JPH0475712B2 - - Google Patents

Description

[Detailed description of the invention]

"Field of Industrial Application" The present invention relates to a color decoding method for a digital color decoder applied to color television receivers, color VTRs, and the like. "Background technology and its problems" Conventionally, color decoding methods for converting NTSC color video signals into RGB three primary color signals were
A Y/C separation filter is used to separate a luminance signal and a carrier color signal, and the separated carrier color signal is demodulated into color difference signals I and Q. Furthermore, the luminance signal and color difference signals I and Q are converted into 3 by matrix conversion. It formed a primary color signal. This conventional color decoding method requires three stages of processing as described above, and has the disadvantage of increasing the number of processing stages. In addition, conventional color decoding methods use, for example, a bandpass filter as a Y/C separation filter to perform separation in the frequency domain, but the frequency characteristics of the carrier color signal and the luminance signal overlap. The drawback was that complete separation could not be performed. Also,
A frequency separation filter using a digital filter or the like as a Y/C separation filter has a problem in that the number of taps must be increased and the luminance signal becomes dull. ``Object of the Invention'' Therefore, an object of the present invention is to provide a method for decoding a color video signal that can directly convert a color video signal such as an NTSC system into an RGB three primary color signal. Another object of the present invention is to provide a color decoding method that can perform complete Y/C separation without dulling the signal by performing optimal separation using sequential processing in the time domain. be. "Summary of the Invention" The present invention provides a color decoding method for decoding digital three primary color signals from a digital color video signal, which includes a step of transitioning the previous estimated values of the three primary color signals based on a state equation model to obtain the current predicted values. , the step of multiplying the predicted value by the observation coefficient, the step of comparing the predicted value multiplied by the observation coefficient with the input color video signal to find the correction amount, multiplying the correction amount by the weighting coefficient, and calculating the correction amount for each component. This color decoding method is characterized by the step of obtaining estimated values of the current three primary color signals by adding them to the corresponding predicted values. "Example" When a composite color video signal based on the NTSC system is digitized using, for example, a sampling pulse with a frequency of 4 _SC ( _SC : color subcarrier frequency), the three primary color signals at the current sample point are highly correlated and continuous. Prediction can be made by introducing a time series model using three primary color signals of other sample points with different characteristics. Sample points with a strong correlation are sample points of neighboring pixels arranged vertically to the pixel of the current sample point, or sample points of neighboring pixels arranged horizontally to the pixel of the current sample point. However, the sample points of pixels arranged in the vertical direction are less continuous than those in the horizontal direction, and are not suitable for predicting the three primary color signals at the current sample points. Therefore,
The three primary color signals at the sample point of the current sample point are predicted using the three primary color signals of the sample points with strong correlation among the sample points of pixels arranged in the horizontal direction. For modeling using a time series model using sample points of pixels lined up in the horizontal direction, the three primary color signals R _k , G _k , B _k of the current sample point are the three primary color signals R of the previous sample point one sample before. _k-1 , G _k-1 ,
A model that assumes that B _k-1 is equal to R _k = R _k-1 G _k = G _k-1 B _k = B _k-1 can be considered because their correlation is strong. However, the inventor of this application has confirmed that modeling errors are strongly colored, and this is not suitable as a model. Therefore, in an optimal embodiment of the present invention, the three primary color signals at the current sample point are
The difference between R _k , G _k , B _k and the three primary color signals R _k-1 , G _k-1 , B _k-1 at the sample point one sample before is 1
Three primary color signals R _k-1 , G _k-1 , B _k-1 and 2 before sample
A model with a constant difference that is equal to the difference between the three primary color signals R _k- , G _k-2 , and B _k-2 before the sample R _k −R _k-1 = R _k-1 − _{R k-2} …… G _k −G _k-1 =G _k-1 −G _k-2 ... B _k −B _k-1 =B _k-1 −B _k-2 ... is used. The inventor of this application has confirmed that the modeling error caused by this model is close to white noise. Using the constant difference model described above, the difference between the three primary color signals at the current sample point and the sample point one sample before is predicted from the difference between the three primary color signals at the sample point one sample before and two samples before. Three primary color signals can be predicted. Based on this predicted value, the input signal of this sample point is
By correcting the observed values of the NTSC composite color video signal, it is possible to estimate the three primary color signals at the current sample point. The predicted value is obtained by creating a state equation from a model with a constant difference. The equation of state shows the above-mentioned constant difference model, by transforming the equation, R _k = 2R _k-1 −R _k-2 G _k = 2G _k-1 −G _k-2 B _k = 2B _{k- 1} −B _k-2 If the system noise is wR _k-1 , wG _k-1 , wB _k-1 respectively, the state equation is

[Table] In the equation, the state variable Xk
Let X ^T _K = [R _k-1 R _k-2 G _k-1 G _k-2 B _k-1 B _k-2 ] and the transition matrix Φ=2 1 0 0 0 0 -1 0 0 0 0 0 0 0 2 1 0 0 0 0 -1 0 0 0 0 0 0 0 2 1 0 0 0 0 -1 0 Then, the equation of state shown by the formula is X _k = ФX _k-1 + W _k-1 ... It is expressed as Since the system noise W _k-1 is close to white noise as mentioned above, the predicted value
X^k' can be predicted as X^ _k '=ФX _k-1 ... The three primary color signals of the current sample point are predicted by the formula. On the other hand, the observed value Zk observed as an input signal is
This is an NTSC composite color video signal for the current pixel. An NTSC composite color video signal can be expressed by a linear equation with three primary color signals. In other words, a color video signal based on the NTSC system is produced by converting the three primary color signals into a luminance signal Y and a color difference signal I as shown in the following equation.
, Q, Y = 0.3R + 0.59G + 0.11B ... I = 0.6R - 0.28G - 0.32B ... Q = 0.21R - 0.52G + 0.31B ... The obtained color difference signals I and Q are converted to frequency _SC A carrier color signal C is formed by performing quadrature two-phase modulation with the color subcarrier (its angular frequency is ω _C ) as shown in the following equation, C = Icos (ω _C t + 33°) + Qsin (ω _C t + 33°) )... It is created by superimposing this carrier color signal C and the luminance signal. Here, the matrix transformation shown by the formula ~ is the matrix A of the following formula, A=0.3 0.6 0.21 0.59 −0.28 −0.52 0.11 −0.32 0.31 ... If the superposition of the luminance signal and carrier color signal is expressed as the following formula, Dk=[1cosω _C t+33°sinω _C t+33°)... NTSC composite color video signal is matrix A
It can be expressed as a linear equation of the three primary color signals by and _Dk . That is, if the three primary color signals of the k-th sample are y _k =R _k G _k B _k , then the composite color signal Z _k of the k-th sample according to the NTSC system can be expressed as Zk = Dk·A·yk. The aforementioned state variable Xk (X ^T _k = [Rk, Rk _-1 , Gk,
Gk _-1 , Bk, Bk _-1 ]) to similarly show a linear equation, if C=1 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 1 0 then Zk=Dk・A・C・Xk... In the equation, if the observation matrix Hk is Hk=Dk・A・C, the observation equation is shown as Zk=Hk・Xk... When the observation matrix Hk is digitized by a sampling pulse with a frequency of 4 _SC , a matrix D _k = [1cos〓 _c t + 33° sin〓 _c t+33°] There are four types of values, and the matrices indicating other transformations are constants, so there are four types of matrices. When calculating the four types of observation matrices H ₁ , H ₂ , H ₃ , and H ₄ , H ₁ = [0.9176 0 0.0720 00.0105 0] H ₂ = [0.1493 0 0.3064 0 0.5443 0] H ₃ = [−0.3176 0 1.10 80 0 The value of 0.2095 0] H ₄ [0.4507 0 0.8736 0−0.3243 0] is obtained. Here, the current predicted value X^K′ obtained from the state equation described above and the current observed value Zk (Zk = Hk・
Xk) is compared. The predicted value X^k' is a predicted value of the current three primary color signals, and the observed value Zk is the current NTSC composite color video signal. In their original form, the two cannot be compared. As shown in the above equation, the observed value of the NTSC composite color video signal is expressed by a linear equation with the state variable Xk using the observation matrix Hk, so the predicted value X^k' is multiplied by the observation matrix Hk. The predicted value by
X^k' is converted into an NTSC composite color video signal, and the amount of correction can be determined by comparing the observed value Zk and the predicted value X^k'. Note that when the observation matrix Hk is digitized using 4 _SC sampling pulses as described above, it takes on four types of values. Based on the amount of correction between the compared observed value Zk and predicted value X^k', the amount of correction for each component is determined by a weighted average using a weight matrix Kk. If there is no observation noise and the variance of system noise is constant, the weight matrix Kk becomes a steady value through repeated processing. As a result of the simulation, the stationary values of the weight matrix K ₁ , K ₂ ,
K ₃ and K ₄ correspond to observation matrices H ₁ to H ₄ , and K ^T ₁ = [1.058 0.560 0.214 0.042 1.324 0.893] K ^T ₂ = [−0.055 −0.050 0.861 0.500 1.367
0.900〕 K ^T ₃ = [ 0.192 0.077 0.978 0.484 −0.108 −
0.294〕 K ^T ₄ = [1.144 0.683 0.424 0.250 −0.351 −
0.286] Four types of values were obtained. By using this stationary value, complex calculations such as inverse matrices are not necessary. From the above results, the correction amount for each component is the predicted value Hk multiplied by the observation matrix for comparison.
Weighted average based on the amount of correction between X^k′ and observed value Zk Ki (Zk−Hi・X^k′) where i=1, 2, 3, 4
It is indicated by... The current estimated value X^k is obtained by correcting each component of the predicted value X^k' using the correction amount for each component determined by the formula. This estimate
The predicted value for the next sample point is obtained in the same way using The estimated value is estimated, and the estimated value is estimated in the same way. In other words, the predicted value by the formula
Since Xk'=ФXk _-1 , the estimated value X^k corrected by the correction amount for each component shown in the equation is shown as the following equation. Based on this estimated value X^k, estimated values Rk, Gk,
We get B^k. X^k=ФX^k _-1 +Ki (Zk−Hi・ФX^k _-1 ) ...where i=1, 2, 3, 4 By realizing the hardware based on the above formula, it is possible to create an NTSC composite. A digital color decoder that directly converts a color video signal into three primary color signals can be realized. An example of a digital color decoder to which the present invention is applied will be described with reference to the drawings. In FIG. 1, reference numeral 1 indicates an input terminal, from which a composite color video signal Zk according to the NTSC system, which is digitized with sampling pulses having a frequency of 4 _SC , is supplied to a subtraction circuit 2. On the other hand, 3, 4, and 5 indicate the output terminals from which the estimated values R^k, G^k, and B^k of the three primary color signals based on the estimated value X^k are taken out, and 6, 7, and 8 indicate the It shows the terminals from which the estimated values Rk _-1 , Gk _-1 and Bk _-1 of the three primary color signals before sampling are taken out. The outputs of output terminals 3, 4, and 5 are output to delay circuits 9 and 1.
0 and 11, respectively. delay circuits 9, 10,
11 is a delay circuit having a delay amount of one sampling period, and the estimated values Rk, Gk, Bk of the three primary color signals based on the current estimated value Xk are taken out from the delay circuits 9, 10, and 11 to the output terminals 3, 4, and 5. , the estimated value Rk _-1 of the three primary color signals based on the previous estimated value Xk _-1 ,
Gk _-1 and Bk _-1 are extracted. Previous estimate Xk _-1
Estimated values Rk _-1 , Gk _-1 and Bk _-1 of the three primary color signals are supplied to adder circuits 33, 35 and 37, respectively, and also to multiplier circuits 12, 13 and 14, respectively. Multiplying circuits 12, 13, 14 double the respective values, and the outputs are sent to subtracting circuits 15, 16, 1.
On the other hand, the three primary color signals Rk -1 and Rk _-1 of the previous sample based on the current estimated values are supplied to terminals 6, 7 and 8, respectively.
Gk _-1 and Bk _-1 are taken out, and the outputs are supplied to delay circuits 18, 19, and 20, respectively. The delay circuits 18, 19, and 20 are delay circuits with one sampling period, and the delay circuits 18, 19, and 20 calculate the estimated values of the three primary color signals two samples before based on the previous estimated values Rk _-2 , Gk _-2 , Bk _{- 2} are taken out respectively. The outputs taken out to the delay circuits 18, 19, 20 are
The signals are supplied to subtraction circuits 15, 16, and 17, respectively. The subtraction circuits 15, 16, and 17 calculate the previous estimated value from the twice the estimated values of the three primary color signals Rk _-1 , Gk _-1 , and Bk _-1 of the previous sample from the previous estimated value Xk _-1 . The estimated values Rk _-2 , Gk _-2 , Bk _-2 of the three primary color signals two samples before by Xk _-1 are subtracted, and based on the state equation, the predicted value of the three primary color signals by the current predicted value Xk' is subtracted. Rk', Gk', and Bk' are obtained by subtraction circuits 15, 16, and 17. Subtraction circuit 1
The outputs of 5, 16, and 17 are added to adder circuits 32, 34, and 3.
6 and multiplier circuits 21 and 2, respectively.
2 and 23, respectively. Coefficients h ₁ , h ₂ , h ₃ of multiplication circuits 21, 22, 23
As shown in Fig. 2, four types of coefficients are used, each corresponding to the four sampling phases within one period of the color subcarrier, and these coefficients are based on the observation matrix H ₁ ~ It is determined by H ₄ . Such coefficients h ₁ , h ₂ , h ₃ are, for example,
Four types of values are written in each ROM, and each coefficient is selected for each sample. The output of the multiplication circuit 21 and the output of the multiplication circuit 22 are supplied to the addition circuit 24, and the output of the addition circuit 24 and the output of the multiplication circuit 23 are supplied to the addition circuit 25. These multiplier circuits 21, 22, 23 and adder circuits 24, 25 generate three primary color signals Rk', Gk', Bk' based on the current predicted value Xk' based on the observation matrix Hk.
Converted to a composite color video signal using the NTSC format. The output of the adder circuit 25 is supplied to the subtracter circuit 2. Subtraction circuit 2 receives the current NTSC signal from input terminal 1.
The value of the composite color video signal Zk according to the method is supplied, and the predicted value is calculated by the subtraction circuit 2.
Xk′ is the value converted to an NTSC composite color video signal and the current NTSC signal based on the input signal.
The values of the composite color video signals of the two systems are compared to determine the amount of correction. The correction amount obtained by the subtraction circuit 2 is applied to the multiplication circuits 26, 27, 28, 2.
9, 30, 31. Coefficients k ₁ , k ₂ , k ₃ , k ₄ ,
For k ₅ and k ₆ , four types of coefficients are used for each sampling phase, as shown in the coefficient table in _{FIG .} It is determined by
Coefficients k ₁ , k ₂ , k ₃ , k ₄ , each having four types of values,
For k ₅ and k ₆ , for example, four types of values are written in the ROM, and each coefficient is selected for each sample. The correction amount for each component obtained by the multiplication circuits 26 to 31 is added to the addition circuits 32, 33, 34, 3.
5, 36, and 37, respectively. Addition circuit 32
~37, the predicted values Rk', Rk' _{-1 of the three primary color signals based on the current predicted value Xk' from the adder circuit 15, delay circuit 9, adder circuit 16, delay circuit 10, adder circuit 17, and delay circuit 11} , Gk', Gk' _-1 , Bk', Bk _-1 are supplied, and the predicted values are corrected by the correction amount for each component by the adder circuits 32 to 37, and the three primary color signals Rk, Rk _-1 are based on the current estimated values. , Gk, Gk _-1 ,
Bk and Bk _-1 are calculated. Addition circuits 33, 35,
37 outputs are taken out to terminals 6, 7, and 8, respectively,
The outputs of the adder circuits 32, 34, and 36 are output to the output terminals 3,
4 and 5, and the estimated value of the current sample point is obtained. "Application Example" This invention is applicable not only to NTSC composite color video signals digitized by 4 _SC sampling pulses, but also to color video signals digitized by any sampling frequency. be able to. "Effects of the Invention" According to the present invention, a color video signal based on the NTSC system can be directly converted into R, G, and B three primary color signals in a single process. Further, according to the present invention, since the separation processing is performed in the time domain based on the equation of state, optimal separation processing is possible, and since the processing is performed sequentially, there is almost no signal distortion. . Further, according to the present invention, it is possible to realize a color decoder that is essentially capable of adaptive processing and noise removal processing.

[Brief explanation of drawings]

FIG. 1 is a block diagram of an example of an NTSC color decoder to which the present invention is applied, and FIGS. 2 and 3 are coefficient tables used to explain this color decoder. 1...Input terminal, 3, 4, 5...Output terminal, 2
1 to 23...Multiplication circuit for multiplying observation coefficients, 26 to
31...Multiplication circuit that multiplies weighting coefficients.

Claims (1)

[Scope of Claims] 1. A color decoding method for decoding digital three primary color signals from an input digital color video signal, including the steps of transitioning past estimated values of three primary color signals based on a state equation model to obtain current predicted values. , a step of multiplying the predicted value by an observation coefficient, calculating a correction amount by comparing the predicted value multiplied by the observation coefficient with the input digital color video signal, and multiplying the correction amount by a weighting coefficient to calculate the amount of correction for each component. A color decoding method comprising: a step of obtaining a correction amount; and a step of obtaining an estimated value of the current three primary color signals by adding the correction amount for each component to the corresponding predicted value.