefficient data Wi is information for converting the DCT coefficient of the MPEG2 stream Vc into a new DCT coefficient for obtaining the image signal Vd.

  The class code CL output from the above-described feature amount extraction unit 123 is supplied to the coefficient memory 126 as read address information. From the coefficient memory 126, the coefficient data Wi of the estimation formula corresponding to the class code CL is read and supplied to the estimated prediction calculation circuit 127. In this case, coefficient data Wi corresponding to the number of DCT coefficients in the target DCT block is supplied to the estimated prediction calculation circuit 127 in order to calculate all DCT coefficients constituting the target DCT block.

  Further, the image signal processing unit 110D has an information memory bank 128. In the estimated prediction calculation circuit 127, the DCT coefficient y constituting the target DCT block is calculated from the pixel data xi of the prediction tap and the coefficient data Wi read from the coefficient memory 126 by the estimation formula of the above-described formula (1). The In the expression (1), n represents the number of prediction taps selected by the first tap selection circuit 241.

The coefficient data Wi of this estimation formula is generated by a generation formula including the parameter p, as shown in the above formula (2). Information in the memory bank 128, the coefficient seed data w _i0 to w _i3 is the coefficient data in the production equation is, for each class, stored. In this case, the coefficient seed data w _i0 to w _i3 of each class is comprised of a plurality of sets of coefficient seed data w _i0 to w _i3 corresponding to each DCT coefficient of each target DCT blocks.

Further, the image signal processing unit 110D uses the coefficient seed data w _{i0 to} w _{i3 of} each class and the value of the parameter p, and the coefficient of the estimation formula corresponding to the value of the parameter p for each class according to the equation (2). A coefficient generation circuit 129 that generates data Wi (i = 1 to n) is included. The coefficient generation circuit 129 is loaded with the above-described class seed coefficient data w _{i0 to} w _i3 from the information memory bank 128. The coefficient generation circuit 129 is supplied with the value of the parameter p from the system controller 101.

  The coefficient data Wi (i = 1 to n) of each class generated by the coefficient generation circuit 129 is stored in the coefficient memory 126 described above. The generation of coefficient data Wi for each class in the coefficient generation circuit 129 is performed, for example, in each vertical blanking period. Thereby, even if the value of the parameter p is changed by the operation of the remote control transmitter 200, the coefficient data Wi of each class stored in the coefficient memory 126 can be immediately changed to one corresponding to the value of the parameter p, The user can smoothly adjust the resolution, block distortion, mosquito noise, noise suppression degree, and the like.

Here, the coefficient seed data w _{i0 to} w _i3 stored in the information memory bank 128 are the same as the coefficient seed data w _{i0 to} w _i3 stored in the information memory bank 128 of the image signal processing unit 110C in FIG. In addition, the MPEG2 stream encoded in the same manner as the MPEG2 stream Vc stored in the stream buffer 109 described above is generated in advance based on the student signal obtained by processing according to the value of the parameter p. Thereby, the resolution, block distortion, mosquito noise, noise suppression degree, and the like of the image obtained from the image signal Vd can be adjusted stepwise according to the value of the parameter p.

  In addition, the image signal processing unit 110D stores, in a memory (not shown), an inverse DCT circuit 243 that performs inverse DCT on the DCT coefficient output from the estimated prediction calculation circuit 127, and image signals of I and P pictures. And a prediction memory circuit 244 that generates and outputs a corresponding reference image signal Vref when an image signal of a P picture or a B picture is output from the inverse DCT circuit 243 using these image signals. Yes.

  In addition, when the inverse DCT circuit 243 outputs an image signal of a P picture or a B picture, the image signal processing unit 110D adds the reference image signal Vref generated by the prediction memory circuit 244 to the image signal. have. Note that when the image signal of the I picture is output from the inverse DCT circuit 243, the reference image signal Vref is not supplied from the prediction memory circuit 244 to the adder circuit 245, and thus is output from the adder circuit 245 from the inverse DCT circuit 243. The image signal of the I picture is output as it is.

  In addition, the image signal processing unit 110D supplies the I-picture and P-picture image signals output from the adder circuit 245 to the prediction memory circuit 244 and stores them in the memory, and also stores each picture output from the adder circuit 245. A picture selection circuit 246 that rearranges the image signals in the correct order and outputs them as an image signal Vd.

  The image signal processing unit 110D has an extraction circuit 227 that extracts encoding control information, that is, picture information PI and motion vector information MI, from the MPEG2 stream stored in the stream buffer 109. The motion vector information MI extracted by the extraction circuit 227 is supplied to the prediction memory circuit 244, and the prediction memory circuit 244 performs motion compensation when generating the reference image signal Vref using the motion vector information MI. The picture information PI extracted by the extraction circuit 227 is supplied to the prediction memory circuit 244 and the picture selection circuit 246, and the prediction memory circuit 244 and the picture selection circuit 246 identify pictures based on the picture information PI. .

Next, the operation of the image signal processing unit 110D will be described.
The MPEG2 stream Vc stored in the stream buffer 109 is supplied to the extraction circuit 221 to extract DCT coefficients as frequency coefficients. This DCT coefficient is variable length encoded, and this DCT coefficient is supplied to the variable length decoding circuit 222 and decoded. Then, the quantized DCT coefficient output from the variable length decoding circuit 222 is supplied to the inverse quantization circuit 224 and subjected to inverse quantization.

  The second tap selection circuit 242 selectively extracts a plurality of class tap DCT coefficients from the DCT coefficients output from the inverse quantization circuit 224. The plurality of DCT coefficients of the class tap are supplied to the feature amount extraction unit 123. In the feature amount extraction unit 123, processing such as 1-bit ADRC is performed on the DCT coefficient of the class tap, and an ADRC code as the class code CL is generated. The class code CL indicates a class to which a DCT block for obtaining a target pixel block of the image signal Vd belongs.

The class code CL is supplied to the coefficient memory 126 as read address information. In the coefficient memory 126, for example, in each vertical blanking period, the coefficient generation circuit 129 corresponds to the value of the parameter p adjusted by the user, and a plurality of DCT coefficients corresponding to each DCT coefficient of the DCT block of interest for each class. Coefficient data Wi (i = 1 to n) of the estimation formula corresponding to each DCT coefficient of the DCT block of interest is obtained and stored using the set of coefficient seed data w _{i0 to} w _i3 (see formula (2)). ).

  As described above, the class code CL is supplied to the coefficient memory 126 as read address information, whereby the coefficient data Wi corresponding to the class code CL is read from the coefficient memory 126 and supplied to the estimated prediction calculation circuit 127. The In this case, coefficient data Wi corresponding to the number of DCT coefficients in the target DCT block is supplied to the estimated prediction calculation circuit 127 in order to calculate all DCT coefficients constituting the target DCT block.

  Further, the first tap selection circuit 225 selectively extracts a plurality of DCT coefficients of the prediction tap from the DCT coefficients output from the inverse quantization circuit 224. The plurality of DCT coefficients xi of the prediction tap are supplied to the estimated prediction calculation circuit 127. In the estimated prediction calculation circuit 127, the DCT block of interest is configured based on the estimation equation (see equation (1)) using the plurality of DCT coefficients xi of the prediction tap and the coefficient data Wi read from the coefficient memory 126. A plurality of DCT coefficients y to be obtained are respectively obtained.

  Further, the inverse DCT circuit 243 performs inverse DCT on the DCT coefficient output from the estimated prediction calculation circuit 127 to obtain an image signal of each picture. The image signal of each picture is supplied to the picture selection circuit 246 via the addition circuit 245. In this case, the reference image signal Vref output from the prediction memory circuit 244 is added by the addition circuit 245 to the image signals of the P picture and the B picture. The picture signals of the pictures are rearranged in the correct order by the picture selection circuit 229 and output as the picture signal Vd.

  Thus, in the image signal processing unit 110D, the coefficient data Wi (i = 1 to n) of the estimation formula corresponding to the value of the parameter p is used to calculate the DCT coefficient y for obtaining the image signal Vd. . Therefore, the user can arbitrarily adjust the resolution, block distortion, mosquito noise, or noise suppression degree of the image by the image signal Vd by changing the value of the parameter p.

The coefficient seed data w _{i0 to} w _i3 for generating the coefficient data Wi is an MPEG2 stream encoded in the same manner as the MPEG2 stream Vc stored in the stream buffer 109 described above according to the value of the parameter p. It is generated in advance based on the student signal obtained by processing, and only the specific quality of the image by the image signal Vd, for example, only the resolution, only the block distortion, only the mosquito noise, or only the noise suppression degree is stepwise. Can be adjusted.

The coefficient seed data w _{i0 to} w _i3 stored in the information memory bank 128 of the image signal processing unit 110D in FIG. 25 are also the coefficient seed data w _i0 to w _i0 stored in the information memory bank 128 of the image signal processing unit 110 in FIG. Similar to w _i3 , it is generated by learning between a plurality of student signals and teacher signals (see FIG. 3).

FIG. 26 shows a configuration of a coefficient seed data generation device 150D for generating the coefficient seed data w _{i0 to} w _i3 stored in the information memory bank 128 of the digital broadcast receiver 100D described above. In FIG. 26, portions corresponding to those in FIGS. 4 and 23 are denoted by the same reference numerals.

  The coefficient seed data generation apparatus 150D obtains an MPEG2 stream that is a student signal SS by encoding the image signal Vd ′ and an input terminal 151 to which the image signal Vd ′ corresponding to the image signal Vd is input. An MPEG2 encoder 152. The MPEG2 encoder 152 outputs DCT coefficients constituting the MPEG2 stream generated from the image signal Vd ′ as the teacher signal ST.

  The coefficient seed data generation device 150D also includes a DCT coefficient extraction circuit 231 that extracts a DCT coefficient as a frequency coefficient from the student signal SS obtained by the MPEG2 encoder 152, and a variable length encoding extracted by the extraction circuit 231. For example, it has a variable length decoding circuit 232 that performs variable length decoding on DCT coefficients that are Huffman encoded.

  The coefficient seed data generation device 150D also includes a quantization characteristic designation information extraction circuit 233 that extracts quantization characteristic designation information from the student signal SS obtained by the MPEG2 encoder 152, and a quantization that is extracted by the extraction circuit 233. And an inverse quantization circuit 234 that performs inverse quantization on the quantized DCT coefficient output from the variable length decoding circuit 232 based on the characteristic designation information.

The value of the parameter p is input to the inverse quantization circuit 234. In the inverse quantization circuit 234, the DCT coefficient is changed in accordance with the value of the parameter p in the process of inverse quantization signal processing in the same manner as described in the coefficient seed data generation device 150C in FIG. A DCT coefficient related to the student signal corresponding to the value of the parameter p is obtained. As a result, the coefficient seed data w _{i0 to} w _i3 generated using the DCT coefficients related to a plurality of student signals has the image resolution, block distortion, mosquito noise, noise suppression degree, and the like obtained from the image signal Vd described above. It can be adjusted stepwise according to the value of the parameter p.

  Further, the coefficient seed data generation device 150D first and second tap selection that selectively extracts and outputs the DCT coefficient corresponding to the target position in the teacher signal ST from the DCT coefficient output from the inverse quantization circuit 234. Circuits 251 and 252 are included. The first and second tap selection circuits 251 and 252 are configured in the same manner as the first and second tap selection circuits 241 and 242 of the image signal processing unit 110D described above, respectively.

  The first tap selection circuit 251 selectively extracts a plurality of DCT coefficients of a prediction tap used for prediction. FIG. 22A shows an example of the DCT coefficient of the prediction tap that is selectively extracted from the DCT coefficient output from the inverse quantization circuit 234 in the first tap selection circuit 251. In the present embodiment, all DCT coefficients of the DCT block corresponding to the target DCT block of the teacher signal ST, that is, 64 (8 × 8) DCT coefficients are extracted as DCT coefficients of the prediction tap.

  The second tap selection circuit 252 selectively extracts a plurality of DCT coefficients of class taps used for class classification. FIG. 22B shows an example of DCT coefficients of class taps that are selectively extracted from the DCT coefficients output from the inverse quantization circuit 234 in the second tap selection circuit 252. In the present embodiment, 320 (8 × 8 × 5) DCTs of a total of five DCT blocks including a DCT block corresponding to the target DCT block and four DCT blocks adjacent in the vertical and horizontal directions. The coefficients are extracted as class tap DCT coefficients.

  The coefficient seed data generation device 150 </ b> D includes a feature quantity extraction unit 156 that extracts a feature quantity from the class tap DCT coefficients selectively extracted by the second tap selection circuit 252. The feature quantity extraction unit 156 is configured in the same manner as the feature quantity extraction unit 123 of the image signal processing unit 110D described above. The feature amount extraction unit 156 generates an ADRC code as the class code CL by performing processing such as 1-bit ADRC on the DCT coefficient of the class tap. This class code CL indicates the class to which the target DCT block in the teacher signal ST belongs.

The coefficient seed data generation device 150D also includes a delay circuit 159 for adjusting the time of the teacher signal (DCT coefficient) ST output from the MPEG2 encoder 152, and the teacher signal ST adjusted in time by the delay circuit 159. The DCT coefficient y of each DCT block of interest, the DCT coefficient xi of the prediction tap selectively extracted by the first tap selection circuit 251 corresponding to each DCT block of interest, the value of the parameter p, A plurality of sets of coefficient seed data w _{i0 to} w _i3 respectively corresponding to each DCT coefficient of the DCT block of interest for each class from the class code CL generated by the feature quantity extraction unit 156 corresponding to each DCT block It has a normal equation generation unit 160 that generates a normal equation (see equation (9)) to obtain.

In this case, one learning data is generated by a combination of one DCT coefficient y and the corresponding DCT coefficient xi of n prediction taps. Depending on the value of the parameter p, the inverse quantization circuit 234 sequentially obtains DCT coefficients related to a plurality of student signals, and learning data is generated between the teacher signal and each student signal. As a result, the normal equation generation unit 160 generates a normal equation in which many pieces of learning data having different values of the parameter p are registered, and the coefficient seed data w _{i0 to} w _i3 can be obtained.

Further, the coefficient seed data generation device 150D is supplied with data of normal equations corresponding to the respective DCT coefficients of the DCT block of interest generated for each class by the normal equation generation unit 160, and solves the normal equations. For each class, a coefficient seed data determining unit 161 for obtaining a plurality of sets of coefficient seed data w _{i0 to} w _i3 corresponding to each DCT coefficient of the DCT block of interest, and a coefficient seed memory 162 for storing the obtained coefficient seed data And have. In the coefficient seed data determination unit 161, the normal equation is solved, for example, by a sweeping method, and coefficient coefficient data is obtained.

The operation of the coefficient seed data generation device 150D shown in FIG. 26 will be described.
An image signal Vd ′ corresponding to the image signal Vd is supplied to the input terminal 151, and the image signal Vd ′ is encoded by the MPEG2 encoder 152 to generate an MPEG2 stream as the student signal SS. The

  The student signal SS obtained by the MPEG2 encoder 152 is supplied to the extraction circuit 231 to extract DCT coefficients as frequency coefficients. This DCT coefficient is variable-length encoded, and this DCT coefficient is supplied to the variable-length decoding circuit 232 and decoded. Then, the quantized DCT coefficient output from the variable length decoding circuit 232 is supplied to the inverse quantization circuit 234 and subjected to inverse quantization.

  The value of the parameter p is input to the inverse quantization circuit 234. In the inverse quantization circuit 234, the signal changes in accordance with the value of the parameter p in the process of inverse quantization signal processing. That is, in the present embodiment, as described above, the DCT coefficient is changed according to the value of the parameter p. Thereby, the DCT coefficient concerning the student signal corresponding to the value of the parameter p is obtained.

  The second tap selection circuit 252 selectively extracts a plurality of class tap DCT coefficients from the DCT coefficients output from the inverse quantization circuit 234. The plurality of DCT coefficients of the class tap are supplied to the feature amount extraction unit 156. In the feature quantity extraction unit 156, the DCT coefficient of the class tap is subjected to processing such as 1-bit ADRC, and an ADRC code as the class code CL is generated. This class code CL indicates the class to which the target DCT block of the teacher signal ST belongs.

  The first tap selection circuit 251 selectively extracts a plurality of DCT coefficients of the prediction tap corresponding to the target DCT block of the teacher signal ST from the DCT coefficients output from the inverse quantization circuit 234.

Then, the DCT coefficient y of each target DCT block of the teacher signal ST time-adjusted by the delay circuit 159, and the prediction tap selectively extracted by the first tap selection circuit 251 corresponding to each target DCT block, respectively. Using the DCT coefficient xi, the value of the parameter p, and the class code CL generated by the feature amount extraction unit 156 corresponding to each target DCT block, the normal equation generation unit 160 uses the target DCT for each class. A normal equation (see equation (9)) for generating a plurality of sets of coefficient seed data w _{i0 to} w _i3 respectively corresponding to each DCT coefficient of the block is generated.

Then, each of the normal equations is solved by the coefficient seed data determination unit 161, and a plurality of sets of coefficient seed data w _{i0 to} w _i3 corresponding to each DCT coefficient of the DCT block of interest are obtained for each class. Data w _{i0 to} w _i3 are stored in the coefficient seed memory 162.

As described above, in the coefficient seed data generation device 150D shown in FIG. 26, it is used in the estimation formula (see formula (1)) for each class stored in the information memory bank 128 of the image signal processing unit 110D in FIG. It is possible to generate coefficient seed data w _{i0 to} w _i3 which are coefficient data in a generation formula (see formula (2)) for determining the coefficient data Wi to be obtained.

  In the image signal processing unit 110D of FIG. 25, the generation formula (2) is used to generate the coefficient data Wi (i = 1 to n). However, a polynomial having a different order or another function is used. It can also be realized with an expression.

  Although detailed description is omitted, the processing in the image signal processing unit 110D in FIG. 25 is realized by software, for example, by the image signal processing device 300 shown in FIG. It is also possible to do this (see the flowchart in FIG. 10). Further, although detailed description is omitted, the processing in the coefficient seed data generation device 150D in FIG. 26 can also be realized by software in the same manner as the processing in the coefficient seed data generation device 150 in FIG. 4 (FIG. 11). (Refer to the flowchart).

26 corresponds to the coefficient seed data generation device 150 shown in FIG. 4, but may also correspond to the coefficient seed data generation device 150A shown in FIG. it can. That is, the coefficient data Wi of each class corresponding to each value of the parameter p is obtained, and thereafter, the coefficient data Wi of each class corresponding to each value of the parameter p is used to calculate the coefficient seed data w _{i0 to} w _{i3 of} each class. Is what you want.

Also, in the image signal processing unit 110D shown in FIG. 25, the coefficient generation circuit 129 uses the coefficient of each class corresponding to the value of the parameter p based on the coefficient seed data w _{i0 to} w _i3 stored in the information memory bank 128. Data Wi is generated, and the coefficient data Wi of each class is stored in the coefficient memory 126 and used. However, like the image signal processing unit 110A of FIG. 13, the coefficient data Wi of each combination of the class and the parameter p is stored in the information memory bank 128A, and the coefficient data of each class corresponding to the value of the parameter p. Wi may be configured to be loaded from the information memory bank 128A into the coefficient memory 126 and used.

  In the above-described embodiment, the MPEG2 stream handling DCT is shown. However, the present invention can be applied to other handling of encoded digital information signals. Further, instead of DCT, encoding with other orthogonal transforms such as a wavelet transform and a discrete sine transform may be used.

  Also, the coefficient seed data generating device 150 shown in FIG. 4, the coefficient seed data generating device 150A shown in FIG. 8, the coefficient data generating device 170 shown in FIG. 14, the difference data generating device 210 shown in FIG. 18, and the coefficient seed shown in FIG. In both of the data generation device 150C and the coefficient seed data generation device 150D shown in FIG. 26, an MPEG2 encoder 152 is provided, and an MPEG2 stream is obtained by the MPEG2 encoder 152. It is good also as a structure which inputs.

  Moreover, although the case where the information signal is an image signal has been described in the above embodiment, the present invention is not limited to this. For example, the present invention can be similarly applied when the information signal is an audio signal.

  It is suitable for application to a digital broadcast receiver or the like.

It is a block diagram which shows the structure of the digital broadcast receiver as embodiment. It is a figure which shows an example of a prediction tap and a class tap. It is a figure for demonstrating an example of the production | generation method of coefficient seed data. It is a block diagram which shows the structural example of a coefficient seed data generation apparatus. It is a figure for demonstrating the change of the gain of AC coefficient at the time of adjusting mosquito noise in steps. It is a block diagram which shows the structure of an MPEG2 decoder. It is a figure for demonstrating the other example of the coefficient seed | species data generation method. It is a block diagram which shows the structural example of a coefficient seed data generation apparatus. It is a block diagram which shows the structural example of the image signal processing apparatus for implement | achieving by software. It is a flowchart which shows an image signal process. It is a flowchart which shows coefficient seed data generation processing (the 1). It is a flowchart which shows coefficient seed data generation processing (the 2). It is a block diagram which shows the structure of the digital broadcast receiver as other embodiment. It is a block diagram which shows the structural example of a coefficient data generation apparatus. It is a flowchart which shows an image signal process. It is a flowchart which shows coefficient data generation processing. It is a block diagram which shows the structure of the digital broadcast receiver as other embodiment. It is a block diagram which shows the structural example of a difference data generation apparatus. It is a flowchart which shows an image signal process. It is a flowchart which shows a difference data generation process. It is a block diagram which shows the structure of the digital broadcast receiver as another embodiment. It is a figure which shows an example of the prediction tap of a DCT coefficient, and a class tap. It is a block diagram which shows the structural example of a coefficient seed data generation apparatus. It is a block diagram which shows the structural example of a coefficient seed data generation apparatus. It is a block diagram which shows the structure of the digital broadcast receiver as another embodiment. It is a block diagram which shows the structural example of a coefficient seed data generation apparatus.

Explanation of symbols

  100, 100A, 100B, 100C, 100D ... Digital broadcast receiver, 101 ... System controller, 102 ... Remote control signal receiving circuit, 105 ... Receiving antenna, 106 ... Tuner unit, 107 ... MPEG2 decoder 108 ... buffer memory 109 ... stream buffer 110,110A, 110B, 110C ... image signal processing unit 111 ... display unit 121,225,241 ... First tap selection circuit, 122, 226, 242 ... second tap selection circuit, 122B ... tap selection circuit, 123 ... feature quantity extraction unit, 124 ... additional information class generation unit, 125 ... Class code generator, 126 ... Coefficient memory, 127 ... Estimated prediction arithmetic circuit, 128, 128A ... Information memory bank, 129... Coefficient generation circuit, 131... Storage table, 132... DCT circuit, 133, 136, changeover switch, 134, adder, 135, 243, reverse DCT Circuit 150 ... Coefficient seed data generator 151 ... Input terminal 152 ... MPEG2 encoder 153 ... MPEG2 decoder 154, 235, 251 ... First tap selection Circuit, 155, 236, 252 ... second tap selection circuit, 156 ... feature quantity extraction unit, 157 ... additional information class generation unit, 158 ... class code generation unit, 159 ... delay Circuit, 160, 163, 165 ... Normal equation generation unit, 161, 166 ... Coefficient seed data determination unit, 162 ... Coefficient memory, 164 ... Coefficient data determination unit, 167 ... Coefficient memory, 170 ... Coefficient data generation device, 171, 173 ... DCT circuit, 172, 174 ... Changeover switch, 175 ... Subtraction unit, 176 ... Accumulation control unit, 177 ... Accumulation Table, 210... Difference data generation device, 221, 231... DCT coefficient extraction circuit, 222, 232... Variable length decoding circuit, 223, 233. 234 ... Inverse quantization circuit, 227, 237 ... Coding control information extraction circuit, 228, 238, 244 ... Prediction memory circuit, 229, 239, 246 ... Picture selection circuit, 239A ... Arithmetic circuit, 245 ... adder circuit, 300 ... image signal processing apparatus

Claims (19)

A first information signal made up of a plurality of information data, which is generated by decoding an encoded digital information signal, is made up of a plurality of information data whose information quality is improved over that of the first information signal. An apparatus for generating coefficient seed data which is coefficient data in a generation formula for generating coefficient data used in an estimation formula used when converting into an information signal to be converted into a second information signal,
Parameter input means for inputting values of a plurality of types of parameters for determining in stages the quality of output information obtained by the student signal corresponding to the first information signal, corresponding to the parameters included in the generation formula; ,
The digital information signal obtained by encoding the teacher signal corresponding to the second information signal is decoded, and the signal is changed according to the parameter value input to the parameter input means in the decoding process. Decoding means for obtaining a student signal corresponding to the value of the parameter;
First data selection means for selecting a plurality of first information data located around the target position in the teacher signal from the student signal output from the decoding means;
Class detection means for detecting a level distribution pattern of the plurality of first information data selected by the first data selection means and detecting a class to which the information data of the target position belongs based on the level distribution pattern; ,
Second data selection means for selecting a plurality of second information data located around the target position in the teacher signal from the student signal output from the decoding means;
The coefficient seed data for each class using the class detected by the class detection means, the plurality of second information data selected by the second data selection means, and the information data of the attention position in the teacher signal And a coefficient seed data generating device. The computing means is
The coefficient seed data for each class using the class detected by the class detection means, the plurality of second information data selected by the second data selection means, and the information data of the attention position in the teacher signal A normal equation generation unit for generating a normal equation for obtaining
The coefficient seed data generation device according to claim 1, further comprising a coefficient seed data calculation unit that solves the normal equation and obtains the coefficient seed data for each class. The computing means is
Using the class detected by the class detecting unit, the plurality of second information data selected by the second data selecting unit, and the information data of the position of interest in the teacher signal, the class and the parameter input unit A first normal equation generation unit that generates a first normal equation for obtaining coefficient data used in the estimation equation for each combination of input parameter values;
A coefficient data calculation unit that solves the first normal equation and obtains coefficient data of the estimation formula for each combination;
A second normal equation generation unit that generates a second normal equation for obtaining the coefficient seed data for each class using the coefficient data for each combination obtained by the coefficient data calculation unit;
The coefficient seed data generation apparatus according to claim 1, further comprising a coefficient seed data calculation unit that solves the second normal equation and obtains coefficient seed data for each class. Additional information used when decoding the digital information signal is added to the digital information signal obtained by encoding the teacher signal.
2. The coefficient seed data generation device according to claim 1, wherein the class detection unit detects a class to which the information data of the target position belongs based on the additional information together with the plurality of first information data. . The encoding is encoding with orthogonal transformation,
2. The decoding unit according to claim 1, wherein at least one of a frequency coefficient and an inverse transform base used for performing inverse orthogonal transform is changed according to a value of the parameter in the decoding process. The coefficient seed data generation device described in 1. The coefficient seed data generation device according to claim 5, wherein the orthogonal transform is a discrete cosine transform. The coefficient seed data generation device according to claim 5, wherein the gain of all AC coefficients of the frequency coefficients is set to the same value, and the gain is changed according to the value of the parameter. 6. The coefficient seed data generation apparatus according to claim 5, wherein the frequency coefficient is quantized, and a quantization step of a DC coefficient among the frequency coefficients is changed according to the value of the parameter. 6. The AC coefficient gain of the frequency coefficient is reduced as the frequency increases, and the degree of decrease in the gain of the AC coefficient in the high frequency is changed according to the value of the parameter. Coefficient seed data generator. 6. The coefficient seed data generation apparatus according to claim 5, wherein random noise is added to at least the frequency coefficient or the inverse transform base, and the level of the random noise is changed according to the value of the parameter. A first information signal made up of a plurality of information data, which is generated by decoding an encoded digital information signal, is made up of a plurality of information data whose information quality is improved over that of the first information signal. An apparatus for generating coefficient data of an estimation equation used when converting into a second information signal,
Parameter input means for inputting values of a plurality of types of parameters for determining in stages the quality of the output information obtained by the student signal corresponding to the first information signal;
The digital information signal obtained by encoding the teacher signal corresponding to the second information signal is decoded, and the signal is changed according to the parameter value input to the parameter input means in the decoding process. Decoding means for obtaining a student signal corresponding to the value of the parameter;
First data selection means for selecting a plurality of first information data located around the target position in the teacher signal from the student signal output from the decoding means;
Class detection means for detecting a level distribution pattern of the plurality of first information data selected by the first data selection means and detecting a class to which the information data of the target position belongs based on the level distribution pattern; ,
Second data selection means for selecting a plurality of second information data located around the target position in the teacher signal from the student signal output from the decoding means;
Using the class detected by the class detecting unit, the plurality of second information data selected by the second data selecting unit, and the information data of the position of interest in the teacher signal, the values of the class and the parameter A coefficient data generating apparatus comprising: an arithmetic means for obtaining the coefficient data for each combination. The computing means is
Using the class detected by the class detecting unit, the plurality of second information data selected by the second data selecting unit, and the information data of the position of interest in the teacher signal, the values of the class and the parameter For each combination, a normal equation generation unit that generates a normal equation for obtaining coefficient data of the estimation equation,
The coefficient data generation device according to claim 11, further comprising: a coefficient data calculation unit that solves the normal equation and obtains the coefficient data for each combination. Additional information used when decoding the digital information signal is added to the digital information signal obtained by encoding the teacher signal.
The coefficient data generation device according to claim 11, wherein the class detection unit detects a class to which the information data of the target position belongs based on the additional information together with the plurality of first information data. The encoding is encoding with orthogonal transformation,
12. The decoding unit according to claim 11, wherein at least one of a frequency coefficient and an inverse transform base used when performing an inverse orthogonal transform is changed according to a value of the parameter in the decoding process. The coefficient data generation device described in 1. The coefficient data generation device according to claim 14, wherein the orthogonal transform is a discrete cosine transform. The coefficient data generation device according to claim 14, wherein the gain of all AC coefficients among the frequency coefficients is set to the same value, and the gain is changed according to the value of the parameter. The coefficient data generation apparatus according to claim 14, wherein the frequency coefficient is quantized, and a quantization step of a DC coefficient among the frequency coefficients is changed according to a value of the parameter. 15. The AC coefficient gain of the frequency coefficient is reduced as the frequency increases, and the degree of decrease in the high frequency gain of the AC coefficient is changed according to the value of the parameter. Coefficient data generator. The coefficient data generation device according to claim 14, wherein random noise is added to at least the frequency coefficient or the inverse transform base, and the level of the random noise is changed according to the value of the parameter.

Images