JPH0514876A - Moving image encoding system - Google Patents

Abstract

PURPOSE:To control the average amount of information which is generated after encoding process to the intended amount per unit time, regardless of the nature of inputted moving image data, and to suppress image quality fluctuation of a reproduced image caused by controlling of information amount to a minimum which is required. CONSTITUTION:A transformation coefficient related with orthogonal transformation of an original image data is, by a quantization circuit 5, quantized at a quantization step width corresponding to the quantization parameter which comes from a quantization parameter control circuit 17, and then supplied to a variable length encoding circuit 6 so as to become an encoded data. This encoded data is supplied to a buffer memory 8 so as to be stored, and then read out at a specified speed. The quantization parameter control circuit 17 sets, in proportion to both the amount of accumulated data in the buffer memory 8 as of after encoding processing and to that as of in the past, a quantization parameter so as to obtain a quantization step width by which an information amount can be made optimum in the next encoding process.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a moving picture coding system for compressing digital moving picture data, and more particularly to a moving picture coding system using a feedback type information amount control technique.

[0002]

2. Description of the Related Art In order to transmit highly efficient encoded digital moving image data (hereinafter referred to as encoded data) through a data communication line or record it on a data recording medium,
Since the transmission speed of the data communication line and the transfer speed of the data recording medium are limited, the information amount control for controlling the generated information amount of the data-compressed encoded data is an essential technique.

In general, in such a moving picture coding system, coded data compressed for data processing is temporarily stored in a buffer memory, and then the transmission speed of a data communication line or the transfer of a data recording medium. The encoded data is read from the buffer memory at a speed according to the speed and transmitted or recorded. Also, the quantization step width of the quantization circuit in the next encoding process is determined according to the amount of untransmitted data stored in the buffer memory, that is, the buffer memory storage amount. When the buffer memory storage amount is large, the quantization step width is increased to control the amount of generated information and reduce the buffer memory storage amount. Conversely, when the buffer memory storage amount is reduced In order to reduce the quantization step width, the generated information amount is increased and the buffer memory storage amount is increased. Then, when the buffer memory overflows, the encoding process is stopped, or the next encoding process is performed with a quantization step width larger than the quantization step width of the determined range,
The amount of generated information is controlled.

Feedback type information generation control for determining the quantization step width based on the buffer memory storage amount is expressed by equation (1) in FIG. Here, Q is the quantization step width, BO is the buffer memory storage amount, OFS is the offset value of the quantization step width, INT () is rounded down toward 0, BM is the buffer size of the buffer memory, and QR is the quantization. It is the number of steps, and S is represented by the equation (2) in FIG.

A system related to such a moving picture coding system is described in, for example, Japanese Patent Application Laid-Open No. 2-222388.

[0006]

SUMMARY OF THE INVENTION The above-mentioned prior art is based on a TV.
It is widely used for high-efficiency compression of digital moving image data in conference systems and TV phones. The image signal in a TV conference or a videophone is composed of an image in which the moving area is small and there are many static areas such as only the person moving, that is, an image in which the correlation between screens is very strong. The amount of generated information after the encoding process is almost uniform. Therefore, according to the above-mentioned conventional technique, the buffer memory storage amount is used so that the average generated information amount becomes the information amount per unit time obtained from the predetermined transmission speed of the data communication line or the predetermined transfer speed of the data recording medium. Can be controlled.

However, in the case of an image signal such as a television signal representing a normal moving image, a camera effect such as panning and zooming, an image with a fine pattern, an image with a large amount of movement, and the like change from moment to moment. Therefore, the amount of generated information also greatly changes according to this change. When the above-mentioned conventional technique is used for such an image signal, the amount of generated information changes from moment to moment, so that the amount of data stored in the buffer memory also changes, and the quantization step width is switched accordingly. Control will be performed frequently. Therefore, it is not only difficult to control the average amount of generated information to be the amount of information per unit time that is suitable for the data communication line and the data recording medium, but also the image quality of the reproduced image due to the change in the quantization step width. Will cause fluctuations in.

An object of the present invention is to solve such a problem, and to determine the average amount of generated information after encoding processing regardless of the nature of input moving image data by a predetermined transmission rate of a data communication line or a predetermined transfer rate of a data recording medium. To provide a moving picture coding method capable of controlling the amount of information per unit time obtained from the above, and suppressing the image quality fluctuation of a reproduced image due to the information amount control to a necessary minimum. It is in.

[0009]

To achieve the above object, the present invention quantizes transform coefficients obtained by orthogonally transforming original image data of a moving image and performs variable length coding to obtain coded data, The coded data is temporarily stored in a buffer memory and read at a predetermined speed, and a storage state determination means for determining a storage state of the coded data in the buffer memory and the storage state determination Quantization from state storage means for holding the determination output of the past accumulation state of the buffer memory from the means, the determination output of the accumulation state determination means, the holding determination output of the state storage means, and the storage amount of the buffer memory A quantization parameter control means including a quantization parameter control determination means for determining a parameter is provided, and the quantization step width for the next quantization is determined according to the quantization parameter. And the.

Further, according to the present invention, transform coefficients obtained by orthogonally transforming original image data of a moving image are quantized and variable length coded to obtain coded data, and the coded data is temporarily stored in a buffer memory. The information amount calculation means for calculating the generated information amount of the encoded data, and the information amount state determination for determining the state of the generated information amount from the information amount calculation means Means, a state storage means for holding the determination output from the information amount state determination means and outputting it as a determination output for the state of the past generated information amount, a determination output of the information amount state determination means, and a state storage means from the state storage means. Quantization parameter control means including a quantization parameter control determination means for determining a quantization parameter from the past determination output and the storage amount of the buffer memory after the encoding processing is provided, Coca step width is assumed in accordance with the quantization parameter.

Further, according to the present invention, the quantization parameter determined by the quantization parameter control means is converted into a more optimal quantization parameter for the next encoding process according to the buffer memory storage state or the generated information amount state. Quantization parameter correcting means for correcting to is provided.

[0012]

The storage state comparison / determination means determines the storage amount of the buffer memory, that is, the storage degree of the untransmitted encoded data to the data communication line or the data recording medium in the buffer memory, and The quantity state determination means assigns the generated information quantity after the coding processing to the coded frame determined by the frame structure in the moving picture coding system and the predetermined transmission rate of the data communication line or the predetermined transfer rate of the data recording medium. The amount of generated information is compared with the amount of information that is generated.

The quantization parameter control means determines the average generated information amount by the buffer memory accumulated state data from the accumulated amount state judging means or the generated information amount state data from the information amount state judging means and the accumulated amount of the buffer memory. Is adaptively processed to control the amount of information per unit time, and the quantization parameter for the next encoding process is set.

The quantization parameter correction means uses the information on the storage state or generated information amount state in the buffer memory after the encoding process to set the above-mentioned quantization parameter to the next encoding process. In order to make it more optimal, it is adaptively corrected.

In this way, the optimum quantization parameter for the next encoding process is set from the state of the state of change in the data storage amount or generated information amount in the buffer memory after the encoding process.

It will be done.

Embodiments of the present invention will be described below with reference to the drawings. FIG. 1 is a block diagram showing an embodiment of a moving picture coding apparatus according to the present invention to which an information amount control method is applied.
Is an input terminal for original image data, 2 is a unit block conversion circuit, 3 is a data subtraction circuit, 4 is an orthogonal conversion circuit, 5 is a quantization circuit, 6 is a variable length coding circuit, 7 is a data multiplexing circuit,
8 is a buffer memory, 9 is an output terminal for encoded data, 1
0 is an inverse quantization circuit, 11 is an inverse orthogonal transformation circuit, 12 is a data addition circuit, 13 is a frame memory, 14 is a motion compensation circuit, 15 is a motion vector detection circuit, 16 is a motion vector coding circuit, and 17 is a quantization. It is a parameter control circuit.

In the figure, original image data which is a digitized video signal is input from an input terminal 1 and converted by a unit block conversion circuit 2 into a block of 8 × 8 pixels which is a unit of encoding processing. The data subtraction circuit 3 subtracts the motion-compensated prediction image data from the motion compensation circuit 14 from the block-shaped original image data to generate motion-compensated prediction error data which is the difference between them. The motion-compensated predicted image data is image data predicted by motion compensation from image data of past frames that have already been encoded.

The obtained motion-compensated prediction error data is subjected to a two-dimensional discrete cosine transform process in which a block of 8 × 8 pixels is performed by the orthogonal transform circuit 4, and the transform coefficient corresponding to the frequency component obtained by this transform process is a quantum. The quantization circuit 5 quantizes with a predetermined quantization step width. And
For the quantized transform coefficient, the variable length coding circuit 6
Code assignment is performed so that short codes are assigned to data with high occurrence frequency and long codes are assigned to data with low occurrence frequency, and encoded data is formed.

The encoded data is the data multiplexing circuit 7
As a result, the motion vector encoded data of the additional information and the quantization step width information are multiplexed, and once the buffer memory 8
After being stored in, the data is read out at a speed according to the transmission speed of the data communication line or the transfer speed of the data recording medium and output from the output terminal 9. The motion vector coded data is the information of the motion vector used for the motion compensation, and is formed by the motion vector coding circuit 16. Further, the information of the quantization step width is obtained from the quantization parameter from the data amount (buffer memory accumulated amount) of the encoded data for one frame stored in the buffer memory 8 at the time when the encoding process for one frame is completed. This is information on the quantization parameter adaptively determined by the control circuit 17, and this information defines the quantization step width in the quantization circuit 5.

The quantized transform coefficient from the quantizer 5 is converted back into a transform coefficient in the inverse quantizer 10 and further converted into motion compensation prediction error data by the inverse discrete cosine transform in the inverse orthogonal transform circuit 11. Will be returned. This locally-decoded motion-compensated prediction error data is added by the data addition circuit 12 to the motion-compensated prediction image data from the motion-compensation circuit 14 used for the subtraction by the data subtraction circuit 3 as described above, and then locally-decoded. The converted image data is written in the frame memory 13. This locally-decoded image data is image data to be reproduced by data expansion in the moving image decoding device paired with the moving image encoding device, and is delayed by one frame in the frame memory 13 and supplied to the motion compensation circuit 14. And is used to generate motion-compensated predicted image data by motion compensation in the next frame.

The motion compensation circuit 14 shifts the locally-decoded image data of one frame before from the frame memory 13 in block units according to the motion vector detected by the motion vector detection circuit 15 to obtain motion-compensated prediction image data. To generate. The motion vector detection circuit 15 compares the original image data of the current block of the current frame from the unit block conversion circuit 2 with the locally encoded image data of the previous frame from the frame memory 13 while shifting the pixel data on a pixel-by-pixel basis. The block shift amount of which the block difference value is reduced is detected as a motion vector. The motion vector coding circuit 16 performs variable length coding on the difference between the motion vector and the preceding block.

Next, the quantization parameter control circuit 17 will be described. FIG. 2 shows the quantization parameter control circuit 1
7 is a block diagram showing a specific example of 7, wherein 18 is a storage state comparison / determination circuit, 19 is a state determination threshold setting circuit, 20 is a delay circuit, 21 is a quantization parameter control selection circuit, and 221 to 22N are parameter control circuits. , 23 are input terminals, and 24 are output terminals.

In the figure, after the accumulation of one frame of encoded data in the buffer memory 8 is completed, the input terminal 2
The information (hereinafter, referred to as accumulated amount information) representing the accumulated amount of data in the buffer memory 8 (FIG. 1) (that is, accumulated amount information in the buffer memory) and the quantization parameter in this frame are input from 3 to select the quantization parameter control. It is supplied to the circuit 21 and the storage state comparison / determination circuit 18. In the storage state comparison / determination circuit 18, the storage state of the buffer memory 8 is determined based on the storage amount information and the threshold value set in the state determination threshold setting circuit 19, and the storage state of the buffer memory 8 is determined. The buffer memory accumulated state data representing the is output to the quantization parameter control selection circuit 21. The buffer memory accumulated state data is delayed by n (where n is an integer of 1 or more) frames by the delay circuit 20 and supplied to the quantization parameter control selection circuit 21 as the buffer memory accumulated state data n frames before. . Quantization parameter control selection circuit 21
Is a parameter control circuit 221 that performs different parameter control by a selection method to be described later based on the buffer memory storage state data from the storage state comparison / determination circuit 18 and the buffer memory storage state data of n frames before from the delay circuit 20. 22N to 22N, a parameter control circuit that performs optimum parameter control for determining the quantization parameter in the encoding process of the next frame is selected.
The parameter control circuit thus selected sets the quantization parameter to be used for the encoding process of the next frame from the storage amount information and the quantization parameter supplied from the buffer memory 8 via the input terminal 23. , From the output terminal 24 to the quantizing circuit 5 and the data multiplexing circuit 7.

Here, an example of the comparison judgment processing of the accumulation state of the buffer memory 8 in the accumulation state comparison judgment circuit 18 is shown in FIG.
Described in. In the figure, the state determination threshold determination circuit 19 includes, for example, three threshold values Th1.
To Th3 (where Th1 <Th2 <Th3 <maximum data storage amount of the buffer memory 8) is set, and the storage amount D of the buffer memory 8 is 0 ≦ D <Th1, the state (a), Th1 ≦ D <When Th2, state (b), Th
When 2 ≦ D <Th3, the state (c) is determined, and when Th3 ≦ D <buffer full, the state (d) is determined. The determination result is supplied to the quantization parameter control selection circuit 21 and the delay circuit 20 as buffer memory accumulated state data.

The quantization parameter control selection circuit 21 is a parameter control circuit 2 corresponding to the buffer memory storage state data from the storage state comparison / determination circuit 18 and the delay circuit 20.
21 to 22N are selected, but the buffer memory 8
In the case where the accumulation state at the time when the accumulation of the encoded data for one frame is finished is divided into the four states shown in FIG. 3 and judged, 16 types of parameter control are performed as shown in FIG. 1 to 16 will be set. However, FIG.
The control i (i = 1, 2, ..., 16) in is the parameter control by the parameter control circuit 22i in FIG. Then, for example, when the data storage state of the buffer memory 8 changes from the state (a) to the state (b), the quantization parameter control selection circuit 21 selects the control 2 as the optimum one for the change of the data storage state. become.

FIG. 5 shows a quantization parameter control selection circuit 21.
It shows a specific example of the parameter control selected in. In this specific example, for the sake of convenience of explanation, two predetermined threshold values are set in the state determination threshold setting circuit 19, and the accumulation states of the buffer memory are A, B, and
It is assumed that the judgment is divided into three states of C state.
Therefore, the outputs of the storage state comparison / determination circuit 18 and the delay circuit 20 enable parameter control of nine kinds of quantization parameters as shown in FIG. The A, B, and C states correspond to the states (a), (b), and (c) of FIG. 4, respectively, and the nine types of parameter control correspond to control 1 to control 9 of FIG.

In FIG. 5, the storage states of the buffer memory 8 are changed from the storage state X represented by the buffer memory storage state data from the delay circuit 20 to the storage state comparison / determination circuit 18.
The storage state Y is represented by the storage state data of the buffer memory from, but if this is expressed as (X → Y), (A →
In the case of A), the control 1 is selected and the minimum value q_min of the quantization parameter is set as the quantization parameter in the next encoding process. As a result, the quantization step width of the quantization circuit 5 is minimized, and the information amount of the encoded data in the next frame is increased. In the case of (A → B), the control 2 is selected, the initial value q_init of the quantization parameter is set as the quantization parameter in the next encoding process, and the quantization step width of the quantization circuit 5 is set to the initial value (for example, the intermediate value). Quantization step width) to maintain the information amount of encoded data in the current state. , (A → C), the control 5 is selected and the quantization parameter q from the input terminal 23 is offset by s2.
The added value q + s2 is used as a quantization parameter in the next encoding process, and the amount of information is reduced by increasing the quantization step width of the quantization circuit 5 by this offset. (B →
In the case of A), control 3 is selected and the quantization parameter q
The value q_s1 obtained by subtracting the offset s1 from is used as the quantization parameter in the next encoding process, and the quantization step width of the quantization circuit 5 is reduced by this offset to increase the information amount. In the case of (B → B), the control 4 is selected, and as shown in FIG. 6A, the quantization parameter determined from the quantization parameter control characteristic according to the buffer memory storage amount from the input terminal 23 is set. It is used as the quantization parameter in the next frame. In this case, even if the B state continues, the quantization step width is controlled therein, and the B state is maintained. In the case of (B → C), the control 6 is selected to obtain the value q + s1 obtained by adding the offset s1 to the quantization parameter q, and in the case of (C → A), the control 7 is selected to offset from the quantization parameter q. In the case of (C → B), the value q_s2 obtained by subtracting s2 is selected as the control 8 and the initial value q_init of the quantization parameter is used as the quantization parameter of the next frame, and the same quantization step width as above is used. Control is performed. In the case of (C → C), the control 9 is selected and the maximum value of the quantization parameter q_ma
Let x be the quantization parameter in the encoding process of the next frame, and maximize the quantization step width in the next frame to reduce the buffer memory storage amount to the maximum.

The offsets s1 and s2 are shown in FIG.
As shown in (b), it may be different according to the buffer memory storage amount input from the input terminal 23.
It may also be constant.

FIG. 7 shows changes in the buffer memory storage amount for each frame when the above quantization parameter control is performed. The arrow in the figure indicates a frame in which the parameter control is changed with respect to the previous frame. By performing the quantization parameter control in this way, the buffer memory storage amount can be controlled to the B state (that is, the storage amount in the middle of the capacity of the buffer memory 8).

As described above, according to this embodiment, the optimum quantization parameter for the encoding process of the next frame is determined from the data accumulation state of the buffer memory 8 after the end of the encoding process and n frames before. Since it can be determined, the buffer memory 8 can be controlled to a state of a predetermined storage amount, and the storage state of the predetermined storage amount can be determined in advance by the transmission speed of the data communication line or the transfer of the data recording medium. By setting the storage state so that the amount of information per unit time obtained from the speed is set, it is possible to control the average amount of generated information after encoding processing to be the amount of information per unit time. Becomes Further, there is an effect that overflow and underflow of the buffer memory 8 can be prevented.

In this embodiment, the information amount control is performed on a frame-by-frame basis. However, a group of several coded blocks may be used as the information amount control unit, or the information may be considered in consideration of an interlaced image. The quantity control may be performed in field units. For past accumulation status,
In this embodiment, the accumulation state of the buffer memory 8 one frame before is used for explanation, but all the accumulation states of the buffer memory 8 up to several frames before may be used.

FIG. 8 is a block diagram showing another embodiment of the moving picture coding apparatus according to the present invention to which the information amount control system is applied. 17 'is a quantization parameter control circuit, which corresponds to FIG. The same reference numerals are given to the parts, and the duplicated description will be omitted.

In the figure, the quantization parameter control circuit 17 ′ has an amount of information (generated information amount) generated in this frame at the time when the encoding process of one frame is completed and the amount of data stored in the buffer memory 8. The quantization parameter in the encoding process of the next frame is adaptively determined according to the (buffer memory storage amount), and the information is supplied to the quantization circuit 5 and the data multiplexing circuit 7.

FIG. 9 is a block diagram showing a specific example of the quantization parameter control circuit 17 ', in which 26 is an information amount addition circuit, 27 is a generated information amount state comparison / determination circuit, 28 is an assigned information amount setting circuit, Reference numeral 29 is a state determination threshold setting circuit, and 30 is a delay circuit. The parts corresponding to those in FIG.

In the figure, first, the data multiplexing circuit 7
The encoded data output from (FIG. 8) is input from the input terminal 25, the information amount is accumulated in the information amount adding circuit 26, and the generated information of the encoded data after the encoding process, that is, for one frame is generated. The amount is detected. This generated information amount is calculated by the generated information amount state comparison / judgment circuit 27 using the equation (3) shown in FIG.
By the comparison / determination processing represented by (4) and (5), the state (a in the above equations (3), (4), and (5) that represents the generated information amount of the encoded data from the data multiplexing circuit 7 is represented. ),
(B) and (c)) are determined, and the generated information amount state data representing the determination result is the quantization parameter control selection circuit 2
It is output to 1. However, in these equations, I is the generated information amount, AN is the assigned information amount for each frame set in the assigned information amount setting circuit 28, Th1, Th2, Th3.
Is a threshold value set by the determination threshold setting circuit 29, and Th1 <Th2 <Th3.

The generated information amount state data output from the generated information amount state comparison / judgment circuit 27 is delayed by the delay circuit 30 having a delay amount of n (where n is an integer of 1 or more) frames, and n frames are obtained. It is supplied to the quantization parameter control selection circuit 21 as past generation information amount state data up to the previous or nth frame. The quantization parameter control selection circuit 21 encodes the next frame from the generated information amount state data after the encoding process from the generated information amount state comparison determination circuit 27 and the past generated information amount state data from the delay circuit 30. The optimum quantization parameter control for determining the quantization parameter for processing (quantization parameter control for making the generated information amount in the next frame equal to the allocated information amount of that frame) is shown in FIG. The parameter control circuits 221 to 22N are selected by the same selection method as in the embodiment. The selected parameter control circuit adaptively sets the quantization parameter to be used for the encoding process of the next frame according to the buffer memory storage amount and the quantization parameter supplied via the input terminal 23, and the output terminal 24 to the quantizing circuit 5 and the data multiplexing circuit 7 (FIG. 8). As a result, in the next frame, the quantization step width of the quantization circuit 5 is set so that the generated information amount of the encoded data from the data multiplexing circuit 7 becomes the allocation information amount of the frame by the set quantization parameter. Is set.
However, since this quantization parameter also corresponds to the buffer memory storage amount, the quantization step width of the quantization circuit 5 changes according to the buffer memory storage amount,
As a result, the quantization step width is determined while also looking at the amount of data stored in the buffer memory 8.

Here, the quantization parameter control selection circuit 2
The quantization parameter control selected by 1 does not uniquely determine the quantization parameter as in the embodiment shown in FIG. 1, and further, different quantization parameters can be obtained according to the buffer memory storage amount. It has a characteristic with respect to the accumulated buffer memory amount. Examples of quantization parameter control according to the generated information amount state data are shown in FIGS.

FIG. 10 shows an example of the quantization parameter control in which the characteristics have the same slope with respect to the buffer memory storage amount. Here, three quantization parameter controls with different characteristics are shown, the broken line is characteristic 1, the solid line is characteristic 2,
The alternate long and short dash lines respectively represent the quantization parameter control of characteristic 3, but they all have the same slope. In each quantization parameter control, if the buffer memory storage amount is different,
The quantization parameters are also different. For example, when the quantization parameter control of the characteristic 2 is selected by the generated information amount state data after the encoding process from the generated information amount state comparison determination circuit 27 and the past generated information amount state data from the delay circuit 30, As the buffer memory storage amount increases by one step, the quantization parameter increases to q ₁ , q ₂ , q ₃ , ..., And the quantization step width of the quantization circuit 5 increases to increase the information amount of encoded data. Decrease.

The quantization parameter control having such a characteristic is based on one quantization parameter, and the other quantization parameter control can be realized by shifting the reference quantization parameter control vertically or horizontally. For example, characteristic 2
On the basis of the quantization parameter control of the characteristic 1, the quantization parameter control of the characteristic 1 is shifted upward by the quantization parameter control of the characteristic 2, and the quantization parameter control of the characteristic 3 is controlled by the quantization parameter control of the characteristic 2. Each is obtained by laterally shifting the activation parameter control.

According to the quantization parameter control of such characteristics, even if the buffer memory storage amount is the same, the generated information amount state data after the encoding process from the generated information amount state comparison / determination circuit 27 and the n from the delay circuit 30. The quantization parameter differs depending on the generated information amount state data before the frame. Therefore, it can be applied when there is little variation in the buffer memory storage amount but there is a difference between the generated information amount and the assigned information amount. .

Further, FIG. 11 shows a quantization parameter according to the generated information amount state data after the encoding process from the generated information amount state comparison / determination circuit 27 and the generated information amount state data n frames before from the delay circuit 30. Another example of control will be described. Here again, three quantization parameter controls having different characteristics are shown. The broken line represents the characteristic 4, the one-dot chain line represents the characteristic 5, and the solid line represents the characteristic 6 quantization parameter control. The quantization parameter control can be realized by changing the slope of the quantization parameter control of characteristic 6.
The characteristics with a steep slope can be applied when the number of quantization parameters is large or when a large change in the quantization parameters is desired in response to a slight change in the buffer memory storage amount. This can be applied when the number of is not small or when it is desired to reduce the change in the quantization parameter even if the buffer memory storage amount changes significantly.

Although all of these characteristics are linear, the quantization parameter control of a nonlinear characteristic like the characteristic 6 may be used. With such a non-linear characteristic, the quantization parameter is finely controlled when the buffer memory storage amount is small, and conversely, the quantization parameter is roughly controlled when the buffer memory storage amount is large. Also,
Although not shown, a non-linear characteristic can be used so that the quantization parameter is coarsely controlled when the buffer memory storage amount is small and the quantization parameter is finely controlled when it is large.

When such an embodiment is applied to a data recording medium such as a CD-ROM or DAT, a special reproduction function such as random access or still reproduction is required, so that the intraframe coding is periodically performed. . The frame structure when the intra-frame encoding cycle is N frames is such that one frame is formed of N frames as shown in FIG. 12A, and the first frame of the N frames is the frame. An intra-coded frame,
The following (N-1) frame is an inter-frame coded frame. When encoding such a frame structure, the above-mentioned allocation information amount is different between the intra-coded frame and the inter-frame coded frame, and the allocation information amount for the intra-frame encoded frame is different. In Equation (6) of FIG. 15, the allocation information amount for the inter-frame coded frame is also expressed by Equation (7). However, in these formulas (6) and (7), AI
Is the allocation information amount for intraframe coding, AP is the allocation information amount for interframe coding, N is the cycle of intraframe coding, DR
Is the average transfer rate (unit is [bit / sec]), F is 1
Number of frames per second (for normal television signals:
F = 30), AR is the ratio (AI / AP) of the allocation information amount for intraframe / interframe coding. FIG. 12B shows the allocation information amounts AI and AP for the frames 1, 2, 3, ... Shown in FIG. Further, the allocation information amounts AI and AP may be updated every time the encoding process of one frame or several blocks is completed.

As described above, according to this embodiment, the state of the generated information amount after the encoding process is compared and determined from the assigned information amount assigned to the frame, and the generation from the determined n frames before is performed. Since the quantization parameter can be adaptively controlled according to the change in the information amount state, the generated information amount after the encoding process can be brought close to the assigned information amount. This corresponds to controlling so that the average generated information amount after the encoding process becomes the information amount per unit time obtained from the transmission speed of the data communication line or the transfer speed of the data recording medium.

In this embodiment, the information amount control is performed on a frame-by-frame basis, but a unit of several blocks of coded data may be used as the information amount control unit.
Information amount control may be performed in field units in consideration of interlaced images. Further, the quantization parameter control in this embodiment is performed by adaptively selecting the characteristics shown in FIGS. 10 and 11, but it is determined in advance as used in the embodiment shown in FIG. The quantization parameter can be set to the obtained value, and conversely, the quantization parameter control in the embodiment shown in FIG. 8 can be used in the embodiment shown in FIG.

FIG. 13 is a block diagram showing still another embodiment of the moving picture coding apparatus according to the present invention to which the information amount control system is applied. Reference numeral 31 is a quantization parameter correction circuit, which corresponds to FIG. The same reference numerals are given to the parts, and the duplicated description will be omitted.

In the embodiment shown in FIG. 1, since the quantization parameter control is selected only by the buffer memory storage amount, even if the allocation information amount of encoded data differs for each frame, the quantum parameter control is performed regardless of this. Parameterization is decided. In the embodiment described here, even if the allocation information amount of encoded data differs for each frame, this can be dealt with.

In FIG. 13, the quantization parameter control circuit 17 has a buffer memory 8 (FIG. 1) at the time when the encoding process of one frame is completed, as described in FIG.
The quantization parameter is adaptively determined from the data storage amount (buffer memory storage amount) stored in 3), and the information is further supplied to the quantization parameter correction circuit 31. In the quantization parameter correction circuit 31, when the encoding process of one frame is completed, the quantization parameter output from the quantization parameter control circuit 17 is the same as that of the encoded data output from the data multiplexing circuit 7. It is corrected and further optimized by the amount of generated information, and the corrected information is supplied to the quantization circuit 5 and the data multiplexing circuit 7.

FIG. 14 is a block diagram showing a specific example of the quantization parameter correction circuit 31 together with the quantization parameter control circuit 17, wherein 32 is an input terminal, 33 is an information amount adding circuit, and 34 is a generated information amount state comparison. A determination circuit, 35 is an allocation information amount setting circuit, 36 is a state determination threshold setting circuit, 37 is a delay circuit, 38 is a parameter correction circuit, and the portions corresponding to those in FIG. To do.

As shown in FIG. 2, the quantization parameter control circuit 17 shown in FIG. 6 stores the amount of data accumulated in the buffer memory 8 (FIG. 13) at the time when the encoding process for one frame is completed (buffer memory). Buffer memory accumulated state data generated from the accumulated amount) and n frames before or n
The quantization parameter for the encoding process of the next frame is determined from the past buffer memory accumulated state data up to the frame before, and this quantization parameter is supplied to the quantization parameter correction circuit 31.

In the quantization parameter correction circuit 31, the encoded data output from the data multiplexing circuit 7 (FIG. 13) is input from the input terminal 32, similarly to the quantization parameter control circuit 17 'described in FIG. The information amount adding circuit 33 accumulates the information amount and detects the generated information amount of the encoded data after the encoding process, that is, for one frame. The generated information amount is compared in the generated information amount state comparison / determination circuit 34 by the allocation information amount for each frame set in the allocation information amount setting circuit 35 and the threshold value set in the determination threshold setting circuit 36. Judgment processing is performed, the state representing the generated information amount of the encoded data from the data multiplexing circuit 7 is determined, and the generated information amount state data representing the determination result is output. This generated information amount status data is
In addition to being supplied to the parameter correction circuit 38, it is delayed by the delay circuit 30 having a delay amount of n (where n is an integer of 1 or more) frames, and the past generated information amount state data up to n frames before or up to n frames before. Is supplied to the parameter correction circuit 38.

The parameter correction circuit 38 corrects the quantization parameter from the quantization parameter control circuit 17 according to the generated information amount state data (for example, simply by adding, subtracting, or scaling an offset), The quantization circuit 5 and the data multiplexing circuit 7 are used as more optimal quantization parameters for the encoding process of the next frame.
(FIG. 13).

As described above, according to this embodiment, the quantization parameter can be adaptively set from the buffer memory accumulation state after the encoding process and before the n frames or up to the n frames before. Similar to the example shown
It is possible to control the average generated information amount of encoded data after encoding processing to the information amount per unit time, and further compare and determine the state of the generated information amount of encoded data from the assigned information amount, and set the quantization parameter. Can be corrected to a more optimal value according to the determined state of generated information, so that the generated information amount of the encoded data after the encoding process can be made closer to the allocated information amount, as in the embodiment shown in FIG. Therefore, the average generated information amount of the encoded data after the encoding process can be controlled to the information amount per unit time. By correcting the quantization parameter set in this way, it is possible to control the average generated information amount of the encoded data after the encoding process to be a more information amount per unit time.

In the embodiment shown in FIG. 13, first,
The quantization parameter is set from the data storage amount state of the buffer memory 8, and then the information amount control is performed by correcting the generated information amount state of the encoded data to a more optimal quantization parameter. On the contrary, the quantization parameter may be set based on the generated information amount state of the encoded data, and thereafter, the quantization parameter may be corrected based on the data storage amount state of the buffer memory 8 so as to be more optimal.

Although the information amount control is performed in frame units, the information amount control may be performed in units of several coded blocks, or in consideration of interlaced images, information amount control is performed in field units. May be performed.

Further, in the embodiment shown in FIG. 13, the same quantization parameter control circuit 17 as that shown in FIG. 2 is used for controlling the quantization parameter, but the same quantization parameter as shown in FIG. 9 is used. The conversion parameter control circuit 17 'may be used.

[0057]

As described above, according to the present invention,
The amount of data accumulated in the buffer memory after encoding processing and the amount of generated information are divided into states, and the change in the information is fed back to set and correct the next quantization parameter. It is possible to control the amount of information per unit time, which is determined by the transmission speed of the communication line or the transfer speed of the data recording medium, and furthermore, it is possible to prevent overflow or underflow of the buffer memory.
Further, since the optimum quantization parameter is set from the feedback of the accumulated state information of the buffer memory and the change of the generated information amount state to control the information amount, it is possible to suppress the image quality deterioration of the reproduced image to a necessary minimum. it can.

[Brief description of drawings]

FIG. 1 is a block diagram showing an embodiment of a moving picture coding apparatus according to the present invention.

FIG. 2 is a block diagram showing a specific example of a quantization parameter control circuit in FIG.

FIG. 3 is a diagram showing a storage state determination process of a storage state comparison / determination circuit in FIG.

FIG. 4 is a diagram showing an operation of the quantization parameter control selection circuit in FIG.

5 is a diagram showing an example of quantization parameter control selected by a quantization parameter control selection circuit in FIG.

FIG. 6 is a diagram showing characteristics of an example of a quantization parameter and an offset.

FIG. 7 is a diagram showing changes in the amount of data stored in the buffer memory in the embodiment shown in FIG.

FIG. 8 is a block diagram showing another embodiment of the moving picture coding apparatus according to the present invention.

9 is a block diagram showing a specific example of a quantization parameter control circuit in FIG.

10 is a diagram showing an example of characteristics of quantization parameter control in the embodiment shown in FIG.

FIG. 11 is a diagram showing another example of characteristics of quantization parameter control in the embodiment shown in FIG.

FIG. 12 is a diagram showing an example of a frame structure in the embodiment shown in FIG.

FIG. 13 is a block diagram showing still another embodiment of the moving picture encoding device according to the present invention.

14 is a block diagram showing a specific example of a quantization parameter control circuit and a quantization parameter correction circuit in FIG.

[Fig. 15] Fig. 15 is a diagram illustrating mathematical expressions representing a process in the moving image encoding device.

[Explanation of symbols]

4 Orthogonal transformation circuit 5 Quantization circuit 6 Variable length coding circuit 7 Data multiplex circuit 8 buffer memory 17, 17 'Quantization parameter control circuit 18 Accumulation state comparison judgment circuit 19 Status judgment threshold setting circuit 20 delay circuit 21 Quantization parameter control selection circuit 221-22N Parameter control circuit 26 Information amount addition circuit 27 Generated information amount state determination circuit 28 Allocation information amount setting circuit 30 delay circuits 31 Quantization parameter correction circuit 38 parameter correction circuit

Claims (4)

[Claims] 1. A transform coefficient obtained by orthogonally transforming original image data of a moving image is quantized and variable length coded to obtain coded data, and the coded data is temporarily stored in a buffer memory and a predetermined speed is obtained. In the moving picture coding method which is read out in step S1, the storage state determination means for determining the storage state of the buffer memory from the storage amount of the encoded data in the buffer memory, and the determination output of the storage state determination means are held. Then, the state storage means for outputting as the determination output of the past accumulation state of the buffer memory, the quantization output of the accumulation state determination means, the past determination output from the state storage means, and the accumulation amount of the buffer memory are quantized. A quantization parameter control means including a quantization parameter control determination means for determining a parameter is provided, and the quantization step width at the time of the quantization is set to the quantization parameter. The moving picture coding method is characterized by controlling the amount of generated information. 2. The information amount calculation means for calculating the generated information amount of a unit period of the encoded data, and the information amount state determination means for determining the state of the generated information amount from the information amount calculation means according to claim 1. A state storage unit for holding the determination output of the information amount state determination unit and outputting it as a determination output of the state of the past generated information amount; and a quantization parameter determined by the parameter control determination unit for the information amount state determination. Means and a parameter correction circuit that corrects according to the past judgment output from the state storage means, and a quantizing parameter correcting means is provided, and the encoding processing is performed according to the corrected quantizing parameter. A moving image coding method characterized in that the amount of generated information is controlled by changing the quantization step width. 3. A transform coefficient obtained by orthogonally transforming original image data of a moving image is quantized and variable length coded to obtain coded data, and the coded data is temporarily stored in a buffer memory and a predetermined speed is obtained. In the moving picture coding method which is read out by the method, an information amount calculating means for calculating the generated information amount of the encoded data in a unit period, and an information amount state for judging the state of the generated information amount from the information amount calculating means. Determination means, state storage means for holding the determination output from the information amount state determination means and outputting it as a determination output for the state of the past generated information amount, determination output of the information amount state determination means and the state storage means Quantization parameter control means for determining a quantization parameter from the past judgment output of the above and the storage amount of the buffer memory after the encoding processing, Saishi the quantization step width is assumed in accordance with the quantization parameter of the moving picture coding method, characterized by controlling the amount of information generated. 4. The accumulation state determination means for determining the accumulation state of the buffer memory from the accumulation amount of the buffer memory, and the determination output of the accumulation state determination means for holding the past accumulation of the buffer memory according to claim 3. A state storage unit that outputs a state determination output, and the quantization parameter determined by the quantization parameter control determination unit is corrected from the determination output of the storage state determination unit and the past determination output from the state storage unit. And a parameter correction circuit for controlling the amount of generated information by changing the quantization step width of the next encoding process according to the corrected quantization parameter. Video coding method.