JPS61264889A - Transmission system for multiple subsample - Google Patents

Abstract

PURPOSE:To improve the accuracy of a moving area detection by performing the inter-field offset subsampling of a video signal and performing the inter- frame offset subsampling with processing by a low-pass filter. CONSTITUTION:An input signal is sampled with 48.6MHz at an A/D converter 1. An inter-field pre-filter 2 is applied as the process of a still picture area and the high-pass component of a picture display in an oblique direction is eliminated. A field offset subsampling 5 is carried out with 24.3MHz and at the next stage, a subsampling frequency conversion 6 from 24.8MHz to 32.4MHz is carried out. The limitation of a frequency band to 12MHz is performed at a within-field prefilter 3 as the process of a moving picture area and the subsampling 5 is carried out with 24.3MHz and the sampling frequency conversion to 32.4MHz is carried out. A motion detection is performed separately and according to its volume, a linear mixture 7 between a still picture system and an animation system is performed. At the next stage, a frame offset subsampling 8 is carried out with 16.2MHz and the low-pass filter of 4MHz is applied and after it is D/A-converted, an analog signal is generated.

Description

Detailed Description of the Invention (Technical Field) The present invention relates to a transmission method in which a television signal is subsampled and transmitted for the purpose of compressing the transmission band. A multiple subsample transmission method using offset subsampling between frames and between fields, such as MUSE (M
ultimate 5ub-Nyquist Sample
There is a current high-definition television signal multiplex subsample transmission method called ingKncodlmg), which effectively realizes band compression.

However, in the case of the current multiple subsample transmission system, there is a problem in detecting a moving image area, which is necessary for performing signal processing on a moving image area and a still image area by distinguishing between the moving image area and the still image area. In other words, since the sub-sampling in the current method goes through one round every two frames, when performing motion detection, it is not possible to use the difference signal between one frame (there is no partner), and the difference signal between two frames must be used. motion detection was incomplete.

(Summary of the Invention) An object of the present invention is to eliminate the above-mentioned drawbacks, to perform motion detection using a complete l-frame difference signal, and to not only simplify the configuration of the receiver, but also to The present invention aims to provide a multiple subsample transmission method that can considerably improve image quality.

That is, the multiplex subsample transmission method of the present invention is a television signal subsample transmission method that compresses the transmission band using interfield and interframe offset subsamples. signal,
Prior to the interframe offset subsampling that follows the interfield offset subsampling,
This is characterized in that a transmission signal is prepared by processing using a low-pass filter having a cutoff frequency lower than the sampling frequency of the interframe offset subsampling.

(Example) FIG. 4 shows a sampling pattern (luminance signal) of this method. With sub-sampling link pattern in 2-frame l order,
This is exactly the same as the current high-definition television signal length x subsampling method (MUSE method). O mark in the diagram,
The 0 mark, ・ mark, 1 mark, and X mark are the 4nth and 4th marks, respectively.
The sub-sampling positions of the (n+1)-th, 4n+2-th, and 4n+8-th fields and sampling points that are not transmitted are shown, respectively. Also, d in the figure is the sampling interval, and the sampling frequency is 1/d, which is 64.8 MH
Corresponds to z. This method uses a combination of inter-field offset subsampling and inter-frame offset subsampling, with the former reducing the resolution in the diagonal direction and reducing information to l/2, and the latter reducing information from one screen to two frames. By transmitting the information over a period of time, the information per frame is halved.

By the way, when the signal transmitted after the above-mentioned signal processing on the transmitting side is played back on the receiving side, for a still image, an information signal that is completed in the order of z frames and l is sent, but in the case of a moving image, This relationship does not hold. Therefore, on the receiving side, intra-field interpolation is performed to reproduce necessary missing signals. Since a certain amount of visual blur is originally acceptable for moving images, it is desirable to transmit information with less information than for still images from the time of signal transmission, and to prevent the introduction of extra noise. Then, depending on the motion information, the transmitting side switches between still image information (wider band) and moving image information (narrower band), or linearly mixes the image information and transmits it, and the receiving side changes the transmitted image. The motion information is detected from the signal and the signal is decoded. In any case, it is necessary to detect motion information with higher accuracy from the transmitted image signal.

However, as mentioned above, in the case of the current multiplex subsample transmission method, sampling takes two frames for one cycle, so motion information must be detected based on the difference between two frames, and fast motion cannot be tracked. Not enough.

If we look at the above relationship from another perspective, we get the following 5. If the sampling pattern in Figure 4 is used to sample the band-limited pattern as shown in Figure 5 (aJ) at 16 MHz, as shown in Figure 5 (b), the frequency within the transmission baseband (below 8 MHz, current high High-frequency components of the signal (8MHz - 16MHz
) wraps around. Of course, at this time, the low frequency components and the folded high frequency components are interleaved so that they do not overlap (see Japanese Patent Application No. 59-80145). An inter-frame difference signal cannot be obtained from this signal format, and the aliased high-frequency component is removed using a two-frame signal (cancellation is performed by focusing on the opposite phase of the high-frequency component between adjacent frames due to frame offset subsampling). ), it is necessary to extract only the low-frequency components to obtain motion information. That is, this is two-frame difference signal processing.

The proposed transmission method improves this point, and FIG. 1 shows an example of the encoding procedure. First, as shown in Figure 1 (al), the signal whose band is limited to 24 MHz is
Inter-field offset sampling is performed at a frequency of MHz to obtain the frequency spectrum shown in FIG. 1(a). Next, as shown in FIG. 1(b), frequency components higher than 12 MHz are removed by a low-pass filter, and interframe offset sampling is performed on them at a frequency of 16 MHz as shown in FIG. 1(c). Then, as shown in the frequency spectrum of Figure 1 (C), the baseband component (0 to 8 MHz)
A spectrum that does not include folded components (components that are inverted between frames) is obtained in the θ to 4 MHz portion. Therefore, by applying a 4 MHz low-pass filter to the signal having the spectrum of Fig. 1 (dl) transmitted every frame, a signal portion without aliasing can be obtained.
Using this, the l-frame difference signal can be obtained.

The signal having the frequency spectrum shown in FIG. 1(d) is the transmission signal according to the method of the present invention, and the procedure for reproducing the original signal by interpolating it between frames and interpolating between fields is essentially This is exactly the same as the decoding of the current high-definition television signal multiple subsample transmission system (MUSE). However, according to the method of the present invention, since an accurate frame-to-frame difference signal is obtained, the moving image area and the still image area can be accurately distinguished, decoding is easy, and the reproduced image has good image quality.

Furthermore, from the decoding standpoint, the signal at the stage shown in FIG. 1(b) corresponds to the stage at which interframe interpolation has been completed, but it does not include the 16 MHz component.

This means that the 16 MHz false signal component that occurs when a moving image area is mistakenly decoded as a still image area can be removed regardless of whether it is a moving image area or a still image area. This shows that even if 16 MHz is processed and removed by a low-pass filter, the original information is not lost. Even if a moving image area is determined to be a still image area and decoded, the 16 MHz component, which is the thickest interference, will not occur at that time, leading to a significant improvement in image quality.

Hereinafter, specific embodiments of the apparatus of the present invention will be described with reference to FIGS. 3 and 8.

As shown in FIG. 3(a), as processing on the sending side, (I
I A/D converter 1 samples the input signal at 48.6 MHz. The signal band is shown in FIG. 8(a). In the figure, the horizontal axis is the horizontal component (7), and the vertical axis is the vertical component (7), and this relationship holds true in FIG.

(2) As a process for the still image area, the inter-field blurring filter 2 as shown in FIG. 8(1)) is used to remove high-frequency components in the diagonal direction of the screen.

(Field offset subsampling 5 is performed at 8124, 13 MHz. At this time, signals of 12.15 MHz or higher are folded back, and the signal band becomes as shown in FIG. 8(C).

(4) Next, perform sampling frequency conversion 6 from 24.8 MHz to 82i4 MHz. The signal band is the 8th
Figure (C1.

(5) As processing of the moving image area, the band is limited to 12 MHz using an intra-field blurring filter 8 as shown in FIG. 8(d).

(6) Next, subsampling is performed at 24.8 MHz, but since the 12 MHz K band is limited in (5), there is no aliasing (the signal band is as shown in FIG. 8(d)).

(7) Perform sampling frequency conversion 6 from 24.8 MI (z to 82.4 MHz. The signal band is the 8th
This is the frame shown in figure (d).

Here, in the processing of (6) and (7), there is no change in the signal band, so from (5) - from 48.6 MHz to 82.
Perform sampling frequency conversion to 4 MHz (
7) may be obtained.

(3) Motion detection 4 is separately performed, and the absolute value of the l-frame difference is subjected to nonlinear processing and the amount of motion is output.

(9) Linearly mix 7 the still image system (4) and the moving image system (7) according to the amount of movement described above.

(10) Next, frame offset subsampling 8 is performed at 16.2 MHz, but at this time, as shown in Figure 8 (C1) for the still image system (C1 is shown in Figure 8 (e)), as shown in Figure 8 (C1) for the moving image system (
dl is folded back at 8.1 MHz as shown in FIG. 8(f). FIG. 8(d) shows a signal band up to 12.15 MHz, and there is no aliasing component below 4 MHz. Therefore 4 M
If you apply a Hz low-pass filter and take the frame difference,
Reliable motion detection is possible without folding back. This part is the most important point of the present invention.

(11) Finally, perform D/A conversion 9 to create an analog signal. At this time, a transmission line filter of 5 MHz is inserted to obtain a cosine roll-off characteristic at 8.1 MHz, and the signal is sent to the transmission line.

On the receiving side, the process is reversed to that on the transmitting side, as shown in FIG. 3 (1) JK.

(12) Resampling is performed by A/D converter l. At this time, the signal band for still images is shown in Figure 8(e), and the signal band for moving images is shown in Figure 8(e).
It becomes 5 as shown in figure (f).

(18) In the still image system, pixels that are not sampled by interframe interpolation are replaced with signals from 17 frames earlier, and the folding of eJ is returned as shown in FIG. 8(c).

(14) 82.4 MHz to 24.8 MHz
Then, sampling frequency conversion 18 is performed. The signal band is as shown in FIG. 8(c).

(15) The interfield interpolation filter 14 restores the folded components. The signal band is J in FIG. 8 (1).

(16) For video systems, perform intra-field interpolation 12,
The folded component in FIG. 8(f) is returned as shown in FIG. 8(d).

(17) Perform sampling frequency conversion 1B from 8L4 MHz to 48.6 MHz. The signal band is the 8th
Mamata in figure (d).

(18) Next, a motion detection command is issued. Here, the signal is band-limited to 4 MHz, a frame difference is obtained, and nonlinear processing is performed on this to obtain the amount of motion.

(19) Perform linear mixing of the still image processing system and the moving image processing system according to the amount of motion.

(20) Finally, perform K D/A conversion 9 to obtain an analog signal. The signal band at this time is calculated, and the still image system is shown in Figure 1s8 (b), and the moving image system is shown in Figure 8 (dJ), and the slightly moved part is in the middle between the two.

(Effects of the Invention) According to the transmission method of the present invention, in a transmission method using inter-field and inter-frame offset subsamples, the difference between one frame can be used for detecting a moving area, so the detection accuracy is improved and the configuration is simpler. It's easy.

Also, even if you make a mistake in detecting the moving area, as mentioned earlier,
This has the advantage that major image quality deterioration does not occur due to this.

[Brief explanation of drawings]

Figure 1 is a diagram showing the frequency spectrum at each stage of the encoding procedure of a television signal (luminance signal) according to the transmission method of the present invention. Figure 3 is the decoding of the transmitting side (aJ) and the receiving side (b) according to the transmission method of the present invention. FIG. 8 is a diagram showing the band characteristics of each stage of decoding and two-encoding in FIG. 3. FIG. 4 is a diagram showing a sample pattern of multiple subsampling of television signals. , FIG. 5 is a diagram showing the frequency spectrum when the sample pattern in FIG. ...Intra-field briny filter 5...Field offset subsampling (24
,8MHz) 6...Frequency conversion (24,8MHz -+ 82.4
MHz)? ... Linear mixing 8 ... Frame offset subsampling (I16
．． 2 MHz) 9...D7. Converter lO...Transmission line filter 11...Inter-frame interpolation 12...Intra-field interpolation 1B-...Frequency conversion (82,4MHz -+
24.8 MHz )14... Interpolation between fields - Sakura→ -#-緘(-A'L/→ Beak--n→v
v

Claims (1)

[Claims] 1. In a television signal subsampling transmission method that compresses the transmission band using interfield and interframe offset subsamples, a television video signal is first subjected to interfield offset subsampling, and the resulting signal is is processed using a low-pass filter having a cutoff frequency lower than the sampling frequency of the inter-frame offset sub-sampling, prior to the inter-frame offset sub-sampling following the inter-field offset sub-sampling, and a transmission signal is prepared. A multiple subsample transmission method that uses 2. The multiple sub-sampling transmission system according to claim 1, wherein the ratio of the sampling frequency of the inter-field offset sub-sampling to the inter-frame offset sub-sampling frequency is 8:2. 3. On the receiving side of the multiple subsample transmission method,
4. The multiple sub-sampling system according to claim 1, wherein when decoding the transmission signal, the sampling frequency component of the inter-frame offset sub-sampling is removed after performing inter-frame interpolation. Transmission method.