JPH0241936B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to an encoding/decoding apparatus that performs unequal length encoding on a prediction error signal obtained by predictively converting an image signal such as a TV (Television) signal. Digitalization of TV signal equipment is progressing, and VTR (Video
Digital VTRs (Tape Recorders) are also being developed. Conventional analog VTRs had the disadvantage that the S/N deteriorated each time dubbing was repeated, making it impossible to dub many times.
On the other hand, digital VTR is used to convert image signals into PCM
If the data is converted into a digital signal such as a (Pulse Code Modulation) signal and recorded on a digital VTR, no deterioration will occur due to dubbing even if dubbing is repeated based on the digital signal. However, when converting analog image signals to PCM signals,
For example, for an NTSC color TV signal, the sampling frequency is
It is said that quantization accuracy of about 8 bits is required at frequencies above 10 MHz, and if a PCM signal is recorded directly onto a digital VTR, the recording speed will be high and the recording capacity will be large. Therefore, if the digital image signal is compressed and encoded and recorded on a digital VTR, the recording capacity can be reduced. For example, if the data is compressed to 4 bits per pixel, the recording capacity can be halved.

DPCM (Differential Pulse Code) is a typical method for compressing and encoding image signals.
Modulation) is well known. Normal DPCM
The method subtracts the prediction signal from the input signal to obtain a prediction error signal, quantizes the prediction error signal with a quantizer with non-uniform quantization characteristics determined using visual characteristics, and then generates the quantized prediction. This is a method of encoding and transmitting an error signal. Therefore, the decoded digital (local) signal becomes a signal containing quantization noise caused by the quantizer and does not match the digital image signal input to the DPCM encoder. Therefore, if this compression encoding is used and dubbing is repeated, in other words, the digital base
When DPCM encoding and decoding are repeated, quantization noise generated by the quantizer is superimposed on the decoded signal, and the image quality gradually deteriorates. As described above, the conventional compression encoding system using a quantizer has the drawback that the decoded and reproduced signal does not match the input signal, and the deterioration of image quality increases as the encoding and decoding process progresses.

SUMMARY OF THE INVENTION An object of the present invention is to provide an encoding/decoding device that does not increase deterioration in image quality of a re-output image even if encoding/decoding is repeated.

The encoding/decoding apparatus of the present invention compresses and encodes an image signal and decodes and reproduces the image signal. a preprocessing means for controlling; a prediction conversion means for converting the preprocessed digital image signal V into a prediction error signal using reversible logic; and a mode signal M representing the preprocessing characteristic of the prediction error signal. means for converting into an unequal length code under control; means for temporarily storing the unequal length code and synchronization and mode signals M necessary for decoding in a buffer memory and sending it out as smoothed compressed data C; and a control means for controlling the preprocessing means and smoothing the amount of generated information by monitoring the amount of generated encoded information stored in the buffer memory. Device and compressed data C
a first decoding device that decodes the digital image signal V and reproduces the mode signal M indicating the processing characteristic; second compression encoding to obtain the compressed data C again by converting the signal V into a prediction error signal, unequal length encoding the prediction error signal under the control of the mode signal M, and smoothing it in a buffer memory; a device;
and a second decoding device that reproduces an image signal V from compressed data C. The preprocessed digital image signal V output from the preprocessing means of the first compression encoding device is reversibly encoded to preserve information, so the compression encoded signal is decoded and reproduced. The signal is equal to the preprocessed digital image signal V. Therefore, when re-encoding the reproduced digital image signal V,
If predictive conversion and unequal length encoding are performed with the same encoding characteristics as the first compression encoding device without adding any preprocessing means, the same signal as the compressed data output from the first compression encoding device will be output. Therefore, by decoding this signal, the preprocessed digital image signal V can be reproduced. That is, according to the encoding/decoding apparatus of the present invention, when the digital image signal V that has been compressed, encoded, decoded, and reproduced is encoded again by the second compression/encoder, the mode signal M representing the preprocessing characteristics is transmitted. Since the image signal V is controlled to be predictively transformed and unequal-length encoded using the same encoding characteristics as the first compression encoding device, there is no deterioration in image quality even if encoding and decoding are repeated. It can be prevented from increasing.

Preprocessing methods include uniform quantization, thinning processing, filter processing, and a combination of these. When uniform quantization is used as preprocessing for the first compression encoding device, for example, the image signal V that is encoded, decoded, and reproduced by performing 8-bit or 7-bit uniform quantization preprocessing is used for the second compression encoding device. When re-encoding with the encoding device, the parts processed using the 8-bit uniform quantization characteristic using the mode signal M are used for 8 bits, and the parts processed using 7 bits are used for the 7-bit code conversion characteristic. If unequal length encoding is instructed, the same encoding as the first compression encoding device can be performed. When re-encoding an image that has been reproduced after being thinned out as preprocessing, a mode signal is used to perform unequal length encoding on only the parts that are not thinned out.

The present invention will be described in detail below with reference to the drawings.

FIG. 1 is a block diagram showing the configuration of a first embodiment of the present invention. An analog image signal input to the input terminal 101 is converted into a digital image signal, such as a PCM signal, by an A/D converter 102 and supplied to a first compression encoding device 103. The first compression encoding device 103 includes preprocessing means for controlling the amount of information according to the city control signal from the control means, and converts the preprocessed digital image signal V into a prediction error signal using reversible logic. means for predictive conversion, means for converting the prediction error signal into an unequal length code under the control of a mode signal M indicating characteristics of the preprocessing means, and a means for converting the prediction error signal into an unequal length code, The preprocessing is controlled by means for temporarily storing necessary synchronization and mode signals M in a buffer memory and sending them out as smoothed compressed data C, and by monitoring the amount of encoded information stored in the buffer memory. and a control means for smoothing control of the amount of information generated by the input image signal, and the data rate of the compressed encoded signal outputted by compressing and encoding the input image signal so as to be equal to the given data rate is kept at a constant value. In some cases, it is determined, and in other cases, it is controlled from outside and varies over time within a certain range. Compressed data C is supplied to a first decoding device 105 via a transmission line or a storage device 104 such as a digital VTR. First decoding device 1
05, the compression encoded signal is decoded, and the digital image signal V obtained by the preprocessing means of the first compression encoding device 103 and the mode signal M indicating the preprocessing characteristics are reproduced. The reproduced digital image signal V and mode signal M are sent to the second compression encoding device 1.
06 and is compressed and encoded again. The second compression encoding device 106 converts the reproduced digital image signal V into a prediction error signal by predictive conversion, and converts the prediction error signal into an unequal length code under the control of the reproduced mode signal M. The compressed data C is obtained again by smoothing with a buffer memory. Even without monitoring and controlling the amount of encoded information stored in the buffer memory, the same encoding as in the first compression encoding device 103 can be achieved by performing encoding under the control of the mode signal M. , the compressed data is equal to that output from the first compression encoding device 103. Since the predictive conversion is reversible, the reproduced digital image signal V is encoded with information preserved.

The compressed data output from the second compression encoding device 106 is supplied to the second decoding device 108 via a transmission path or a storage device 107 such as a digital VTR. In other words, the second decoding device 108 decodes the compressed data and decodes the compressed data.
The same digital image signal V that was decoded by the decoding device 105 is reproduced. The reproduced digital image signal V is supplied to the D/A converter 109, converted into an analog signal, and then outputted from the output terminal 110. That is, even if encoding and decoding are repeated, no image quality deterioration occurs due to the repetition.

Further, the compressed data output from the second compression encoding device 106 is stored in the storage device 10 of the digital VTR.
7 and then read out and supplied to the first decoding device 105, the digital image signal V decoded and reproduced from the compressed data by the first decoding device and the mode showing the preprocessing characteristics. Even if the signal M is supplied to the second compression encoding device 106 and encoded again, the same encoding can be performed using the mode signal M, so no deterioration in image quality occurs. In other words, no matter how many times encoding and decoding are repeated, image quality does not deteriorate.

Another set of a digital VTR storage device 104 and a first decoding device 105 is prepared, and an image signal V and a mode signal M to be reproduced and output from one of the first decoding devices 105 are selected. By supplying the compressed data C to the second compression encoding device 106, encoding it again, and recording the compressed data C in the storage device 107 of the digital VT, editing using the digital VTR can be performed without deteriorating the image quality.

The difference between the first compression encoding device 103 and the second compression encoding device 106 is that the first compression encoding device 10
3 has a preprocessing means, and controls the preprocessing means by monitoring the generated amount of encoded information stored in the buffer memory, and performs unequal length encoding of the prediction error signal in response to the preprocessing. However, the difference is that the compression encoding device 106 does not have a preprocessing means and performs unequal length encoding of the prediction error signal under the control of a mode signal M representing preprocessing characteristics. The difference between the first decoding device 105 and the second decoding device 108 is whether or not they output a mode signal M representing the preprocessing characteristics. Therefore, if a control mode for performing the second compression encoding is added to the first compression encoding device, the first compression encoding will be performed according to the mode switching information.
By switching between the first compression encoding mode and the second compression encoding mode, an encoding device having both functions can be constructed.

If an encoding device having a first compression encoding mode and a second compression encoding mode is used instead of the second compression encoding device 106, the data rate at which the reproduced image signal is encoded, that is, the compressed data If the data rate matches the previously encoded data rate, encoding is performed in the second compression encoding mode, and if the data rate is insufficient, encoding is performed in the first compression encoding mode. . In addition to switching the encoding mode using a mode switching signal, there is also a method of adding information indicating a data rate to the mode signal and using this information to make a switching decision.

Next, initialization of the buffer memory will be explained.

The second compression encoding device 106 performs the same encoding as the first compression encoding device 103 and outputs the same compressed data. In other words, it outputs the same amount of encoded information, but when starting encoding The initial value of the amount of information stored in the buffer memory is indefinite unless determined by some method, and does not match the initial value of the first compression encoding device 103. Therefore, in the second compression encoding device 106, unless the initial value of the amount of information stored in the buffer memory is controlled or the amount of information stored in an appropriate portion of the image signal matches, two compression codes will be generated. The amount of information stored between the converting devices may not change even if they match, resulting in overflow or underflow. In order to prevent this, there is a method to initialize the amount of information stored in the buffer memory, in which the amount of information stored in the buffer memory is reset at the beginning of encoding or at every appropriate period, and an indirect method is to reset the amount of information stored in the buffer memory. method for increasing the second compression encoding device 1
06, if the buffer memory is likely to overflow, stop encoding according to the mode signal M, control the amount of information using preprocessing means, and smooth the amount of information generated. There are methods to prevent overflow from occurring. If underflow is likely to occur, add dummy information. The amount of information stored can be reset by forcibly resetting it to a certain value, by transmitting the value of the amount of information stored, by estimating the amount of information stored using the mode signal M, or by using the encoding device. There is a method of initializing the amount of information stored in the buffer memory by limiting the amount of encoded information by limiting the reproduced image signal input to the buffer memory. First compression encoding device 1 describing an initialization method using mode signal M
In No. 03, the preprocessing characteristics are determined using the amount of information stored in the buffer memory, so that there is a certain correspondence between the amount of information stored and the mode signal M. Therefore, in the second compression encoding device 106, if the difference between the amount of information accumulated from the mode signal M and the amount of information accumulated in the buffer memory exceeds a certain range, a dummy signal is generated or the code is The amount of stored information is controlled to match by stopping the generation of information. Figures 2a and b are the first embodiment of the present invention.
The first compression encoding device 103, the first encoding device 105, and the second compression encoding device 106 of the embodiment
FIG. 3 is a block diagram showing an example of the configuration of a second decoding device. In this embodiment, a quantization circuit for uniform quantization is provided as a preprocessing means, and a mode signal M is used to initialize the buffer memory.

The image signal X converted into a digital signal by the A/D converter 102, for example, an 8-bit PCM signal having a level value of -128 to 127, is sent to the front via the input terminal 4 of the first compression encoding device 103. The signal is supplied to the quantization circuit 7 of the processing means. The quantization circuit 7 has several predetermined uniform quantization characteristics, that is, preprocessing characteristics, and converts the image signal X according to the quantization characteristics selected by the control signal from the control circuit 13. A digital image signal V which has been quantized and preprocessed is output and supplied to the subtracter 10 and the predictor 8 of the non-recursive type predictive encoder 64. The predictor 8 outputs a prediction signal according to predetermined prediction characteristics and supplies it to the subtracter 10. The subtracter 10 subtracts the prediction signal from the image signal V output from the quantization circuit 7 to obtain a prediction error signal, which is supplied to the unequal length encoding circuit 11 . The unequal length encoding circuit 11 has several types of code conversion characteristics that correspond to the quantization precision of the image signal V input to the predictive encoder 64, and therefore correspond to the quantization characteristics of preprocessing. Then, a code conversion characteristic corresponding to the quantization characteristic is selected by the control signal, and the prediction error signal is converted into an unequal length code according to the selected code conversion characteristic, and the code required for decoding is converted into an unequal length code. synchronization and preprocessing properties, and hence quantization properties,
A compressed encoded signal consisting of a mode signal M representing the mode signal M is supplied to the buffer memory 12. The amount of information in the compressed encoded signal sent to the buffer memory 12 changes from moment to moment depending on the image signal input to the first compression encoded device 103. Therefore, in the buffer memory 12, the unequal length encoding circuit 1
The compressed encoded signal sent from 1 is temporarily stored, smoothed to match the transmission speed of the transmission line, and then sent out as compressed data C from the output terminal 14 to the transmission line. The transmission path includes an actual transmission path and writing to a storage device such as a digital VTR. The buffer memory 12 monitors the amount of information stored in the buffer memory, and the amount of information stored is supplied to the control circuit 13.

The control circuit 13 uses the amount of accumulated information supplied from the buffer memory 12 to control switching of preprocessing characteristics at appropriate intervals. If the amount of information stored is small, it does not matter if the amount of generated information increases, so a decision is made to select a finer quantization characteristic, and if the amount of information stored is large, coarser quantization is selected to reduce the amount of generated information. Determine to select a characteristic. Based on the determination result, a control signal for selecting and switching a preprocessing characteristic, that is, a quantization characteristic, is sent to the quantization circuit 7, and a control signal for selecting a code conversion characteristic corresponding to the selected quantization characteristic is sent to the unequal length encoding circuit. 11
supply to The above is the first compression encoding device 103
This is an explanation of the operation.

In the first decoding device 105, the compressed data C sent from the first compression encoding device 103 via the transmission line is supplied from the input terminal 15 to the buffer memory 7 and is temporarily stored therein. The buffer memory 17 monitors the amount of information stored in the buffer memory, and the amount of information stored is supplied to the control circuit 19. The data of the compressed encoded signal once stored is read out in accordance with a request from the unequal length decoding circuit 18.
8. The unequal length decoding circuit 18 supplies control information such as synchronization and mode signal M in the compressed encoded signal to the control circuit 19, and the unequal length code information is decoded by the unequal length decoding circuit 18. . Control circuit 1
9 uses the information storage amount supplied from the buffer memory 17 and the control information supplied from the unequal length decoding circuit 18 to control readout of encoded information and select code inverse conversion characteristics corresponding to the quantization characteristics. A control signal for switching the quantization characteristic is supplied to the unequal length decoding circuit 18, a control signal for switching the quantization characteristic is supplied to the quantization circuit 21, and the quantization corresponding to the reproduced digital image signal V output from the output terminal 23 is performed. A mode signal M representing the characteristic is supplied at the output terminal 24.

The unequal length decoding circuit 18 has code inverse conversion characteristics corresponding to the code conversion characteristics of the unequal length encoding circuit 11 of the first compression encoding device 103, and is selected by a control signal from the control circuit 19. The information of the unequal-length code is inversely converted according to the code conversion characteristics, and the first
The prediction error signal supplied from the predictive encoder 64 of the compression encoding device 103 to the unequal length encoding circuit 11 is adjusted to the accuracy of quantization with the image signal input to the predictive encoder 64, so that the mode signal M is A prediction error signal equal to the quantized signal, in other words, a quantized prediction error signal e _Q with the quantization characteristic expressed,
Play. The reproduced prediction error signal e _Q is supplied to the adder 20 of the prediction decoder 65. From the quantization circuit 21, a prediction signal quantized to the same precision as the reproduced prediction error signal is supplied to the adder 20, and the adder 20 adds the prediction signal and the prediction error signal, and outputs the output of the adder 20. The decoded image signal is output. The reproduced image signal becomes a signal having the same precision as the image signal supplied to the predictive encoder 64 of the first compression encoding device 103, that is, a signal that matches the image signal V supplied to the predictive encoder 64. The reproduced digital image signal V is supplied to an output terminal 23 and a predictor 22. Predictor 22
has the same function as the predictor 8 of the first compression encoding device 103, and obtains and outputs the next prediction signal from the image signal V reproduced according to the prediction characteristics.The quantization circuit 2
Supply to 1. The quantization circuit 21 has the same function as the preprocessing quantization circuit 7 of the first compression encoding device 103, and quantizes the predicted signal according to the quantization characteristic selected by the control signal. A prediction signal quantized to the same precision as the prediction error signal output from the unequal length decoding circuit 18 is output and supplied to the adder.

The above is an explanation of the operation of the first decoding device 105.

The second compression encoding device 106 in FIG. 2b is different from the first compression encoding device 103 in FIG. 2a in that it does not have a quantization circuit for preprocessing and performs control using the mode signal M. different. The second decoding device 108 in FIG. 2b is exactly the same as the first decoding device 105 in FIG. 2a, except that it does not have a terminal for outputting a mode signal.

Reference numbers 4, 8, 10, 64, 1 in Figure 2b
1, 12, 14, 15, 17, 18, 19, 2
0, 21, 22, 23 and 65 have the same functions as the reference numerals in FIG. 2a and perform similar operations. The reproduced image signal V sent from the first decoding device 105 is transmitted to the second compression encoding device 1.
06 input terminal 4, mode signal M is input terminal 5
is supplied to The image signal V supplied to the input terminal 4 is sent to the quantization circuit 7 of the first compression encoding device 103.
equal to the image signal V preprocessed by
Similarly to the compression encoding device 103, the signal is supplied to the predictive encoder 64, which performs predictive conversion, outputs a prediction error signal, and supplies it to the unequal length encoding circuit 11.

The mode signal M supplied to the input terminal 5 is supplied to the control circuit 13. The control circuit 13 controls the unequal length encoding circuit 11 using the information storage amount and the mode signal M supplied from the buffer memory 12. In a normal case, control is performed to perform unequal length encoding by selecting a code conversion characteristic corresponding to the quantization characteristic represented by the mode signal M in accordance with the mode signal M. In this case, the unequal length encoding circuit 11 performs the same encoding as that performed by the first compression encoding device 103. If the information storage amount differs from the estimated information storage amount estimated from the mode signal M by more than a certain range, control is performed to initialize the buffer memory. If the amount of stored information is small, a control signal is sent to the unequal length encoding circuit 11 to generate dummy information, and if the amount of stored information is large, a control signal is sent to stop generating compression encoded information. Once the initialization has been performed, the initialization will not be performed unless the amount of stored information deviates due to a transmission path error or the like, and normal control will be performed according to the mode signal M.

The unequal length encoding circuit 11 encodes the prediction error signal in unequal length according to the control signal from the control circuit 13, outputs a compressed encoded signal, and supplies the compressed encoded signal to the buffer memory 12. The buffer memory 12 temporarily stores the compressed encoded signal and smoothes the compressed data C.
The information stored in the buffer memory 12 is output to the output terminal 14 and the amount of information stored in the buffer memory 12 is supplied to the control circuit 13. Compressed data C output from the second compression encoding device 106 is equal to that output from the first compression encoding device 103. The above is an explanation of the operation of the second compression encoding device 106.

The second decoding device 108 is the first decoding device 1
Since it performs the same operation as 05, the explanation is omitted. Since the second decoding device 108 does not output the mode signal M, the control circuit 19 does not output the mode signal.

FIG. 3 is a diagram showing a specific circuit example of the quantization circuit 7 of the first compression encoding device 103 in the first embodiment of the present invention. This circuit example uses an 8-bit image signal X (LSB (Least
Significant Bit) is x ₁ and its size is 1. ) is quantized and output by truncating the 6- to 8-bit image signal V according to the quantization control signal. The 8-bit image signal X input to the quantization circuit 7 is an input terminal.
The upper 6 bits from x ₃ to _{x 8} are used as output terminals.
The bits _{of x1} and _x2 are supplied to AND circuits 29 or 28 _, respectively. The quantization control signal QC supplied from the control circuit 13 is a 2-bit signal and is supplied to the code inverter 27, where it is inversely transformed and outputs 2-bit quantization selection signals indicated by QS1 and QS2. When the control signal QC is 1, the quantization selection signal QS1 is "0" and QS2 is "1", and when the control signal QC is 2, the quantization selection signal
When QS2 is “0”, QS1 is “1”, and the control signal
When QC is 0, both quantization selection signals QS1 and QS2 are "1". Quantization selection signal QS1 is supplied to AND circuit 29, and quantization selection signal QS2 is supplied to AND circuits 28 and 29. AND circuits 28 and 29 output a signal obtained by ANDing the respective input signals, and supply the signal to output terminals y ₂ and y ₁ .
Therefore, the output of the quantization circuit 21 has an accuracy of 8 bits when the quantization control signal QC is 0, and an accuracy of 8 bits when the quantization control signal QC is 0.
When QC is 2, the image signal Y is uniformly quantized with 7-bit accuracy and when QC is 2, 6-bit accuracy is output.

The quantization circuits 21 of the first decoding device 105 and the second decoding device 108 are similarly configured.

FIG. 4 is a diagram showing a specific circuit example of the predictor 8 of the first compression encoding device 103 in the first embodiment of the present invention. In this circuit example, when the sampling frequency s is three times the subcarrier sc,
Two prediction functions are defined as prediction functions that can efficiently predict an NTSC color TV signal, and the two prediction functions are adaptively switched and selected to obtain a prediction signal.

The first prediction function P ₁ (Z) is a one-dimensional function and is expressed by the following Z function (Z=e ^−j2 〓 ^{/ s} ).

P ₁ (Z)=0.5Z ^-1 +Z ^-3 -0.5Z ^-4 (1) The second prediction function P ₂ (Z) is a function that predicts from two lines in advance and is expressed by the Z function of the following equation.

P ₂ (Z) = Z ^-2H (2) However, H indicates the number of samples in one horizontal scanning period s
When =3sc, H=682.5.

The selection method is based on a prediction signal using two prediction functions and a locally decoded signal. The one corresponding to the decoded signal is the predictive encoder 64
matches the input image signal to. ), and the prediction function that outputs a prediction signal close to the input image signal is used for the next prediction.

The image signal supplied to the predictor 8 is sent to a first prediction circuit 30 having the characteristics of the prediction function of equation (1), a second prediction circuit 31 having the characteristics of the prediction function of equation (2), and a determination circuit 32. Sent. The first prediction signal outputted from the first prediction circuit 30 is sent to the terminal a of the switching circuit 34 and the determination circuit 32, and the second prediction signal outputted from the second prediction circuit 31 is sent to the terminal b of the switching circuit 34.
and is sent to the determination circuit 32. The determination circuit 32 determines which predicted signal is closer to the image signal, and outputs a selection signal of "0" if it is close to the first predicted signal, and outputs a selection signal of "1" if the second predicted signal is close. ” outputs a selection signal. The selection signal is delayed by one sampling clock cycle in the register 33 and then sent to the switching circuit 34. In the switching circuit 34, when the selection signal is "0", the first predicted signal at terminal a is selected, and the selection signal is When is "1", the second predicted signal at terminal b is selected and output to obtain a predicted signal. The second prediction circuit 31 is composed of a delay circuit that delays the input signal by 1365 sampling clock periods and outputs the delayed signal.

FIG. 5 is a diagram showing a specific circuit example of the first prediction circuit 30 in FIG. 4. The first prediction circuit 30
Multipliers 35 and 40 with a coefficient of 0.5 and an input signal having a prediction characteristic expressed by the function of equation (1)
a register 36 for outputting with a delayed sampling period;
38, 39 and 42, subtractor 37 and adder 41
This is a non-recursive type digital filter consisting of.

The predictors 8 or 22 of the first decoding device 105, the second compression encoding device 106, and the second decoding device 108 are similarly configured.

Next, code conversion by the unequal length encoding circuit 11 and code inverse conversion by the unequal length decoding circuit 8 shown in FIG. 2a will be explained. The image signal that has undergone quantization preprocessing that is input to the predictive encoder 64 is input to V _Q (Q represents the type of quantization characteristic), the predicted signal output from the predictors 8 and 22 is input to P, and the quantization circuit If the predicted signal quantized to the same precision as V _Q in 21 is P _Q , the quantization noise q generated in the quantization circuit 21 is q = P - P _Q
(Since the quantization circuit 21 performs quantization by rounding down as shown in FIG. 3, q takes a smaller positive value depending on the quantization precision of the quantization characteristic represented by Q.). Subtractor 10 of predictive encoder 64
Then, the prediction signal P is subtracted from the quantized image signal V _Q to output a prediction error signal e=V _Q -P, which is supplied to the unequal length encoding circuit 11. Using q=P-P _Q , the prediction error signal is expressed as e=V _Q -P _Q -q.

Therefore, to perform code conversion, first convert the prediction error signal e (=V _Q −P _Q −q) into a quantized image signal.
Quantization is performed to make the accuracy equal to the precision of V _Q to obtain a quantized prediction error signal e _Q =V _Q −P _Q , and then code conversion is performed, while the predictive decoder 65 reproduces it by inverse code conversion. Adding the predicted error signal e _Q and the quantized prediction signal P _Q (e _Q + P _Q = V _Q )
An image signal equal to the image signal V _Q input to the predictive encoder 64 is reproduced. Prediction error signal e to -q
Quantization excluding the above results in quantization by rounding up. That is, the unequal length encoding circuit 11 quantizes the prediction error signal e by rounding up to be equal to the quantization precision of the image signal V _Q , and then converts the quantized prediction error signal e _Q to unequal length. It consists of a function to convert the code. Note that since the predictive encoder 64 does not use a quantizer, reversible encoding can be performed even if the subtracter 10 of the predictive encoder 64 and the adder 20 of the predictive decoder 65 perform modulo calculations while ignoring carry. Therefore, the quantized image signal input to the predictive encoder 64 and the image signal decoded by the predictive decoder 65 match. Therefore, modulo calculation can also be performed on the prediction error signal. An example of how to create a code conversion characteristic for converting a quantized prediction error signal e _Q into an unequal length code when the image signal V _Q has an 8-bit error length will be described. One type of unequal-length code consisting of 256 code words of unequal length is determined from the statistics of the distribution of the prediction error signal e _Q quantized to 8-bit accuracy, and the accuracy of quantization of the prediction error signal e _Q is determined. That is, the quantization characteristic of the quantization circuit 7 of the preprocessing means is provided with conversion characteristics for converting each prediction error signal into the unequal length code, and the prediction error is calculated by the selected code conversion characteristics. Convert the signal to an unequal length code. In other words, when the quantization precision is 8 bits, for a prediction error signal e _Q of an integer between -128 and 127,
Allocate 256 codewords, -128 when 7 bits
128 codewords are assigned from the shortest code length to even prediction error signals e _Q up to 126, and similarly when the code length is 6 bits, prediction difference signals e _Q that are multiples of 4 from -128 to 124 are assigned. The code conversion characteristics are determined so that 64 code words are assigned to .

Another example is to divide the prediction error signal _eQ by 1 for 8 bits, 2 for 7 bits, and 4 for 6 bits according to the quantization characteristics, and then divide it by 4 for 8 bits. is converted into an unequal-length code using the code conversion characteristics, and on the decoding side, the output of the code inverse conversion characteristics for 8 bits is multiplied by 1, 2, or 4 according to the quantization characteristics, and a prediction error signal e _Q is output. . Another example is to use three types of unequal length codes consisting of 256, 128, and 64 code words corresponding to the types of quantization characteristics of 8 bits, 7 bits, and 6 bits. Performs equal length encoding and decoding.

FIG. 6a is a block diagram showing another example of the configuration of the first compression encoding device 103 according to the first embodiment of the present invention. In this embodiment, among the two functions of finishing quantization and code conversion, the unequal length encoding circuit 11 of the first compression encoding device 103 shown in FIG. 64.

Reference numbers 4, 7, 8, 10, 64, 12, 13
and 14 have the same functions as the parts indicated by the respective reference numerals in FIG. 2a or b, and perform similar operations. The quantization circuit 9 has the same function as the quantization circuit 7, and selects the same quantization characteristic as the quantization circuit 7 according to the control signal from the control circuit 13, and quantizes and outputs the predicted signal from the predictor 8. and is supplied to the subtracter 10. The subtracter 10 subtracts the quantized prediction signal P _Q from the quantized image signal V _Q to obtain a quantized prediction signal e _Q and supplies it to the unequal length encoding circuit 11 .
The unequal length encoding circuit 11 has several types of code conversion characteristics corresponding to the quantization characteristics, and performs code conversion corresponding to the quantization characteristics selected by the quantization circuit 7 in accordance with a control signal from the control circuit 13. Select the characteristics and convert the prediction error signal e _Q into an unequal length code. The second compression encoding device can also have a similar configuration.

FIG. 6b is a block diagram showing another example of the configuration of the first decoding device 105 according to the first embodiment of the present invention. In this embodiment, the first decoding device 10 shown in FIG.
The configuration is such that the position of the quantization circuit 21 of No. 5 is equivalently changed.

Reference numbers 15, 16, 17, 18, 19, 2
0, 21, 22, 23, 24 and 65 have the same functions as the reference numerals in FIG. 2a and perform similar operations.

FIG. 7 is a block diagram showing an example of the configuration of the first compression encoding device 103 and the first decoding device 105 when thinning processing is used as a preprocessing means in the first embodiment of the present invention. be. The thinning process can be performed as long as it is performed equivalently. Either perform thinning first to reduce the number of samples and perform predictive conversion, or simply determine the pixels to be thinned and then perform predictive conversion as is. There is a method in which code conversion is performed by actually thinning out data in step 11. In this embodiment, the latter configuration is adopted.

Reference numbers 4, 8, 10, 12, 13, 14, 1
5, 16, 17, 19, 20, 22, 23, 2
4, 64 and 65 have the same functions as the reference numerals in FIG. 2a and perform similar operations.

The image signal supplied from the A/D converter 102 to the input terminal 4 of the first compression encoding device 103 is supplied to the thinning compensation processing circuit 43. The thinning and interpolation processing circuit 43 has several types of predetermined thinning characteristics, and determines which pixels will not be thinned out and which pixels will be thinned out according to the thinning characteristics selected by the control signal from the control circuit 13. Note that the predictor 8 of the next predictive encoder 64 is an example of a circuit that performs prediction using pixels in the vicinity of the pixel to be thinned out, so the signal of the pixel to be thinned out is based on the signal of the surrounding pixels that are not thinned out. Interpolated in advance using The image signal subjected to interpolation processing by the thinning interpolation processing circuit 43 is supplied to a predictive encoder 64,
The predictive encoder 64 obtains a predictive error signal and supplies it to the unequal length encoding circuit 11 . The unequal length encoding circuit 11 has a predetermined code conversion characteristic, for example, the same code conversion characteristic for the 8-bit quantization characteristic having the unequal length encoding circuit 11 of the second embodiment. In response to the thinning characteristics selected by the thinning interpolation processing circuit 43, only the prediction error signals corresponding to pixels that are not thinned out are converted into unequal length codes according to the code conversion characteristics, and the unequal length codes and the unequal length codes necessary for decoding are Mode signal M representing synchronization and thinning characteristics
Buffer memory 1 stores the compressed encoded signal consisting of
Send to 2. The buffer memory 12 smoothes the encoded information and outputs it as compressed data C to the output terminal 14.
A control circuit 13 outputs information and controls the amount of information accumulated.
supply to The control circuit 13 uses the amount of accumulated information supplied from the buffer memory 12 to control switching of the preprocessing characteristics of the thinning process, and performs smoothing control of the amount of information generated.

A control signal output from the control circuit 13 is supplied to the thinning-out interpolation processing circuit 43 and the unequal length encoding circuit 11. The above is the first compression encoding device 103
This is an explanation of the operation.

In the first decoding device 16, compressed data C sent from the compression encoding device 103 at a given data rate is supplied to the input terminal 15. The compressed data C supplied to the input terminal 15 is temporarily stored in the buffer memory 17, and then inversely transformed in the unequal length decoding circuit 18 according to the code inverse transform characteristic, and a prediction error signal is output and added. Supplied to the container 20. For pixels that have been thinned out, a prediction error signal having an appropriate value, for example, a value of 0, is output. The adder 20 adds the prediction signal sent from the predictor 22 and the prediction error signal.
A decoded signal is output to the output of 0 and supplied to the interpolation processing circuit 44. The interpolation processing circuit 44 has the same function as the decimation interpolation processing circuit 43, and interpolates the decimated pixels according to the decimation characteristics selected by the control signal using the decoded signals of the surrounding non-thinned pixels. A decoded image signal that has been subjected to interpolation processing is output. The decoded signal subjected to the interpolation process is supplied to the output terminal 23 and the predictor 22.

The control circuit 19 uses the information storage amount supplied from the buffer memory 17 and the control information supplied from the unequal length decoding circuit 18 to control the readout of encoded information and perform code inverse conversion in accordance with the thinning characteristics. A control signal for performing interpolation processing corresponding to the thinning characteristics is supplied to the unequal length decoding circuit 18, a control signal for performing interpolation processing corresponding to the thinning characteristics is supplied to the interpolation processing circuit 44, and a mode signal M indicating the thinning characteristics according to the preprocessing characteristics is supplied. is supplied to the output terminal 24. The above is the first decoding device 1
This is an explanation of the operation of 05.

The second compression encoding device 106 and the second decoding device can be similarly configured.

8a to 8c are diagrams showing the arrangement of sampled pixels for expressing the thinning characteristics in the interpolation processing circuit 43 of the first compression encoding device of FIG. 7. The pixel arrangement of a part of the screen from the (l-3) line to the l-th line of the 1st field when the sampling frequency is s = 3sc, and the characteristics without thinning are as follows.
It will look like the circle shown in figure a. FIG. 8b is a diagram showing the first thinning characteristic. Pixels marked with × indicate pixels to be thinned out, and in the horizontal direction, every 3 samples.
In the vertical direction, thinning is performed every other line, and the number of effective samples is reduced to 5/6.

FIG. 8c is a diagram showing the second thinning characteristic.
All lines are thinned out every 3 samples in the horizontal direction, and the number of effective samples is reduced to 2/3. Interpolation of the signal of pixel Y, which is to be thinned out, is performed using pixels A to E which are not thinned out, as shown in FIG. 8c. The signal of pixel Y is interpolated using the following equation.

Y = 0.5A + 0.25B - 0.5C + 0.5D + 0.25E (3) If expressed by the interpolation filter Hy (Z) of the Z function, the following equation is obtained.

Hy(Z)=0.5Z ^-1 +0.5Z ^-681 -0.5Z ^-682 +0.5Z ^-684 +0.25 ^-1365 (4) Figure 9 shows the first compression encoding device 103 in Figure 7.
3 is a diagram showing a specific circuit example of a thinning-out interpolation processing circuit 43; FIG. The image signal supplied to the input terminal 51 of the thinning-out interpolation processing circuit 43 is supplied to the terminal a of the switching circuit 53 and the interpolation filter circuit 52. The interpolation filter circuit 52 is composed of a non-recursive type digital filter having the characteristics of an interpolation filter shown by equation (4), and outputs an interpolation signal to the output of the interpolation filter circuit 52, and the switching circuit 5
3 to terminal b. A control signal is supplied from the control circuit 13 to the input signal 54 in order to perform thinning processing according to the thinning characteristics selected from among the thinning characteristics shown in FIGS. 8a to 8c. For pixels that are not thinned out, the signal at terminal a is selected, and for pixels that are thinned out, the signal at terminal b is selected and output from the output terminal 55.
Note that if the predictor 8 does not need the image signal for the pixel to be thinned out, there is no need to perform interpolation. The interpolation processing circuit 44 is the thinning interpolation processing circuit 43
is configured the same as.

As is clear from the above explanation, the first method consists of preprocessing, reversible predictive transformation, and unequal length encoding.
The digital image signal V that is reproduced by decoding the compressed data encoded using the compression encoding device is the same as the output signal of the preprocessing means. Therefore, in order to encode the reproduced image signal V again, the second
The compression encoding device 106 does not perform preprocessing, but performs encoding with the same encoding characteristics as the first compression encoding device using the mode signal M representing the preprocessing characteristics, and outputs the same compressed data. The second decoding device 108 can decode the image signal V again, and no image quality deterioration occurs due to repeated encoding. Therefore, if the encoding/decoding device of the present invention is used, encoding can be performed repeatedly, and if the compressed data C encoded and output by the encoding/decoding device of the present invention is recorded on a digital VTR, dubbing can be performed. It is possible to prevent image quality deterioration from increasing no matter how many times the process is performed.

[Brief explanation of drawings]

FIG. 1 is a block diagram showing the configuration of a first embodiment of the present invention, and FIG. 2 a and b are a first compression encoding device 103 and a first decoding device of the first embodiment of the present invention. 105, a block diagram showing an example of the configuration of the second compression encoding device 106, and the second decoding device 108, FIG. 3 is a diagram showing an example of a specific configuration of the quantization circuit 7, and FIG. A diagram showing an example of a specific configuration of the predictor 8, FIG. 5 shows the first prediction circuit 30.
FIGS. 6a and 6b are block diagrams showing other examples of the configurations of the first compression encoding device 103 and the first decoding device 105,
FIG. 7 is a block diagram showing another example of the configuration of the first compression encoding device 103 and the first decoding device 105 of the first embodiment of the present invention, FIG. 8 is a diagram showing thinning characteristics, FIG. 9 is a diagram showing an example of a specific configuration of the thinning-out interpolation processing circuit 43. 101, 4, 5, 15, 51 and 54 are input terminals, 110, 14, 23, 24 and 55 are output terminals, 102 is an A/D converter, 103 is a first compression encoding device, 104 and 108 are storage device or transmission line; 105 is a first decoding device; 10
6 is a second compression encoding device, 108 is a second decoding device, 109 is a D/A converter, 7, 9 and 2
1 is a quantization circuit, 64 is a predictive encoder, 8 and 2
2 is a predictor, 10 is a subtracter, 11 is an unequal length encoding circuit, 12 and 17 are buffer memories, 1
3 and 19 are control circuits, 18 is an unequal length decoding circuit, 65 is a predictive decoder, 20 is an adder, 27 is a code inverse converter, 28 and 29 are AND circuits, 30
is the first prediction circuit, 31 is the second prediction circuit, 32
33 is a register, 34 is a switching circuit, 35 and 40 are multipliers, 36, 38, 39
42 is a register, 37 is a subtracter, 41 is an adder, 43 is a thinning-out interpolation processing circuit, 44 is an interpolation processing circuit, 52 is an interpolation filter circuit, and 53 is a switching circuit.

Claims (1)

[Claims] 1. In an encoding/decoding device that compresses and encodes an image signal and decodes and reproduces it, a preprocessing means for controlling the amount of information according to a control signal from a control means, and a digital image signal subjected to the preprocessing. a prediction conversion means for converting into a prediction error signal using reversible logic; a means for converting into an unequal length code under the control of the control signal; and synchronization and control necessary for the unequal length code and decoding. The amount of information generated by controlling the preprocessing means by monitoring the generation amount of coded information stored in the buffer memory, and the amount of encoded information stored in the buffer memory. a first compression encoding device comprising a control means for performing smoothing control; a first decoding device for decoding the compressed data and reproducing the digital image signal and the control signal; When compressing and decoding the reproduced digital image signal again, the image signal is converted into a prediction error signal, and the prediction error signal is unequal length encoded under the control of the control signal and smoothed by a buffer memory. The original information is preserved in the process of repeating encoding and decoding by using a second compression encoding device that obtains the compressed data again by doing so, and a second decoding device that reproduces the image signal from the compressed data. An encoding/decoding device characterized in that it is configured to perform.