JP2870588B2 - Moving picture prediction encoding apparatus, moving picture prediction encoding method, and moving picture prediction decoding apparatus - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION Used in recording, transmission, and display devices for processing digital signals, the signals are efficiently coded with a smaller code amount, and in particular, the coding of moving picture signals capable of random access and search. TECHNICAL FIELD The present invention relates to a device, an encoding method, and a decoding device.

[0002]

2. Description of the Related Art In high-efficiency coding of a moving image, a frame in which only a prediction error is coded by predicting a frame to be coded by a coded frame by utilizing a correlation between frames of an image signal. There is inter prediction coding. In recent years, motion-compensated inter-frame prediction, in which an image is predicted by moving the image in accordance with the motion, has become more common.

On the other hand, in coding for storage media, an intra-frame independent coded image (hereinafter, also simply referred to as an I-frame or an independent frame) that performs intra-frame independent coding without performing inter-frame prediction every several frames. It is set every few frames to support random access and high-speed search. There is an inter-frame prediction coded image between the I frames, and inter-frame prediction coding is performed using inter-frame prediction. (Hereinafter, these frames are also simply referred to as P frames.) For example, as shown in FIG. 11A, a frame includes an intra-frame independent coded image (I) and an inter-frame predictive coded image (P). . In the following description, independent encoding that performs intra-frame independent processing without performing inter-frame prediction is also simply referred to as I-encoding or independent encoding, and prediction encoding using inter-frame prediction is simply referred to as P-encoding. Write. Next, a configuration example of the above-described conventional encoding device will be described with reference to FIG.

An image signal input from an image signal input terminal 1 is led to a prediction subtractor 2 where the prediction signal is subtracted. The output of the predictor 2, which is the prediction residual, becomes data encoded and compressed by the intra-frame encoder 6. The data is output from the data output 9 and becomes the prediction residual reproduced by the intra-frame decoder 18. The prediction signal is added by the inverse prediction adder 16 to become a reproduced image signal. The reproduced image signal is stored in the frame memory 15, delayed by one frame, and then guided to the P terminal of the changeover switch 35.

The changeover switch 35 is controlled by a synchronization signal separated from the input. The movable terminal S is connected to the fixed terminal I in an independent frame once every N frames, and connected to the P side in other predicted frames. Here, since the reproduced image signal of the previous frame is input to P, if P is selected, inter-frame prediction is performed. I has a fixed value (0)
Is connected, the output of the predictive subtractor 2 becomes the same as the input signal, and the intra-frame independent encoding is performed.

FIG. 8 shows the configuration of a decoding device. The data input from the data input terminal 30 is decoded by the intra-frame decoder 18 and the prediction signal is added by the inverse prediction adder 16 to become a reproduced image signal. The signal is output from the reproduced image signal output terminal 34 and is stored in the frame memory 15.
After being delayed by the frame, it is input to the inverse prediction adder 16 through the switch 35. The switch 35 is controlled by the frame synchronization signal separated from the input data in the same manner as the encoder.

In the intra-frame encoder 6 of the encoding device, DCT (Discrete Cosine Transform) is first performed, the transform output is quantized, and variable-length encoded by Huffman code or the like. The intra-frame decoder 18 of the decoder first decodes the variable-length code, replaces the fixed-length code with a quantized representative value, and further performs inverse DCT to obtain a reproduced signal.

[0008]

In such a conventional encoding / decoding apparatus, an independent frame is periodically provided, and the amount of data generated from the independent frame is larger than that of the predicted frame. Become.

In the high-speed search, only the independent frame is reproduced. However, since the amount of data in the independent frame is large, if the amount of data that can be decoded by the high-speed search is the same as that in the normal reproduction, all the independent frames corresponding to the search speed are used. Cannot be decoded, and the same frame is repeatedly output. Therefore, if frequency components are divided into bands and only low-frequency components are decoded, the amount of data that must be decoded for each frame is reduced, so that a smooth search image can be obtained even at a low resolution. However, in this case, high frequency components are wasted.

The present invention has been made in view of the above points. Only the low frequency components are made independent, and the inter-frame prediction is always performed for the high frequency components. It is an object of the present invention to provide a moving picture predictive coding apparatus capable of improving coding efficiency and reducing a data amount while corresponding to a function to be performed, and a decoding apparatus for the same.

[0011]

SUMMARY OF THE INVENTION The present invention relates to a predictive coding apparatus for a moving picture signal, which performs inter-frame or inter-field coding on a signal having a high frequency component of a first frame or field. It is an object of the present invention to provide a moving picture predictive encoding apparatus comprising means for performing intra-frame or intra-field encoding on a low frequency component signal of the first frame or field on a frame or field basis.

The present invention also relates to a method for predictive encoding of a moving picture signal, wherein inter-frame or inter-field encoding is performed on a signal of a high frequency component of a first frame or a field. Another object of the present invention is to provide a video predictive encoding method for performing intra-frame or intra-field encoding on a signal of a low frequency component of a field on a frame or field basis.

Further, the present invention is a video predictive encoding apparatus according to claim 1 or a predictive decoding apparatus for a video signal encoded by the video predictive encoding method according to claim 2, wherein Inter-frame or inter-field decoding of a signal of a high frequency component of a frame or a field, and intra-frame or intra-field decoding of a signal of the first frame or a low frequency component in a frame or a field unit It is an object of the present invention to provide a moving picture prediction decoding device comprising means for performing the decoding.

[0014]

According to the predictive coding apparatus and the decoding apparatus of the present invention configured as described above, for a certain frame,
Independent encoding (I encoding) for performing intra-frame independent processing without using inter-frame prediction is performed only on low-frequency components, and prediction using inter-frame prediction is performed on high-frequency components. Encoding (P encoding) is performed.

By doing so, low frequency components are handled in the same manner as in the prior art, and high frequency components are improved in coding efficiency by prediction processing, and the amount of coded output data is reduced as a whole. When a recording medium on which an encoded signal is recorded is reproduced, in a random access or the like, a reproduced image at an access point is composed of only low-frequency components, and high-frequency components are prediction residual components as a frame advances. Is gradually regenerated. In the high-speed search, a smooth search image can be obtained with an image having a low frequency component.

[0016]

BRIEF DESCRIPTION OF THE DRAWINGS FIG. 1 is a block diagram showing an embodiment of the present invention; <First Embodiment> FIG. 1 shows a first embodiment of a predictive coding apparatus according to the present invention.
FIG. 3 is a block diagram showing an embodiment. The same parts as those in the conventional example of FIG. As compared with the encoding apparatus shown in FIG. 7, the present encoding apparatus is characterized in that a spatial LPF 13 and a subtractor 12 for a prediction signal are provided.

In FIG. 1, an image signal input from an image signal input terminal 1 is supplied to a predictive subtractor 2. The prediction subtractor 2 subtracts the prediction signal output from the subtractor 12 from the image input signal, obtains a prediction residual, and supplies the prediction residual to the intra-frame encoder 6. The intra-frame encoder 6 intra-frame encodes the input signal, outputs the encoded data from which the redundancy has been removed from the data output terminal 9, and also outputs the encoded data to the intra-frame decoder 18. Intra-frame encoder 6
Is the same as in the conventional example.

On the other hand, the prediction signal in the encoding device needs to be generated from the intra-frame encoded signal in order to make the same signal as the decoding device. Therefore, the intra-frame decoder 18 decodes the intra-frame coded signal, obtains a reproduction prediction residual, and supplies it to the inverse prediction adder 16. The inverse prediction adder 16 adds the reproduction prediction residual and a prediction signal corresponding thereto to obtain a reproduction image signal, and
Supply to 15. The frame memory 15 obtains an output signal obtained by delaying the input signal by one frame, and supplies the output signal to an LPF (low-pass filter) 13 and a subtractor 12. LP
The output of F13 is connected to the fixed terminal I of the switch.

The changeover switch 36 is controlled by a synchronization signal separated from the input. The movable terminal S is connected to the fixed terminal I in an independent frame once every N frames, and connected to the P side in other predicted frames. Where I is L
A low frequency component of the reproduced image signal one frame before, which is the output of the PF 13, is input, and P is connected to a fixed value (0). The output of the subtractor 12 is a switch
When the switch 36 is on the P side, the output of the frame memory 15 remains as it is, so that the same processing as in the conventional example is performed in the prediction subtractor 2. When the changeover switch 36 is on the I side, the low frequency component from the LPF 13 is subtracted. Therefore, even if the prediction signal is subtracted from the input signal by the prediction subtractor 2, the low frequency component remains in the prediction residual, and the intra-frame encoder 6 performs the intra-frame independent processing. .

The LPF 13 is a two-dimensional filter, but may be a cascade connection of a vertical filter and a horizontal filter.
The configuration in that case is as shown in FIG. The numbers in the figure are examples of tap coefficients.

FIG. 10 shows the frequency characteristics of each signal. The input signal shown in (A) has all frequency components, whereas the predicted signal shown in (B) is a signal in which low frequencies are suppressed. As shown in (C), in the prediction residual from which such a prediction signal has been subtracted, a low frequency component of the input signal remains as it is, and only a high frequency component corresponding to the deviated prediction. Since the reproduction signal shown in (D) is an addition of the prediction signal (B) and the prediction residual (C), all frequency components are obtained.

Here, if the prediction of the high frequency component is assumed to be complete (prediction coefficient 1), the prediction residual of the high frequency component is completely eliminated, so that the reproduction of the high frequency component after the random access is restored to the normal state. It takes a long time. If this poses a problem, if the prediction coefficient is reduced from 1 to 0.8, for example, the coding efficiency is slightly reduced, but a normal image can be obtained relatively quickly. The realization of this processing does not completely suppress high-frequency components in the LPF 13, but rather uses 0.
It can be realized by leaving about two.

When motion compensation with a precision finer than one pixel is used in inter-frame prediction, resampling processing is performed in motion compensation, and high frequencies are slightly suppressed by the characteristics of the filter. Need not be reduced below 1. In this case, for example, to shift by 1/2 pixel with 1/2 pixel accuracy, use a 4-tap filter to set -1/16, 9/16, 9/16, -1
It becomes a coefficient such as / 16, and high frequencies are suppressed. However, if the pixel is not shifted, it becomes 0,1,0,0 and the prediction coefficient becomes 1, so that 1/16, 14/16, 1 / 16,0, etc. are used to suppress high frequencies. It is a coefficient.

FIG. 2 is a block diagram showing a first embodiment of the moving picture prediction decoding apparatus according to the present invention, and is a block diagram showing an embodiment of a decoding apparatus corresponding to the encoding apparatus shown in FIG. . 1 and 8 described above are denoted by the same reference numerals. In FIG. 2, an encoded data signal transmitted from an encoding device via a data input terminal 30 is received and supplied to an intra-frame decoder 18. The intra-frame decoder 18 decodes the intra-frame coded signal, obtains a reproduction prediction residual, and supplies it to the inverse prediction adder 16. The inverse prediction adder 16 adds the reproduction prediction residual and the prediction signal supplied from the subtractor 12, obtains a reproduction image signal, outputs the signal via the reproduction image signal output terminal 34, and also outputs the signal to the frame memory 15. Supplying. The frame memory 15 obtains an output signal obtained by delaying the input signal by one frame, and
And to the subtractor 12. The output of LPF 13 is connected to fixed terminal I of switch 36. The configuration and operation of the switch are the same as those of the encoder, and the output of the subtractor 12 is input to the adder 16 as the same prediction signal as that of the encoder.

<Second Embodiment> FIG. 3 is a block diagram showing a second embodiment of the predictive coding apparatus according to the present invention. 1 are given the same reference numerals. The difference from the first embodiment shown in FIG. 1 is that low frequency components are encoded separately,
The point is that the processing is not changed for each frame, and low-frequency components in all frames are independently coded (I-coded).
The input signal is first divided into a band division subtractor 3 and a sub-sampler 4
Is input to After performing the same filtering as the LPF 13 in the sub-sampler 4, the sub-sampler corresponding to the frequency band is performed. In the example of FIG. 9 described above, both the vertical and horizontal directions are set to ２.

The output of the sub-sampler 4 is input to the intra-frame encoder 7 and the over-sampler 5. The encoded data of the low frequency component, which is the output of the intra-frame encoding, is transmitted to the decoder via the data output terminal 10. The intra-frame encoder 7 is for a subsampled image, and the number of pixels to be processed is 1/4 of the input signal.

On the other hand, the oversampling unit 5 interpolates and makes the pixels that are no longer the subsampling unit 4 to have the same number of pixels as the original image. The output of the oversampler 5 is input to the subtraction input of the band division subtractor 3, and the band division subtracter 3 subtracts the input signal from the input signal to obtain a signal with a high frequency component. The output of the band division subtractor 3 is led to the prediction subtractor 2, and the prediction signal output from the frame memory 15 is subtracted. The output of the predictive subtracter 2 which becomes the prediction residual is intra-coded by the intra-frame encoder 8, and the output is transmitted to the decoding device via the data output terminal 11 and input to the intra-frame decoder 20. Is done. The intra-frame encoder 8 is for the prediction residual of the high frequency component, and the number of pixels to be processed is the same as that of the input signal.

The residual reproduced by the intra-frame decoder 20 is added to the corresponding prediction signal by the inverse prediction adder 16, delayed by one frame in the frame memory 15, and supplied to the prediction subtractor 2 as a prediction signal. .

FIG. 4 is a block diagram showing a second embodiment of the moving picture prediction decoding apparatus according to the present invention, and is a block diagram showing an embodiment of a decoding apparatus corresponding to the encoding apparatus shown in FIG.
The difference from the decoding apparatus of FIG. 2 is that there are two intra-frame decoders, and an oversampler 5 and an adder 17 for adding two intra-frame decoder outputs are added. On the other hand, since the output of the intra-frame decoder 20 is only high frequency components, the LPF 13 for forming a prediction signal,
No subtractor 12 is required. The data input signal of the low frequency component coming from the data input terminal 31 is intra-frame decoded by the intra-frame decoder 19 and input to the oversampler 5. The oversampler 5 generates a signal having the same number of pixels as the original image, and supplies the signal to the adder 17.

On the other hand, a data input signal of a high frequency component coming from the data input terminal 32 obtains a reproduction prediction residual by the intra-frame decoder 20 and is input to the inverse prediction adder 16. The inverse prediction adder 16 adds the reproduced prediction residual and the prediction signal supplied from the frame memory 15 to obtain a reproduced image signal of a high frequency component, inputs the reproduced image signal to the adder 17, and supplies the reproduced image signal to the frame memory 15. ing. The frame memory 15 delays the input signal by one frame and inputs it to the inverse prediction adder 16 as a prediction signal. The adder 17 adds the low-frequency component signal supplied from the oversampler 5 and the high-frequency component signal supplied from the adder 16 to obtain a reproduced image signal, which is output via the reproduced image signal output terminal 34. .

<Third Embodiment> FIG. 5 is a block diagram showing a third embodiment of the moving picture predictive coding apparatus according to the present invention. The difference from the second embodiment shown in FIG. 3 is that the high frequency component and the low frequency component are divided before the encoding process in the embodiment of FIG. 3, whereas the embodiment of FIG. Frequency division is performed on the inter prediction residual. Therefore, both frequency components are subjected to intra-frame decoding, and a prediction signal is created by adding the results. In the prediction, motion compensation is performed.

The input signal is input to a band division subtracter 3 and a sub-sampler 4 after the prediction signal is subtracted by the prediction subtractor 2. The sub-sampler 4 performs the same processing as in FIG. 3, and its output is supplied to the intra-frame encoder 7. Further, the output of the band division subtracter 3 which is a high frequency component is supplied to the intra-frame encoder 8. The intra-frame encoder 7 is for the sub-sampled low frequency component, and the intra-frame encoder 8 is for the high frequency component. The outputs of the respective intra-frame encoders are transmitted to the decoding device via the data output terminals 10 and 11, and are also input to the intra-frame decoders 19 and 20.

The output of the intra-frame decoder 19 is input to the oversampler 5, and the output of the intra-frame decoder 20 is input to the adder 17. The output of the oversampler 5 is input to the adder 17 and the subtraction input of the band division subtractor 3. An adder 17 adds both components to obtain a reproduction prediction residual, and performs an inverse prediction addition.
Supply 16 Inverse prediction adder 16, frame memory 15,
The operations of the LPF 13, the switch 36, and the subtractor 12 are the same as those in FIG.

On the other hand, the motion compensator 14 has a frame memory
A motion vector is detected for each block using the reproduced image signal and the input image signal delayed by one frame, which is the output of 15, and a one-frame delayed signal motion-compensated by the vector (accurate one frame by the motion compensation) Is not supplied) and supplied to the LPF 13 and the subtractor 12. Further, since the motion vector information is also required on the decoding side, it is output from the motion vector output terminal 21 and transmitted to the decoding side.

FIG. 6 is a block diagram showing a third embodiment of the moving picture prediction decoding apparatus according to the present invention, and is a block diagram showing an embodiment of a decoding apparatus corresponding to the encoding apparatus shown in FIG. This is a form in which the decoding device of FIG. 2 and the decoding device of FIG. 4 are combined. The data input signals input from the data input terminals 31 and 32 are input to the intra-frame decoder 1 respectively.
In 9 and 20, it is intra-coded. The low frequency component is input to the oversampling unit 5 and has the same number of pixels as the original image, and both components are added by an adder to produce a reproduction prediction residual. Inverse prediction adder 16, frame memory 15, LP
The operations of F13, switch 36 and subtractor 12 are the same as in FIG. The output of the inverse prediction adder 16 is output as a reproduced image via a reproduced image signal output terminal 34. On the other hand, the motion compensator 37
Performs motion compensation on the output of the frame memory 15 based on the motion vector value coming from the motion vector information input 35, and supplies the result to the LPF 13 and the subtractor 12.

<Modification of Frame Structure> FIG. 11B
(E) shows the frame structure of each embodiment. The squares are frames and those separated by broken lines are high frequency components at the top and low frequency components at the bottom. I indicates that I coding is independently coded within a frame, P indicates P coding that is coded by inter-frame prediction, and B indicates that B coding that is predicted from both front and rear frames is performed. I have. Therefore, as shown in (E) of the figure, as a fourth embodiment, B encoding is performed on a frame in which only P encoding is performed in the first embodiment, and P encoding is performed by P encoding. The prediction may be performed between frames to be encoded, and the B encoding may be predicted from two frames before and after the I and P encoding is performed.

According to the moving picture predictive coding apparatus and its decoding apparatus described in detail above, for a certain frame (or field), only low frequency components are independently coded in a frame, and the same By performing inter-frame predictive coding on high frequency components of a frame,
Although random access and search can be performed using the image of the low frequency component as in the related art, the encoding efficiency of the high frequency component is improved by the prediction process, so that the encoded output data amount can be reduced as a whole.

When the recording medium on which the coded signal is recorded is reproduced, in a random access or the like, the reproduced image of the access point has only low frequency components. Since the image is reproduced gradually by the residual component, there is no waste on the visual characteristics. That is, the human visual characteristic is such that the visual ability is considerably lower than usual immediately after the image changes abruptly, so that there is not much problem even if the reproduction of such high frequency components is delayed. Further, in the high-speed search, only low frequency components are used, but a smooth search image can be obtained.

[0039]

According to the moving picture predictive coding apparatus, the moving picture predictive coding method and the moving picture decoding apparatus of the present invention, it is less necessary to independently code a certain frame (or field) in a frame. Since only the frequency component is used and the inter-frame prediction coding is performed using the inter-frame prediction for the high frequency component of the same frame, the low frequency component can be randomly accessed and searched as in the related art, The frequency component has an extremely excellent effect in practical use, for example, the encoding efficiency is improved by the prediction processing, and the encoded output data amount can be reduced as a whole.

[Brief description of the drawings]

FIG. 1 is a block diagram illustrating a first embodiment of a moving picture prediction encoding apparatus according to the present invention.

FIG. 2 is a block diagram showing a first embodiment of the moving picture prediction decoding apparatus of the present invention.

FIG. 3 is a block diagram showing a second embodiment of the video predictive coding apparatus of the present invention.

FIG. 4 is a block diagram showing a second embodiment of the moving picture prediction decoding apparatus of the present invention.

FIG. 5 is a block diagram showing a third embodiment of the video predictive coding apparatus of the present invention.

FIG. 6 is a block diagram showing a third embodiment of the moving picture prediction decoding apparatus of the present invention.

FIG. 7 is a block diagram showing a conventional example of a moving picture prediction encoding device.

FIG. 8 is a block diagram showing a conventional example of a moving picture prediction decoding apparatus.

FIG. 9 is a diagram illustrating a configuration and coefficients of an LPF according to the embodiment.

FIG. 10 is a diagram illustrating a state of frequency characteristics of a signal according to the embodiment.

FIG. 11 is a diagram showing a state of a prediction structure between frames.

[Description of Signs] 1 ... Image signal input terminal, 2 ... Predictive subtractor, 3 ... Band division subtractor, 4 ... Sub-sampler (SS), 5 ... Over-sampler (OS), 6, 7, 8 ... Frame Inner encoder (ENC), 9, 10, 11 ... data output terminal, 12 ...
Subtractor, 13 LPF, 14, 37 Motion compensator (M
C), 15: frame memory (FM), 16: inverse prediction adder, 17: adder, 18, 19, 20 ... intra-frame decoder (DEC), 21: motion vector output terminal, 3
0, 31, 32: data input terminal, 34: reproduced image signal output terminal, 35, 36: switch, 40: input terminal, 41: vertical LPF, 42: horizontal LPF,
43 ... Output terminal.

────────────────────────────────────────────────── ─── Continuation of the front page (56) References JP-A-2-192378 (JP, A) IEICE Transactions, J71-A [2] (1988-2) 488-496 IEEE TRANSACTIONS ON CIRCUITS AND SSYSTEMS FOR VIDEO ECHNOLOGY VOL. 1 [2] (1991.6) p. 174-183 IEEE GLOBECOM'88 P.E. 743-749 (58) Field surveyed (Int. Cl. ⁶ , DB name) H04N ^7/ 24-7/68

Claims (3)

(57) [Claims] An apparatus for predictive coding of a moving image signal, comprising: performing inter-frame or inter-field coding on a signal of a high frequency component of a first frame or field; A moving picture predictive coding apparatus comprising means for performing intra-frame or intra-field coding on a signal of a low frequency component on a frame or field basis. 2. A predictive encoding method for a moving image signal, comprising: performing inter-frame or inter-field encoding on a signal of a high frequency component of a first frame or field; A moving picture predictive coding method characterized by performing intra-frame or intra-field coding on a signal of a low frequency component on a frame or field basis. 3. A predictive decoding apparatus for a moving picture signal encoded by the moving picture predictive coding apparatus according to claim 1 or the moving picture predictive coding method according to claim 2, wherein: Inter-frame or inter-field decoding of a signal of a high frequency component of a frame or a field, and intra-frame or intra-field decoding of a signal of a low frequency component of the first frame or a field in frame or field units A video predictive decoding device comprising means for performing: