JPH0562874B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to a band compression signal transmission method suitable for use when transmitting signals having a wide frequency band, such as high-definition television signals.

[Conventional technology]

For example, a video signal for reproducing high-definition images occupies a much wider band than the current video signal band, so such a wideband video signal can be digitized, recorded on, for example, a VTR, and played back from it. In this case, the recording band becomes wider,
Hardware implementation is not easy.

To solve this problem, a method of band compression of a wideband video signal can be considered. In this method, sampling data obtained by sampling a wideband video signal at a sampling frequency f ₀ that satisfies the Nyquist theorem is output every other time, and only the extracted sampling data is used to transmit, that is, to store in this example. It's a method.

According to this method, as shown in FIG. 10, it is almost equivalent to sampling a wideband video signal at a sub-sampling frequency f _s that is 1/2 of the sampling frequency f ₀ , and the frequency band is reduced to 60
This is because it can be compressed to about %.

However, when using this subsampling frequency _fs , it is not possible to faithfully reproduce a wideband video signal, so it is necessary to interpolate the sampling data using some means in the reproduction system. Data interpolation may be performed by generating average value interpolation data from a pair of mutually adjacent sampling data of the reproduced sampling data.

For example, as shown in Figure 11, the sampling point
P ₁ , P ₂ ... are sampling points when using a sampling frequency f ₀ that satisfies the Nyquist theorem, and the wideband video signal shown by the broken line is sampled at this sampling interval, and the odd-numbered sampling points P ₁ , P ₃ , ... (sampling points when sub-sampling frequency f _s is used) are recorded on the VTR, and during playback, the average value data of adjacent sampling data of these odd-numbered sampling data (Fig. The average value data is used as interpolation data.

In this way, since the reproduction system also satisfies the Nyquist theorem, it is possible to reproduce a wideband video signal.

[Problem that the invention seeks to solve]

By the way, although such a band compression signal transmission method can reproduce a wideband video signal, the signal reproducibility is extremely poor.

That is, the wideband video signal after the average value interpolation becomes as shown by the chain line in FIG. 11, and is restored as an envelope that is quite different from the envelope of the original wideband video signal.

This is because the average value is simply interpolated without referring to the non-extracted sampling data of the original sampling data.

Therefore, the present invention proposes a band compressed signal transmission method that improves restoration of a signal that is band compressed and transmitted using the above-described method, and can reproduce a signal with high fidelity.

[Means for solving problems]

In order to solve the above-mentioned problem, the present invention forms average value data from the reproduced sampling data, and adds error data (average value at multiple sampling points) obtained by taking non-extracted sampling data into consideration to this average value data. ) is used as interpolation data.

Specifically, sampling data sampled at a sampling frequency that satisfies the Nyquist theorem is extracted every other piece, and average value data is obtained from at least mutually adjacent sampling data among the extracted sampling data. Interpolation error data and data indicating its poles are obtained from the non-extracted sampling data corresponding to this average value data and the average value data. A plurality of pieces of this error data are used to obtain average value error data, and the average value error data, polarity data, and the extracted sampling data are transmitted.

On the receiving side, at least the average value data at the interpolation point is obtained from mutually adjacent transmission sampling data, and the interpolated data is obtained by adding the average value error data to this average value data, so that the original signal can be reproduced. Reproduce.

[Effect]

Since the average value error data is added to the average value data for interpolation in this way, the data error at each interpolation point is reduced compared to the case where the average value data alone is interpolated. Furthermore, since average value error data is used, errors in interpolated data are averaged.

〔Example〕

Next, the band compression signal transmission method according to the present invention is used to record and reproduce the above-mentioned wideband video signal.
The application to a VTR will be explained in detail with reference to FIG. 1 and subsequent figures.

First, a description will be given of how to form a band-compressed wideband video signal (digital signal).

In FIG. 1, the signal indicated by the broken line is a wideband video signal to be band-compressed, and the "o" mark is the point sampled at the sampling frequency f ₀ , A, B, C, . . . I, J. , K, . . . are the sampling data (quantization bits are, for example, 8 bits). Among these sampling data, odd numbered sampling data, therefore sampling data A, B, C, ... are extracted and recorded, and even numbered sampling data I,
J, K, . . . are not used as recording data.

Even numbered sampling data I, J, K,...
Interpolated data I'', J'', K'', . . . corresponding to ... are calculated by interpolation processing in the reproduction system.

Extracted subsampling data A, B, C
Average value data (interpolation point data) I′, J′, K′, ... found from mutually adjacent data of ...
Error data (absolute value) ε _i at the interpolation point and the polarity Q _i at that time are calculated from the difference between the true sampling data I, J, K, . . . corresponding to the interpolation point. Each error data ε _i and polarity Q _i of the average points i, j, k, . . . are as follows.

ε ₁ = |A+B/2-I|, Q ₁ =0...(1) ε ₂ =|B+C/2-J|, Q ₂ =0...(2) ε ₃ =|C+D/2-K| ,Q ₃ =1 ...(3) ... ... In the case of Fig. 1, I>I' (=A+B/2) and J
>J'(=B+C/2) and K<K'(=C+D/2)
Therefore, ε ₁ and ε ₂ are negative, and ε ₃ is positive.

In this example, when ε _i <0, the bit Q _i indicating the polarity is "0", and conversely, when ε _i >0, the bit Q _i indicating the polarity is "1". The polarity data Q _i in the case shown in the figure is as described above.

Now, such error data ε _i and polarity data Q _i
is determined over n sampling points, and then the average value of the error data ε _i is determined. n is given as a power of 2 to facilitate digital signal processing. Although n can take a numerical value such as 8, 16, . . . , in this example, n=8. Therefore, the average value error data (8-bit data) ε is as follows: ε=ε ₁ +ε ₂ + +ε ₈ /8 (4).

The polarity data Q _i is 8 bits (Q ₁ to _{Q 8} are collectively 1
This is taken as sample data Q.

Average error data ε obtained in this way
, polarity data Q, and each sampling data A, B...H at eight sampling points are put together to form one block of data S _D. A collection of such block data becomes the transmission form of the television signal after band compression. An example of the data arrangement of block data S _D is shown in FIG. Subsampling A, B, ...H is converted into video data block S _VD
shall be.

In the playback system, a TV data block consisting of sub-sampling data A to H at sub-sampling points is generated from the block data S _D shown in Fig. 2.
While decoding _SVD , interpolated data I'', J', K'', . . . are calculated from this video data block _SVD , average value error data ε, and polarity data Q.

Therefore, first, average value data I', J', K', . . . at interpolation points i, j, k, . . . are calculated from mutually adjacent sub-sampling data A to H, and then these average values Average value error data ε having the polarity set by the polarity data Q _i is synthesized with the data I′, J′, K′, . In other words, when the polarity data Q _i is “0”, the average value error data ε
is added, and subtracted when it is "1".

Therefore, in the case of FIG. 1, the polarity data of interpolation points i, j, k are Q ₁ = "0", Q ₂ = "0", Q ₃ = "1", so the interpolation data I'', J″, K″ are as follows.

I″=A+B/2+ε ……(5) J″=B+C/2+ε ……(6) K″=C+D/2−ε ……(7) Illustrating such interpolated data I″, J″, K″ Then, the result will be as shown in FIG. 1, and it can be seen that by adding the average value error data ε, the waveform of the restored television signal (shown by the solid line) becomes even closer to the original waveform.

If the encoding method using the average error data ε is called the average error encoding method, the band compression rate when using the band compression method that adopts this average error encoding method is 1 as described above. If the block data consists of sub-sampling data (8 bits) A to H of 8 sub-sampling points, average value sampling data ε (8 bits) and polarity data (8 bits) Q, the sampling points is 16, so compression ratio = (number of bits of error data ε) + number of bits of polarity data Q) / (sampling point) (
Number of bits) * *+ Sampling point) × (Number of bits) / ...(8) = 8 + 8 + 8 x 8/8 x 16 = 0.62
5...(9) In other words, the wideband video signal will be compressed to 62.5% and transmitted.

Of course, if n is 16, the wideband video signal can be reduced to 50
It can be compressed to about %.

3 and 4 show a specific example of realizing the above-mentioned band compression signal transmission method according to the present invention, and FIG. 3 shows an encoder 10 for forming a wideband video signal.
This is an example.

Terminal 1 receives sampling data A, B, ..., I, J, ... that has been digitally converted to 8-bit data.
... is supplied to the sampling data storage circuit 2, where it is stored in units of eight sub-sampling data, ie, video data blocks _SVD .

In addition, the sampling data A, B, ... supplied to the terminal 1 is passed through the first and second delay circuits (the delay time is 1/sampling frequency) 4 and 5 connected in series to form average value data. It is supplied to circuit 6. This forming circuit 6 includes data adders 7 and 1.
It is composed of a data shift circuit 8 that shifts data by bits. The adder 7 is supplied with the sampling data supplied to the terminal 1 and the output data of the second delay circuit 5, and the data shift circuit 8 outputs average value data I, J, .

These average value data _I , J, . By supplying this to the average value error data ε forming circuit 11, the average value is calculated based on the eight interpolated data ε ₁ to _{ε 8} .

The error data ε _i is further supplied to the polarity determining circuit 12 to determine the polarity Q _i of each error data ε _i . The polarity data Q, average value error data ε, and video data block S _VD are supplied to the data conversion circuit 15.
, the data is arranged as shown in FIG.

The band compressed video signal thus formed is supplied to a decoder 30 provided in the reproduction system, and the band is expanded to its original state.

FIG. 4 shows an example of a decoder 20 for band-compressed video signals that performs such operations, and this circuit system is also provided with an average value data forming circuit 21. This forming circuit 21 is composed of a data adder 22 and a data shift circuit ₂₃ .
B, . . . and the subsampling data A, B, . Average value data is formed.

On the other hand, of the block data _SD , average value error data ε is supplied to the register 27, and polarity data Q is supplied to the polarity register 28, where 8-bit polarity Q is stored. As for the polarity data Q, polarity data Q _i corresponding to the subsampling data is output at each subsampling timing. This polarity data Q _i is given its polarity to the average value error data ε sent out from the register 27 in a polarity giving circuit 29 .

The polarized average error data ±ε is converted into average value data I′, J′, K′, . . . by the data adder 30.
The target interpolated data I″, J″,
... is formed. This interpolated data I″, J″, ...
is supplied to the data inverse conversion circuit 32 together with the sub-sampling data A, B, C, . ，J″，……
Sampling data A, I″, B, where are inserted
J″,... is formed.

Since this sampling data includes interpolation data, its sampling frequency is _f0 , and a television signal with the band expanded to its original state is obtained.

By the way, although the above embodiment is applied to one-dimensional data interpolation using line correlation, the present invention can also be applied to two-dimensional data interpolation using vertical correlation.

In this case, subsampling data I, J, . . . on a certain horizontal line as shown in FIG.
The data I″, J″, K″, ... corresponding to ... include two sub-sampling data A, B, C, ... adjacent to the sampling data I, J, K, ..., and this horizontal line. Data interpolation processing is also performed using sub-sampling data X and Y existing on the previous and subsequent horizontal lines.

Therefore, the error data ε _i is obtained as follows.

ε ₁ = |A+B+X+Y/4-
I｜, Q ₁ = α ₁ (α ₁ 0 or 1)……(10) ε ₂ =｜B+C+V+W/4−
J｜, Q ₂ = α ₂ (α ₂ 0 or 1) ... (11) ... Therefore, the average value error data ε is: ε = ε ₁ + ε ₂ + ... ε ₈ /8 ... (12) Error An example of data ε ₁ is shown in FIG. The arrows in the figure indicate the signal level at each sampling point, a ₁ is the average value of A and B, a ₂ is the average value of X and Y, and the error data ε ₁ with respect to the true sampling data I is not shown. become that way.

As described above, the polarity data Q _i is obtained from the difference between the true sampling at the interpolation point and the average value data.

Interpolated data I″, J″, K″,… formed by the reproduction system
...It is determined from the average value of the four sub-sampling data and the average value error data ε, and the interpolated data I'' is as follows.

I''=A+B+X+Y/4±ε...(13) The polarity of the average error data ε is the polarity data Q ₁
= determined by α ₁ , + when α ₁ = 0, α ₁ = 1
When , it is -.

In the two-dimensional interpolation process described above, the upper and lower sampling data seen from the interpolation point on the preceding and following horizontal lines were used, but as shown in Figure 7, diagonal sampling data was also used, resulting in a total of six sub-sampling data. If you use the interpolated data
True sampling data I of I″, J″, K″, ...
The error rate for J, K, . . . can be further improved. If only ε ₁ of the error data ε _i at this time is shown, it is given by equation (14).

ε ₁ = |S+T+U+A+B+X+Y+Z/8-I...(14) Figures 8 and 9 show specific examples of implementing interpolation processing using the two-dimensional sampling data shown in Figure 7, and Figure 8 shows This is an example of an encoder 10 for forming a band compressed video signal.

In this case as well, the average value data forming circuit 40 is provided,
This is a data adder 41 and a bit shift circuit 42.
The adder 41 is an adder that adds a plurality of sampling data obtained from three horizontal lines.

Therefore, the sampling data supplied to the terminal 1 is passed through the first and second delay circuits 44 and 45, each having a delay time of one horizontal period, through a pair of series-connected delay circuits (the delay time of which is 1 of the sampling frequency) is supplied to 46, 47,
Subsampled data output from each is supplied to an adder 41. In addition, the first delay circuit 4
The output data of 4 is transmitted through three delay circuits connected in series (the delay time is 1/sampling frequency)5.
0 to 52, and is supplied to the first and final stage delay circuits 5
Each sampling data of 0 and 52 is supplied to an adder 41. Similarly, the sampling data supplied to terminal 1 is transmitted through a pair of series-connected delay circuits (the delay time is 1/subsampling frequency).
54 and 55, the input/output data of the first stage delay circuit 54 and the final stage output data are input to the adder 41.
is supplied to

As shown in FIG. 7, using a plurality of delay circuits, predetermined sampling data on three horizontal lines is supplied to the adder 41, so that data can be input from the bit shift circuit (in this example, a three-bit shift) 42. are the data I′, J′, ... shown in equation (14)
is output.

These average value data I', J _' , . By being supplied to the value data ε forming circuit 58, the average value thereof is calculated based on the eight pieces of error data ε ₁ to ε ₈ .

The error data ε _i is further supplied to the polarity determining circuit 59, and the polarity of each error data ε _i is determined.
Among the sampling data A, B, . . . , I, J, . . . which is the output data output from the delay circuit 44, the sampling data A _, B, . and stored in units of eight sub-sampled data. This video data block S
VD, polarity data Q, and average value error data ε are arranged in the data conversion circuit 61 into data as shown in FIG.

The block data _SD , which is a band compressed television signal thus formed, is supplied to a decoder 20 provided in the reproduction system, and the band is expanded to its original value.

FIG. 9 shows an example of a decoder 20 for compressed band television signals that performs such an operation, and this circuit system is also provided with an average value data forming circuit 70. This forming circuit 70 is composed of a data adder 71 and a data shift circuit 72, and the adder 2 is supplied with a plurality of sampling data.

Therefore, the block data S _D supplied to the terminal 24 is transmitted to the 1H delay circuits 74 and 75 connected in series.
A pair of delay circuits connected in series (the delay time is 1/subsampling frequency) 7
7 and 78, and their respective delayed output data are supplied to an adder 71. Similarly, delay circuit 74
The output of the signal is supplied to a delay circuit (its delay time is 1/subsampling frequency) 78, and its input/output data is supplied to an adder 71. Also, terminal 24
A pair of delay circuits in which TV signals are connected in series (the delay time is 1/subsampling frequency)
80 and 81, and the respective input/output data are supplied to an adder 71.

By supplying multiple pieces of output data at different sampling times to the adder 71, the sampling data for three horizontal lines as shown in FIG. 7 are supplied simultaneously in time, and the added output is sent to the bit shift circuit By shifting the data by 3 bits at step 72, the average value data I', J',
K', ... is calculated, and this is the data adder 86
supplied to

On the other hand, among the output data of the delay circuit 74, the average value error data ε and the polarity data Q are stored in the corresponding registers 83 and 84. As for polarity data Q, polarity data Q _i corresponding to sub-sampling data is output at each sampling timing. This polarity Q _i is assigned to the average value error data ε sent out from the register 83 by a polarity assigning circuit 85 .

The polarized average error data ±ε is converted into average value data I′, J′, K′, . . . by the data adder 86.
The target interpolated data I″, J″,
... is formed. This interpolated data I″, J″, ...
is supplied to the data inverse conversion circuit 87 together with the sub-sampling data output from the delay circuit 78, and is the sampling data in which interpolated data I'', J'', . . . are inserted between the sub-sampling data A, B, . A, I'', B, J'', ... are obtained. Since this sampling data includes interpolation data, its sampling frequency is _f0 , and a television signal with the band expanded to its original state is obtained.

According to this configuration, more subsampling data is used than in the data interpolation described above, so the error rate in data interpolation is improved.

In the above, this invention was applied to the case where a wideband video signal is band-compressed and recorded and played back on a VTR, but the invention can also be applied to other signal transmission systems, and the invention can also be applied to other signal transmission systems. can be applied not only to television signals but also to other wideband signals.

〔Effect of the invention〕

As explained above, according to the present invention, when a wideband signal is compressed and transmitted, when forming an interpolated signal on the receiving side, instead of the conventional method of simply interpolating the average value, it is formed from subsampled data. Since an average value error encoding method is adopted in which average value error data is added to the average value data obtained while taking into account its polarity, the error of the interpolated data with respect to the true sampling data at the interpolation point is minimized.

In this case, since the error data added to the average value data uses the average value of error data at a plurality of sub-sampling points, the errors in the interpolated data are averaged.

Therefore, the reproducibility of the transmitted signal is greatly improved, and when applied to a television signal recording/reproducing system, it is possible to obtain a signal with good image quality.

Moreover, according to the present invention, the configuration of the specific means for realizing the transmission method according to the present invention is relatively simple, so that it is only necessary to add some hardware to the conventional hardware.

For this reason, the present invention is extremely suitable for application to cases where wideband video signals for obtaining high-quality images are recorded and reproduced.

[Brief explanation of the drawing]

FIG. 1 is an explanatory diagram of one-dimensional processing for explaining the band compression signal transmission method according to the present invention, FIG. 2 is a diagram showing an example of encoded block data, and FIG.
Fig. 4 is a system diagram showing an example of an encoder for band compression television signal formation, Fig. 4 is a system diagram showing an example of the decoder, Fig. 5 is an explanatory diagram when extending the present invention to two-dimensional processing, and Fig. 6 is a system diagram showing an example of an encoder for forming a band compression television signal. The figure is an explanatory diagram of the interpolation error at that time, Fig. 7 is an explanatory diagram showing another example of two-dimensional processing, and Fig. 8 is a band compression TV used when realizing the two-dimensional processing shown in Fig. 7. FIG. 9 is a system diagram showing an example of an encoder for signal formation; FIG. 9 is a system diagram showing an example of the decoder; FIG. 10 is a sampling frequency diagram for explaining the present invention;
FIG. 1 is an explanatory diagram of a conventional band compression signal transmission method. A, B, C, ..., I, J, K, ... are sampling data, I″, J″, K″, ... are interpolated data,
ε _i is error data, Q _i is polarity data, 10 is an encoder, 20 is a decoder, 6, 21, 40, 70 are average value data formation circuits, 11, 58 are average value error data formation circuits, 12, 59 are polarity This is a discrimination circuit.

Claims (1)

[Claims] 1 On the transmitting side, every other sampling data sampled at a sampling frequency that satisfies the Nyquist theorem is subsampled to extract subsampling data, and among these sampling data, at least mutually adjacent subsampling data The average value data is calculated from the above, and the interpolation error data and its Obtain polarity data indicating polarity, use a plurality of the above interpolation error data to obtain the average value error data, transmit this average value error data, the above polarity data, and the above subsampling data, and on the receiving side, the transmission The average value data of at least the adjacent sub-sampling data is calculated from the data, and the average value error data obtained by multiplying the average value data by the polarity data is added to the average value data. A band compression signal transmission method characterized in that interpolated data corresponds to non-subsampled data.