JPH0248885A - System and apparatus for hybrid decoding of moving picture signal - Google Patents

Abstract

PURPOSE:To round off an orthogonal transform coefficient at coding and to prevent deterioration in the picture quality by eliminating a signal component corresponding to a part of the orthogonal transform coefficients to an inter-frame prediction signal of a decoding picture signal at a decoder side. CONSTITUTION:At the decoder side, a waveform 519 corresponding to a coefficient rounded off by the coder side from waveforms 518, 519 corresponding to each orthogonal transform coefficient included in inter-frame prediction signal 517 is rounded off as a waveform 514 and a waveform 522 being the inversion of the orthogonal transform coefficient corresponding to the waveforms 508, 509 from the coder side is added to obtain a picture signal of waveform 515. A waveform 507 corresponding to the transform coefficient rounded off by a quantizer 16 from component waveforms 506, 507 included in a prediction signal 505 at the coder side is rounded off to form a waveform 510 and it is added to a waveform 520 being the result of inverting the orthogonal transform coefficient corresponding to the waveforms 508, 509 obtained by the quantizer 16 to form a local decoding signal with a waveform 512. Through the constitution above, the level difference between the output waveform 515 and the input waveform 501 is decreased to suppress the deterioration in the decoded picture signal.

Description

DETAILED DESCRIPTION OF THE INVENTION (Field of Industrial Application) The present invention relates to a method and apparatus for decoding a moving image signal compressed and encoded with high efficiency.

(Prior Art) Conventionally, as a coding method for video signals that combines interframe predictive coding and orthogonal transform coding, the 1986 Image Coding Symposium, 4.12. pp, 53-54.
The one described in September 1986 (Reference 1) is known. An encoding block diagram of Document 1 is shown in FIG.

In this encoding method, a motion compensated interframe prediction error signal of a moving image signal is orthogonally transformed encoded. In Document 1, in order to reduce the amount of generated information while suppressing the deterioration of the image quality of the decoded image signal as much as possible, when quantizing orthogonal transform coefficients, high-frequency coefficients are truncated based on the magnitude of movement, Two steps are performed: truncation of isolated significant coefficients.

(Problems to be Solved by the Invention) In orthogonal transform encoding, the amount of information can be compressed by cutting down orthogonal transform coefficients to be encoded to reduce the number. At this time, in the conventional method of orthogonal transform coding of the interframe prediction error, in the decoded signal, the signal component corresponding to the orthogonal transform coefficient that was truncated on the coding side is predicted from the decoded signal of the previous frame. The signal components included in the signal will appear as they are in the decoded signal of the current frame. For example, in the block diagram of FIG. 7, a case as shown in FIG. 6(b) can be considered. Figure 6 is an example of using one-dimensional orthogonal transformation, and the two waveforms shown in the square frame are two of the orthogonal transformation coefficients.
ν) indicates a waveform corresponding to one coefficient, and the amplitude of the waveform represents the magnitude of the coefficient. Note that the DC component is not shown. In FIG. 6(b), it is assumed that a locally decoded signal such as a waveform 505 is obtained when the previous frame is encoded, and a decoded signal such as a waveform 517 is obtained on the decoding side. If a method of simply using the previous frame signal as is is used as the interframe prediction method, the waveform 505 becomes the current frame prediction signal on the encoding side, and the waveform 517 becomes the interframe prediction signal on the decoding side. In encoding the current frame, when an input signal such as waveform 501 is encoded, an inter-frame difference is taken between waveform 501 and waveform 505 on the encoding side, and error waveform 502 is obtained. This waveform 502 is orthogonally transformed to obtain orthogonal transform coefficients corresponding to waveforms 503 and 504. In the quantizer 16, the orthogonal transform coefficients corresponding to the waveform 503 are saved, and the orthogonal transform coefficients corresponding to the waveform 504 are truncated.
Orthogonal transform coefficients corresponding to waveforms 508 and 509 will be encoded. In the quantizer 16, the symbol ◯ indicates that the transform coefficient is retained immediately afterward, and the symbol X indicates that it is truncated. On the decoding side, first, orthogonal transform coefficients corresponding to waveforms 508 and 509 are inversely transformed to obtain waveform 522.

By adding the interframe prediction signal 517 to this waveform 522, a decoded signal 523 is obtained. To predict the signal of the next frame, the encoding side uses a waveform 520 that is obtained by inversely transforming the orthogonal transform coefficients corresponding to waveforms 508 and 509, and a waveform 521 that is the sum of the interframe prediction signal 505. , a waveform 523 is used on the decoding side. Comparing the decoded waveform 523 with the input waveform 501 and the decoded waveform 517 of the previous frame, the falling position of the slope in the decoded waveform 523 is shifted to the right compared to the input waveform 501 due to the effect of coefficient truncation, It has a shape similar to the waveform 517 of . As described above, in the conventional method, the decoded waveform of the previous frame corresponding to the truncated orthogonal transform coefficient remains as it is in the decoded signal of the current frame, resulting in significant image quality deterioration. Particularly for a moving part, if the high-frequency orthogonal transform coefficients of the frame difference are discarded, the wide-range signal component, that is, the outline of the object before it moves, remains in the background.

The purpose of the present invention is to eliminate the above-mentioned drawbacks of the conventional method, prevent unnecessary effects from the previous frame signal from remaining on the decoded image signal even if the orthogonal transform coefficients are truncated during encoding, and improve image quality. An object of the present invention is to provide a decoding device for a moving image signal with little deterioration.

(Means for Solving the Problems) The decoding method of the present invention, on the encoding side, calculates the difference between the input image signal and the interframe prediction signal on the encoding side, orthogonally transforms the difference signal, and then encodes the signal. A locally decoded signal is obtained by adding a signal obtained by inverse orthogonal transformation of the encoded image signal and the interframe prediction signal, and from the locally decoded signal,
The signal of the next frame is predicted and used as the interframe predicted signal on the encoding side, and on the decoding side, the encoded image signal is inversely orthogonally transformed, and the inversely orthogonally transformed signal and the interframe predicted signal on the decoding side are In this video signal encoding method, the next frame is predicted from the decoded image signal and the interframe predicted signal on the decoding side is calculated. It is characterized in that signal components corresponding to some orthogonal transform coefficients are removed from the predicted signal or the decoded image signal.

The configuration of a first decoding device that realizes the present invention includes means for inversely orthogonally transforming a coded image signal, means for adding an output signal of the inversely orthogonal transformer and an interframe prediction signal, and a means for adding the output signal of the inversely orthogonal transformer and an interframe prediction signal, A means for outputting the output signal as a decoded image signal, and a means for predicting the signal of the next frame from the output signal of the adding means, and calculating a part of the signal based on the mode information generated by the encoding device from the Kono prediction signal. Either the signal component corresponding to the orthogonal transform coefficient is truncated and output as an interframe prediction signal, or the signal component corresponding to the orthogonal transform coefficient is handled based on the mode information generated by the encoding device from the output signal of the adding means. The present invention is characterized by comprising an inter-frame predictor that truncates signal components that are truncated, predicts the next frame signal from the truncated signal, and outputs the signal as an inter-frame predicted signal.

Further, a second decoding device realizing the present invention includes means for inversely orthogonally transforming a coded image signal, means for adding an output signal of the inversely orthogonal transforming means and an interframe prediction signal, and a means for adding an output signal of the inversely orthogonal transforming means and an interframe prediction signal, means for cutting off signal components corresponding to some of the orthogonal transform coefficients based on mode information generated by the encoding device from the output signal; and means for outputting the output signal of the component cutting means as a decoded image signal. , and an interframe predictor that predicts and outputs the signal of the next frame from the output signal of the component truncation means (Operation) In FIG. 6(b), the output signal 523 of the decoding device is ,
The shape that is similar to the decoded signal 517 of the previous frame is as follows.
This is because, among the signal components included in the waveform 517, the signal component corresponding to the orthogonal transform coefficient that was truncated before encoding was not modified during frame decoding of the current frame. The image quality can be improved by truncating the signal components corresponding to the orthogonal transform coefficients truncated on the encoding side during interframe decoding in the decoding device as well. In the present invention, in order to prevent signal components corresponding to orthogonal transform coefficients cut off on the encoding side from appearing in the decoded image signal, the signal components included in the interframe predicted signal or the interframe decoded signal on the decoding side are prevented from appearing in the decoded image signal. Among the signal components, the signal components corresponding to the orthogonal transform coefficients that were discarded on the encoding side are discarded. In this case, the example shown in FIG. 6(b) becomes as shown in FIG. 6(a). Figure 6(a)
1 shows signals of various parts of an encoding/decoding device that is a combination of an example of an encoding device corresponding to the decoding device of the present invention and a decoding device of the first configuration of the present invention. On the decoding side, among the signal components (waveforms 518 and 519) corresponding to each orthogonal transformation coefficient included in the interframe prediction signal 517, the signal component (waveform 519) corresponding to the orthogonal transformation coefficient that was truncated on the encoding side is is truncated to form waveform 514, and this waveform 514 and waveforms 508 and 5 sent from the encoding side are
A decoded image signal of waveform 515 is obtained by adding the waveform 522 obtained by inverse orthogonal transformation of the orthogonal transform coefficient corresponding to 09.

Furthermore, on the encoding side, signal components (waveform 506.5) corresponding to each orthogonal transform coefficient included in the interframe prediction signal 505
07), the signal component (waveform 5o7) corresponding to the orthogonal transform coefficient truncated by the quantizer 16 is truncated to form a waveform 510, and this waveform 510 corresponds to waveforms 508 and 509 obtained by the quantizer 16. By adding the waveform 520 obtained by inverse orthogonal transformation of the orthogonal transform coefficients, waveform 5 is obtained.
12 locally decoded signals are obtained.

Compared to the waveform 523, the waveform 515 is closer to the manual waveform 501 in terms of the position where the waveform rises and crosses level zero;
The level difference with the human power waveform 501 is small. This tendency becomes more pronounced when the difference between the value of the coefficient to be discarded in the current frame and the value of the corresponding coefficient of the interframe prediction signal is large. As described above, in the present invention, deterioration of a decoded image signal can be suppressed compared to the conventional method in which signal components corresponding to orthogonal transform coefficients are not truncated on the decoding side during encoding.

The difference between the first configuration and the second configuration of the decoding device of the present invention is that in the first configuration, the decoding signal used for interframe prediction or the interframe prediction signal is In the second configuration, the signal components of the decoded signal are truncated.

(Example) FIGS. 1(a) and (b) are block diagrams showing the configuration of a decoding device according to the present invention, and FIGS. 8(a) and (b) are block diagrams of the corresponding encoding device. An example is shown.

Before starting to explain the embodiments of the decoding device of the present invention, the encoding device side will first be explained.

In FIG. 8(a), the subtracter 3 calculates the difference between the image signal input from the input terminal 100 and the interframe prediction signal output from the interframe predictor 2. In FIG. This difference signal is
After being orthogonally transformed by the orthogonal transform circuit 1, it is output from the output terminal 200 as an encoded image signal.

Further, the output signal of the orthogonal transform circuit 1 is inversely transformed by the inverse orthogonal transform circuit and then added to the adder 4. The component truncation circuit 6 truncates signal components corresponding to some orthogonal transform coefficients from the interframe prediction signal output from the interframe predictor 2 based on the mode information generated by the truncation component determination circuit 9. The adder 4 adds the output signal of the inverse orthogonal transform circuit 8 and the output signal of the component truncation circuit 6. The interframe predictor 2 predicts and outputs the image signal of the next frame from the output signal of the adder 4. The truncation component determination circuit 9 generates mode information that instructs which signal component corresponding to the orthogonal transform coefficient is to be truncated.

Mode information is output from output terminal 210.

In FIG. 8(b), the subtracter 3 calculates the difference between the image signal input from the input terminal 100 and the output signal of the interframe predictor 7. This difference signal is orthogonally transformed in the orthogonal transform circuit 1, and then sent to the output terminal 20 as a coded image signal.
Output from 0. Further, the output signal of the orthogonal transform circuit 1 is subjected to inverse orthogonal transform in an inverse orthogonal transform circuit 8 and then added to an adder 4 . The adder 4 adds the output signal of the inverse orthogonal transform circuit 8 and the output signal of the interframe predictor 7. The interframe predictor 7 truncates signal components corresponding to some orthogonal transform coefficients from the output signal of the adder 4 based on the mode information generated by the truncated component determination circuit 9, and uses this signal to generate the next frame. The image signal of the next frame is predicted and outputted from the output signal of the adder 4, and the image signal of the next frame is predicted and output from this predicted signal based on the mode information generated by the truncated component determination circuit 9. The signal component corresponding to the orthogonal transform coefficient of the part is truncated and output.

Next, embodiments of the present invention will be explained.

In FIG. 1(a), an encoded image signal input manually from an input terminal 300 is subjected to inverse orthogonal transformation in an inverse orthogonal transformation circuit 13, and then applied to an adder 11. In the adder 11,
The output signal of the inverse orthogonal transform circuit 13 and the interframe predictor 1
The two output signals are added together and outputted from the output terminal 400 as a decoded image signal. In the interframe predictor 12,
Based on the mode information input from the input terminal 310, signal components corresponding to some orthogonal transform coefficients are discarded from the decoded image signal of the previous frame, and the image signal of the current frame is predicted from this signal. Alternatively, the image signal of the current frame is predicted from the decoded image signal of the previous frame, and based on the mode information input from the input terminal 310 from this predicted signal, it corresponds to some orthogonal transform coefficients. Truncates and outputs signal components.

In FIG. 1(b), an encoded image signal inputted from an input terminal 300 is subjected to inverse orthogonal transformation in an inverse orthogonal transformation circuit 13, and then applied to an adder 11. In the adder 11,
The output signal of the inverse orthogonal transform circuit 13 and the interframe predictor 1
and the output signal of No. 5. In the component truncation circuit 14,
Based on the mode information input from the input terminal 310, signal components corresponding to some of the orthogonal transform coefficients are cut off from the output signal of the adder 11, and this signal is output from the output terminal 400 as a decoded image signal. . Interframe predictor 15
Then, the image signal of the current frame is predicted from the decoded image signal of the previous frame and output.

The difference between Figures 1 (a) and (b) and the conventional method is as shown in Figure 1 (
In a), in the interframe predictor 12, as shown in FIG.
In the component truncation circuit 14, signal components corresponding to some orthogonal transform coefficients are truncated.

The mode information inputted from the input terminal 310 of the decoding device includes information that directly instructs which signal component corresponds to the orthogonal transform coefficient to be truncated, as well as information that outlines the pattern of the orthogonal transform coefficient of the signal component to be truncated. It is also possible to prepare a pattern and use the pattern number as mode information. The mode information does not have to be exclusive information that indicates which signal component to be truncated, and may be a signal that can be derived from other encoded information.

The component truncation circuit 14 shown in FIG. 1(b) can be realized, for example, by the circuit shown in the block diagram of FIG. 2(a). In FIG. 2(a), an input signal is orthogonally transformed into orthogonal transform coefficients, the negative part of the orthogonal transform coefficients is discarded based on the mode information, and then inverse orthogonal transform is performed to obtain an output signal.

The interframe predictor 15 in FIG. 1(a) can also be realized, for example, by the circuits shown in the block diagrams in FIGS. 2(b), (c), (d), and (e). Figure 2 (b), (
In c), (d), and (e), an interframe predictor 5o is inserted in the block diagram of FIG. 2(a), and (b)
, (c), (d), and (e), the insertion position of the interframe predictor 50 is different. The interframe predictor 50 performs prediction at the amplitude level of the image signal in FIGS. 2(b) and 2(e), and performs prediction at the orthogonal transform coefficient level in FIGS. 2(c) and 2(d). However, as shown in FIGS. 2(b), (c), (d), and (e), each circuit can be made to have the same transfer characteristic as a whole.

As shown in FIG. 2(b), the interframe predictor 12 in FIG. 1(a)
), (c), and (d), the inverse orthogonal transform circuits 13 and 53 can be placed after the adder 11 and combined. An example in this case is shown in FIG. 3(a). In FIG. 3, in place of the inverse orthogonal transform circuits 13 and 53, there is an inverse orthogonal transform circuit 57 after the adder ll. In addition, the interframe predictor 33 performs the following functions as shown in FIG.
The last inverse orthogonal transform circuit 53 in the block diagrams c) and (d)
It consists of an interframe predictor 50 except for the above, an orthogonal transform circuit 51, and a coefficient truncation circuit 52.

As shown in FIG. 2(a) as the component truncation circuit 14 in FIG. 1(b),
When using the circuit shown in the block diagram, it can also be realized by incorporating an orthogonal transform circuit into the interframe predictor 15 as shown in FIG. can. The interframe predictor 34 in FIG. 3(b) is composed of the interframe predictor 15 and an orthogonal transform circuit. You may go.

FIG. 4 is a block diagram showing an embodiment of an encoding/decoding apparatus using the encoding apparatus of FIG. 8(a) and the decoding apparatus of FIG. 1(a) of the present invention. In FIG. 4, a quantizer/inverse quantizer and a multiplexer/demultiplexer are added to the embodiment of the encoding/decoding apparatus shown in FIG.

FIG. 4(a) is a block diagram of the encoding device, where loo is an input terminal, l is an orthogonal transform circuit, 2 is an interframe predictor,
3 is a subtracter, 4 is an adder, 6 is an orthogonal transform component truncation circuit, 9 is a truncation component determination circuit, 16 is a quantizer, 17 is an inverse quantizer, 18 is an encoder output from the quantizer 16 This multiplexer multiplexes the image signal, information on the quantization characteristics in the quantizer 16, and mode information output from the truncated component determination circuit 9, and outputs the multiplexed information to the output terminal 290.

FIG. 4(b) is a block diagram of the decoding device, where 390 is an input terminal, 19 is a demultiplexer that separates the signal from the input terminal 390 into a coded image signal, quantization information, and mode information, and 2o is an inverse Quantization amount, 11 is an adder, 13 is an inverse orthogonal transform circuit, and 12 is a circuit for truncating the inter-frame predicted orthogonal transform component.

Demultiplexer 1 in the decoding device of FIG. 4(b)
9 and the inverse quantizer 29 can also be incorporated into the decoding device shown in FIG. 1(b).

An example of the interframe prediction function included in the interframe predictor 2 or 7 can be realized by simply delaying the input signal until the next frame and outputting the delayed signal. At this time, the interframe predictor 1 of the decoding device
5 or the interframe prediction function included in the interframe predictor 12, the human signal is simply delayed until the next frame and output.

In addition, a motion compensation method can also be used as the interframe predictor 2. FIG. 5(a) shows an embodiment in which a motion compensation technique is introduced as the interframe predictor 2 in the encoding apparatus of FIG. 8(a). The motion detection circuit 2 calculates the amount of image movement from the input image signal and outputs it to the circuit 22 and the multiplexer 18 as motion information.

The circuit 22 corrects the output signal of the adder 4 by the amount of image movement detected by the motion detection circuit 21 and outputs it as a predicted signal for the next frame. The multiplexer 18 multiplexes the motion information output from the motion detection circuit 21 with other signals and outputs the multiplexed signal to an output terminal 290. Figure 5 (
b) shows an embodiment in which a motion compensation technique is introduced as the interframe predictor 15 in the decoding device of FIG. 1(b). The demultiplexer 19 separates motion information from the signal input from the input terminal 390 and transmits it to the circuit 23. The circuit 23 corrects the decoded image signal of the previous frame based on the motion information and outputs it as a predicted signal of the current frame.

Further, the motion compensation method is implemented as the interframe predictor 7 in the encoding device of FIG. 8(b), as shown in FIG. 2(b),
2(b) as the interframe predictor 12 in the decoding device of FIG. 1(a),
It can also be applied when using the configuration (e).

When the configurations of FIGS. 2(C) and (d) are used as the interframe predictor 7 of the encoding device of FIG. 8(b), or the interframe predictor of the decoding device of FIG. 1(a) 2nd as 12
When using the configurations in Figures (c) and (d), interframe prediction is performed on orthogonal transform coefficients, so the motion compensation method described above cannot be applied as is. Tix Speech and Signal Processing (
International Conference
on Acoustucs.

5peech, and Signal Processes
sing) 4.89.1. pp157-160.19
Motion compensation can be performed using the motion compensation method using orthogonal transform coefficients described in April 1986 (Reference 2).

(Effect of the invention) In the conventional method, when orthogonal transformation coefficients are truncated during encoding, the signal components corresponding to the truncated orthogonal transformation coefficients are not corrected, and the signal components of the previous frame remain in the decoded image signal. However, in the present invention, such a phenomenon can be suppressed and room deterioration can be avoided.

[Brief explanation of the drawing]

FIGS. 1(a) and (b) are block diagrams of a decoding device of the present invention, FIGS. 2(a) to (e) are block diagrams showing an embodiment of a component truncation circuit of the present invention, and FIG. a), (b), 4th
Figures (a) and (b) and Figures 5 (a) and (b) are block diagrams showing embodiments of the encoding device and decoding device of the present invention, and Figures 6 (a) and (b) are block diagrams showing embodiments of the encoding device and decoding device of the present invention. FIG. 7 is an explanatory diagram of a conventional method, and FIGS. 8(a) and 8(b) are block diagrams showing an example of an encoding device corresponding to the decoding device of the present invention. In the figure, 1 is an orthogonal transform circuit, 2 and 15 are interframe predictors, 3 is a subtracter, 4 and 11 are adders, 6 and 14 are component truncation circuits, and 7 and 12 perform component truncation and interframe prediction. The circuits 8 and 13 are inverse orthogonal transform circuits, and 9 is a truncated component determination circuit.

Claims (3)

[Claims] (1) On the encoding side, the difference between the input image signal and the interframe prediction signal on the encoding side is taken, the difference signal is orthogonally transformed to obtain a encoded image signal, and the encoded image signal is inversely orthogonally transformed. and the inter-frame predicted signal to obtain a locally decoded signal, predict a signal of the next frame from the locally decoded signal and use it as the inter-frame predicted signal on the coding side,
On the decoding side, the encoded image signal is inversely orthogonally transformed,
A moving image in which the sum of an inverse orthogonal transformed signal and an interframe predicted signal on the decoding side is obtained as a decoded image signal, and the next frame is predicted from the decoded image signal to become the interframe predicted signal on the decoding side. In a signal encoding method, decoding of a moving image signal is characterized in that signal components corresponding to some orthogonal transform coefficients are removed from an interframe prediction signal or a decoded image signal on the decoding side. Method. (2) means for inversely orthogonally transforming an encoded image signal; means for adding the output signal of the inverse orthogonal changing means and the interframe prediction signal; and means for outputting the output signal of the adding means as a decoded image signal. Then, the next signal is predicted from the output signal of the addition means, and the signal components corresponding to some orthogonal transform coefficients are discarded based on the mode information generated by the encoding device from this predicted signal, and the signal components corresponding to the orthogonal transform coefficients are A signal that is output as a prediction signal, or is obtained by truncating signal components corresponding to some orthogonal transform coefficients based on mode information generated by an encoding device from the output signal of the adding means, and truncating the components. 1. A decoding device for a moving image signal, comprising an interframe predictor that predicts a signal of a next frame from a predicted signal and outputs the predicted signal as an interframe prediction signal. (3) means for inversely orthogonally transforming an encoded image signal, means for adding the output signal of the inverse orthogonal transformer and the interframe prediction signal, and a mode generated by the encoding device from the output signal of the adding means means for cutting off signal components corresponding to some of the orthogonal transform coefficients based on the information; means for outputting the output signal of the component cutting means as a decoded image signal; A moving image signal decoding device comprising an interframe predictor that predicts and outputs a frame signal.