JP2000115803A - Image signal processing unit and its method, learning device and its method, and distributing medium - Google Patents

Abstract

PROBLEM TO BE SOLVED: To reduce the scale of a processing unit. SOLUTION: A control circuit 14 supplies a CS signal to select a Y±I memory 13A, when an NTSC signal of a notice pixel is a Y-I signal and a Y+I signal and to select a Y±Q memory 13B when the NTSC signal of the noted pixel is a Y-Q signal and a Y+Q signal to a prediction coefficient memory section 13. A code inversion circuit 17 inverts the code of each of U, V signals inputted from an arithmetic circuit 16 (executes code inversion processing), when the notice pixel is the Y-I signal or the Y-Q signal, based on the CS signal inputted from the control circuit 14 and does not apply the code sign inversion processing to the U, V signals, when the notice pixel is the Y+I signal or the Y+Q signal and output them.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image signal processing apparatus and method, a learning apparatus and method, and a providing medium, and more particularly to, for example, NTSC (National Television System).
The present invention relates to an image signal processing apparatus and method for converting a composite signal such as a television signal of a Committee) system into a component signal such as a YUV signal, a learning apparatus and method, and a providing medium.

[0002]

2. Description of the Related Art For example, an NTSC television signal includes a luminance signal (Y) and a chrominance signal (C) (I component and Q component).
Are multiplexed by quadrature modulation, that is, a signal having four kinds of phases of a YI signal, a YQ signal, a Y + I signal, and a Y + Q signal. When an image is displayed, it is necessary to separate a luminance signal and a chrominance signal from the television signal (Y / C separation), and further perform matrix conversion to a component signal such as a YUV signal.

[0003]

SUMMARY OF THE INVENTION However, the conventional
In the Y / C separation, for example, a luminance signal or a color signal of a target pixel is calculated using a composite signal of a pixel (including the target pixel) located in the vicinity of the target pixel and a fixed filter coefficient. Was asked by. For this reason, if the coefficient is not appropriate for the pixel of interest, dot interference, cross color, and the like occur, and there is a problem that image quality is degraded.

Further, in the conventional Y / C separation, it is necessary to set the above-described four types of phases, filter coefficients corresponding to each of the four types, and the scale of a circuit for performing an operation using the filter coefficients becomes large. there were.

The present invention has been made in view of such circumstances, and it is an object of the present invention to reduce the scale of a device by using the same filter coefficient in different phases.

[0006]

According to a first aspect of the present invention, there is provided an image signal processing apparatus, comprising: detecting means for detecting four types of phases of a composite signal of a target pixel; and a prediction coefficient corresponding to two types of phases of the composite signal. Means for calculating a component signal of the pixel of interest using a prediction coefficient stored in the storage means corresponding to the phase of the composite signal of the pixel of interest detected by the detection means; And correcting means for correcting the component signal of the pixel of interest calculated by the calculating means in accordance with the phase of the composite signal of the pixel of interest detected by the control means.

According to a sixth aspect of the present invention, there is provided an image signal processing method comprising: a detecting step of detecting four types of phases of a composite signal of a target pixel; and a storing step of storing prediction coefficients corresponding to two types of phases of a composite signal. Calculating the component signal of the pixel of interest using the one corresponding to the phase of the composite signal of the pixel of interest detected in the detection step among the prediction coefficients stored in the storage step; A correction step of correcting the component signal of the pixel of interest calculated in the calculation step in accordance with the phase of the composite signal.

According to a seventh aspect of the present invention, in the providing medium, a detecting step of detecting four types of phases of the composite signal of the pixel of interest, a storing step of storing prediction coefficients corresponding to two types of phases of the composite signal, and a storing step A calculating step of calculating a component signal of the pixel of interest using a phase coefficient of the composite signal of the pixel of interest detected in the detection step among the prediction coefficients stored in the step; and a composite signal of the pixel of interest detected in the detection step A computer-readable program that causes the image signal processing device to execute a process including a correction step of correcting the component signal of the pixel of interest calculated in the calculation step in accordance with the phase of

The learning device according to the present invention has four types of conversion means for converting a learning component signal into a learning composite signal, and a learning composite signal for a pixel of interest converted by the conversion means. Detecting means for detecting the phase; correcting means for correcting the component signal for learning corresponding to the phase of the composite signal for learning of the pixel of interest detected by the detecting means; and a composite signal for learning and a prediction coefficient. A component signal calculated using
A calculating means for performing a calculation for obtaining a prediction coefficient for reducing an error with a learning component signal corrected by the correcting means.

According to a twelfth aspect of the present invention, there are provided a learning method, comprising: a conversion step of converting a learning component signal into a learning composite signal; and a learning composite signal of the pixel of interest converted in the conversion step. A detection step of detecting a phase, a correction step of correcting a learning component signal corresponding to the phase of the learning composite signal of the pixel of interest detected in the detection step, and a learning composite signal and a prediction coefficient. The method is characterized by including an operation step of performing an operation for obtaining a prediction coefficient for reducing an error between the component signal calculated using the above and the learning component signal corrected in the correction step.

According to a thirteenth aspect of the present invention, there are provided four types of media: a conversion step of converting a component signal for learning into a composite signal for learning, and a composite signal for learning of a pixel of interest converted in the conversion step. A detection step of detecting a phase, a correction step of correcting a learning component signal corresponding to the phase of the learning composite signal of the pixel of interest detected in the detection step, and a learning composite signal and a prediction coefficient. A computer-readable program that causes a learning device to execute a process including an operation step of performing an operation for obtaining a prediction coefficient that reduces an error between a component signal calculated using the above and a learning component signal corrected in the correction step. A unique program is provided.

In the image signal processing apparatus according to the first aspect, the image signal processing method according to the sixth aspect, and the providing medium according to the seventh aspect, four phases of the composite signal of the target pixel are detected. The prediction coefficients corresponding to the two types of phases of the composite signal are stored, and the component signal of the pixel of interest is calculated using one of the stored prediction coefficients corresponding to the phase of the composite signal of the detected pixel of interest. Is done. Further, the calculated component signal of the target pixel is corrected in accordance with the detected phase of the composite signal of the target pixel.

[0013] In the learning apparatus according to the eighth aspect, the learning method according to the twelfth aspect, and the providing medium according to the thirteenth aspect, the component signal for learning is converted into a composite signal for learning, and is converted. The four types of phases of the composite signal for learning of the pixel of interest are detected, and the component signal for learning is corrected in accordance with the detected phase of the composite signal for learning of the pixel of interest, and the composite composite for learning is obtained. An operation is performed to obtain a prediction coefficient that reduces an error between the component signal calculated using the signal and the prediction coefficient and the corrected learning component signal.

[0014]

DESCRIPTION OF THE PREFERRED EMBODIMENTS Embodiments of the present invention will be described below, but before that, the correspondence between each means of the invention described in the claims and the following embodiments will be clarified. For this reason, the features of the present invention are described as follows by adding the corresponding embodiment (however, an example) in parentheses after each means. However, of course, this description does not mean that each means is limited to those described above.

The image signal processing apparatus according to the first aspect of the present invention includes a detecting means (for example, the control circuit 14 in FIG. 2) for detecting four types of phases of the composite signal of the target pixel, and two types of phases of the composite signal. A storage unit for storing the corresponding prediction coefficient (for example, the prediction coefficient memory unit 13 in FIG. 2), and a prediction coefficient stored in the storage unit corresponding to the phase of the composite signal of the pixel of interest detected by the detection unit. A calculating means (for example, the calculating circuit 16 in FIG. 2) for calculating the component signal of the target pixel by using the target pixel calculated by the calculating means in accordance with the phase of the composite signal of the target pixel detected by the detecting means. A correction means (for example, the sign inversion circuit 17 in FIG. 2) for correcting the component signal is provided.

According to a fifth aspect of the present invention, there is provided an image signal processing apparatus comprising a class classification means (for example, the class classification circuit in FIG. 2) for classifying a target pixel based on the spatial and temporal characteristics of the target pixel. Further, the storage means stores the prediction coefficients corresponding to the two types of phases of the composite signal for each class.

In the learning device according to the present invention, a conversion means for converting a learning component signal into a learning composite signal (for example, the NTSC encoder 22 in FIG. 6).
Detecting means for detecting four types of phases of the composite signal for learning of the pixel of interest converted by the converting means (for example,
Control means 24 in FIG. 6) and correcting means for correcting the learning component signal corresponding to the phase of the learning composite signal of the pixel of interest detected by the detecting means (for example, sign inverting circuit 25 in FIG. 6). Calculating means for calculating a prediction coefficient for reducing an error between the component signal calculated using the composite signal for learning and the prediction coefficient and the component signal for learning corrected by the correction means (for example, , An arithmetic circuit 27A) of FIG.

The learning apparatus according to the eleventh aspect further includes class classification means (for example, the class classification circuit 23 in FIG. 6) for classifying the pixel of interest based on the spatial and temporal characteristics of the pixel of interest. , The calculation means calculates prediction coefficients corresponding to two types of phases of the composite signal for each class.

FIG. 1 shows an example of the configuration of a television receiver to which the present invention is applied. The tuner 1 detects and demodulates an NTSC television signal received by an antenna (not shown), and supplies an A / D converter 2 with an image signal (hereinafter, appropriately referred to as an NTSC signal) as a composite signal. At the same time, an audio signal is supplied to the amplifier 5. The A / D converter 2 samples the NTSC signal input from the tuner 1 at a predetermined timing, thereby sequentially outputting a so-called YI signal, YQ signal, Y + I signal, and Y + Q signal. It has been made like that. Digital NTSC signals output from the A / D converter 2 (YI signal, YQ signal, Y + I signal, and Y + Q
Signal) is supplied to the component signal conversion circuit 3. Here, for example, assuming that the phase of the YI signal is 0 degree, the YQ signal, the Y + I signal, and the Y + Q signal are signals having phases of 90 degrees, 180 degrees, and 270 degrees, respectively.

The component signal conversion circuit 3 outputs the NTSC signal of the pixel of interest and the NTSC signal of the pixel spatially or temporally adjacent to the pixel of interest among the NTSC signals input thereto.
Based on the signal, the target pixel is classified into a predetermined class, and a calculation is performed using a prediction coefficient (described later) corresponding to the class of the target pixel. For example, a YUV signal is obtained. The Y, U, and V signals obtained by the component signal conversion circuit 3 are CRT (Cathode
Ray Tube) 4. CRT4
Is the YUV input from the component signal conversion circuit 3.
An image corresponding to the signal is displayed.

The amplifier 5 amplifies the audio signal input from the tuner 1 and supplies it to the speaker 6. The speaker 6 reproduces an audio signal input from the amplifier 5.

In the television receiver configured as described above, when a user operates a remote commander (not shown) to select a predetermined channel, the tuner 1 detects a television signal corresponding to the channel and detects the television signal. , Demodulated, and the NTSC signal which is the image signal is A /
The audio signal is supplied to the D converter 2 and the amplifier 5. In the A / D converter 2, the NTS input from the tuner 1
The C signal is A / D converted and supplied to the component signal conversion circuit 3. In component signal conversion circuit 3, A / D
As described above, the NTSC signal input from the converter 2
It is converted into a V signal, supplied to the CRT 4 and displayed. On the other hand, in the amplifier 5, the audio signal input from the tuner 1 is amplified, supplied to the speaker 6, and output.

FIG. 2 shows a detailed configuration example of the component signal conversion circuit 3 of FIG. The NTSC signal input from the A / D converter 2 is supplied to the field memory 11. The field memory 11 stores the supplied NTSC signal under the control of the control circuit 14,
Further, the stored NTSC signals are read out by a predetermined number of fields and supplied to the classifying circuit 12 and the prediction tap forming circuit 15.

The class classification circuit 12 is provided with a predetermined number of NTSCs which are supplied from the field memory 11 and which are temporally different from each other.
The spatial or temporal characteristics of the pixel of interest are detected using the signal, the class of the pixel of interest is classified based on the characteristic, and the class of the pixel of interest is stored in the prediction coefficient memory unit 13.
To be supplied.

The prediction coefficient memory unit 13 stores the Y ± I memory 1
3A and a Y ± Q memory 13B,
Each is supplied with the class of the pixel of interest output from the class classification circuit 12 as an address, and supplied with a CS (ChipSelect) signal output from the control circuit 14. Y ± I memory 13A and Y ± Q
The memory 13B stores a prediction coefficient for each class used for an operation of converting the NTSC signal of the target pixel into a YUV signal.

FIG. 3 shows pixels constituting a predetermined field of the NTSC signal. In FIG. 3, a mark O represents a Y-I signal having a phase of 0 degree, a mark represents a YQ signal having a phase of 90 degrees, and a mark represents Y + I signal having a phase of 180 degrees. ■ indicates a Y + Q signal whose phase is 270 degrees. In FIG. 3, in the horizontal direction, the YI signal, YQ signal, Y + I signal, Y +
In the vertical direction, a column in which Y-I signals and Y + I signals are alternately arranged and a column in which Y-Q signals and Y + Q signals are alternately arranged are alternately arranged. ing.

Returning to FIG. Y ± I memory 13A and Y ± Q memory 13B (hereinafter, memories 13A, 13
B) is to convert the Y + I signal or the Y + Q signal
The prediction coefficients for each class used for converting into YUV signals are stored. Then, the memory 13A, 1
Of the prediction coefficients for each class stored in the 3B, which are selected by the CS signal input from the control circuit 17, those corresponding to the class of the target pixel input from the class classification circuit 12 are as follows: The data is read out from this and supplied to the arithmetic circuit 16.

In each of the memories 13A and 13B, an NTSC signal is used as a prediction coefficient for each class.
A set of prediction coefficients for each of Y, U, and V, which are used to convert the signals into U and V signals, are stored.

The control circuit 14 controls reading and writing from and to the field memory 11. That is, the control circuit 14 performs various controls using the field already stored in the field memory 11 as a field of interest, and when the processing on the field of interest is completed, sets the next field as a new field of interest. Further, a field newly supplied is stored in the field memory 11 in place of the field which has been set as the field of interest up to now. Further, the control circuit 14
The pixels constituting the field of interest are sequentially set as the pixels of interest, for example, in line scan order, and the pixels necessary for processing the pixel of interest are read out from the field memory 11, the classification circuit 12 and the prediction tap formation. Circuit 1
5

The control circuit 14 outputs a CS signal based on the phase of the NTSC signal of the pixel of interest to the prediction coefficient memory unit 13 and the sign inversion circuit 17, and outputs the CS signal to the prediction coefficient memory unit 13.
Causes the memory 13A, 13B to select the one corresponding to the phase of the pixel of interest, and causes the sign inversion circuit 17 to execute a sign inversion process (described later) corresponding to the phase of the pixel of interest.

That is, the control circuit 14 determines the NT of the target pixel.
When the SC signal is a Y-I signal and a Y + I signal, Y ±
The I-memory 13A is selected, and the Y-Q signal and Y + Q
In the case of a signal, a CS signal for selecting the Y ± Q memory 13B is supplied to the prediction coefficient memory unit 13.

A pixel read from the field memory 11 is supplied to the prediction tap forming circuit 15. The prediction tap forming circuit 15 outputs the NTSC signal of the pixel of interest from the pixel to the YUV signal. A prediction tap used to convert the data into a data is formed and supplied to the arithmetic circuit 16.

That is, for example, the prediction tap forming circuit 1
5 are pixels b to e adjacent to the pixel a of the target field shown in FIG.
Pixels f to i adjacent to the lower left and lower right, a pixel j on the left of the pixel d on the left of the target pixel, a pixel k on the right of the pixel e on the right of the target pixel, and FIG. 4B In the field two fields before the field of interest shown in (1), pixels a ′ to k ′ at positions corresponding to pixels a to k are used as prediction taps and supplied to the arithmetic circuit 16.

The operation circuit 16 calculates the YUV signal of the pixel of interest by performing an operation using the prediction coefficient supplied from the prediction coefficient memory unit 13 and the prediction tap supplied from the prediction tap forming circuit 15. It has been made like that. That is, as described above, the arithmetic circuit 16 uses the Y, U, and V signals from the prediction coefficient memory unit 13 to convert the NTSC signal of the target pixel into Y, U, and V signals.
A set of prediction coefficients for U and V is supplied, and a prediction tap (FIG. 4) formed for the pixel of interest is supplied from the prediction tap forming circuit 15. Then, the pixels constituting the prediction tap are shown in FIG.
As described in the above, with the pixels a to k and the pixels a 'to k', the prediction coefficients for the Y signal are w _{Ya to} w _Yk and w _{YA to} w _YK, and the prediction coefficients for the U signal are
_{Assuming that} w _{Ua to} w _Uk and w _{UA to} w _UK, and the prediction coefficients for the V signal are w _{Va to} w _Vk and w _{VA to} w _VK , the arithmetic circuit 16 performs the following attention in accordance with the following linear linear expression. The Y, U, and V signals of the pixel are calculated.

Y = w _Ya a + w _Yb b + w _Yc c + w _Yd d + w _Ye e + w _Yf f + w _Yg g + w _Yh h + w _Yi i + w _Yj j + w _Yk k + w _YA a '+ w _YB b' + w _YC c '+ w _YE c' + w _{YD YF f '+ w YG g'} + w YH h '+ w YI i' + w YJ j '+ w YK k' + w Yoffset U = w Ua a + w Ub b + w Uc c + w Ud d + w Ue e + w Uf f + w Ug g + w Uh h + w Ui i + w Uj j + w Uk _{k + w UA a '+ w} UB b' + w UC c '+ w UD d' + w UE e '+ w UF f' + w UG g '+ w UH h' + w UI i '+ w UJ j' + w UK k '+ w Uoffset V = w _{Va a + w Vb b + w} Vc c + w Vd d + w Ve e + w Vf f + w Vg g + w Vh h + w Vi i + w Vj j + w Vk k + w VA a '+ w VB b' + w VC c '+ w VD d' + w VE e '+ w VF f' + w VG g '+ W _VH h' + w _VI i '+ w _VJ j' + w _VK k '+ w _Voffset (1)

Note that w _Yoffset , w _Uoffset , and w
_Voffset is the offset value for correcting the difference in bias between the NTSC signal and the YUV signals (constant term), the constant term w _Yoffset, w _Uoffset, and w _Voffset also
It is included in a set of prediction coefficients for each of Y, U, and V.

Here, as described above, the processing performed by the arithmetic circuit 16 using the coefficient (prediction coefficient) corresponding to the class of the pixel of interest, that is, the prediction coefficient corresponding to the property (characteristic) of the pixel of interest is applied. The processing performed by using this method is called adaptive processing. Hereinafter, this adaptive processing will be briefly described.

For example, the predicted value E [y] of the component signal y of the pixel of interest is now converted to a pixel (including the pixel of interest) at a position spatially or temporally close to the pixel of interest.
Composite signal (hereinafter referred to as learning data as appropriate)
x ₁ , x ₂ ,... and predetermined prediction coefficients w ₁ , w ₂ ,.
Is determined by a linear first-order combination model defined by a linear combination of In this case, the predicted value E [y] can be expressed by the following equation.

E [y] = w ₁ x ₁ + w ₂ x ₂ +... (2)

Therefore, for generalization, a matrix W composed of a set of prediction coefficients w, a matrix X composed of a set of learning data, and a matrix Y ′ composed of a set of predicted values E [y] are

(Equation 1) Defines the following observation equation.

XW = Y ′ (3)

Then, it is considered that a least square method is applied to this observation equation to obtain a predicted value E [y] close to the component signal y of the target pixel. In this case, a matrix Y (different from the above-described luminance signal Y), which is a set of true component signals y of the target pixel serving as teacher data, and a set of residuals e of predicted values E [y] for the component signals y. Matrix E

(Equation 2) From equation (3), the following residual equation holds.

XW = Y + E (4)

In this case, the prediction coefficient w _i for obtaining the predicted value E [y] close to the component signal y is a square error

(Equation 3) Can be obtained by minimizing.

Therefore, the above square error is calculated by the prediction coefficient w
_{When the value obtained} by differentiating with _i becomes 0, that is, the prediction coefficient w _i that satisfies the following equation is an optimum value for obtaining a predicted value E [y] close to the component signal y.

[0046]

(Equation 4) ... (5)

Therefore, first, equation (4) is used to calculate the prediction coefficient w _i
By differentiating with, the following equation is established.

[0048]

(Equation 5) ... (6)

From equations (5) and (6), equation (7) is obtained.

[0050]

(Equation 6) ... (7)

Further, the learning data x, the prediction coefficient w, the teacher data y, and the residual e in the residual equation of the equation (4) are obtained.
In consideration of the relationship, the following normal equation can be obtained from Expression (7).

[0052]

(Equation 7) ... (8)

The normal equation of the equation (8) can be set by the same number as the number of prediction coefficients w to be obtained. Therefore, by solving the equation (8) (however, in order to solve the equation (8), it is necessary that the matrix formed by the coefficients related to the prediction coefficient w in the equation (8) is regular). The prediction coefficient w can be obtained. In solving equation (8), for example, a sweeping method (Gauss-Jordan elimination method)
Etc. can be applied.

As described above, the optimum prediction coefficient w is obtained, and using the prediction coefficient w, the following equation (2) is used.
The process of obtaining the predicted value E [y] close to the component signal y is the adaptive process (however, the process of obtaining the predicted coefficient w in advance and obtaining the predicted value from the predicted coefficient w is also included in the adaptive process). .

In the present embodiment, as described above, the constants w _Yoffset ,
including but w _Uoffset, and w _Voffset, the constant term is to extend the above method, it can be obtained by solving the normal equation corresponding to the equation (8).

The Y ± I memory 13A and the Y ± Q memory 13B of the prediction coefficient memory unit 13 shown in FIG. 2 have a class obtained by solving the normal equation of Expression (8) by learning as will be described later. For each prediction coefficient, NTSC
Y, corresponding to the Y + I signal and the Y + Q signal,
U and V are stored.

The U and V signals of the Y, U and V signals of the target pixel calculated by the calculation circuit 16 are output to the sign inversion circuit 17.

As described above, the prediction coefficients stored in the Y ± I memory 13A and the Y ± Q memory 13B correspond to the Y + I signal and the Y + Q signal of the NTSC signal. When the NTSC signal is the YI signal or the YQ signal, the values of the U and V signals calculated by the arithmetic circuit 16 using the prediction coefficients corresponding to the Y + I signal or the Y + Q signal of the NTSC signal are: , The sign is inverted with respect to the true value. Therefore, in the sign inversion circuit 17, the input from the control circuit 14 is performed.
Based on the CS signal, the pixel of interest is a Y-I signal or a Y-
If the signal is a Q signal, the U, V
Are inverted (the sign inversion processing is executed), and when the target pixel is the Y + I signal or the Y + Q signal, the signal is output without performing the sign inversion processing.

Next, the processing of the component signal conversion circuit 3 shown in FIG. 2 will be described with reference to the flowchart of FIG. When the NTSC signal is stored in the field memory 11 and the control circuit 14 selects a predetermined field as the field of interest, in step S1, the control circuit 14 selects a predetermined pixel of the field of interest as the pixel of interest. In step S <b> 2, the control circuit 14 reads out the pixels required for classifying the target pixel from the field memory 11 and supplies the pixels to the classifying circuit 12.

The class classification circuit 12 uses the pixels input from the field memory 11 to determine the spatial and
Then, a temporal characteristic is detected, the class is classified based on the characteristic, and the obtained class of the target pixel is output to the prediction coefficient memory unit 13 as an address.

In step S3, the control circuit 14
The CS signal indicating the phase of the NTSC signal of the target pixel is supplied to the prediction coefficient memory unit 13 and the sign inversion circuit 17. The prediction coefficient memory unit 13 stores the NTSC of the pixel of interest based on the CS signal.
It is determined whether the signal is a Y-I signal or a Y + I signal. If it is determined that the NTSC signal of the target pixel is the Y-I signal or the Y + I signal, the process proceeds to step S4.

In step S4, the prediction coefficient memory unit 13 uses the addresses of the Y ± I memory 13A (addresses corresponding to the class of the pixel of interest inputted from the class classification circuit 12) to calculate the prediction for each of Y, U, and V. The set of coefficients is read and supplied to the arithmetic circuit 16.

In step S5, the prediction tap forming circuit 15 reads out a pixel from the field memory 11 and
The prediction tap described with reference to FIG. 4 is formed for the pixel currently set as the target pixel. The prediction tap is supplied to the arithmetic circuit 16.

The processing in step S5 is performed in step S5.
It is also possible to perform in parallel with the processing of 2 to S4.

The operation circuit 16 receives the prediction coefficients from the prediction coefficient memory section 13 and also executes the prediction tap forming circuit 1
When receiving the prediction tap from No. 5, in step S6, an adaptive process using the prediction coefficient and the prediction tap is performed. That is, specifically, the linear primary equation shown in the equation (1) is calculated, whereby the Y, U, and V signals of the target pixel are obtained and output to the subsequent stage.

In step S7, the sign inversion circuit 17
Is based on the CS signal supplied from the control circuit 14, the NTSC signal of the target pixel is changed to the YI signal or the YQ signal.
It is determined whether the signal is a signal. The NTSC signal of the pixel of interest is
If it is determined that the signal is the YI signal or the YQ signal, the process proceeds to step S8.

In step S8, the sign inversion circuit 17
Inverts the sign of each of the U and V signals of the pixel of interest input from the arithmetic circuit 16 in step S6 and outputs the inverted signal.

In step S9, the control circuit 14
It is determined whether or not all the pixels constituting the field of interest stored in the field memory 11 have been processed as the pixel of interest. If it is determined that all the pixels constituting the field of interest have not yet been processed as the pixel of interest,
Returning to step S1, among the pixels constituting the field of interest, a pixel that has not yet been set as the pixel of interest is newly set as the pixel of interest, and the processing from step S2 is repeated.
If it is determined in step S9 that all the pixels forming the field of interest have been processed as the pixel of interest, the process ends.

In step S3, the target pixel
When it is determined that the NTSC signal is not the YI signal or the Y + I signal (the YQ signal or the Y + Q signal), the process proceeds to step S10.

In step S10, the prediction coefficient memory unit 13 calculates the prediction for each of Y, U, and V from the address of the Y ± Q memory 13B (the address corresponding to the class of the pixel of interest input from the class classification circuit 12). The set of coefficients is read and supplied to the arithmetic circuit 16.

In step S7, the target pixel
If it is determined that the NTSC signal is not the YI signal or the YQ signal (the signal is the Y + I signal or the Y + Q signal), the step S8 is skipped because the sign inversion processing does not need to be performed.

The processes of steps S1 to S9 shown in the flowchart of FIG. 5 are repeatedly performed each time a new field is set as a field of interest.

As described above, according to the present embodiment, when the target pixel is the Y-I signal and when the target pixel is Y + I
In the case of a signal, processing can be performed using the same prediction coefficient. When the pixel of interest is a YQ signal and when the pixel of interest is a Y + Q signal, processing can be performed using the same prediction coefficient. it can. However, when the target pixel is a YI signal or a YQ signal, a correction process such as the above-described sign inversion process is required.

FIG. 6 shows an example of the configuration of a learning device that performs learning for obtaining a prediction coefficient to be stored in the prediction coefficient memory unit 13 in FIG.

The field memory 21 of the learning device stores a learning YUV signal (learning component signal).
Are supplied and stored for a predetermined number of fields. Further, the YUV signals of the pixels constituting the learning image stored in the field memory 21 are:
The data is read out under the control of the control circuit 24 and supplied to the NTSC encoder 22 and the control circuit 24.

The NTSC encoder 22 encodes (converts) the YUV signal of each pixel input from the field memory 21 into an NTSC signal and supplies the NTSC signal to the class classification circuit 23 or the control circuit 24.

The class classification circuit 23 is configured in the same manner as the class classification circuit 12 of FIG. 2. The class classification circuit 23 classifies the pixel of interest and supplies the class to the learning data memory unit 26 as an address. It has been done.

The control circuit 24 sequentially sets the fields already input and stored in the field memory 21 as a field of interest, and further sets the pixels constituting the field of interest as, for example, pixels of interest in the order of line scan. The YUV signal of the pixel required for processing the target pixel is read from the field memory 21 and supplied to the NTSC encoder 22 and the control circuit 24 itself.

Further, the control circuit 24 causes the YUV signal of the pixel of interest and the YUV signal of the pixel forming the prediction tap corresponding to the pixel of interest to be read from the field memory 21, and the control circuit 24 reads the YUV signal of the pixel forming the prediction tap. The YUV signal is supplied to the NTSC encoder 22. As a result, the YUV signal of the pixel constituting the prediction tap is transmitted to the NTSC encoder 22 by N
The signal is converted into a TSC signal (composite signal for learning) and supplied to the control circuit 24.

Further, when the control circuit 24 receives the NTSC signal of the pixel constituting the prediction tap from the NTSC encoder 22 as described above, the control circuit 24 sets it as learning data, and reads the YUV of the pixel of interest read from the field memory 21. Using the signal as teacher data, the learning data and the teacher data are combined and supplied to the learning data memory unit 26 via the sign inversion circuit 25. That is, with respect to the YUV signal of the pixel of interest and the pixel of interest, FIG.
Is combined with the NTSC signal of the pixel forming the prediction tap having the positional relationship as described in
Is supplied to the learning data memory unit 26 via the.

Further, the control circuit 24 controls the NTSC of the pixel of interest.
A CS signal indicating whether the signal is a YI signal, a YQ signal, a Y + I signal, or a Y + Q signal is supplied to the learning data memory unit 26 and the sign inversion circuit 25. .

The sign inverting circuit 25 inverts the sign of each of the U and V signals input from the control circuit 24 in accordance with the phase of the pixel of interest based on the CS signal input from the control circuit 24. The other signals (Y signal and NTSC signal of the pixel constituting the prediction tap) are not processed as they are,
The data is supplied to the learning data memory unit 26. That is, the sign inversion circuit 25 refers to the CS signal input from the control circuit 24, and sets the NTCS signal of the target pixel to
If the signal is a YI signal or a YQ signal, the control circuit 2
Invert the sign of each of the U and V signals input from No. 4 (perform sign inversion processing) so that the NTCS signal of the target pixel is Y + I
In the case of a signal or a Y + Q signal, the signal is supplied to the learning data memory unit 26 together with other signals (the Y signal and the NTSC signal of the pixel forming the prediction tap) without performing the sign inversion processing.

The learning data memory section 26 is composed of a Y ± I memory 26A and a Y ± Q memory 26B, and the class of the pixel of interest output from the class classification circuit 23 is supplied to each of them as an address. Along with
The CS signal output from the control circuit 24 is supplied. Further, the combination of the above-described teacher data and learning data output from the control circuit 24 is supplied to the learning data memory unit 26 via the sign inversion circuit 25. Then, referring to the CS signal input from the control circuit 24, if the NTSC signal of the target pixel is the YI signal or the Y + I signal, the Y ± I memory 26A
Is selected, and when the NTSC signal of the target pixel is the YQ signal or the Y + Q signal, the Y ± Q memory 26B is selected. The addresses of the selected Y ± I memory 26A or Y ± Q memory 26B (addresses corresponding to the class of the pixel of interest output from the classifying circuit 23) include the teacher data and the learning data input from the sign inverting circuit 25. Is stored.

Therefore, in the Y ± I memory 26A, when the NTSC signal of the target pixel is the Y-I signal or the Y + I signal, the YUV signal (teacher data) of the target pixel and the target pixel are formed. The combination with the NTSC signal of the pixel constituting the prediction tap is stored, and Y ± Q
When the NTSC signal of the pixel of interest is a YQ signal or a Y + Q signal, the memory 26B stores the YUV signal (teacher data) of the pixel of interest and the pixels forming the prediction tap formed for the pixel of interest. Is stored.

In each of the memories 26A and 26B, a plurality of pieces of information can be stored at the same address, whereby the same address stores the learning data of the pixels classified into the same class. And a plurality of combinations of data and teacher data.

The operation circuits 27A and 27B are provided in the memory 26
A combination of the NTSC signal of the pixel forming the prediction tap as the learning data stored in each address of A and 26B and the YUV signal as the teacher data is read out, and the YUV signal is calculated by using the least square method. And a prediction coefficient for minimizing an error between the predicted value of the target data and the teacher data. That is, the operation circuit 27A,
In 27B, a normal equation corresponding to Expression (8) is established for each class and for each signal of Y, U, and V, and by solving this, Y, U, and V for each class are obtained. Prediction coefficients (prediction coefficients w _{Ya to} w _Yk , prediction coefficients w _{YA to} w _YK , and w _Yoffset for _Y , prediction coefficients w _{U a to} w _{Uk for U} , prediction coefficients w _{UA to} w _UK , and w _Uoffset , And prediction coefficients w _{Va to} w _Vk , prediction coefficients w _{VA to} w _VK , and w _Voffset for V are obtained, respectively.

Here, in the arithmetic circuits 27A and 27B, the processing is performed using the data stored in the memories 26A and 26B, respectively. Therefore, the prediction coefficients obtained in the arithmetic circuits 27A and 27B respectively correspond to the phase of the NTSC signal. The corresponding signal, that is, the signal for converting the Y + I signal or the Y + Q signal into the YUV signal, respectively.

The prediction coefficients obtained by the arithmetic circuits 27A and 27B are
Is supplied to the Y ± I memory 13A or the Y ± Q memory 13B of the prediction coefficient memory unit 13 and is stored at an address corresponding to each class.

Next, a learning process performed by the learning device will be described with reference to a flowchart of FIG. When the YUV signal of the learning image is stored in the field memory 21, in step S21, the control circuit 2
4 selects a predetermined pixel constituting the image for learning as a pixel of interest, causes the pixel necessary for performing the class classification to be read from the field memory 21 for the pixel of interest, and supplies the pixel to the NTSC encoder 22. . In the NTSC encoder 22, the YUV signal of the pixel input from the field memory 21 is converted into an NTSC signal.
Supplied to

In step S22, the class classification circuit 23 classifies the pixels using the pixels input from the NTSC encoder 22 in the same manner as the class classification circuit 12 in FIG. To the data memory unit 26 for use.

In step S23, the control circuit 24
Is a CS signal indicating the phase (Y-I signal, Y-Q signal, Y + I signal, or Y + Q signal) of the NTSC signal of the pixel of interest.
It is supplied to the sign inversion circuit 25 and the learning data memory unit 26.

Further, the control circuit 24 controls the YUV
A signal and a YUV signal of a pixel forming a prediction tap corresponding to the pixel of interest are read from the field memory 21, and the YUV signal of the pixel of interest is supplied to the control circuit 24 itself, and a signal of the pixel forming the prediction tap is generated. The YUV signal is supplied to the NTSC encoder 22. In this case, the NTSC encoder 22
Converts the YUV signal of the pixel constituting the prediction tap input from the field memory 21 into an NTSC signal and supplies the signal to the control circuit 27.

The control circuit 24 uses the NTSC signal of the pixel constituting the prediction tap input from the NTSC encoder 22 as learning data,
The YUV signal of the pixel of interest input from
The combination of the learning data and the teacher data is output to the sign inversion circuit 25.

Note that the processing in step S23 can be performed in parallel with the processing in step S22.

In step S24, the sign inversion circuit 2
5 refers to the CS signal input from the control circuit 24,
It is determined whether the NTSC signal of the target pixel is a YI signal or a YQ signal. The NTSC signal of the target pixel is Y-
If it is determined that the signal is an I signal or a YQ signal, the process proceeds to step S25.

In step S25, the sign inversion circuit 2
5 inverts the sign of each of the U and V signals input from the control circuit 24 (performs sign inversion processing), and together with other signals (Y signal and NTSC signal of a pixel constituting a prediction tap). The data is supplied to the learning data memory unit 26.

In step S26, the learning data memory unit 26 refers to the CS signal input from the control circuit 24 and determines whether the NTSC signal of the target pixel is a YI signal or a YI signal.
It is determined whether the signal is a + I signal. When it is determined that the NTSC signal of the target pixel is the Y-I signal or the Y + I signal, the process proceeds to step S27.

In step S27, the learning data memory unit 26 stores the teacher data input from the sign inversion circuit 25 at the address of the Y ± I memory 26A corresponding to the class of the pixel of interest supplied from the class classification circuit 23. And the combination of the learning data.

At the step S28, the control circuit 24
Determines whether all pixels constituting the learning image stored in the field memory 21 have been processed as the pixel of interest. If it is determined that all the pixels forming the learning image have not been processed as the target pixel, the process returns to step S21, and a pixel that has not been set as the target pixel is newly set as the target pixel. Is repeated.

Thereafter, if it is determined in step S28 that all the pixels constituting the learning image have been processed as the pixel of interest, the process proceeds to step S29.

In step S29, the arithmetic circuit 27
A and 27B respectively read the combination of the learning data and the teacher data stored in the memories 26A and 26B for each address, and for each signal of Y, U and V, a normal equation corresponding to the equation (8). Is generated, and the generated normal equation is solved to obtain a set of prediction coefficients for each class for Y, U, and V used for the operation of converting the Y + I signal or the Y + Q signal into the YUV signal. And the Y + I signal,
Alternatively, the set of prediction coefficients for each class corresponding to the Y + Q signal is supplied to and stored in the Y ± I memory 13A or the Y ± Q memory 13B of the prediction coefficient memory unit 13, and the processing is terminated.

As described above, the target pixel is classified into classes based on the spatial and temporal characteristics of the target pixel, and the prediction coefficient corresponding to the class obtained thereby, that is, the prediction coefficient suitable for the target pixel. The coefficient is used to convert the NTSC signal of the pixel of interest into a YUV signal, thus reducing dot interference caused by edges due to luminance, and cross color in which the color is changing with luminance. It becomes possible.

In this embodiment, the NTSC signal is converted to a YUV signal by calculating a linear equation with the prediction coefficient.
Although the signal is converted into a signal, it is also possible to convert the NTSC signal into a YUV signal by calculating, for example, a nonlinear arithmetic expression.

Further, in the present embodiment, the prediction tap is formed from the pixels described with reference to FIG. 4, but the prediction tap may be formed from other pixels.

Further, in the present embodiment, the adaptive processing and the learning processing are performed in accordance with the phase of the NTSC signal. However, the adaptive processing and the learning processing can be performed regardless of the phase of the NTSC signal. It is. However, it is more accurate to perform adaptive processing and learning processing for each NTSC signal phase.
YUV signals and prediction coefficients can be obtained.

In the present embodiment, the NTSC signal is converted to a YUV signal.
Converts signals such as AL signals to YUV signals, NTSC
It is also possible to convert the signal into an RGB signal or a YIQ signal. That is, the composite signal before conversion and the component signal after conversion are not particularly limited.

In the present embodiment, the processing is performed in units of fields. However, it is also possible to perform processing in units of frames, for example.

Further, the present invention is applicable to, for example, a VTR (Video Tape Recorder) and other devices that handle images other than the television receiver. Further, the present invention is applicable to both moving images and still images.

Further, in this embodiment, by sampling the NTSC signal, the YI signal, the YQ signal,
Y + I signal and Y + Q signal, but NT
The sampling of the SC signal may be performed at any timing as long as an in-phase signal is obtained every four samplings. However, in this case, it is necessary to use signals of the same phase also in learning.

Note that the computer program for performing each of the above processes can be provided to the user via a network medium such as the Internet or a digital satellite in addition to a medium such as a magnetic disk or a CD-ROM. it can.

[0111]

As described above, according to the image signal processing apparatus according to the first aspect, the image signal processing method according to the sixth aspect, and the providing medium according to the seventh aspect, the composite signal of the target pixel is provided. Since the four types of phases are detected, and the calculated component signal of the target pixel is corrected in accordance with the detected phase of the composite signal of the target pixel,
The scale of the device can be reduced.

According to the learning device of the eighth aspect, the learning device of the twelfth aspect, and the providing medium of the thirteenth aspect, four types of the composite signal for learning the converted target pixel are provided. Detecting the phase, correcting the component signal for learning corresponding to the phase of the composite signal for learning of the detected pixel of interest, a component signal calculated using the composite signal for learning and the prediction coefficient, Since the calculation for obtaining the prediction coefficient for reducing the error from the corrected component signal for learning is performed, it is possible to obtain the prediction coefficient for reducing the scale of the apparatus.

[Brief description of the drawings]

FIG. 1 is a block diagram illustrating a configuration example of a television receiver to which the present invention has been applied.

FIG. 2 is a block diagram illustrating a configuration example of a component signal conversion circuit 3 of FIG. 1;

FIG. 3 is a diagram illustrating a configuration example of a field of an NTSC signal.

FIG. 4 is a diagram for explaining processing of a prediction tap forming circuit 15 of FIG. 2;

FIG. 5 is a flowchart illustrating an operation of the component signal conversion circuit 3 of FIG. 2;

FIG. 6 is a block diagram illustrating a configuration example of a learning device to which the present invention has been applied.

FIG. 7 is a flowchart illustrating an operation of the learning device in FIG. 6;

[Explanation of symbols]

1 tuner, 2 A / D converter, 3 component signal conversion circuit, 4 CRT, 5 amplifier, 6 speaker, 11 field memory, 12 class classification circuit, 13 prediction coefficient memory section, 13A Y ± I memory, 13B Y ± Q memory , 14 control circuit, 1
5 prediction tap formation circuit, 16 operation circuit, 17 sign inversion circuit, 21 field memory, 22 NTSC
Encoder, 23 class classification circuit, 24 control circuit, 25 sign inversion circuit, 26 learning data memory section, 26A Y ± I memory, 26B Y ± Q memory, 27A, 27B arithmetic circuit

Continued on the front page (72) Inventor Takaya Hoshino 6-7-35 Kita-Shinagawa, Shinagawa-ku, Tokyo Inside Sony Corporation (72) Inventor Takeharu Nishikata 6-35-35 Kita-Shinagawa, Shinagawa-ku, Tokyo Sony Corporation Inside the company (72) Inventor Ken Ken Inoue 6-7-35 Kita-Shinagawa, Shinagawa-ku, Tokyo Sony Corporation (72) Inventor Naoki Kobayashi 6-35, Kita-Shinagawa, Shinagawa-ku, Tokyo Sonny Corporation (72) Inventor Tetsushi Kokubo 6-35 Kita Shinagawa, Shinagawa-ku, Tokyo F-term in Sony Corporation (reference) 5C066 AA03 BA02 CA01 CA07 DC01 DD07 EE01 GA02 GA04 GA05 HA02 KD07 KE03 KE04 KE09 KE17 KE19 KF01

Claims (13)

[Claims] 1. An image signal processing apparatus for converting a composite signal into a component signal, comprising: detecting means for detecting four types of phases of a composite signal of a target pixel; and predicting coefficients corresponding to two types of phases of the composite signal. Storage means for storing, and calculating means for calculating the component signal of the pixel of interest using the one corresponding to the phase of the composite signal of the pixel of interest detected by the detection means among the prediction coefficients stored by the storage means, An image signal processing apparatus comprising: a correction unit that corrects a component signal of the pixel of interest calculated by the calculation unit in accordance with a phase of a composite signal of the pixel of interest detected by the detection unit. 2. The storage unit stores a prediction coefficient corresponding to a Y + I signal or a Y + Q signal of the composite signal, and the correction unit determines that the composite signal is a YI signal;
2. The image signal processing apparatus according to claim 1, wherein when the signal is a YQ signal, the component signal of the target pixel calculated by the calculation unit is corrected. 3. The image signal processing device according to claim 1, wherein the component signal is a YUV signal or an RGB signal. 4. The method according to claim 1, wherein the calculating means calculates a component signal of the pixel of interest by a linear linear combination of the prediction coefficient and a pixel spatially or temporally adjacent to the pixel of interest. Item 2. The image signal processing device according to item 1. 5. The method according to claim 1, further comprising: classifying means for classifying the pixel of interest based on spatial and temporal characteristics of the pixel of interest, wherein the storage means corresponds to two types of phases of the composite signal. 2. The image signal processing apparatus according to claim 1, wherein a prediction coefficient is stored for each class. 6. An image signal processing method for converting a composite signal into a component signal, comprising: a detecting step of detecting four types of phases of a composite signal of a pixel of interest; and a prediction coefficient corresponding to two types of phases of the composite signal. A storage step of storing, a calculation step of calculating the component signal of the pixel of interest using a phase coefficient of the composite signal of the pixel of interest detected in the detection step of the prediction coefficients stored in the storage step, A correcting step of correcting the component signal of the pixel of interest calculated in the calculating step in accordance with the phase of the composite signal of the pixel of interest detected in the detecting step. 7. An image signal processing apparatus for converting a composite signal into a component signal, a detecting step of detecting four types of phases of the composite signal of the target pixel, and a prediction coefficient corresponding to the two types of phases of the composite signal. A storage step of storing, a calculation step of calculating the component signal of the pixel of interest using a phase coefficient of the composite signal of the pixel of interest detected in the detection step of the prediction coefficients stored in the storage step, A computer-readable program for executing a process including a correction step of correcting the component signal of the pixel of interest calculated in the calculation step in accordance with the phase of the composite signal of the pixel of interest detected in the detection step. A providing medium characterized by being provided. 8. A learning device for calculating a prediction coefficient used for an operation for converting a composite signal into a component signal, comprising: conversion means for converting a learning component signal into a learning composite signal; Detecting means for detecting four types of phases of the learning composite signal of the detected pixel of interest, and the learning component corresponding to the phase of the learning composite signal of the pixel of interest detected by the detecting means. Correction means for correcting a signal; a component signal calculated using the composite signal for learning and the prediction coefficient; and the prediction coefficient for reducing an error between the component signal for learning corrected by the correction means. A learning device comprising: a calculation unit for performing a calculation for obtaining the learning value. 9. The learning device according to claim 8, wherein the component signal is a YUV signal or an RGB signal. 10. The method according to claim 1, wherein the calculating means calculates a component signal of the pixel of interest by a linear linear combination of the prediction coefficient and a pixel spatially or temporally adjacent to the pixel of interest. 9. The learning device according to 8. 11. The image processing apparatus further comprises a classifying unit that classifies the pixel of interest based on spatial and temporal characteristics of the pixel of interest, wherein the arithmetic unit corresponds to two types of phases of the composite signal. The learning device according to claim 8, wherein a prediction coefficient is calculated for each of the classes. 12. A learning method for obtaining a prediction coefficient used in an operation for converting a composite signal into a component signal, comprising: a conversion step of converting a learning component signal into a learning composite signal; A detecting step of detecting four types of phases of the learning composite signal of the target pixel, and a learning component corresponding to the phase of the learning composite signal of the target pixel detected in the detecting step. A correction step of correcting a signal; a component signal calculated using the composite signal for learning and the prediction coefficient; and the prediction coefficient for reducing an error between the component signal for learning corrected in the correction step. And an operation step of performing an operation for obtaining Learning method. 13. A learning device for calculating a prediction coefficient used for an operation for converting a composite signal into a component signal, a conversion step of converting a learning component signal into a learning composite signal, and a conversion step in the conversion step. A detecting step of detecting four types of phases of the learning composite signal of the target pixel, and a learning component corresponding to the phase of the learning composite signal of the target pixel detected in the detecting step. A correction step of correcting the signal; a component signal calculated using the composite signal for learning and the prediction coefficient; and the prediction coefficient for reducing an error between the component signal for learning corrected in the correction step. And a calculation step for performing a calculation for obtaining Providing medium characterized by Yuta to provide readable program.