JP2002232894A - Data rate converter - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a data rate converter that can properly prevent leakage deterioration caused by cyclic prediction of a reproduced image even when a data rate is decreased. SOLUTION: An AC code reduction device 21 receives data of a B picture decoded by a VLD(variable length code decoder) 13 via a SW 15. The AC code reduction device 21 reduces a code quantity instructed by a code quantity controller 23, that is, a VLC(variable length coding) code of the AC coefficient so as to obtain an object code quantity per picture. On the other hand, data of an I or P picture decoded by the VLC 13 are decoded via an SW 15 and given to a quantizer 4 via a subtractor circuit 1 and a DCT device 3. The I and P picture use a reference image obtained by locally decoding an output signal of a quantizer 4 whose step width is controlled corresponding to a degree of occupancy of a buffer 25.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a data rate conversion apparatus, and more particularly to a data rate conversion apparatus that suitably prevents leak deterioration caused by cyclic prediction of a reproduced image even when the data rate is reduced.

[0002]

2. Description of the Related Art Today, digital technology is actively used in computers, broadcast media, communication media, and storage media. The most important role in these information infrastructures is MPEG (Moving Pic)
Nature Experts Group), which will be briefly described. MPEG was established in 1988 by the ISO / IEC JTC
1 / SC2 (International Organization for Standardization / International Electrotechnical Commission / Technical Committee 1 / Subcommittee 2, current SC29) This is the abbreviation of the name of the organization that considers the moving picture coding standard.

[0003] MPEG has MPEG1, MPEG2 and other standards. MPEG1 (MPEG phase 1)
Is a standard for storage media of about 1.5 Mbps. It uses JPEG for the purpose of still image coding and video compression for low transfer rates of video conferencing and video telephony of Integrated Services Digital Network (ISDN). The target H. 26
1 (CCITT SGXV, current ITU-T SG1
5 standardized) and introduced a new technology for storage media. These are 199
It was enacted as ISO / IEC 11172 in August 3rd. MPEG2 (MPEG Phase 2) is an ISO / I / O in November 1994, aimed at a general-purpose standard so as to support various applications such as communication and broadcasting.
EC 13818, H.E. 262.

[0004] The encoding part of MPEG is created by combining several techniques. FIG. 6 is a block diagram showing an example of an image compression / encoding device using MPEG. In the figure, an input image is decoded by a motion compensation predictor 2, and a difference between the motion compensated prediction image and the input image is obtained by a subtraction circuit 1 to reduce a time redundant portion. There are three modes of prediction from the past, the future, and both. These are 16
It can be switched and used for each MB (macroblock) of pixels × 16 pixels. The prediction direction is determined by the picture type given to the input image.

[0005] Picture types include P pictures, B pictures and I pictures. There are two modes in which prediction from the past and two modes in which the MB is independently encoded without performing prediction are P
It is a picture. B-pictures have four modes for prediction from the future, prediction from the past, prediction from both, and independent encoding. It is an I picture that all MBs are independently encoded.

[0006] Motion Compensation (MC)
Predicts a motion vector with half-pel accuracy by performing pattern matching on a motion area for each MB, and shifts the motion vector by an amount corresponding to the motion for prediction. The motion vector has a horizontal direction and a vertical direction, and is transmitted as additional information of the MB together with the MC mode indicating where to predict the motion vector. GOP (Group Of) from the I picture to the picture before the next I picture
Picture), when used in storage media or the like, generally about 15 pictures are used.

The difference image signal extracted from the subtraction circuit 1 is subjected to orthogonal transformation in the DCT unit 3. Discrete Cosine Transform (DCT) is an orthogonal transform that discretely transforms an integral transform using a cosine function as an integral kernel into a finite space. In MPEG, two-dimensional DCT is performed on an 8 × 8 DCT block obtained by dividing an MB into four. In general, since a video signal has many low-frequency components and few high-frequency components, when DCT is performed, coefficients concentrate on low frequencies.

[0008] DCT-processed image data (DCT coefficients)
Are quantized by the quantizer 4. In this quantization, a value obtained by multiplying an 8 × 8 two-dimensional frequency called a quantization matrix by a visual characteristic by a value called a quantization scale for multiplying the whole by a scalar is used as a quantization value.
Divide the T coefficient by its quantized value. When inverse quantization is performed by the decoder, the original DCT is obtained by multiplying by the quantization value.
You will get a value that is close to the coefficient.

The quantized data is variable-length coded by a VLC unit 9. A direct current (DC) component of the quantized value is a DPCM (Differential Pull
se Code Modulation). In addition, Huffman coding in which an alternating current (AC) component performs a zigzag scan from a low frequency band to a high frequency band, assigns a run length of zero and an effective coefficient value to one event, and assigns a code having a short code length to a code having a high probability of occurrence. Is performed. The variable-length coded data is stored in a temporary buffer 10 and output as coded data at a predetermined transfer rate.

The generated code amount of the output data for each macroblock is supplied to a code amount controller 11.
The code amount is controlled by feeding back the error code amount between the target code amount and the generated code amount to the quantizer 4 and adjusting the quantization scale. The quantized image data is inversely quantized by an inverse quantizer 5, inversely DCTed by an inverse DCT unit 6, and temporarily stored in an image memory 8 through an adder 7, and then, by a motion compensation predictor 2. , Is used as a reference decoded image for calculating a difference image. The output signal of the motion compensation predictor 2 is input to the subtraction circuit 1 and the adder 7.

An encoded bit stream output from the buffer 10 has a variable length code amount for each picture in the case of video. This is because MPEG uses DCT, quantization,
At the same time as using the information conversion called Huffman coding, it is necessary to adaptively change the code amount allocated to each picture in order to improve image quality. This is because since the motion compensation prediction is performed, the entropy of the coded image itself greatly changes, for example, in some cases, the input image is coded as it is, and in some cases, the difference image, which is the difference between the predicted images, is coded.

In this case, in many cases, the code amount is controlled while allocating to the entropy ratio of the image and keeping the buffer limit. The limitation of this buffer is that encoding is performed so that the buffer on the decoding device side does not cause overflow or underflow. VBV (Video)
Buffering Verifier). Details on this are provided by the International Organization for Standardization (ISO).
11172-2 and ISO13818-2. If this rule is adhered to, the rate in the VBV buffer changes locally, but if the observation time is lengthened, the transfer rate becomes fixed, and MPEG defines this as the fixed transfer rate.

FIG. 7 is a block diagram showing an example of an apparatus for decoding encoded data which has been compression-encoded by MPEG. In the figure, encoded data compressed and encoded by MPEG is input to a VLD unit 33 through a buffer 32, where it is subjected to variable length decoding and then multiplied by a quantization width by an inverse quantizer 34, After having been approximated to the original DCT coefficient, it is supplied to the inverse DCT
Is locally decoded.

The motion vector and the prediction mode extracted from the VLD unit 33 are supplied to the motion compensation predictor 38 together with the decoded data from the image memory 37, and output the motion compensated image data. Adder 3
6 is the data from the inverse DCT unit 35 and the motion compensation predictor 38
By adding the motion-compensated and predicted image data, the image data equivalent to the image data input to the encoding device is decoded, and the decoded image data is stored in the image memory 37 as decoded data.
And output to the outside.

[0015] In such an MPEG system, for example, when coded data of a high coding rate from a video source (digital broadcasting or the like) is recorded on a recording medium having a limited capacity, or even if the image quality is slightly degraded. If the user wants to further compress the program compressed data or edits and connects the compressed and encoded data, and wants to adjust the VBV buffer occupancy value on the bit stream, the encoding rate is lower than at the time of input. Rate conversion is performed.

In order to convert the compression coding rate using the conventional coding and decoding techniques, the original compression coded data is once decoded to obtain an image memory, which is expanded in the image memory, It is conceivable to re-compress the data again at a desired rate.

Some conventional data rate converters perform decoding up to DCT coefficients without decoding the original compressed and coded data, and remove high-frequency components of AC coefficients in the DCT coefficient area. There is also known an apparatus for converting AC coefficients by reducing or resetting the quantization scale to a large value (Japanese Patent Application Laid-Open No. 8-251587).

[0018]

However, the conventional data rate conversion apparatus for performing the above-mentioned recompression requires a processing capability capable of simultaneously performing decoding and encoding, and expands the data once decoded into an image memory. Must have
Extra memory needs to be added.

In the conventional data rate conversion device described in Japanese Patent Application Laid-Open No. Hei 8-251587, if the AC coefficient is reduced or the Q (quantization) scale is changed in an I picture or a P picture, the original data rate conversion is originally performed. When a coded picture is created, the image quality of the past I picture or P picture used as a reference picture for motion compensation has been changed, and an error occurs in the prediction residual component.
The errors are accumulated in the course of predicting cyclically in the forward prediction from MPEG-specific I-pictures to several P-pictures, and the accumulation of the errors may cause remarkable image quality degradation in a reproduced image. is there.

The present invention has been made in view of the above points,
It is an object of the present invention to provide a data rate conversion device capable of suitably preventing leak deterioration caused by cyclic prediction of a reproduced image even when the data rate is reduced.

It is another object of the present invention to provide a data rate conversion device capable of suitably converting a data rate without adding a memory.

[0022]

According to the present invention, in order to achieve the above object, variable length decoding means for decoding a variable length code of MPEG which is input coded data, and each picture of the input coded data comprises: A picture type detecting means for detecting which picture type is coded, and an input signal is output to a first terminal if the picture type detected by the picture type detecting means is an I picture or a P picture, In the case of a first switch means for performing a selection operation of outputting an input signal to the second terminal,
If the picture type detected by the picture type detection means is an I picture or a P picture, the input signal of the first terminal is selected and output, and if the picture type is a B picture, the second signal is output.
A second switch means for selecting and outputting an input signal from a terminal of the terminal, and temporarily storing a signal taken out from the second switch means and outputting the same as encoded data obtained by rate-converting the input encoded data. And a B picture selected by the first switch means and subjected to variable length decoding by the variable length decoding means is input as an input signal, and re-encoded to reduce the AC code of the input signal. Re-encoding means for inputting to a second terminal of the switch means;
An I-picture or a P-picture selected by the first switch and subjected to variable-length decoding by the variable-length decoding unit is input as an input signal, and the input signal is subjected to an MPEG-specified decoding operation to obtain decoded data. Decryption means;
An encoding unit including a quantizer, encoding the decoded data output from the decoding unit according to the MPEG standard to obtain encoded data, and inputting the encoded data to a first terminal of a second switch unit; According to the buffer sufficiency, the amount of AC code reduction of the input signal is controlled by the re-encoding means at the time of B-picture detection so that the target code amount per picture is obtained. Code amount control means for controlling the quantization width of the quantizer of the quantization means.

According to the present invention, the AC code is directly reduced by the re-encoding means for the B picture of the input encoded data, and the I and P pictures are decoded and then re-encoded. Therefore, even if the AC coefficient is reduced or the quantization scale is changed in the I picture or the P picture, no error occurs in the prediction residual component. For an I picture or a P picture, MP
Since the decoded data is obtained by performing the decoding operation specified by the EG, and then the decoded data is subjected to the encoding specified by the MPEG to obtain the encoded data, the decoding and the encoding can be performed simultaneously. No capability is required, and no memory is needed to store the data once decrypted.

Here, the re-encoding means applies the AC coefficient to the B picture selected by the first switch means and variable-length decoded by the variable-length decoding means while keeping the quantization scale as it is. Of the VLC code according to an approximate ratio with the target code amount, or BLC selected by the first switch means and variable-length decoded by the variable-length decoding means. By performing requantization on the picture using a quantization scale changed to a value equal to or greater than the quantization scale used when quantizing the input coded data of the variable length decoding means, A
It is characterized by comprising a re-quantizer for reducing the C code.

[0025]

Next, embodiments of the present invention will be described with reference to the drawings. FIG. 1 shows a block diagram of a first embodiment of a data rate conversion device according to the present invention. 6, the same components as those in FIG. 6 are denoted by the same reference numerals.
The first embodiment shown in FIG. 1 is characterized in that the code amount conversion of the B picture is performed by the AC code reducer 21.

The input coded data is first input to the VLD unit 13. The VLD unit 13 is a variable length code decoder (Vari
It is a part that decodes a variable-length code that has been compression-encoded by the MPEG method. The VLD device 13 decodes the input coded data, that is, the variable length code, and supplies the picture type to the picture type detector 14.

The picture type detector 14 detects which picture type of each picture of the input coded data is coded, and supplies the detection signal to the switch circuits (SW) 15 and 24, respectively. For switching control. That is, if the picture type is I
In the case of a picture or a P picture, SW15 and SW24
Is switched to IP in the figure, and is switched to B in the figure for a B picture.

First, a B picture is detected, and SW15
The operation when the switches 24 and 24 are switched to the B side will be described. In this case, the data of the B picture decoded by the VLD unit 13 is input to the AC code reduction unit 21 via the SW 15. In the AC code reducer 21, the code amount controller 23
, The AC code is reduced so as to be the target code amount per picture obtained by an algorithm described later.

The AC code code of the macro block of each picture is coded by the Huffman code (VLC code) specified in MPEG shown in Table 1. In Table 1,
The code length indicates the code length of the Huffman code (VLC code) of the AC coefficient.

[0030]

[Table 1] Some macroblocks specified by MPEG include 8 × 8
There are a total of six pixel blocks, four for luminance signals and one for each of two types of color difference signals. An AC code exists in each of these blocks. As shown in FIG. 3, when the DCT coefficients are arranged in a zigzag manner, the Huffman code event in these blocks is a combination of the number of 0s (0 run length) and the effective coefficients until an effective coefficient other than 0 is detected. The event is represented by

That is, assuming that these are developed, there is a difference between the case of the intra in FIG. 3 and the case of the non-intra shown in FIG. The VLD unit 13 detects the “0 run length and effective coefficient” event of the AC coefficient and the code length thereof, and
Addresses indicating the order in which the AC coefficient codes are transmitted as shown in FIGS. 3 and 4, that is, the accumulated code amount of the code length when arranged from the low-frequency signal and the coefficient position when zigzag scanning is performed. And supplies the information as VLD information to the AC code reducer 21 together with the encoded data.

The AC code reducer 21 reduces the AC coefficient code based on the encoded data from the VLD unit 13 and the VLD information, and the target code amount Tb from the code amount controller 23. VL
The D information includes an AC coefficient code and its code length. As shown in FIG. 5, by arranging from the low-frequency signal of the AC coefficient code, the approximate ratio of the original coded picture data to the target code amount (to the total code amount of 278 bits of the cumulative code amount of the code length ( 1-The value multiplied by the deletion rate Rc) is regarded as valid, and the subsequent symbols are deleted. An EOB (END OF BLOCK) code is transmitted as the AC code after the deletion.

Therefore, for example, when Rc = 0.5, the cumulative code length of the original coded picture data is a value obtained by multiplying 278 bits, which is the total code amount, by 0.5 (= 1-0.5). The VLC code of the AC coefficient until it reaches 139 bits is left as it is, and the VLC code after that is deleted.
The code is deleted (value is set to 0). The numerical values in FIG. 5 indicate the values of the AC coefficients.

In this way, the coded data processed for all blocks in the AC code reducer 21 is switched to the SW side when the SW 24 is switched to the B side.
The data is supplied to the buffer 25 for one picture through the buffer 24 and stored.

The buffer 25 supplies the generated code amount to the code amount counter 22. Since the code amount counter 22 counts the generated code amount for each picture, the code amount controller 2
The rate of deletion controlled by 3 may be controlled by feeding back within the picture so as to reach the target code amount when processing the macroblock in one picture.

Next, an operation when an I picture or a P picture is detected by the picture type detector 14 and the SWs 15 and 24 are switched to the IP side will be described. In this case, the data of the I picture or the P picture decoded by the VLD unit 13 is supplied to the inverse quantizer 17 via the SW 15 and is subjected to inverse quantization. The inversely quantized data is subjected to inverse DCT by an inverse DCT unit 18. The difference image data obtained by the inverse DCT is added to the previous motion-compensated image data from the motion compensation predictor 16 by the adder 19 and supplied to the subtraction circuit 1.

The image data added by the adder 19 at the same time is supplied to and stored in the image memory 20, and is again used as a reference image (reference image) for the next motion compensation. Is input to The motion compensation predictor 16 motion-compensates the image data stored in the image memory 20 based on the VLD-based motion vector for each macroblock, and adds the result to the next inverse DCT differential image data to the adder 19. To calculate the image to be added.

In the subtraction circuit 1, in the case of an I picture, the input image data is directly subjected to DCT without performing a subtraction operation.
To the vessel 3. In the case of a P picture, the VLD unit 13
The motion compensation prediction is performed by the motion compensation predictor 2 on the basis of the motion vector from, and the difference from the predicted image is calculated by the subtraction circuit 1. The difference image data is subjected to DCT in the DCT unit 3 to obtain DCT coefficients.

The DCT coefficients are quantized by the quantizer 4, variable-length coded (VLC) by the VLC unit 9 together with the motion vector and the coding mode, supplied to the buffer 25 through the SW 24, and temporarily stored therein. It is output as an MPEG video stream that has been rate-converted.

At this time, the code amount controller 23
5 is monitored from the value of the code amount counter 22. Basically, the quantization is made coarser when the degree of sufficiency of the buffer 25 is larger, and finer when the degree of sufficiency is smaller. The quantization width of the unit 4 is controlled.

Since the I picture and the P picture need to be used later as a reference screen for motion compensation prediction,
The information quantized by the quantizer 4 is supplied to an inverse quantizer 5 and an inverse DC
The inverse quantization and the inverse DC are performed by the T unit 6 and the motion compensation predictor 2.
T, motion compensation is performed, local decoding is performed, and the same image as the decoder is restored and stored in the image memory 8. This image is used as a reference screen for the next motion compensation prediction.

As described above, in the present embodiment, the I-picture and the P-picture are locally decoded from the output signal of the quantizer 4 whose step width is controlled in accordance with the sufficiency of the buffer 25 (accumulated code amount). On the other hand, the B picture directly removes the AC code, so that even if the AC coefficient is reduced in the I picture or P picture or the quantization scale is changed, an error occurs in the prediction residual component. It is possible to solve the problem that errors accumulate in the operation of predicting cyclically without occurrence, and the error accumulation remarkably occurs in a reproduced image.

Further, after the decoding from the dequantizer 17 to the adder 19 is performed according to the MPEG specification, the circuit from the subtraction circuit 1 to the VLC unit 9 performs the coding according to the MPEG. Therefore, there is no need for a processing capability capable of performing decoding and encoding at the same time, and there is no need for a memory for storing once-decoded data.

Next, a method of controlling the code amount will be described. The encoded data is stored in the temporary buffer 25 and output as encoded data at a predetermined transfer rate. The generated code amount for each macro block of the output data is supplied to the code amount controller 23 through the code amount counter 22, and the difference between the generated code amount and the target code amount is fed back to the quantizer 4. The code amount is controlled. For example, if the short section is 1 GOP (Group of Picture), the code can be controlled by maintaining the image quality to some extent by the following method.

(A) Step S1 First, assuming that the target code amount after rate conversion of each GOP is R, in step S1, the code amount allocated to each picture of the GOP is calculated for a picture that has not been coded in the GOP yet. Allocate with a certain weight. Xi = Si × Qi Xp = Sp × Qp Xb = Sb × Qb Here, X is called a global complexity measure, and S (S) of the previous encoding result of the same picture type Generated code amount) and Q
(Average quantization scale) is a parameter indicating the complexity of the screen.

The quantization scale for achieving the ideal image quality is such that the ratio with the P picture is Kp = 1.0 and the ratio with the B picture is Kb = 1.
Assume 4. Here, the code amounts Ti, Tp allocated to the I picture, P picture and B picture in the GOP.
And Tb are represented by the following equations. Ti = MAX {R / (1+ (NpXp / XiKp) + NbXb / XiKb)}, bit_rate /
(8 * picture_rate)} Tp = MAX {R / (Np + (NbKpXb / KbXp)), bit_rate / (8 * picture
_rate)} Tb = MAX {R / (Nb + (NpKbXp / KbXb)), bit_rate / (8 * picture
_rate)} where Np and Nb are the numbers of uncoded pictures of P and B in the GOP. The initial value of R is the code amount given to the GOP. Bit_rate is a target rate.

Each time a picture is coded based on the code amount thus obtained, the target code amount R to be allocated to the uncoded picture in the GOP is as follows. Will be updated to R = R-Si, p, b

(B) Step S2 In step S2, in order to make the code amount (Ti, Tp, Tb) of each picture allocated in step S1 equal to the actual generated code amount, the generated code amount is set for each macroblock (MB). And the difference between the target code amount and the predicted target code amount in the middle is fed back to the quantization scale in MB units. Here, prior to encoding of the j-th MB, I,
The occupation amounts dji, djp, and djb of each virtual buffer at the j-th position from the beginning of each of the P and B pictures are obtained by the following equations.

Dji = d0i + B _j−1 − (Ti · (j−1) / MB_cnt) djp = d0p + B _j−1 − (Tp · (j−1) / MB_cnt) djb = d0b + B _j−1 − (Tb · ( j-1) / MB_cnt) where d0i, d0p, and d0b are I, P, B
The initial occupancy of each virtual buffer of the picture, B _j, is j from the head of each picture counted by the code amount counter 22.
The generated code amount up to the MB, MB_cnt, is the number of MBs in one picture.

The quantization scale Q of the I and P pictures is obtained by the following equation. Q = dj × 31 / r (1 ≦ Q ≦ 31) r = 2 × bit_rate / picture_rate where r is a parameter that determines the response speed of feedback. In this way, it is possible to control the code amount.

In the case of a B picture, since the code amount control is performed by reducing the VLC code of the AC coefficient, the quantization scale is fixed, and
Although the code amount is controlled by deleting the C coefficient, the code amount of the B picture can be controlled by the following equation.

Rc = dj × 0.9 / r (0.1 ≦ Q · Rc ≦
0.9) r = 2 × bit_rate / picture_rate where Rc is a reduction rate. In this way, it is possible to control the code amount. Also, here
Although Rc is shown to be fed back by the dj buffer, Rc may be fixed and AC coefficient reduction may be performed.

Next, a second embodiment of the present invention will be described. FIG. 2 shows a block diagram of a second embodiment of the bit rate conversion device according to the present invention. In the figure, the same components as those in FIG. 1 are denoted by the same reference numerals. In FIG. 2, input coded data is first input to a VLD unit 13, where a variable length code is decoded, and a picture type is supplied to a picture type detector 14, and each of the input coded data is It is detected in which picture type the picture is encoded.

As in the first embodiment, when the picture type is an I picture or a P picture, the SWs 15 and 29 are switched to the IP side. In the case of a B picture, it is switched to the B side.

First, a description will be given of a case where the SWs 15 and 29 are switched to the B picture. The B picture data decoded by the VLD unit 13 is input to the requantizer 28 via the SW 15. The requantizer 28 is a code amount controller 2
The quantization scale is controlled so that the code amount indicated by 3, that is, the target code amount per picture obtained by an algorithm described later. In this case, the quantization is performed with a value equal to or larger than the quantization scale for each macroblock of the target B picture described in the original encoded data.

That is, based on the code amount counter value from the buffer 30 by the code amount counter 22, a value smaller than the code amount of the entire original picture before rate conversion is set as the target code amount. The quantization scale should be controlled in the larger direction, but locally the buffer 30
May be smaller than the original quantization scale depending on the feedback control state of. In this case, since the quality of the image is not improved from the original image quality, the quantization scale controlled when requantization is performed by the requantizer 28 is performed by applying a limiter to the original quantization scale. Then, the quantization is changed to a value larger than the original quantization scale, and thereby the quantization scale takes a uselessly small value, thereby preventing an increase in the code amount that does not lead to an improvement in image quality.

In the requantizer 28, B from the VLD unit 13
Victure coded data and VLD information, code amount controller 2
The quantization scale is controlled as described above based on the target code amount Tb from No. 3. The method of controlling the code amount may use the same algorithm as that for the I picture and the P picture described below.

In this way, the coded data obtained by requantizing all blocks by the requantizer 28 is transferred to the buffer 30 for one picture via the SW 29 connected to the B side. Supplied and accumulated. Buffer 3
0 supplies the generated code amount to the code amount counter 22. As described above, since the code amount counter 22 counts the generated code amount for each picture, the rate of deletion controlled by the code amount controller 23 is determined by processing a macroblock in one picture. May be controlled by feeding back in the picture so that the target code amount is obtained.

Next, the operation when the picture type detector 14 detects the picture type as an I picture or a P picture will be described. The encoded data of the I picture or P picture decoded by the VLD unit 13 is supplied to the inverse quantizer 17 through the SW 15 connected to the IP side, where the inverse quantization is performed. The inversely quantized data is subjected to inverse DCT by an inverse DCT unit 18 and then supplied to an adder 19, where the motion compensated predictor 16
Is added to the previous motion-compensated image data, and supplied to the subtraction circuit 1.

At the same time, the image data from the adder 19 is supplied to and stored in the image memory 20, and is again input to the motion compensation predictor 16 as a reference image for the next motion compensation. The motion compensation predictor 16 motion-compensates the image stored in the image memory 20 based on the motion vector for each macroblock subjected to VLD, and performs the next inverse DC.
An image to be added by the adder 19 to the T difference image data is calculated.

In the case of an I picture, the subtraction circuit 1 transmits the input data to the DCT unit 3 without performing a subtraction operation. In the case of a P picture, motion compensation prediction is performed by the motion compensation predictor 2 based on the motion vector from the VLD unit 13, and the difference from the prediction image from the motion compensation predictor 2 is calculated by the subtraction circuit 1. You.

The difference image data from the subtraction circuit 1 is DCT
DCT is performed in the unit 3. The DCT coefficients are quantized by the quantizer 4, variable-length coded (VLC) by the VLC unit 9 together with the motion vector and the coding mode, and then supplied to the buffer 30 via the SW 29 connected to the IP side. After being stored, it is output as an MPEG video stream.

The code amount controller 23 monitors the sufficiency (accumulated code amount) of the buffer 30 through the code amount counter 22. Basically, when the sufficiency of the buffer 30 increases, the quantization is coarse, and when the sufficiency decreases, the quantization is reduced. The quantization width of the quantizer 4 is controlled so as to make.

Since the I picture and the P picture need to be used later as a reference screen for motion compensation prediction, the quantized information is supplied to the inverse quantizer 5, the inverse DCT unit 6 and the motion compensation predictor 2. , Inverse quantization, inverse DCT, and motion compensation prediction, local decoding is performed, and the same image as the decoder is restored and stored in the image memory 8. This image is used as a reference screen for the next motion compensation prediction.

Next, a method of controlling the code amount common to each of the I picture, P picture, and B picture will be described. The encoded data is stored in the temporary buffer 30 and output as encoded data at a predetermined transfer rate. The generated code amount for each macro block of the output data is output to the code amount controller 23, and the difference between the generated code amount and the target code amount is fed back to the quantizer 4 to control the code amount. . For example, a short section is 1 GO
If P is set, the code can be controlled by the following method while maintaining the image quality to some extent.

That is, first, assuming that the target code amount after rate conversion of each GOP is R, the code amount allocated to each picture of the GOP is determined in step S1 by the above-described first code amount.
Similar to step S1 in the embodiment, a certain weight is assigned to a picture which has not been coded in the GOP and distributed. This distribution is repeated in the order of the coded pictures in the GOP, and the same code amount allocation operation as the operation in step S1 of the first embodiment described above is performed. Also, R is updated as follows each time encoding proceeds in a GOP. R = R-Si, p, b

Next, in step S2, in order to make the allocated code amounts Ti, Tp and Tb for each picture obtained in step S1 coincide with the actual generated code amounts, first, the coding of the j-th macroblock is performed. First, the occupancy amounts dji, djp, and djb of the virtual buffer are obtained by the same expressions as the expressions dji, djp, and djb described in the first embodiment, and then the quantization scale Q for the j-th macroblock is calculated as follows. It is determined by the formula.

Q = dj × 31 / r (prevQ ≦ Q ≦ 31) r = 2 × bit_rate / picture_rate where r is a parameter that determines a feedback response speed, and prevQ is a macroblock before rate conversion. Is the quantization scale used in. In this way, it is possible to control the code amount.

According to the present embodiment, I-picture data and P-picture data are locally decoded to be re-encoded as reference data for the next predicted picture, and B-pictures are re-quantized by requantizer 28. Since direct re-encoding is performed, even if the AC coefficient is reduced or the quantization scale is changed in an I picture or a P picture, prediction is performed cyclically without any error occurring in the prediction residual component. During operation, it is possible to prevent errors from accumulating and significantly occurring in a reproduced image.

The present invention is not limited to the above embodiment. For example, in FIGS. 1 and 2, instead of supplying the output signal of the quantizer 4 to the inverse quantizer 5, An inverse VLC device is provided on the output side of the VLC device 9 and the inverse VLC
The signal from the LC unit may be supplied to the inverse quantizer 5. Therefore, in this case, the data after VLC for rate conversion is locally decoded.

[0071]

As described above, according to the present invention,
For I picture data and P picture data, MP picture
After decoding according to the EG rules, the decoded data is quantized for rate conversion, or the data after VLC is locally decoded to perform re-encoding as reference data of the next predicted image. Since direct re-encoding is performed, even if the AC coefficient is reduced or the quantization scale is changed in an I picture or a P picture, prediction is performed cyclically without any error occurring in the prediction residual component. In some operations, it is possible to solve the problem that errors are accumulated and the errors are significantly accumulated in a reproduced image, and the quality of the reproduced image after rate conversion can be improved.

Further, according to the present invention, unlike the conventional recompression method, there is no need for a processing capability capable of simultaneous decoding and encoding, so that an extra memory for storing data once decoded is used. No additional work is required, and costs can be reduced.

[Brief description of the drawings]

FIG. 1 is a block diagram of a rate conversion device according to a first embodiment of the present invention.

FIG. 2 is a block diagram of a rate conversion device according to a second embodiment of the present invention.

FIG. 3 is an explanatory diagram illustrating a method of scanning an intra MB of a DCT coefficient.

FIG. 4 is an explanatory diagram illustrating a method for scanning a non-intra MB with DCT coefficients.

FIG. 5 is an explanatory diagram showing an example of generation of DCT coefficients.

FIG. 6 is a block diagram illustrating an example of an MPEG encoder.

FIG. 7 is a block diagram illustrating an example of an MPEG decoder.

[Explanation of symbols]

 Reference Signs List 1 subtraction circuit 2, 16, 38 motion compensation predictor 3 DCT unit 4 quantizer 5, 17, 34 inverse quantizer 6, 18, 35 inverse DCT unit 7, 19, 36 adder 8, 20, 37 image memory 9 VLC device 10, 25, 30, 32 Buffer 11, 23 Code amount controller 13, 33 VLD device 14 Picture type detector 15, 24, 29 Switch circuit (SW) 21 AC code reducer 22 Code amount counter 28 Requantity Chemist

 ────────────────────────────────────────────────── ─── Continued on the front page (72) Inventor Kazuhiko Morita 3-12 Moriyacho, Kanagawa-ku, Yokohama-shi, Kanagawa Japan F Co., Ltd. F-term (reference) 5C059 KK01 KK34 KK35 MA00 MA23 MC11 MC38 ME02 NN15 PP05 PP06 PP07 PP16 SS01 SS06 SS11 TA46 TA49 TA60 TA71 TB04 TC10 TC18 TC24 TC38 TD06 UA02 UA05 UA33 5J064 AA01 AA04 BA09 BB05 BC01 BC05 BC16 BD02 BD03

Claims (3)

[Claims] 1. A variable length decoding means for decoding an MPEG variable length code which is input coded data, and a picture type for detecting which picture type each picture of the input coded data is coded Detecting means for outputting an input signal to a first terminal if the picture type detected by the picture type detecting means is an I picture or a P picture, and outputting an input signal to a second terminal if the picture type is a B picture A first switch for performing a selection operation; and selecting and outputting an input signal of a first terminal when the picture type detected by the picture type detection means is an I picture or a P picture; A second switch for selecting and outputting an input signal of a second terminal; and a signal extracted from the second switch. A buffer that accumulates and outputs as encoded data obtained by rate-converting the input encoded data; and a B picture selected by the first switch means and variable-length decoded by the variable-length decoding means. A re-encoding unit which is input as an input signal and reduces the AC code of the input signal, and which is input to a second terminal of the second switch unit, and selected by the first switch unit The I-picture or the P-picture, which has been variable-length decoded by the variable-length decoding means, is input as an input signal.
A decoding means for obtaining decoded data by performing a decoding operation specified by the PEG; and a quantizer, performing encoded coding specified by the MPEG on the decoded data output from the decoding means to obtain encoded data, Encoding means for inputting to the first terminal of the second switch means; and the re-encoding means at the time of detecting the B-picture, so as to have a target code amount per picture according to the sufficiency of the buffer. Controls the amount of AC code reduction of the input signal by
A data rate conversion device comprising: a code amount control unit that controls a quantization width of the quantizer of the encoding unit when the I picture or the P picture is detected. 2. The re-encoding means, for a B picture selected by the first switch means and subjected to variable-length decoding by the variable-length decoding means, retains a quantization scale and maintains a VLC of an AC coefficient. 2. The data rate conversion device according to claim 1, further comprising a reducer for deleting a part of the code according to an approximate ratio to a target code amount. 3. The re-encoding means, for a B picture selected by the first switch means and subjected to variable-length decoding by the variable-length decoding means, encodes input encoded data of the variable-length decoding means. 2. A re-quantizer for reducing an AC code by performing re-quantization on a quantization scale changed to a value equal to or larger than a quantization scale used when quantization is performed. Data conversion device as described.