JP3567479B2 - Adaptive filter device and adaptive filter processing method - Google Patents

Description

[0001]
[Industrial applications]
The present invention relates to an adaptive filter device that classifies video signals including a luminance signal and a chrominance signal and performs adaptive filter processing according to the classification, and an adaptive filter processing method thereof.
[0002]
[Prior art]
For a color television signal used in a general color television or the like, an NTSC system, a PAL system, or the like is adopted.
[0003]
The color television signals of the NTSC system and the PAL system are an I (wide band) signal and a Q (narrow band) signal, which are a luminance signal (Y signal) and a chrominance signal, respectively. A color signal (C signal) consisting of a Y signal) is a decoded color television signal multiplexed by frequency interleaving. This decoded color television signal is called a composite signal.
[0004]
On the other hand, a color television signal separated into three signals of a luminance signal Y, a color difference signal I, and a Q signal (a luminance signal Y, a color difference signal RY, and a BY signal in the PAL system) in the NTSC system is Called component signal. To generate a component signal from the composite signal, the composite signal is separated into a luminance signal (Y signal) and a chrominance signal (C signal) (Y / C separation) using a band-pass filter, and further, a chrominance signal is generated. It can be obtained by converting C into an I signal and a Q signal.
[0005]
As a method of Y / C-separating the composite signal, instead of the band-pass filter, several band-limiting filters are used to adaptively select outputs from these band-limiting filters, There is also proposed a method of mixing the outputs. It is also known that the luminance signal band and the chrominance signal band of these composite signals are not constantly multiplexed with respect to position and time.
[0006]
[Problems to be solved by the invention]
By the way, in the conventional Y / C separation process for digital signal data by the adaptive filter process, the number of processes of the adaptive filter and the number of the mixing coefficients are limited to a few types and a few types. This kind of limitation cannot improve the accuracy of the Y / C separation processing even if the filter coefficients of the taps obtained by the learning processing are used.
[0007]
Therefore, the present invention has been made in view of the above situation, and has an adaptive filter device and an adaptive filter processing method capable of improving filter performance so as to obtain a filter characteristic adapted to an input signal. The purpose is to provide.
[0008]
[Means for Solving the Problems]
In order to solve the above-described problem, an adaptive filter device according to the present invention is a adaptive filter device that changes a filter characteristic according to an input signal, wherein the input signal is classified into one of a plurality of classes. Classifying means, filter mixing coefficient output means for outputting a filter mixing coefficient learned in advance according to the class classified by the class classification means, and a plurality of filters; A filter processing unit that outputs a plurality of output signals that have been subjected to different filtering processes; and a mixing unit that mixes each output signal from the filter processing unit in accordance with the filter mixing coefficient from the filter mixing coefficient output unit. The filtering means filters the video signal in which the luminance signal Y and the chrominance signal C are multiplexed. A Y / C separation filter for separating a luminance signal Y and a chrominance signal C, wherein the filter mixing coefficient is determined for each of the classes from the video signal, the luminance signal Y and the chrominance signal C before being multiplexed. It is characterized by being learned in advance.
[0009]
Here, since the classification unit blocks the input signal into two-dimensional data, the filter processing of the filter processing unit is performed on the blocked data.
[0010]
Also, a plurality of filter mixing coefficients output from the weighting coefficient ROM and output signals from the filter processing units are respectively associated one-to-one with each other, and block-divided data are respectively output to the filter mixing unit for the luminance signal Y and the color signal C. Are mixed in the filter mixing section.
[0011]
Further, the filter mixing section includes a filter mixing coefficient multiplication section that multiplies each output signal from the filter processing section by a filter mixing coefficient from the weight coefficient ROM, and an adder that sums the output signals from the filter mixing coefficient multiplication section. It consists of.
[0012]
Further, in order to solve the above-described problem, the adaptive filter processing method according to the present invention, in the adaptive filter processing method for changing a filter characteristic in accordance with an input signal, includes any one of a plurality of classes for an input signal. Classifying step, a filter mixing coefficient output step of outputting a plurality of filter mixing coefficients previously learned according to the class classified by the class classification step, and a plurality of filters for the input signal, A filtering step of outputting a plurality of output signals that have been subjected to different filtering processes; and a mixing step of mixing each output signal from the filtering step according to the filter mixing coefficient from the filtering mixing coefficient output step. In the filtering process, the video signal in which the luminance signal Y and the chrominance signal C are multiplexed is filtered. The filter mixing coefficient is separated into a degree signal Y and a chrominance signal C, and the filter mixing coefficient is learned in advance for each class from the video signal and the luminance signal Y and the chrominance signal C before being multiplexed. It is characterized by. Here, in the class classification step, since the input signal is divided into two-dimensional data blocks, the filtering processing step is performed on the blocked data. In addition, the plurality of filter mixing coefficients output in the filter mixing coefficient output step and the output signal output in the filter processing step correspond to each other in a one-to-one correspondence. The signal C is mixed in the mixing step in the filter mixing section. Further, in the mixing step, a filter mixing coefficient multiplication step of multiplying each output signal output in the filtering processing step by a plurality of filter mixing coefficients output in the filter mixing coefficient output step, and an output signal output in the filter mixing coefficient multiplication step And a summation step for obtaining a sum.
[0013]
[Action]
In the adaptive filter device according to the present invention, the class code that is classified into the corresponding class by the class classification unit and supplied is supplied as an address from the requantization processing unit to the filter mixing unit. The filter mixing section supplies the filter mixing coefficient supplied from the weight coefficient ROM to the filter mixing processing section of the filter mixing section in accordance with the output from the requantization processing section, and outputs each output from the filter processing section of the filter mixing section. The signal is supplied to a filter mixing processing unit, and the result obtained by multiplying the filter mixing coefficient by each output signal from the filter processing unit is mixed by the filter mixing processing unit. We are increasing.
[0014]
By using a Y / C separation filter for the filter processing unit, the luminance signal Y and the chrominance signal C in the video signal can be separated.
[0015]
Also, a plurality of filter mixing coefficients output from the weighting coefficient ROM and output signals from the filter processing units are respectively associated one-to-one with each other, and block-divided data are respectively output to the filter mixing unit for the luminance signal Y and the color signal C. , The filter characteristics for the luminance signal Y and the chrominance signal C are improved.
[0016]
The filter mixing unit also multiplies a plurality of filter mixing coefficients from the weight coefficient ROM by an output signal from the filter processing unit in a filter mixing and multiplying unit, and sums them in the mixing unit. The separation from the signal C is performed.
[0017]
Further, in the adaptive filter processing method according to the present invention, the input signal is classified into any of a plurality of classes in a class classification step, and a plurality of filter mixing according to the class classified in the filter mixing coefficient output step is performed. By outputting the coefficients, performing different filtering processes on the input signal in the filtering process, and mixing the multiplied output of the filter output signal and the filter mixing coefficient in the mixing process, a desired filter characteristic can be obtained. become.
In the filtering process, the video signal can be separated into a luminance signal Y and a chrominance signal C.
In addition, the plurality of filter mixing coefficients output in the filter mixing coefficient output step and the output signal output in the filter processing step correspond to each other in a one-to-one correspondence. By mixing in the mixing step in the filter mixing section for the signal C, the filter characteristics for the luminance signal Y and the chrominance signal C are improved.
In the filter mixing step, a plurality of filter mixing coefficients output in the filter mixing coefficient output step and the output signal output in the filter processing step are multiplied in the filter mixing multiplication step, and the sum is obtained in the mixing step. The signal Y and the color signal C are separated.
[0018]
【Example】
Hereinafter, an embodiment of an adaptive filter device and an adaptive filter processing method according to the present invention will be described with reference to the drawings.
Here, as a preferred example of an adaptive Y / C separation device to which the adaptive filter device of the present invention is applied, a frequency-multiplexed digital composite signal data including a luminance signal Y and a chrominance signal C as input signals (hereinafter referred to as signal data). ) Will be described with reference to, for example, block diagrams shown in FIGS. 1 to 3.
[0019]
As shown in FIG. 1, for example, the adaptive Y / C separating apparatus has a class classifying unit 1 for classifying digital composite signal data into one of a plurality of classes, and a block from the class classifying unit 1. A filter mixing unit 2 which receives the converted signal data B and the class code P of the class classification, performs a filter mixing process on the signal data B and outputs a luminance signal Y and a color signal C, and converts the color signal C into an I signal, It comprises an I / Q signal separation processing unit 3 for separating into Q signals.
[0020]
This class classification unit 1 is configured by a signal processing unit 1a, a correlation processing unit 1b, and a requantization processing unit 1c, for example, as shown in FIG. .
[0021]
The block processing section 1a converts the frequency-multiplexed, scan-converted, one-dimensional signal data of the NTSC system supplied via the input terminal 4 into two-dimensional signal data obtained by blocking, and further forms the two-dimensional signal data. The dimensional signal data is converted to be near-time data. Blocked two-dimensional signal data B is determined according to the processing contents of the subsequent correlation processing section 1b and filter mixing section 2. As described above, the blocking processing unit 1a supplies the signal data B to the correlation processing unit 1b and the filter mixing unit 2, respectively.
[0022]
The correlation processing unit 1b converts the block signal data B into correlation data indicating the strength of image correlation. The data representing the strength of the image correlation is obtained, for example, by a correlation determination method based on the absolute value of the data difference at the separation tap. The correlation processing unit 1b supplies the correlation data to the requantization processing unit 1c.
[0023]
The requantization processing unit 1c performs a bit reduction process to represent the number of bits of the supplied correlation data with a smaller number of bits. The bit-reduced correlation value directly corresponds to the class code P by the class classification. The requantization processing unit 1c outputs the class code P to the filter mixing unit 2.
[0024]
Next, the filter mixing section 2 shown in FIG. 1 is composed of a luminance signal Y filter mixing section 2a and a chrominance signal C filter mixing section 2b for outputting, for example, a luminance signal Y and a color signal C, respectively. I have. Here, the luminance signal Y filter mixing section 2a and the chrominance signal C filter mixing section 2b both have a basic configuration, for example, as shown in FIG. And a unit 23.
[0025]
The weight coefficient ROM 21 is supplied with the class code P output from the requantization processing unit 1c as address data, and outputs the weight coefficient stored at the corresponding address to the filter mixing processing unit 23 as a filter mixing coefficient.
[0026]
In addition, the filter processing unit 22 includes a plurality of filter processing circuits 22a to 22n. Signal data B is input to each of the filter circuits 22a to 22n. Each of the filter circuits 22a to 22n supplies the result of the product-sum operation of various prepared filter coefficients for separation and the supplied signal data B to the filter mixing processing unit 23.
[0027]
The filter mixing processing unit 23 includes filter mixing coefficient multiplying units 23a to 23n that multiply output signals from the filter circuits 22a to 22n by filter mixing coefficients from the weight coefficient ROM 21, and output signals from the filter mixing coefficient multiplying units 23a to 23n. And a mixing unit 23M for mixing signals.
[0028]
The filter mixing processing unit 23 multiplies each output signal from the filter circuits 22a to 22n by the filter mixing coefficient multiplying units 23a to 23n, applies the filter mixing coefficient to the output signals, and mixes the output signals by a mixing unit 23M including an adder. To produce a separate output. The luminance signal Y and the chrominance signal C are separated from the digital composite signal data by configuring the luminance signal Y filter mixing unit 2a and the chrominance signal C filter mixing unit 2b of the filter mixing unit 2 in this manner. .
[0029]
The color signal C filter mixing section 2b of the filter mixing section 2 outputs the separated color signal C to the I / Q signal separation processing section 3. In the I / Q signal separation processing unit 3, the signal is separated into an I signal and a Q signal. The I / Q signal separation processing unit 3 outputs I and Q signals based on the separated color signal C from output terminals 5 and 6. The luminance signal Y filter mixer 2 a of the filter mixer 2 outputs the separated luminance signal Y from the output terminal 7.
[0030]
With this configuration, it is possible to mix a large number of separation taps, further improve the separation of the luminance signal Y and the chrominance signal C, and improve the image quality.
[0031]
Next, the operation of the adaptive Y / C separation device will be described with reference to the flowchart of FIG. 4, the taps of correlation determination and class generation in FIG. 5, and the schematic diagram of the separation tap relationship in the filter processing of FIG.
Start adaptive Y / C separation. In step S10, one-dimensional signal data is divided into two-dimensional signal data. For example, as shown in FIG. 5, the signal data block includes five lines ((V−2) to (V + 2)) in the vertical direction in field 0 (# 0) and nine pixels ((H−4) to (H−4)) in the horizontal direction. The area of (H + 4)) is regarded as one block.
[0032]
Further, in order to correctly perform the adaptive Y / C separation processing, the separation value Y0 at the tap of the target pixel (H0, Y0) and the taps (H-4, V0), (H + 4, V0), separated values of taps (H0, V-2) and (H0, V + 2) having the same phase in the vertical direction are Y1 to Y4. The separation value Y0 at the tap of the target pixel (H0, Y0) and the tap whose color signal phase is inverted by 180 ° are (H−2, V0), (H + 2, VO), and the separation values Y5, Y6. The taps are (H−2, V−1), (H + 2, V + 1), (H−2, V + 1), (H + 2, V−1) on the lines adjacent to the target pixel (H0, Y0), respectively. Separation values Y7 to Y10 are set. Here, the dashed circles and squares indicate the tap positions displayed to indicate the relationship with the taps described above.
[0033]
In step S11, a correlation process is performed using the above-described separation values Y0 to Y10. In the correlation processing, for example, | Y0-Y1 |, | Y0-Y2 |, | Y0-Y3 |, | Y0-Y4 |, | Y5-Y6 |, | Y7-Y8 |, and | Y9-Y10 | Is the absolute value of the difference.
[0034]
In step S12, a requantization process is performed. In this requantization process, in order to perform bit reduction on the absolute value of each difference indicating the strength of the correlation, the set threshold value TH is compared with the absolute value of the difference. When | Y0−Y1 | is smaller than the threshold value TH, bit 0 is set to level “1”; otherwise, bit 0 is set to level “0”. When | Y0-Y2 |, | Y0-Y3 |, | Y0-Y4 |, | Y5-Y6 |, | Y7-Y8 | and | Y9-Y10 | are also smaller than the threshold value TH, bits 1 to 6 are respectively The level is set to "1", and in other cases, the bits 1 to 6 are set to the level "0". By this comparison, the absolute value of each difference is binarized and represented by one bit. The seven types of one-bit correlation amounts binarized represent a class code. Therefore, the class code P is
(Bit 6, bit 5, bit 4, bit 3, bit 2, bit 1, bit 0). Note that a level “1” indicates that there is a correlation, and a level “0” indicates no correlation.
[0035]
In step S13, a filter mixing process is performed on the luminance signal Y.
Next, in step S14, a filter mixing process for the color signal C is performed.
In the filter mixing process performed in step S13 or step S14, for example, as shown in FIG. 6, the tap used in the filter mixing process is a tap indicated by a symbol (○) having the same phase as the data X0 of the target pixel (H0, V0). (H-2, V0), (H + 2, V0), (H0, V-1), (H0, V + 1) correspond to data X1 to X4, and tap (H-1) indicated by other symbols (□). , V0), (H + 1, V0), (H-1, V-1), (H + 1, V + 1), (H-1, V + 1), and (H + 1, V-1) respectively correspond to data X5 to X10. .
[0036]
In the case of the filter mixing process of the luminance signal in step S13, for example, 0.5 is set to the coefficient of the filter processing unit. Here, the separation types a0y to a6y are the same seven types as in the case of the correlation determination described above:
a0y = 0.5 * X0 + 0.5 * X1
a1y = 0.5 * X0 + 0.5 * X2
a2y = 0.5 * X0 + 0.5 * X3
a3y = 0.5 * X0 + 0.5 * X4
a4y = 0.5 * X5 + 0.5 * X6
a5y = 0.5 * X7 + 0.5 * X8
a6y = 0.5 * X9 + 0.5 * X10
Is obtained from two taps. The luminance signal Y is separated from the signal data by performing a product-sum operation in which these seven types of filter processing outputs are multiplied by k0y to k6y output from the weighting coefficient ROM and added. The product-sum operation is
Y = k0y * a0y + k1y * a1y + k2y * a2y + k3y * a3y + k4y * a4y + k5y * a5y + k6y * a6y
It is.
[0037]
In the case of the color signal filter mixing processing in step S14, for example, 0.5 and 0.25 are set as the coefficients of the filter processing unit. Here, the types of separation a0c to a6c are the same seven types as in the case of the correlation determination described above:
a0c = 0.5 * X0-0.5 * X1
a1c = 0.5 * X0−0.5 * X2
a2c = 0.5 * X0−0.5 * X3
a3c = 0.5 * X0−0.5 * X4
a4c = 0.5 * X0-0.25 * X5-0.25 * X6
a5c = 0.5 * X0-0.25 * X7-0.25 * X8
a6c = 0.5 * X0-0.25 * X9-0.25 * X10
Is obtained from two taps or three taps. The color signal C is separated from the signal data by performing a product-sum operation in which these seven types of filter processing outputs are multiplied by k0c to k6c output from the weighting coefficient ROM and added. The product-sum operation is
C = k0c * a0c + k1c * a1c + k2c * a2c + k3c * a3c + k4c * a4c + k5c * a5c + k6c * a6c
It is.
[0038]
Next, in step S15, the color signal obtained in step S14 is further separated into an I signal and a Q signal. This series of processing is performed, and the adaptive Y / C separation ends.
[0039]
Further, a method of learning the filter mixing coefficient will be briefly described with reference to FIG.
At the start of the learning process, signal data is input in step S20.
In step S21, the signal data supplied in step S20 is subjected to exactly the same class classification as in the respective steps from step S10 to step S12 performed by the above-described adaptive Y / C separation device.
[0040]
In step S22, outputs of various Y / C separation filters are calculated for the input signal data. Also in this step S22, the same processing as the filtering processing in steps S13 and S14 described above is performed.
[0041]
Thereafter, in step S23, a normal equation is generated. This normal equation makes the filter mixing coefficient unknown and uses, for example, a luminance signal Y or color signal C immediately before frequency multiplexing and a composite signal according to the NTSC system after multiplexing used for signal data, and output signals of various Y / C separation filters. Is calculated sequentially using
[0042]
In step S24, for example, it is checked whether the input of the signal data for one screen has been completed. If the input of the signal data has not been completed yet, the process shifts to step S20 to repeat the above steps. When the input of the signal data has been completed, the process proceeds to step S25.
[0043]
In step S25, the mixing equation of the filter is determined by solving the normal equation generated in step S24 according to the solution of the system of linear equations.
[0044]
In step S26, the filter mixing coefficient obtained in step S25 is stored in the weight coefficient ROM.
[0045]
By learning the filter mixing coefficient in accordance with such a procedure, the optimum filter mixing coefficient can be used.
[0046]
By increasing the number of filter processes and the mixing coefficients in the related art using the above configuration, it is possible to improve the accuracy of the filter processes and improve the image quality.
[0047]
The image quality can be improved by separating the luminance signal Y and the color signal C in the video signal by using a Y / C separation filter in the filter processing unit.
[0048]
Also, a plurality of filter mixing coefficients output from the weighting coefficient ROM and output signals from the filter processing units are respectively associated one-to-one with each other, and block-divided data are respectively output to the filter mixing unit for the luminance signal Y and the color signal C. Even if they are mixed by the filter mixing section, the luminance signal Y and the chrominance signal C can be sufficiently separated to improve the accuracy of the filter processing and improve the image quality.
[0049]
The filter mixing unit multiplies a plurality of filter mixing coefficients from the weight coefficient ROM by an output signal from the filter processing unit by a filter mixing and multiplying unit, and sums them by the mixing unit to obtain a sufficiently appropriate luminance signal Y and color signal C. And the image quality can be improved.
[0050]
Also, the input signal is classified into any of a plurality of classes in a class classification process using an adaptive filter processing method, and a plurality of filter mixing coefficients corresponding to the class classified in the filter mixing coefficient output process are output. Even if the input signal is subjected to different filtering processes in the filtering process and the sum of the multiplication results of the output signals filtered and the filter mixing coefficients is obtained in the mixing process, the desired filter is obtained. A signal having characteristics can be output, and image quality can be improved.
[0051]
【The invention's effect】
In the adaptive filter device according to the present invention, the class code that is classified and supplied to the corresponding class by the class classification unit is supplied as an address from the requantization processing unit to the weight coefficient ROM of the filter mixing unit. The filter mixing coefficient learned in advance for each class from the video signal, the luminance signal Y and the color signal C before being multiplexed according to the class code is supplied to the filter mixing processing unit of the filter mixing unit. The filter processing unit of the filter mixing unit including a plurality of filters supplies each output signal obtained by performing a plurality of filtering processes on one input signal to the filter mixing processing unit. In the filter mixing processing unit, the result of multiplying the filter mixing coefficient by each output signal from the filter processing unit is mixed, and the number of the conventional filtering processing and the mixing coefficient are increased, thereby improving the accuracy of the filtering processing and improving the image quality. Can be improved.
[0052]
Also, a plurality of filter mixing coefficients output from the weighting coefficient ROM and output signals from the filter processing units are respectively associated one-to-one with each other, and block-divided data are respectively output to the filter mixing unit for the luminance signal Y and the color signal C. Even if they are mixed by the filter mixing section, the luminance signal Y and the chrominance signal C can be sufficiently separated to improve the accuracy of the filter processing and improve the image quality.
[0053]
The filter mixing unit multiplies a plurality of filter mixing coefficients from the weight coefficient ROM by an output signal from the filter processing unit by a filter mixing and multiplying unit, and sums them by the mixing unit to obtain a sufficiently appropriate luminance signal Y and color signal C. And the image quality can be improved.
[0054]
Also, the input signal is classified into any of a plurality of classes in a class classification process using an adaptive filter processing method, and a plurality of filter mixing coefficients corresponding to the class classified in the filter mixing coefficient output process are output. Even if the input signal is subjected to different filtering processes in the filtering process and the sum of the multiplication results of the output signals filtered and the filter mixing coefficients is obtained in the mixing process, the desired filter is obtained. A signal having characteristics can be output, and image quality can be improved.
In addition, the plurality of filter mixing coefficients output in the filter mixing coefficient output step and the output signal output in the filter processing step correspond to each other in a one-to-one correspondence. Even if the signals are mixed in the mixing step in the filter mixing unit for the signal C, the luminance signal Y and the chrominance signal C can be sufficiently separated to improve the accuracy of the filtering process and improve the image quality.
In the filter mixing step, a plurality of filter mixing coefficients output in the filter mixing coefficient output step and the output signal output in the filter processing step are multiplied in the filter mixing multiplication step, and the sum is calculated in the mixing step to obtain a sufficiently appropriate luminance signal Y. Separation from the color signal C can be performed to improve image quality.
[Brief description of the drawings]
FIG. 1 is a schematic block diagram of an adaptive Y / C separation device to which an adaptive filter device according to the present invention is applied.
FIG. 2 is a block diagram showing a configuration of a class classification unit of the adaptive Y / C separation device.
FIG. 3 is a block circuit diagram showing a configuration of a filter mixing unit of the adaptive Y / C separation device.
FIG. 4 is a flowchart illustrating an operation in the adaptive Y / C separation device.
FIG. 5 is a schematic diagram illustrating an arrangement of pixels related to a tap for correlation determination and class generation in the adaptive Y / C separation device.
FIG. 6 is a schematic diagram illustrating an arrangement of pixels related to a separation tap used in a filter process in the adaptive Y / C separation device.
FIG. 7 is a flowchart illustrating a learning process of a filter coefficient in the adaptive filter processing method according to the present invention.
[Explanation of symbols]
1 Classification section
2 Filter mixing section
3 I, Q signal separation processing unit
4 Input terminal
5, 6, 7 output terminals
1a Blocking processing unit
1b Correlation processing unit
1c Requantization processing unit
2a Filter Mixer for Luminance Signal Y
2b Color signal C filter mixing unit
21 Weight coefficient ROM
22 Filter processing section
22a to 22n filter circuits
23 Filter mixing section
23a to 23n filter mixing coefficient multiplier
23M mixing unit (adder)

Claims (6)

In an adaptive filter device that changes a filter characteristic according to an input signal,
Class classification means for classifying the input signal into any of a plurality of classes;
A filter combining coefficient output means for outputting previously learned filter combining coefficient according to the classified class by the class classification means,
Filter processing means comprising a plurality of filters, and outputting a plurality of output signals obtained by performing different filter processing on the input signal by the plurality of filters,
The output signals from the filtering means and a mixing means for mixing according to the filter combining coefficients from the filter combining coefficient output means,
The filter processing means is a Y / C separation filter for filtering a video signal in which a luminance signal Y and a chrominance signal C are multiplexed to separate the video signal into a luminance signal Y and a chrominance signal C,
The adaptive filter device, wherein the filter mixing coefficient is learned in advance for each class from the video signal, a luminance signal Y and a chrominance signal C before being multiplexed. The plurality of filter mixing coefficients output by the filter mixing coefficient output means and the output signals from the filter processing means correspond to each other on a one-to-one basis. 2. The adaptive filter device according to claim 1, wherein the filter is mixed by a filter mixing means for the filter. The mixing means is a filter mixing coefficient multiplying means for multiplying each output signal from the filter processing means by a filter mixing coefficient from the filter mixing coefficient output means,
2. An adaptive filter device according to claim 1, further comprising an adding means for calculating a sum of output signals from said filter mixing coefficient multiplying means. In an adaptive filter processing method for changing a filter characteristic according to an input signal,
A classifying step of classifying the input signal into one of a plurality of classes;
A filter combining coefficient output step of outputting a plurality of filter mixing coefficients previously learned in accordance with the classes classified by the classification step,
A filtering process of outputting a plurality of output signals obtained by performing different filtering processes on the input signal by a plurality of filters,
A mixing step of mixing each output signal from the filtering step according to the filter mixing coefficient from the filter mixing coefficient output step,
In the filtering process, the video signal in which the luminance signal Y and the color signal C are multiplexed is filtered to be separated into the luminance signal Y and the color signal C,
The adaptive filter processing method, wherein the filter mixing coefficient is learned in advance for each class from the video signal, a luminance signal Y and a chrominance signal C before being multiplexed. The plurality of filter mixing coefficients output in the filter mixing coefficient output step and the output signal output in the filter processing step correspond to each other in a one-to-one correspondence. 5. The adaptive filter processing method according to claim 4, wherein mixing is performed in a mixing step by a C filter mixing unit. In the mixing step, a filter mixing coefficient multiplying step of multiplying each output signal output in the filter processing step by a plurality of filter mixing coefficients output in the filter mixing coefficient output step,
5. The adaptive filter processing method according to claim 4, further comprising an addition step of calculating a sum of output signals output in the filter mixing coefficient multiplication step.

Images