JP2591438B2 - Moving image signal encoding method and apparatus - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a moving picture signal encoding method and apparatus which combine inter-frame prediction and orthogonal transformation.

[0002]

2. Description of the Related Art In a quantizer for orthogonal transform coefficients, a configuration described in, for example, JP-A-3-10486 is known as a configuration capable of reducing a code amount while suppressing visual deterioration. In this configuration, a threshold is determined according to the position of the transform coefficient, and the transform coefficients below the threshold are set to zero before quantization. In the two-dimensional conversion, the code amount can be reduced while suppressing visual deterioration by increasing a threshold value of a conversion coefficient representing a signal component in an oblique direction.

[0003]

In the above-described conventional configuration, since the difference in the effect of each block of the orthogonal transform on the decoded image is not taken into account, the effect of reducing the code amount while suppressing visual deterioration is considered. We have not been able to fully demonstrate.

[0004] It is an object of the present invention to reduce the amount of codes while suppressing visual deterioration in consideration of the difference in the effect of each block of the orthogonal transform on a decoded image.

[0005]

Means for Solving the Problems The movement of the first invention of the present invention
An image signal encoding method digitally compresses a moving image signal through inter-frame prediction and orthogonal transform processing, and encodes a transform coefficient of an input signal for each orthogonal transform block.
Or the conversion coefficient or conversion coefficient of the inter-frame prediction error signal.
Selects whether to encode a number of inter-frame prediction errors and selects
In the moving image signal encoding method for quantizing and encoding the encoded signal to be encoded, threshold processing is performed before the quantization, and the threshold used for the threshold processing is set to a value whose transform coefficient is in a high band.
And the selected signal to be encoded is the input signal
Of the inter-frame prediction error signal
From the case of inter-frame prediction error of coefficients or transform coefficients
Is also set to be large .

According to a second aspect of the present invention, there is provided an apparatus for encoding a moving image signal, comprising: a frame memory for storing a local decoded signal in a previous frame; and a prediction signal generated from the image signal stored in the frame memory. A subtractor that calculates a prediction error signal that is a difference between the input image signal and the input image signal, a determination circuit that determines whether to encode any of the input image signal and the prediction error signal, A first switch for selecting one of the prediction error signal and the input image signal, an orthogonal transformation circuit for orthogonally transforming a signal selected by the first switch, and a notification from a threshold memory to an output signal of the orthogonal transformation circuit A threshold processing circuit that performs threshold processing according to the threshold value to be processed; a quantizer that quantizes an output signal of the threshold processing circuit; And a second switch for selecting whether the prediction signal or the zero value according to the determination result of the determination circuit, and a signal selected by the second switch and an output signal of the inverse conversion circuit. And an adder for generating a local decoded signal in the current frame. The threshold value notified from the threshold value memory to the threshold value processing circuit is such that a transform coefficient processed by the threshold value processing circuit is in a high band. About
Large and select the input image signal with the first switch
Is larger than when the prediction error signal is selected.
Characteristic value is set.

According to a third aspect of the present invention, there is provided an apparatus for encoding a moving image signal, comprising: an orthogonal transformation circuit for performing an orthogonal transformation on an input image signal; a frame memory for storing a local decoded signal in a previous frame; A subtractor that calculates a prediction error signal that is a difference between a prediction signal generated from a signal stored in a memory and an output signal of the orthogonal transform circuit; and a subtractor that calculates an output signal of the orthogonal transform circuit and the prediction error signal. And a first switch that selects one of the prediction error signal and the output signal of the orthogonal transform circuit according to the determination result, and a first switch that is selected by the first switch. A threshold processing circuit that performs threshold processing on the signal in accordance with a threshold value notified from a threshold memory; a quantizer that quantizes an output signal of the threshold processing circuit; A second switch for selecting between the predicted signal and the zero value, and an adder for adding a signal selected by the second switch and an output signal of the quantizer to generate a local decoded signal in a current frame. And a threshold value notified from the threshold value memory to the threshold value processing circuit,
The higher the conversion coefficient processed by the threshold processing circuit is in the higher band,
Large and the output of the orthogonal transform circuit is
When the force signal is selected, the
It is characterized in that a value larger than the case is set.

[0008]

Next, the present invention will be described with reference to the drawings.

FIG. 1 is a block diagram of one embodiment of the present invention. The orthogonal transformation circuit 41 performs orthogonal transformation on the image signal input from the terminal 10, and the inter-frame prediction processing circuit 42 calculates the inter-frame prediction error and then performs the orthogonal transformation or performs the orthogonal transformation. Then, an inter-frame prediction error is calculated. The decision circuit 43 decides which of the transmission signals of the orthogonal transformation circuit 41 and the inter-frame prediction processing circuit 42 is to be coded. And then to the threshold processing circuit 12. The threshold processing circuit 12 has the input / output characteristics shown in FIG. 2 according to whether or not the value of the signal selected by the switch 44 is larger than the threshold Th given from the threshold memory 14 (absolute value).
If the absolute value of the input signal is less than the threshold Th, it is replaced with zero. The signal of the processing result of the threshold processing circuit 12 is output from the terminal 20 after being quantized by the quantizer 13. The output signal of the quantizer 13 is also input to the inter-frame prediction processing circuit 42, and a local decoded signal is calculated according to the determination result of the determination circuit 43, and is used for prediction in a subsequent frame.
In the threshold memory 14, an address is specified by the determination result of the determination circuit 43, the quantization step size Q, and the coefficient position of the orthogonal transform coefficient processed by the threshold processing circuit 12,
The threshold Th is read. An example of the characteristic of the threshold will be described later.

FIG. 5 is a block diagram showing a configuration example of the inter-frame prediction processing circuit 42 in the embodiment of FIG.
The frame memory 17 stores the local decoded signal of the previous frame, and the subtracter 15 calculates the difference between the local decoded signal and the image signal input from the terminal 10, and calculates a difference between the frames as a prediction error signal between frames. Is output. The determination circuit 21 determines which of the input image signal and the prediction error signal is to be encoded, and the switch 22 selects the corresponding signal according to the determination result of the determination circuit 21. The signal selected by the switch 22 is orthogonally transformed by the orthogonal transformation circuit 16, subjected to threshold processing by the threshold processing circuit 12, quantized by the quantizer 13, and output from the terminal 20 as a coded signal. The output signal of the quantizer 13 is subjected to an inverse transform of the orthogonal transform by the orthogonal transform circuit 16 in the inverse transform circuit 18, and is output to the adder 19. On the other hand, the switch 23 selects the output signal of the frame memory 17 according to the determination result of the determination circuit 21 when the determination result indicates selection of the prediction error signal, and the determination result indicates selection of the input image signal. If so, a zero value is selected and output to the adder 19. The adder 19 adds the output signals of the inverse conversion circuit 18 and the switch 23 and outputs the result as a local decoded signal of the current frame. Threshold memory 1
In 4, the threshold Th is determined according to the coefficient position of the transform coefficient processed by the threshold processing circuit 12, the determination result of the determination circuit 21, and the quantization step size, and the threshold Th is notified to the threshold processing circuit 12.

FIG. 5B is a block diagram showing another configuration example of the inter-frame prediction processing circuit 42 in the embodiment of FIG. The image signal input from the terminal 10 is first orthogonally transformed by the orthogonal transformation circuit 24 and then inputted to the subtracter 25. The subtractor 25 subtracts the output signal of the frame memory 26 from the output signal of the orthogonal transform circuit 24 and outputs the result as a prediction error signal between frames. The frame memory 26 stores the locally decoded signal of the previous frame. The determination circuit 31 determines whether to encode the output signal of the orthogonal transform circuit 24 or the prediction error signal, and the switch 32 selects the corresponding signal according to the determination result. The signal selected by the switch 32 is subjected to threshold processing by the threshold processing circuit 12 and then quantized by the quantizer 13.
It is output from the terminal 20 as an encoded signal. On the other hand, the switch 33 selects the output signal of the frame memory 26 when the prediction error signal is selected based on the determination result of the determination circuit 31, and selects the output signal of the orthogonal transform circuit 24 based on the determination result. Is selected and output to the adder 27. The adder 27 adds the output signals of the quantizer 13 and the switch 33 and outputs the result as a local decoded signal of the current frame. The threshold memory 14 determines a threshold Th based on the coefficient position of the transform coefficient processed by the threshold processing circuit 12, the determination result of the determination circuit 31, and the quantization step size, and notifies the threshold processing circuit 12 of the threshold Th.

As the threshold value Th stored in the threshold value memory 14, for example, a characteristic as shown in FIG. 3 can be given as an example. The vertical axis of FIG. 3 represents the magnitude of the threshold value Th, and the horizontal axis corresponds to the level of the frequency represented by the transform coefficient. The transform coefficient representing the low frequency component is on the left, and the transform coefficient representing the high frequency component is on the right. Suppose there is. In FIG. 3, the characteristic curve c takes a value a in a low frequency range, rapidly rises as the frequency increases, and becomes a value b. The curves d and e are obtained by making the rising of the curve c gentle. As the curves become c, d, and e, the number of transform coefficients whose values are replaced by zero in the threshold processing circuit 12 increases. Less.

By selecting the characteristic curves c, d, and e in accordance with the coding type of the block, it is possible to take into account the difference in the visual deterioration that each block gives to the decoded signal. For example, in the configuration example of FIG. 5, a signal obtained by orthogonally transforming an inter-frame prediction error signal is normally encoded. If the inter-frame prediction error signal is large, a signal obtained by orthogonally transforming the input image signal itself is encoded. When encoding a signal obtained by orthogonally transforming the inter prediction error signal, the curve e is selected, and a signal obtained by orthogonally transforming the input signal is encoded.
To select the curve c at the time. When the curve c is selected when encoding a signal obtained by orthogonally transforming the input signal,
Instead of reducing the code amount, the high-frequency transform coefficient with small amplitude becomes zero, and the decoded image is blurred. That is, conversion
In the threshold processing of numbers, increase the threshold as the frequency increases.
As a result, the high-frequency transform coefficient with a small amplitude becomes zero,
The image will be blurred. The signal to be encoded is the input signal
Conversion coefficient of the inter-frame prediction error signal
Than the case of inter-frame prediction error of numbers or transform coefficients
By increasing the threshold, the code amount can be reduced instead of being restored.
The number image will be blurred. However, such a block has a large inter-frame prediction error, in other words, a large motion, so that even if the decoded image is blurred, there is no visual deterioration. The same applies to the configuration example of FIG. 6 except that the order of the inter-frame prediction and the orthogonal transformation is changed in the configuration example of FIG.

As a method for setting the thresholds a and b in FIG. 3, there is a method for setting the thresholds a and b in accordance with the ratio to the quantization step size Q. For example, when the orthogonal transformation is an eight-row, eight-column two-dimensional transformation,
Values such as those shown in FIGS. 4A and 4B are set. FIG.
In (a) and (b), the threshold value Th at each position in the block is represented by a scaling factor with respect to the quantization step size Q. The higher the frequency, the higher the frequency from left to right in the horizontal direction, and the lower the frequency in the vertical direction. Represents a threshold value for a transform coefficient such that the higher the frequency, the higher the frequency. The threshold values shown in FIGS. 4A and 4B have the same value on a diagonal line from the upper right to the lower left. However, the present invention is not limited to the method shown in FIG.

[0015]

As described above, according to the present invention, the code amount can be reduced while suppressing the visual deterioration in consideration of the difference in the effect of each block of the orthogonal transform on the decoded image.

[Brief description of the drawings]

FIG. 1 is a configuration diagram of an embodiment according to the present invention.

FIG. 2 is a characteristic diagram illustrating input / output characteristics of a threshold processing circuit 12 in FIG. 1;

FIG. 3 is a threshold Th stored in a threshold memory 14 in FIG. 1;
FIG. 6 is a characteristic diagram showing a setting example of FIG.

FIGS. 4A and 4B show a threshold Th stored in a threshold memory 14 in the case of two-dimensional orthogonal transformation in the embodiment of FIG.
FIG. 9 is a screen block diagram showing an example of setting a scaling factor for a quantization step size Q of FIG.

FIG. 5 is a block diagram showing a detailed configuration example of the embodiment of FIG. 1;

FIG. 6 is a block diagram showing another detailed configuration example of the embodiment of FIG. 1;

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 10 Input terminal 12 Threshold processing circuit 13 Quantizer 14 Threshold memory 15, 25 Subtractor 16, 24, 41 Orthogonal transformation circuit 17, 26 Frame memory 18 Inverse transformation circuit 19, 27 Adder 20 Output terminal 21, 31, 43 Judgment Circuit 22, 23, 32, 33, 44 Switch 42 Inter-frame prediction processing circuit

Claims (3)

(57) [Claims] 1. A video signal is digitally compressed by inter-frame prediction and orthogonal transform processing, and transform coefficients of an input signal are encoded for each block of the orthogonal transform.
Transform coefficient of inter prediction error signal or frame of transform coefficient
Selects whether to encode the prediction error between
In a moving image signal encoding method for quantizing and encoding a target signal , a threshold value process is performed before the quantization, and a threshold value used for the threshold value process increases as the transform coefficient becomes higher.
And the selected encoding target signal is the transform coefficient of the input signal.
In the case of, the conversion coefficient of the inter-frame prediction error signal or
Coding method of the moving image signal, characterized by large set <br/> constant than in the case of inter-frame prediction error of the transform coefficients. 2. A frame memory for storing a local decoded signal in a previous frame, and a prediction error signal which is a difference between a prediction signal generated from an image signal stored in the frame memory and an input image signal. A subtractor, a determination circuit for determining which of the input image signal and the prediction error signal is to be encoded, and a first circuit for selecting one of the prediction error signal and the input image signal according to the determination result A switch, an orthogonal transform circuit for orthogonally transforming the signal selected by the first switch, a threshold processing circuit that performs threshold processing according to a threshold value notified from a threshold memory to an output signal of the orthogonal transform circuit, A quantizer that quantizes an output signal of the threshold processing circuit, an inverse transform circuit that performs an inverse transform of the orthogonal transform circuit on an output signal of the quantizer, and a decision result of the decision circuit. A second switch for selecting between the predicted signal and the zero value in response thereto, and an adder for adding a signal selected by the second switch and an output signal of the inverse transform circuit to generate a local decoded signal in the current frame. And
The threshold value notified from the threshold value memory to the threshold value processing circuit is such that the transform coefficient processed by the threshold value processing circuit is in a high band.
And the input image signal is controlled by the first switch.
Is selected, the prediction error signal is selected.
A moving image signal encoding apparatus, wherein a large value is also set. 3. An orthogonal transformation circuit for performing an orthogonal transformation on an input image signal, a frame memory storing a local decoded signal in a previous frame, and a prediction signal generated from a signal stored in the frame memory and the orthogonal signal. A subtractor that calculates a prediction error signal that is a difference from an output signal of the transformation circuit, a decision circuit that decides which of the output signal of the orthogonal transformation circuit and the prediction error signal is to be encoded, and a decision result thereof. A first switch for selecting one of the prediction error signal and the output signal of the orthogonal transform circuit in response to the threshold value, and performing a threshold process on the signal selected by the first switch in accordance with a threshold value notified from a threshold value memory A threshold processing circuit, a quantizer that quantizes an output signal of the threshold processing circuit, and a second switch that selects the prediction signal or a zero value according to a determination result of the determination circuit. An adder that adds the signal selected by the second switch and the output signal of the quantizer to generate a local decoded signal in the current frame, and notifies the threshold processing circuit from the threshold memory. The threshold to be processed is such that the conversion coefficient processed by the threshold processing circuit is in a high frequency range.
And a quadrature conversion circuit using the first switch.
If you select the output signal of
A moving image signal encoding apparatus, wherein a larger value is set than in the case where