JPS6132877B2 - - Google Patents

Description

[Detailed description of the invention]

The present invention samples a composite color television signal in which a color signal such as an NTSC signal is modulated at a subcarrier frequency ( _fsc ) and is superimposed on a luminance signal, and extracts the original composite color television signal from this sampled signal. It relates to a circuit that reproduces signals. To sample and encode a composite color television signal, the sampling frequency f _S is usually chosen to be three times the color subcarrier frequency f _SC , so the sampling frequency is high and the bit rate of the encoded output is high. There was a problem. For this reason, several so-called sub-Nyquist sampling methods have been proposed in which the sampling frequency f _S is twice or less the band f _C of the composite color television signal. The outline and problems are described below. FIG. 1 shows the transmission frequency band of a composite color television signal of the NTSC system on a two-dimensional frequency plane in which the horizontal frequency is represented on the horizontal axis and the vertical frequency is represented on the vertical axis. Here, the horizontal frequency F _H indicates the cycle of change on the scanning line of the image as the number of cycles per horizontal scanning period, and the normal frequency corresponds to this multiplied by the horizontal scanning frequency f _H . On the other hand, the vertical frequency F _V indicates the cycle of vertical change of the image signal as the number of cycles per vertical scanning period,
In the case of the NTSC method, the number of vertical sampling cycles is 525 scanning lines. In this case, the subcarrier frequency f _SC of the NTSC signal corresponds to a position of 227.5 cycles in the horizontal direction and 131.25 cycles in the vertical direction. When this signal is sampled at frequency f _S , its harmonics appear symmetrically around f _S on the diagram. In the same figure, f _S
= 3f The harmonic components in the case of _SC are also shown. If the sampling frequency is lowered to less than twice the band f _C of the original image signal, the band of the original image and the lower sideband of the harmonic component (this will be referred to as a folded component) will overlap. A region in which this overlap does not occur is determined in a two-dimensional frequency domain, and is hereinafter referred to as a transmission band. The first conventional example of sub-Nyquist sampling of an NTSC signal is to change the sampling frequency f _S to f _S =2f _SC ±1/2f _H (that is, horizontal direction F _SH ≒
455, vertical direction F _SV =131.25), and the shaded area is the transmission band. In this case, since the band on the horizontal axis of the luminance signal is limited to about 3 MHz, there is a problem that the resolution of the raw image deteriorates. The second conventional example converts f _S to f _S =2f _SC ±1/4f _H (that is, F _sH
≒455, F _SV =65.625), and the shaded area in FIG. 2b is set as the transmission band. In this case, since the vertical band of the aliased portion is limited to 1/4 of the original image, there is a problem that image deterioration increases in images that contain many high-frequency components in the vertical direction (i.e., edge portions in the horizontal direction). It was hot. In the third conventional example, as shown in FIG. 2c, f _S =
Choose nf _H (i.e. F _sH = n, F _SV = 0) and set 0
~1/2f _S transmits the vertical low frequency component, and 1/2f _S ~f _S transmits the vertical high frequency component. In this case, there was a problem in that the resolution deteriorated significantly because the vertical transmission band from 0 to 1/ _2fs was narrow. The present invention has been made to solve the above-mentioned problems, and an object of the present invention is to provide a sub-Nyquist sampling reproduction circuit with less deterioration in image quality. In order to achieve the above object, in the sampling reproduction circuit of the present invention, the sampling frequency f _S is set higher than the highest frequency band f _C of the color television signal, and 1/2f _S is the lower limit frequency of the carrier color signal (2f _SC − f _C , the carrier color signal band is approximately equal to or less than f _SC ±(f _C − f _SC )), and f _S is approximately an integer multiple of the horizontal scanning frequency f _H (F _SV = 0 or F _SV =
262.5), and from the signal sampled at such a sampling frequency f _S , components in the vicinity of integral multiples of f _H are passed at least up to DC ~ 1/2f _S , and (of f _H For components in the vicinity of integral multiples + 1/2f _H ), the bands of DC ~ (f _S - f _C ) and 1/2 _{f S} ~ f _C are passed, and the band of (f _S - f _C ) ~ 1/2 f _S is passed. The composite color television signal is reproduced by processing the color to prevent it from occurring. That is, DC ~ (f _S − f
Since there is no aliasing component in the frequency band _C ), both integral multiples of f _H and the vicinity of integral multiples of f _H + 1/2f _H are passed, and the frequency band from (f _S − f _C ) to 1/2 f _S is passed. The carrier color signal exists centered on a point whose horizontal frequency and vertical frequency are equal to (f _S − f _SC ) and f _SC , respectively, and its frequency component, that is, (integer multiple of f _H + 1/2 f _H ) Block nearby components. Since there is a carrier color signal in the band 1/2f _S to f _C , it is necessary to pass at least that part.
Let the passband be an integer multiple of _fH + 1/ _2fH . Specific examples will be described below. First, the range of sampling frequencies that satisfy the above conditions will be described. In the case of NTSC system, f _SC =3.58MHz, f _C =
4.2MHz, for narrowband color signals, the band is
0.5MHz (for normal home color receivers)
Therefore, f _S is in the range of 4.2MHz<f _S 6.2MHz. As a typical example within this range,
In the following embodiments, f _S ≈1.5f _SC will be explained. FIG. 3 shows the desired sampling frequency for f _S ≈1.5f _SC . Figure a shows the case where the number of horizontal cycles of the sampling frequency f _S is F _sH ≒ 341, F _SV = 262.5,
The phase of the sampling frequency f _S with respect to the current scanning line is
This indicates that at least the scanning lines near the same field have the same phase, and the phases of the scanning lines near the previous and succeeding fields differ by 180 degrees from each other. Fig. 3b shows the case where the sampling frequency is the same as that in Fig. 3a, or when the sampling frequency is completely an integer multiple of _fH , or when it is almost an integer multiple, and the phase is the same at least in nearby scanning lines. (f _S ' in the figure). A two-dimensional frequency region in which the aliasing components generated by sampling do not overlap with the original signal for each sampling frequency is indicated by diagonal lines in the figure. Comparing this two-dimensional transmission area with the two-dimensional transmission area shown in Figure 2 when the sampling frequency of the conventional system is moved to the same horizontal frequency position, the second
In figure a, the number of horizontal transmission cycles is 227.5/4 = 56.875
Therefore, the visually important horizontal transmission band is f _SC /4
Limited to ≒0.9MHz. In the case of Figure 2b, the number of stopband cycles closest to the origin is

[Formula] The vertical band from horizontal direction f _SC /4 to f _C is 1/2 that of the method of the present invention. In the case of Figure 2c,
The visually important number of vertical cycles is limited to 131.25/2, resulting in significant image quality deterioration. FIG. 4 shows the configuration of a television signal transmission device to which the present invention is applied. In the same figure, the composite color television signal input to terminal 1 is converted to a frequency 2f _S which is twice the sampling frequency f _S in the analog/digital conversion circuit 2.
sampled and digitally encoded. This repeated digital code of 2f _S is filtered by a filter 3 whose transmission characteristic is shown by the shaded area in FIG _. : 1 and repeatedly obtain a signal of f _S . This signal is transmitted to the transmission path 5 where it is encoded by an encoding circuit (not shown) as required. On the receiving side, the code of the repeated f _S decoded by a decoding circuit (not shown) is alternately switched with the "0" code by the switching circuit 6 as necessary to obtain the code of the repeated frequency 2f _S , which is then passed through the filter. Filter 7 with similar characteristics to 3
, remove the aliasing component, and obtain the obtained 2f _S
A composite color television signal can be obtained at the terminal 9 by converting the code of . Figure 5 shows the frequency f _S or f _S ' shown in Figure 3.
When sampled in the original signal (area A, B, ...,
H) and its aliasing noise components (areas A', B',...
…, H′) is shown. In other words, the arrangement of the folded components obtained by sampling in Figure 3a is shown in Figure 5a,
The arrangement of folded components due to f _S and f _S ' in FIG. 3B are shown in FIGS. 5B-1 and 5B-2, respectively. Filters 3 and 7 in the block diagram shown in FIG. 4 remove these areas A', B', C', D' or E', F', G', H'. Next, the specific configuration of the filter will be described.
In the characteristics shown in Figure 5a, the transmission characteristics of the filter whose passbands are regions A and C' are The transfer characteristic of a filter with passbands B and D is expressed as the product of It is expressed as the product of Also, area C is It is expressed as the product of Of these, the filter 3 indicating area C is an 8th-order filter and has a complicated configuration, and even if area C is removed, the image quality will not be affected much. Therefore, in practice, if a filter is constructed whose passbands are regions A, B, D, and C' as transmission characteristics, a good reproduced image can be obtained since there is no aliasing component in C'. Part 2
The dimensional impulse response (the gain is multiplied by 16 so that the coefficients are integers) is shown in FIG. 6a. In the same figure, the 〇 or △ mark indicates the sample position according to f _S , and the combination of 〇 and △ marks
2f Describe _{the S} sampling. The transfer characteristic of the filter whose passbands are E and G in Figure 5b-1 is Therefore, the transfer characteristic of a filter whose passband is Fig. 3b is the product of equation (4) and equation (2).
The impulse response is shown in FIG. 6b. FIG. 7 shows an example of the configuration of the filter 3 or 7 whose impulse response is shown in FIG. 6b. In the figure, a repeating 2f _S signal inputted to a terminal 10 is inputted to two continuously connected delay circuits 11 and 12 having a delay time of one horizontal scanning period.
When the three inputs and outputs are added by an adder circuit 13 that adds coefficients 1, 2, and 1, and an adder circuit 14 that adds coefficients -1, 2, and -1, the output has a vertical response of 4cos. ² (π/f _S f), 4sin ² (π/f _S f). These outputs are added by the adder circuit 20 to the delay circuits 15 and 16 of 2 clock cycles, the adder circuit 17, the delay circuit 18 of 6 clocks, and the adder circuit 19 using the 2f _S clock, and a coefficient circuit with a coefficient of 1/8 is added. By adjusting the gain at , and adding it to the output of the delay circuit 3 of 3 clock cycles in the adder circuit 23, a filtered signal of 2f _S can be obtained repeatedly at the terminal 24. That is, adder 1
The output of 9 becomes the first interpolation signal of the interline difference signal of the sample value having the same carrier color signal phase as the carrier color signal phase of the interpolation signal (signal at the position indicated by the center coefficient 8 in FIG. 6b). , the output of the adder 17 becomes a second interpolation signal which is an interline sum signal of neighboring sample values including the interpolation signal. The sum of these first and second interpolation signals is generated as an interpolation signal at a frequency _fS ,
The adder 23 adds the input sample values so that they are inserted in the middle of the signal, resulting in a signal with a sampling frequency of _2fS . Note that the filter 7 can be simplified by slightly modifying the configuration shown in FIG. That is, the capacity of delay circuits 11, 12, 15, 16, 17, 18, and 22 is halved, and the operating frequency is changed from 2f _S to f _S (22 may be f _S with a delay of one clock delay). By making the adder circuit 23 a switching circuit that repeatedly operates at 2f _S and alternately switching the outputs of the delay circuit 22 and the coefficient circuit 21, a desired signal of 2f _S can be obtained. In this case, the switching circuit 6 shown in FIG. 4 becomes unnecessary. The filter shown in FIG. 6a can be similarly realized by using a delay circuit of 262H (H is the horizontal scanning period). In particular, when the present invention is applied to a so-called still image transmission device that temporarily stores a still image in a frame memory, slowly transmits it via a narrowband transmission path, and also installs a frame memory on the receiving side to reproduce the image. This frame memory and the delay circuit of the field period (262H) of the filters 3 and 7 can be shared, and can be realized without increasing the number of circuits. Although the present invention has been described in the case of image transmission as a specific embodiment, it is also possible to replace the transmission line in FIG. 4 with a storage circuit. As described in detail above, according to the present invention, a composite color television signal is sampled at a subcarrier frequency of less than twice its subcarrier frequency f _SC , particularly around 1.5 times f _SC , and although there is a slight decrease in resolution in the diagonal direction, conventionally known According to the present invention, by greatly reducing the sampling frequency, the transmission band of the image transmission device or the amount of data stored in the image storage can be greatly reduced. Compression becomes possible.

[Brief explanation of the drawing]

1, 2, 3, and 5 are diagrams showing transmission frequency bands on a two-dimensional frequency plane for explaining the operation of the conventional method and the present invention, and FIG. 4 is a television signal to which the present invention is applied. Diagram showing the configuration of the transmission device, No. 6
The figure shows the relationship between sample surface elements on the screen, and FIG. 7 is a circuit diagram showing the configuration of filters 3 and 7 in FIG. 4.

Claims (1)

[Claims] 1. A composite color television signal in the frequency band 0 to f _C in which the color signal is modulated with a subcarrier signal (f _SC ) and superimposed on the luminance signal is sampled at a sampling frequency f lower than the frequency 2f _C. In the circuit that samples and reproduces with _S ,
The above sampling frequency f _S is higher than f _C and 1/2f _S
The composite color television signal is sampled at a sampling frequency f _S selected such that it is approximately equal to or less than (2f _SC - f _C ) and such that f _S is approximately an integer multiple of the horizontal scanning frequency f _H . The _frequency band of the output of the sampling circuit and the output of the sampling circuit from DC to (f _S - f _C ) passes through the vicinity of an integer multiple of f _H /2, and
The frequency band from f _C ) to 1/2f _S blocks components in the vicinity of at least an integral multiple of f _H + 1/2 f _H , and the frequency band from 1/2 f _S to f _C
The frequency band is at least an integer multiple of f _H + 1/2 f _H
1. A sampling and reproducing circuit for a composite color television signal, comprising a reproducing circuit that passes components in the vicinity of the composite color television signal, and reproducing the composite color television signal. 2. In the circuit according to claim 1, the sampling frequency f _S is selected to be approximately 1.5 times f _SC , and the reproducing circuit generates an interpolation signal between two adjacent input sampling signals. It consists of an interpolation circuit and a circuit that adds the interpolation signal and the input sampling signal to obtain a sampling signal with a sampling frequency of 2f _S , and the interpolation circuit adds a color feeding signal having the same phase as the carrier color signal phase of the interpolation signal. The first signal is a line-to-line difference signal of sampled values having a phase.
and a second interpolation signal which is an interline sum signal of neighboring sample values including the interpolation signal, which outputs the sum as an interpolation signal.