JPH0782705B2 - High efficiency coding system - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION (Industrial field of application) In recording, transmission equipment, and other various equipments that perform signal processing of digital signals, it is possible to efficiently encode signals with a smaller amount of information. The present invention relates to an efficient coding system.

(Prior Art) As a basic encoding means when a signal such as an image signal or an audio signal is to be digitized, a signal level is equally divided with respect to each sample value, and a value included in each range is included. Linear quantization (equal quantization) that replaces with one representative value is used.

Then, in order to make the difference between the representative point and the original value unclear when the above-described uniform quantization is used for digitizing the signal, generally, 6 bits (32th floor) are used for the natural image. Key) and 8 bits (256 gradations) and 8 bits to 16 bits (65536 gradations) are required for audio. Therefore, image signals or audio signals can be digitalized by equal quantization as described above. Transmit the converted signal,
When recording, it is necessary to handle a large amount of information as described above for each sample value, so it is used when transmitting and recording the image signal or audio signal in an analog system. The problem is that a device with a larger scale than that of the device is required.

Therefore, even if a signal is encoded with a smaller amount of information, it is sensitive to changes in the part where the signal changes little, and even if there is some error in the part where the signal changes rapidly, it can be detected. It is well known that various high-efficiency coding methods have been conventionally proposed in which the amount of information per sample is reduced by utilizing the human visual and auditory properties that are difficult.

As one specific example of the above-described high-efficiency coding method, a signal to be coded is divided into blocks of a certain fixed section, and a mode of signal level distribution of signals in the divided individual blocks , But each block can be converted (normalized or normalized) so as to be in a substantially common state.

Then, according to the above-mentioned conventional method, the mode of distribution of the signal level in the block including a portion in which the change in the signal level of the signal to be encoded is large is widely distributed in the block, and The distribution of the signal level in the block including the portion of the signal to be encoded in which the signal level hardly changes is biased to a part of the signal level, so that the region having no signal is deleted, and eventually, Since the signal to be coded is in a compressed state, the signal obtained by this conventional method is a quantization representative point (level) when quantizing the signal to be coded. Even if the quantization is performed with a smaller number of quantization representative points (grayscales) than the number of tones, the error is large for the signal that is inversely transformed by the decoding system for the part that changes drastically, but the part that changes little Since the error is reduced for the inverse transformed signal in the decoding system with even less problematic when viewed from the nature of the visual and auditory aforementioned human.

When encoding is performed by the above-described conventional method, information about what kind of conversion (standardization) has been performed on the signal for each block, that is, information indicating the state of normalization (normalization However, since it is possible to reduce the amount of quantized information for each sample as described above, it is possible to reduce the overall amount of information by making the block larger than a certain amount. is there.

By the way, a block obtained by dividing a signal to be encoded into certain constant sections is a one-dimensional block or image such as a scanning line of a voice or an image appearing on the time axis depending on the type of the signal. A two-dimensional block for handling in both the horizontal direction and the vertical direction, and a three-dimensional block considering the time axis direction like a moving image are conceivable. Further, as a standardization method, for example, Something like is known.

When the signal within the block is normalized (normalized) using the maximum value and the minimum value of the sample values within the “1” block, Yi = k (Xi−Xmin) / (Xmax−Xmin). 1) where i = 1,2,3, ... N Xi is the i-th input sample value (one-dimensional) Yi is the i-th normalized sample value (one-dimensional) Xmin is the minimum sample value in the block Xmax Is the maximum sample value in the block N is the number of samples in the block k is the normalization constant According to the above equation (1), normalization (normalization) is performed using the maximum value and the minimum value of the sample values in the block. The i-th standardized sample value Yi is 0 when the i-th input sample value Xi is the smallest sample value Xmin in the block, and the i-th input sample value Xi is the largest sample value in the block. When the sample value is Xmax, it becomes k, so N inputs in the block All normalized sample values Y1 to Yn for samples X1 to Xn will take values between 0 and k.

“2” When the average value of the sample values in the block is used to normalize (normalize) the signal in the block, Yi = k (Xi−Xmean) / D (2) where i = 1, 2,3, ... N D indicates the degree of change, which is shown in the following 1) to 3).

1) Maximum value of D = | Xi−Xmean | (3) Note that the block is one-dimensional here.

Next, it is assumed that the sequential 8 (N = 8) samples corresponding to each of the constant intervals in the one-dimensional signal belong to one sequential block A, B, respectively. For example, in accordance with the above formula (1), that is, Yi = k (Xi−Xmin) / (Xmax−Xmin) (1), as shown in the above “1”, Referring to FIG. 3, the case where the sample value Xi in the block is normalized (normalized) using the maximum value Xmax of the sample values and the minimum value Xmin of the sample values in the block will be described below. It is as follows.

FIG. 3 (a) shows signals (sample values) belonging to individual blocks A and B obtained by dividing a discrete signal to be coded into certain intervals in the horizontal direction of the figure. For example, it is a diagram in which, while taking time, on the other hand, the signal level is taken in the vertical direction, and successive sample values of the signal are shown by white circles, and XmaxA is 8 in block A in FIG. 3 shows the maximum sample value among the eight samples and the corresponding signal level, and XminA shows the minimum sample value among the eight samples in the block A and the corresponding signal level, and FIG. Where XmaxB is 8 in block B
The maximum sample value of the samples and the corresponding signal level are shown, and XminB is 8 in block B.
The signal level corresponding to the smallest sample value among the samples is shown.

In the example shown in FIG. 3 (a), the eight samples belonging to block A are approximately in the full width of the signal level.
Exists within a width of about 30% and belongs to block B 8
Samples are shown to be present over substantially the entire width of the signal level.

For each sample belonging to each block A, B obtained by dividing the discrete signal to be encoded shown in FIG. When the blocks A and B are subjected to the normalization (normalization) process for each of the blocks A and B by normalizing the distribution in the amplitude direction of the signal according to the above-described equation (1), FIG. Each block as shown in (b) of
The sample value for each A and B is normalized to 0 for the sample having the sample value corresponding to the minimum signal level in each block A and B before normalization shown in (a) of FIG. Further, the sample having the sample value corresponding to the maximum signal level in each of the blocks A and B before normalization shown in FIG. 3A is a standardized signal in a state standardized to k. You can get it.

As for the sample value of each sample belonging to the individual blocks A and B obtained by dividing the discrete signal that is the object of encoding as described above at regular intervals, the individual blocks described above
For each of A and B, the distribution of the signal in the amplitude direction is standardized according to the equation (1), and the individual blocks are
FIG. 4 shows an example of the structure of a normalizer used to perform normalization processing on each of A and B.

In FIG. 4, 1 is an input terminal of a discrete signal to be encoded, and 2 to 5 are time intervals of the above-mentioned discrete signal (sampling period used to generate the discrete signal). A delay circuit for giving a delay time equal to the signal, 6 is a maximum value detection circuit, 7 is a minimum value detection circuit, 8 and 9 are quantization circuits, 10 to 15 are subtractors, 16 to 20 are dividers, 21 and 22. Is an output terminal of information indicating the standardized state (data indicating the standardized state), and 23 to 27 are output terminals of the quantized data, which are discrete signals which are supplied to the input terminal 1 and are to be encoded. The signals are respectively supplied to a series circuit of delay circuits 2 to 5, which are equal to the time interval (the sampling period used to generate the discrete signal) of the discrete signal described above.

The discrete signal X1 is output from the delay circuit 5 and the maximum value detection circuit 6 and the minimum value detection circuit 7 are output.
At the time when it is applied to the subtracter 11, the delay circuit 4 outputs the discrete signal X2, which is applied to the maximum value detection circuit 6, the minimum value detection circuit 7, and the subtractor 12,
Further, the delay circuit 3 outputs a discrete signal X3, which is given to the maximum value detection circuit 6, the minimum value detection circuit 7, and the subtracter 13, and further, from the delay circuit 2 described above, the discrete signal X ( n-1) is output, which is the maximum value detection circuit 6
, The minimum value detection circuit 7 and the subtractor 14, and the discrete signal Xn supplied to the input terminal 1 is further supplied to the maximum value detection circuit 6, the minimum value detection circuit 7, and the subtractor 15. Given to.

As a result, the maximum value detection circuit 6 has a maximum value Xmax in the sample values X1 to Xn of N samples in one block.
Is detected and given to the quantizing circuit 8, and in the quantizing circuit 8, the quantized signal obtained by quantizing the maximum value Xmax with appropriate precision is supplied to the subtractor 10 as the minuend signal. .

Further, the minimum value detection circuit 7 detects the minimum value Xmin in the sample values X1 to Xn of N samples in one block and gives it to the quantization circuit 9, and in the quantization circuit 9, the above-mentioned minimum value Xmin. The quantized signal obtained by performing the quantization with an appropriate precision is supplied to the subtractors 10 to 15 as a subtraction signal and is also sent to the output terminal 22.

The signal output from the subtracter 10 is a signal obtained by subtracting the quantized signal for the minimum value Xmin from the quantized signal for the maximum value Xmax, that is, the maximum sample value of N samples in one block. A signal representing the difference (Xmax-Xmin) between the value and the minimum value, which signal is sent to the output terminal 21 and supplied to the dividers 16 to 20 as a divisor signal.

Since the output of the corresponding one of the subtractors 11 to 15 is supplied to the dividers 16 to 20 as a dividend signal, the output terminals 23 to 27 are output from the dividers 16 to 20. , Yi = k (Xi−Xmin) / (Xmax−Xmin) (1) i = 1,2, ... N Normalized sample value Y1 to which standardized signal processing is performed according to the above equation (1) Yn is output.

As described above, the sample values of the samples belonging to the individual blocks A and B obtained by dividing the discrete signal to be encoded into the predetermined intervals from the encoder shown in the fourth diagram , A standardized signal (normalized quantized data) obtained by normalizing the distribution in the amplitude direction of the signal for each of the individual blocks A and B described above according to the equation (1) is supplied to the terminal 23. 27 to 27, a signal indicating the standardized state of the signal to be set corresponding to the standardized signal (data indicating the standardized state) is output to the output terminals 21 and 22.

When the signal to be encoded is a multidimensional encoder, the parallel distribution of the input is configured accordingly to configure the block in that dimension, and the signals in the block are It should be obtained. Also the maximum value Xmax
And the minimum value Xmin are the normalized sample values Yi
It is necessary to round up the maximum value Xmax and round down the minimum value Xmin in order to allow the value of 0 to fall between 0 and k.

In order to decode the signal encoded in this way, it is natural that the inverse transform processing of the equations (1) and (2) should be performed.

(Problems to be Solved by the Invention) However, in the above-mentioned conventional high-efficiency coding method, since the standardization within a block is performed only once, how to determine the number N of samples in one block There are various advantages and disadvantages depending on whether it is set.

That is, in the above-mentioned conventional high efficiency coding system,
For example, if the number of samples N in one block is increased,
The amount of standardized data required for each block for decoding {the signal indicating the standardized state of the signal to be set corresponding to the standardized signal (data indicating the standardized state)} is The amount of each sample is small, but since the above-described normalization processing is performed uniformly in one block, if the number N of samples in one block is large, one Since the state of the distribution of sample values in the block in the amplitude direction tends to be diversified, and proper standardization processing is difficult to perform, it is necessary to considerably increase the number of quantization steps in quantizing the sample values. , A quantization error having a correlation with the block shape is likely to occur.

The quantization error that occurs in the above-described state is easy to detect in view of the characteristics of human visual characteristics and auditory characteristics, and thus is likely to appear as a large deterioration in image quality and sound quality. Therefore, it is necessary to increase the amount of quantized data in order to prevent the above-mentioned problems from occurring. However, if this is done, the compression of the amount of data will naturally be insufficient.

On the other hand, if the number of samples N in one block is reduced and the local characteristics of the signal in one block are to be captured, the normalized data amount for each sample increases, so By reducing the number of samples N in a block, even if the quantized data of standardized sample values can be reduced, the total data amount cannot be reduced so much, and the encoding Even when the number of samples N in one block may be large, as in the case where the signal of interest has a small change in signal level over a wide range, the number of samples N in one block is reduced as described above. If the encoding process is performed after that, the encoding process becomes redundant.

As an example of two kinds of discrete signals having different numbers of samples in one block, the signal to be encoded is standardized by the above-mentioned conventional method according to the above-mentioned equation (1), and FIG. The above problems will be described with reference to.

FIG. 5 shows an example in which the number of samples N in one block is 8 ((a), (b) in FIG. 5), and the number of samples in one block is large. N is 4
In the case of {(c), (d) in FIG. 5 is an example when the size of one block is small.

Further, (a) and (c) of FIG. 5 are intended for encoding the same first signal, and (b) and (d) of FIG. The signal of is targeted for encoding.

The shaded portion in FIG. 5 is the portion that is excluded from the subject of quantization during the normalization process because there is no signal that is the subject of encoding. The quantization is performed on the remaining portion not shaded, but if the number of quantization levels (the number of quantization steps) for the standardized sample value is made equal, naturally, the above-mentioned The narrower the width of the remaining portion, the smaller the error caused by the quantization. This is because the quantization of each sample value is performed regardless of the standardization process.

Then, when the signal normalization is performed by the signal processing according to the equation (1), the width of the remaining portion is the maximum sample value and the minimum sample value of the samples in one block. The smaller the number N of samples in one block, the smaller the difference.

Regarding the first signal, as is clear from (a) and (c) of FIG. 5, when the number of samples N is 8, it becomes half or less when the number of samples N is 4, and the second signal As is clear from FIGS. 5 (b) and 5 (d), the signal on the right side of FIG. 5 (d) is the same as that in FIG. 5 (b), but the right block in FIG. The left block in (d) is reduced to about 1/4 as compared with the block shown in (b) of FIG.

From the above, when the number of quantization levels (the number of quantization steps) with respect to the standardized sample value is made equal, it is possible to reduce the number of samples in one block to achieve good quantization with less error. Recognize.

The data amount of the standardized signal (quantized data amount of the standardized signal) and the standardized signal are set in the case where the data is standardized and encoded as in the above example. The following Table 1 shows the data amount of the signal indicating the normalization state of the signal (the data indicating the normalization state ... The standardized data amount).

However, in Table 1 above, the input sample is 8bi.
It is assumed that the quantization after t and the standardization processing is performed with an accuracy of 3 bits per value.

As can be seen from Table 1 above, the signal to be coded is sampled when the number N of samples in one block is 8
When the signal is 4 and the signal standardization is described above (1)
When signal processing is performed according to the formula, the standardized data amount required for normalization is 8 bits for each parameter of 2 tumors, and when the number of samples N is 8, per block The total amount of data is 40 bits, and 5 bits per sample, but the original amount of data is 8 bits.
However, the amount of data is compressed to 5/8.

When the number of samples N is 4, the total data amount per block is 28 bits, which is 7 bits per sample.
Most of the data will not be compressed. As described above, the conventional high-efficiency encoding method has a problem in that the amount of data cannot be compressed so much when an encoding with a relatively small quantization error is attempted.

(Means for Solving the Problems) The present invention relates to a signal belonging to each block obtained by dividing a discrete signal to be encoded for each constant section, for each of the individual blocks, Means for obtaining a standardized signal in a state in which the distribution in the amplitude direction of each signal is standardized and each block is standardized, and the standardized signals obtained in correspondence with the individual blocks described above. For the signals belonging to the new individual blocks obtained by dividing the signal into the individual blocks used to obtain each of the standardized signals and further dividing the individual blocks into predetermined blocks, For each individual block,
Means for sequentially repeating at least one or more times to standardize the distribution in the amplitude direction of each signal and obtain the standardized signal in the state in which the standardization is applied to each of the new individual blocks. In the high-efficiency coding method and the high-efficiency coding method, the standardization state of the signal to be set corresponding to the standardized signal in each of the stages described above is shown. As for the information, the data amount obtained by summing the data amount indicating the standardized state of the signal set corresponding to the standardized signal for each one block at each stage described above over all the stages is 1 In order to solve the above-mentioned conventional problems by providing a high-efficiency coding method that is equal to the amount of data indicating the standardized state of the signal to be set when it is performed only once. It is.

(Example) Hereinafter, specific contents of the high-efficiency coding method of the present invention will be described in detail with reference to the accompanying drawings. The high-efficiency coding method of the present invention is such that, for a signal belonging to an individual block obtained by dividing a discrete signal to be coded into certain intervals, the amplitude of each of the individual blocks is increased. Means for standardizing the distribution in the direction to obtain a standardized signal in a state in which standardization has been performed for each individual block, and the standardized signal obtained corresponding to each of the individual blocks described above, Of the individual blocks used to obtain each of the standardized signals of each of the new individual blocks obtained by dividing each of the blocks into a predetermined interval in a state of being further divided. For each of the above, it is possible to perform normalization of the distribution in the amplitude direction of each signal to obtain a standardized signal in a state in which the standardization is applied to each of the new individual blocks, at least sequentially. Information indicating the standardization state of the signal to be set in correspondence with the standardized signal at each stage in the high efficiency coding method described above and the high efficiency coding method described above. Is the data amount obtained by summing the data amount indicating the standardized state of the signal set corresponding to the standardized signal for each block in each of the above-mentioned stages over all the stages. This is a high-efficiency coding method in which the amount of data indicating the standardized state of the signal to be set when it is performed only once is made equal to the high-efficiency coding method. As described above, the standardization process is not performed once on the block of the signal to be encoded, but after the standardization process is performed on the block, the block is gradually subdivided. State of It is obtained to perform the normalization process of the multi-stage that sequentially straining facilities the normalization process with respect to the lock.

Then, if the standardization of the signal by the multistage standardization process performed in the high-efficiency encoding system of the present invention is equivalent to the standardization performed only once as a result of the standardization process through all the stages thereof. Therefore, the standardized data amount in each stage of the standardization process can be smaller than the standardized data amount in the case where the standardization process is performed only once.

This is to ensure that the same quantization accuracy can be obtained even when the signal amplitude is very small and the standardization accuracy is required. However, since the quantization error eventually becomes large in such a case, the above-mentioned deterioration of the standardization accuracy becomes negligible and is not a serious problem.

In case of normalizing the signal multiple times,
Since the aspect of the change of the signal is generally grasped at the time of normalizing the signal of the first stage, the standardization of the signal in the subsequent stage does not cause a problem even if it is fairly rough. Therefore, the amount of standardized data when the signal is standardized in only one stage is
If the maximum value Xmax and the minimum value Xmin of the sample values in one block are 8 bits, for example, when the signal is subjected to the standardization process twice, the standardized warp of the signal is 1 In the first stage, 6 bits each for the maximum value Xmax and the minimum value Xmin of the sample values in one block, and in the second stage of the signal normalization processing, the maximum value Xmax of the sample values in one block at that time And the minimum value Xmin are about 2 bits each.

In the second stage of the signal standardization process, a plurality of standardizers are used for the standardization of the signal, so that the overall standardized data amount is only one stage of the signal standardization process. A little more than if.

FIG. 1 is a block diagram showing a configuration example of an apparatus used for multistage standardization of signals in implementing the high-efficiency coding system of the present invention. In the apparatus shown in FIG. Reference numerals 28 to 30 denote normalizers, and the normalizers 28 to 30 having the same configuration as that described with reference to FIG. 4 can be used as the normalizers. Further, 31 is a quantizer.

In FIG. 1, the first-stage normalizer 28 is for normalizing a signal such that the number N of samples in one block is 8, and the second-stage normalizer 29, 30 is 1
The signal is normalized such that the number of samples N in the block is 4, and the normalizer 28 of the first stage is used.
The signal standardized by is divided into two and supplied to the second-stage normalizers 29 and 30, and further standardized there, and then supplied to the quantizer 31 and quantized and output. To be done. The standardized data is output from a total of three normalizers 28 to 30 including the first-stage normalizer 28 and the second-stage two normalizers 29 and 30, and the quantized data is the quantum data. It is output from the converter 31.

Next, the state of the standardized signal processing over two stages of the standardization of the signal as described above will be described with reference to FIG. 2 which is shown in correspondence with FIG. 5 already described. FIG. 2 shows signals (sample values) belonging to individual blocks C, E1, and E2 obtained by dividing a discrete signal to be encoded at regular intervals in the horizontal direction of the figure, for example, time. On the other hand, it is a diagram in which the signal level is taken in the vertical direction and sequential sample values of the signal are indicated by white circles.

In FIG. 2, the number N of samples in one block C of the signal to be standardized by the normalizer 28 in the first stage in FIG.
And the two normalizers 29, 3 in the second stage in FIG.
A case where the number of samples N in one block E1, E2 of the signal to be standardized is 0 is 0.

XmaxC in FIG. 2 is the maximum sample value of the eight samples in block C and the corresponding signal level,
XminC indicates the signal level corresponding to the minimum sample value among the eight samples in block C, XmaxE1 indicates the signal level corresponding to the maximum sample value among the four samples in block E1, and XminE1 indicates the signal level corresponding to block E1. Shows the signal level corresponding to the smallest sample value of the four samples in X, and XmaxE2 represents the signal level corresponding to the largest sample value of the four samples in block E2, and XminE2 represents the signal level of the four samples in block E2. The signal level corresponding to the smallest sample value in the sample is shown.

Further, FIG. 2 (a) is the same as FIG. 5 (a),
It is assumed that the first signal described with reference to (c) is the signal to be encoded, while (b) in FIG. 2 corresponds to (b) and (d) in FIG. The second signal described with reference is assumed to be the signal to be encoded.

Furthermore, the shaded portion in FIG. 2 indicates that there is no signal to be encoded,
In the normalization process by the normalizer 28 of the first stage, the part that is excluded from the target of quantization, and the part shown by the dotted line in FIG. 2 is the target of encoding. Since there is no signal, this is a part that is excluded from the target of quantization in the normalization processing by the second-stage normalizers 29 and 30.

As shown in FIG. 2, the portion left by the signal normalization processing by the first-stage normalizer 28 is the standard compared to the case of the conventional example in which the signal normalization processing is performed only one stage. Although it is slightly increased by reducing the amount of normalized data, it is significantly narrowed by the signal normalization processing by the standardizers 29 and 30 that perform the second-stage signal normalization processing. The size is almost the same as when the number of samples N is 4, and the quantization error of each sample value is also the same.

Signal normalization by the second-stage normalizers 29, 30 is performed by the maximum value Xmax of sample values and the minimum value of sample values in one block.
Even if it is fairly rough, such as 2 bits (4 levels) for each Xmin, the portion left by the signal normalization processing by the first-stage normalizer 28 is considerably narrowed, so It is sufficient to perform the standardization to the above degree.

Correspondence between the standardized signal (normalized quantized data) and the standardized signal in the case where the two-step standardization process for the signal to be encoded is performed in the above-described manner. The following table 2 shows the data amount with the signal indicating the standardized state of the signal to be set (data indicating the standardized state).

If the quantized data amount of each sample value is the same as in the case of the conventional method, the difference from the case of the conventional method is the standardized data, and 6 bits each for the first stage of the signal standardization processing, In the second stage of the standardization process, two sets of 2 bits each make 20 bits for 8 samples. As a result, the amount of data per sample is 5.5 bits, which is slightly larger than the case where the number of samples N is 8 in the conventional method.

As is clear from comparing Tables 1 and 2, the high-efficiency coding method of the present invention has a quantization error equivalent to that when a relatively small block is used in the above-described conventional high-efficiency coding method. It has characteristics and can significantly reduce the amount of data required.

(Effect) As is apparent from the above description in detail, the high-efficiency coding method of the present invention belongs to each block obtained by dividing a discrete signal to be coded into constant intervals. With respect to the signal, a means for standardizing the distribution in the amplitude direction of the signal for each of the individual blocks to obtain a standardized signal in a state in which the standardization has been performed for each individual block, and Of the standardized signal obtained in correspondence with each block of the above-mentioned blocks, the individual blocks used for obtaining the above-mentioned standardized signals are further divided into fixed intervals. For the signals belonging to the individual blocks, the distribution of the signals in the amplitude direction is standardized for each of the new individual blocks, and the standardization is performed for each of the new individual blocks. A high-efficiency coding method including means for sequentially repeating at least one or more times to obtain a standardized signal in the above state, and standardization at each of the above-mentioned stages in the high-efficiency coding method The information indicating the standardization state of the signal to be set in correspondence with the completed signal indicates the standardization state of the signal set in correspondence with the standardized signal for each one block at each stage described above. A high-efficiency code in which the total amount of data over all stages is equal to the amount of data indicating the signal normalization state that should be set when the signal is standardized only once. The high-efficiency coding method of the present invention is ideal because the standardization can be performed in accordance with a change in a relatively small range even when the degree of signal change changes in a complicated manner. Naive signal Since there is no wasted area in the standardized sample value distribution, it is possible to reduce the quantization error even if the quantization is relatively simple and the amount of data is small. The improvement effect is remarkable at the part where the signal changes gently and the small change part adjacent to the large change part. In addition, since the error in this case can be reasonably distributed according to the visual characteristics and the auditory characteristics, it is possible to make it difficult to detect even with an equivalent error amount. The quality of the decoded signal can be maintained even if the amount of data to be converted is reduced, and a high data compression rate can be realized.
Furthermore, the standardized sample values can be processed efficiently by a simple quantizer, so that the device can be relatively simple.

[Brief description of drawings]

FIG. 1 is a block diagram of an example configuration of a multi-step signal standardization device used in a high efficiency coding system of the present invention, FIGS. 2 and 3 and FIG. 5 are explanatory diagrams of signal processing, and FIG. FIG. 1 is a block diagram showing an example configuration of a signal normalizer. 1 ... Input terminal of discrete signal to be encoded,
2 to 5 ... delay circuit for giving a signal a delay time equal to the time interval of discrete signals (sampling period used to generate discrete signals), 6 ... maximum value detection circuit, 7 ... minimum value Detecting circuit, 8,9 Quantizing circuit, 10-15, Subtractor, 16-2
0 …… divider, 21,22 …… output terminal for information indicating normalization status (data indicating normalization status), 23 to 27 …… quantized data output terminal, 28 to 30 …… normalization Vessel, 31 ... Quantizer,

Claims (2)

[Claims] 1. A standard of distribution in the amplitude direction of a signal belonging to each block obtained by dividing a discrete signal to be coded into constant intervals. Means for obtaining a standardized signal in a state where standardization is performed for each individual block, and a standardized signal obtained in correspondence with each of the individual blocks described above For the signals belonging to the new individual blocks obtained by dividing each of the individual blocks used to obtain each, in a state of further dividing, for each of the new individual blocks,
Means for sequentially repeating at least one or more times to standardize the distribution in the amplitude direction of each signal and obtain the standardized signal in the state in which the standardization is applied to each of the new individual blocks. And a high-efficiency coding method characterized by 2. A standard of distribution in the amplitude direction of a signal belonging to each block obtained by dividing a discrete signal to be coded into constant intervals. Means for obtaining a standardized signal in a state where standardization is performed for each individual block, and a standardized signal obtained in correspondence with each of the individual blocks described above For the signals belonging to the new individual blocks obtained by dividing each of the individual blocks used to obtain each, in a state of further dividing, for each of the new individual blocks,
Means for sequentially repeating at least one or more times to standardize the distribution in the amplitude direction of each signal and obtain the standardized signal in the state in which the standardization is applied to each of the new individual blocks. In the high-efficiency coding method including the above, the information indicating the standardization state of the signal to be set corresponding to the standardized signal in each of the above-mentioned stages is the standardization of one block in each of the above-mentioned stages. The signal standard that should be set when the signal standardization is performed only once, with the total data amount indicating the standardized state of the signal set corresponding to the completed signal over all stages. High-efficiency coding method that equals the amount of data indicating the coding status