JPH04172887A - Video signal encoder and decoder - Google Patents

Abstract

PURPOSE:To efficiently encode video signals by dividing the band of the video signals by performing filtration in an oblique direction after forming a sample pattern to an oblique grate. CONSTITUTION:Video signals inputted from an input terminal 50 are divided into sampling points of two kinds of five-mesh grids by means of a five-mesh grid division circuit 400 and respectively sent to the 1st oblique dividing filters 410 and 420. The filters 410 and 420 respectively make band division by performing filtration in an oblique direction on the sampling points of the two kinds of five-mesh grids. Similarly, the 2nd oblique dividing filters 450-480 make band division by performing filtration in an oblique direction which is different from that of the 1st filters 410 and 420. The band-divided signals are sent to a multiplex circuit 600 after the signals are respectively encoded suitably to the frequency bands by means of encoders 500-570. The circuit 600 multiplexes the signals on the time base and the multiplexed signals are outputted from a terminal 60.

Description

DETAILED DESCRIPTION OF THE INVENTION Field of the Invention The present invention relates to a video signal encoding device and decoding device used when a video signal is digitized and transmitted or recorded.

2. Description of the Related Art In recent years, efforts have been made to improve the quality of television images, and a high-definition television system that is completely different from the current television system has been proposed by the Japan Broadcasting Corporation (for example, in the literature [Special feature "High Definition Television", Journal of the Televisier Society, Vol. 36S, No. 10, 1982).

This high-definition television system has about five times the amount of information compared to the current television system, and the transmission circuit has a wide band.How to efficiently transmit high-definition television signals with a large amount of information. How to perform recording is considered to be one of the important issues, and various high-efficiency encoding methods have been proposed.

One of the effective coding methods that has been proposed in the past is the subband coding method (for example, "Subband Extrapolation and Interpolation Predictive Coding Method J Television Society National Conference Proceedings, p.
p457-458.1989), this subhand encoding method uses a quadrature mirror filter (QMF).
etc. (for example, a filter family
Foreuse in quadrature mirror filter (
A Filter FarxrIy Desjgned
for use 1nQuadra-ture Mi
rror Filter BanKs J) 1980
Icy Ninis SP (ICASSP), pp2
91 to 294), the video signal is divided into several different frequency bands, and optimal encoding and quantization are performed for each frequency band.

FIG. 4 is a diagram showing the configuration of a conventional subband encoding system.

In FIG. 4, a video signal is input from an input terminal 10, passes through a horizontal division filter (H division filter) 100 that performs band division according to the frequency in the horizontal direction, and is then vertically divided to perform band division according to the frequency in the vertical direction. Encoder 1 for inputting a filter (V-divided filter) divided into 110° and 120 frequency bands and performing encoding in each divided frequency band.
50,. 160, 170, and 180, and is applied to a multiplexing circuit 200 that temporally multiplexes the signals encoded by the respective encoders, and the encoded video signal is output from the output terminal 20. .

The above is the configuration of the encoding device. Also, in the same figure,
The encoded video signal is input to an input terminal 30, passes through a demultiplexer circuit 300 that separates the temporally multiplexed signal, and then is sent to a decoder 3 which encodes each frequency band. It is added to ゜320.330,340.

Furthermore, it passes through vertical synthesis filters (■ synthesis filters) 350 and 360 that synthesize vertical frequency components, and is synthesized into horizontal frequency components by horizontal synthesis filters (H synthesis filters) 370, and then is decoded as a video signal. It is output from the output terminal 40. The above is the configuration of the decoding device.

The operation of the conventional encoding device and decoding device configured as described above will be described below with reference to FIGS. 4, 5, and 6. Here, FIG. 5 is a sample pattern in the encoding device shown in FIG. 4, and FIG. 6 is a graph showing characteristics of band division in FIG. 4. Further, the horizontal division filter 10 shown in FIG.
0 and the horizontal synthesis filter 370, and the vertical division filters 110 and 120 and the vertical synthesis filter 350 and 360 are the quadrature Mira filters (Q
MF), etc.

The encoding device will first be described below. In FIG. 4, a video signal input from an input terminal 10 of the device is sent to a horizontal division filter 100, and is divided into a low frequency component (L) and a high frequency component (H) at the horizontal frequency. The divided horizontal frequency low frequency components (L) and constituent components (H) are sent to vertical division filters 110 and 120, respectively. The vertical division filters 110 and 120 divide the vertical frequency into low frequency components and high frequency components, and the divided video signal is sent to the encoder 150.
．． Send to 160, 170.180. Here, the transition of sample points and the state of band division will be explained using FIGS. 5 and 6. The horizontal division filter 100 and the vertical division filters 110 and 120 shown in FIG. 4 are composed of QMF filters, etc., and during band division, sample points are filtered (passing through a filter is hereinafter referred to as filtering). Thinned in half in the direction. FIG. 5 shows the transition of sample points at this time, as shown in FIG. 5(a). The grid-like original sample points of the original image are processed by the horizontal division filter 100 as shown in FIG.
As shown in Figure 5(C), the pixels are halved in the horizontal direction, and then by the vertical division filter 110°120, the pixels are further halved in the vertical direction as shown in Figure 5(C). The point is 1/4 of the original image.

Also, FIG. 6 shows the state of band division at this time.
This is shown on the dimensional frequency, and band division is performed using a horizontal filter and a vertical filter, so the sixth
As shown in the figure, each is divided by a straight line parallel to the horizontal frequency axis and the vertical frequency axis. Hereinafter, the explanation will be given again using FIG. 6. Encoder 1.50, 160, 170.180
The video signal sent to is encoded here. The encoders 150, 160, 1.70, and 180 each include a circuit including an encoding circuit and a quantization circuit, and encode with encoding characteristics and quantization characteristics suitable for each frequency band. In the multiplexing circuit 200, encoders 150, 16
Multiplexing is performed by temporally switching video signals coded at 0, 170, and 180. The above configuration constitutes an encoding device, and an encoded video signal is output from the output terminal 20 of the device.

Next, the decoding device will be explained. In FIG. 4, an encoded video signal input manually from an input terminal 30 of the device is sent to a demultiplexing circuit 300. Demultiplexing circuit 3
00, the encoding side divides the temporally multiplexed signals for each frequency band and sends them to the decoders 310, 320, and 33.
Sent to 0.340. Decoder 310.320.33
0°340 is a block that performs the opposite process to that of the encoder on the encoding side, and is composed of a circuit including a decoding circuit and an inverse quantization circuit, and according to each encoding method performed on the encoding side, Decoding is performed for each frequency band. In the vertical synthesis filter 350° 360, the decoder 310.3
The video signals for each frequency band decoded at 20, 330° and 340 are frequency-combined using a vertical frequency. Thereafter, the horizontal synthesis filter 370 performs frequency synthesis at the horizontal frequency, restores the original video signal, and outputs it from the output terminal 40 of the device. The decoding device is configured with the above configuration.

Problems to be Solved by the Invention In the conventional subhand encoding system configured as described above, a video signal is divided into multiple frequency bands at horizontal and vertical frequencies, and appropriate encoding is performed according to each frequency band. It has been devised in the past due to its simple circuit configuration and high encoding efficiency.

However, in such a conventional configuration, since band division is performed at horizontal and vertical frequencies, the band division pattern on the two-dimensional frequency becomes a square grid, and up to relatively high frequency components in the diagonal direction are divided into horizontal frequencies. Both horizontal and vertical frequencies are divided as low-frequency components, and although the high-frequency components in the diagonal direction are not visually noticeable, both the horizontal and vertical frequencies, which are important for images, are divided into low-frequency components. The problem is that encoding is performed in the same way as the components, and sufficient encoding efficiency cannot be achieved.

The present invention solves the above-mentioned problems by forming a band division pattern in a diagonal lattice pattern on two-dimensional frequencies, and for frequency components in diagonal directions that are visually inconspicuous, up to relatively low components, horizontal frequencies and vertical Similar to the high frequency component,
It is an object of the present invention to provide an encoding device and a decoding device with high encoding efficiency by performing encoding with coarse quantization steps.

Means for Solving the Problems The present invention provides an encoding device for achieving the above object.
By classifying the square grid sample points into two types of quincunx lattice sample points and performing two diagonal filters on these quincunx lattice sample points, the video signal is converted to a two-dimensional frequency. The band is divided into diagonal lattice shapes and then encoded.

In addition, the decoding device inputs the video signal encoded as described above, decodes it for each divided frequency band on the encoding side, and then synthesizes the signals for each frequency band into two types. The original signal is restored by restoring the sample points in the quincunx grid pattern, and then combining the two types of sample points in the quincunx grid pattern.

According to the above-described configuration, the present invention classifies the square grid sample points into two types of quincunx grid sample points, and performs diagonal filtering on the diagonal grid sample points to diagonally convert the video signal. This method performs band division based on frequency in the direction, and for frequency components in diagonal directions that are not visually noticeable, it is possible to encode relatively low components with coarse quantization steps similar to the high-frequency components of horizontal and vertical frequencies. This makes it possible to increase encoding efficiency.

Embodiment Hereinafter, an encoding device and a decoding device according to an embodiment of the present invention will be described with reference to the drawings.

FIG. 1 is a block diagram showing the configuration of an encoding device and a decoding device according to an embodiment of the present invention.

In FIG. 1, a video signal is input manually from an input terminal 50 and is applied to a quincunx lattice dividing circuit 400 for dividing a square lattice sample point into two types of quincunx lattice sample points. After that, it is divided into two parts and passes through first diagonal dividing filters 410 and 420 that divide the frequency in a first diagonal direction.
It is divided into a second diagonal division filter 450.460.470 degrees that divides the frequency in a second diagonal direction different from the first direction. Encoders 500, 510° 520, 530, 54 for further encoding the band-divided video signal
0.550.560° 570 and is applied to a multiplexing circuit 600 that temporally multiplexes the signals encoded by the respective encoders. The encoded video signal is output from the output terminal 60. The above is the configuration of the encoding device of this embodiment. Further, in the figure, an encoded video signal is inputted from an input terminal 70, and a temporally multiplexed signal is separated by a demultiplexing circuit 700. After that, the encoded signals for each frequency band are sent to decoders 710 and 72.
0,730°740.750,760.770.780
decrypted with . It passes through the second diagonal synthesis filters 800, 810, 820, 830, which recombines the frequency bands divided by the second diagonal division filter on the encoding side, and is further divided by the first diagonal division filter on the encoding side. First diagonal synthesis filters 850, 860 for resynthesizing frequency bands
After the signals are divided into two types of quincunx lattice sample points on the encoding side, they are combined again and added to the quincunch lattice synthesis circuit 900 which outputs square lattice sample points for decoding. 9 or more of the video signals outputted from the output terminal 80 are the configuration of the decoding apparatus of this embodiment.

The operation of the encoding device and decoding device configured as described above will be explained below with reference to FIGS. 1, 2, and 3. Here, FIG. 2 is a pattern diagram showing changes in sample points of the encoding device shown in FIG. 1, and FIG. 3 is a graph showing band division in FIG. 1. In addition, in FIG. 1, first diagonal division filters 410 and 420 and first diagonal synthesis filter 8
50.860, second diagonal dividing filter 450°460.
470.480 and second diagonal synthesis filter 800.810
, 820, and 830 perform band division and synthesis by filtering in the same direction on the sample pattern, respectively, and are similar to the quadrature mirror filter (Q
MF), etc.

The encoding device of this embodiment will be described below first. In FIG. 1, a video signal inputted from the input terminal 50 of the device W is divided into two types of quincunx grid sample points by a quincunx grid dividing circuit 400, and each is sent to a first diagonal dividing filter 410 and 420. Sent. First diagonal division filter 4
10.420 performs band division by performing diagonal filtering on the sample points of each quincunx grid. Similarly, the second diagonal dividing filter 450°460.4
In 70.480, band division is performed by filtering sample points of each quincunx grid in a diagonal direction different from that of the first diagonal culture filter. Here, the transition of sample points will be explained using FIG. 2.

First, the original sample points in the square lattice shape shown in FIG.
) is divided into two types of sample points in a quincunx grid pattern. Next, the first diagonal dividing filter 41o.

In 420, the sample points are thinned out in half in the filtering direction based on the principle of the QMF filter described above, so the second
The sample points are shown in Figure (C). Furthermore, in the second diagonal dividing filter 450.460°470.480, the sample points are thinned out in half in the filtering direction, resulting in sample points as shown in FIG. 2(d). Figure 30 shows the band division at this time on a two-dimensional frequency.The frequency domain of the original image shown in Figure 3(a) is first divided into diagonal high-frequency components by quincunx grid division. By folding back into the area and further filtering at diagonal sample points, the band division shown in Fig. 3 fb) is obtained. At this time, components with very high horizontal and vertical frequencies are folded back to the low frequency range, but such components generally have little power.

Hereinafter, the description of the encoding device of this embodiment will be continued using FIG. 1 again. 2nd diagonal split filter 450° 460.470
, 480 is processed by an encoder 500.5.
10.520,530°540.550.560,57
0, encoding suitable for each frequency band is performed and sent to the multiplexing circuit 600. In the multiplexing circuit 600, eight encoded signals are multiplexed on the time axis and output from the output terminal 60 of the device. The above configuration constitutes the encoding device according to the present invention, and the output terminal 60 of the device
A video signal encoded for each band divided by frequencies in the diagonal direction is output from the .

Note that during encoding, due to aliasing due to quincunx grid division, components with very high horizontal and vertical frequencies are encoded in the same way as low-frequency components, but such components are As mentioned above, the power is small and the encoding efficiency does not decrease.

Next, the decoding device of this embodiment will be explained using FIG. 1. In FIG. 1, an encoded video signal inputted from an input terminal 70 of the device is sent to a demultiplexing circuit 700.
sent to. The demultiplexing circuit 700 separates the signals of each frequency band temporally multiplexed on the encoding side and sends them to decoders 710°720.730, 740, 750, 760°770.780. decoder 710°720,
730. 740. 750. 760°7707.
At 780, the encoded signals for each frequency band are decoded, and second diagonal synthesis filters 800 and 810
．． Sent to 820.830. Second diagonal synthesis filter 8
00,810°820.830 corresponds to the second diagonal division filter 450, 460, 470.480 on the encoding side, and the signal divided by the frequency in the second diagonal direction on the encoding side is The signals are synthesized again and sent to first diagonal synthesis filters 850 and 860.

The first diagonal synthesis filters 850 and 860 correspond to the first diagonal division filters 410 and 420 on the encoding side, and the signals divided at the frequency in the first diagonal direction on the encoding side are synthesized again here. and quincunx grid synthesis circuit 90
Sent to 0. The quincunx lattice synthesis circuit 900 corresponds to the quincunch lattice dividing circuit 400 on the encoding side.
The sample points divided into five types of lattices are combined again and output as square lattice sample points. The decoding device of this embodiment is configured with the above configuration, and the decoding is performed for each frequency band, and the original video signal is restored and outputted from the output terminal 80 of the device.

Further, in the above embodiment, the number of band divisions is two in both the first diagonal direction and the second diagonal direction, but it can be easily inferred that this number can be expanded to three or more.

In addition, in the above embodiment, filtering was performed by dividing the original image into a quincunx grid. After removing such components using a fixed filter, filtering may be performed by dividing into a quincunx lattice.

Effects of the Invention As is clear from the above embodiments, in the encoding device and decoding device according to the present invention, the sample pattern is formed into a diagonal lattice shape and then diagonal filtering is performed to perform band division, so that the video signal can be divided. Since it is divided into diagonal frequency components, the diagonal frequency components are encoded with a coarse quantization step to relatively low frequencies, and the horizontal and vertical frequency components are quantized to relatively high frequencies. Since detailed step encoding can be performed,
High-frequency components in diagonal directions that are not visually noticeable can be encoded with a reduced number of quantization bits, allowing efficient encoding.

[Brief explanation of drawings]

FIG. 1 is a block diagram showing the configuration of an encoding device and a decoding device according to an embodiment of the present invention, and FIG. 2 is a sample of the encoding device and decoding device according to an embodiment of the present invention shown in FIG. FIG. 3 is a pattern diagram showing the transition of patterns, FIG. 3 is a graph showing the frequency division of the encoding device and decoding device according to an embodiment of the present invention shown in FIG. 1, and FIG. FIG. 5 is a block diagram showing the configuration of the decoding device. FIG. 5 is a pattern diagram showing the transition of sample patterns in the conventional encoding device shown in FIG. 4 and the decoding device. FIG. It is a graph shown about frequency division of an encoding device and a decoding device. 400...Gentacle lattice division circuit, 410°42
0...First diagonal division filter, 450.464
60.470,480...Second diagonal division filter, 800,810.820,830 Old...Second diagonal synthesis filter, 850,860...First diagonal synthesis filter, 900・・・・・・Gotoku lattice synthesis circuit. Name of agent: Patent attorney Haruaki Ogata Two idiots 1 1
. 1!

Claims (2)

[Claims] (1) A quincunx lattice dividing means for dividing a square lattice sample pattern into two types of quincunx lattice sample patterns, and a first diagonal direction for the two types of quincunch lattice sample patterns. a first diagonal division filter means that performs filtering in a second diagonal direction and divides the frequency; and a second diagonal division filter means that performs filtering in a second diagonal direction and divides the frequency with respect to the two types of quincunx grid sample patterns. A video signal encoding device comprising: (2) Input the video signal band-divided from two types of quincunx grid sample patterns, perform filtering in the first diagonal direction for the two types of quincunx grid sample patterns, and synthesize frequencies. a first diagonal synthesis filter means for performing frequency synthesis by filtering the two types of quincunx grid-like sample patterns in a second diagonal direction; What is claimed is: 1. A video signal decoding device comprising a quincunx lattice synthesis means for synthesizing sample patterns in the form of a square lattice to form a sample pattern in the form of a square lattice.