JP3277425B2 - Digital signal encoding method, encoding table generation method, encoding apparatus, and encoding method - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a digital signal encoding method, an encoding table generating method, and an encoding method suitable for transmitting, recording, and reproducing various information such as video and audio. The present invention relates to an encoding device and an encoding method.

[0002]

2. Description of the Related Art Conventionally, various video signal encoding apparatuses have been proposed in order to transmit or record an image at a bit rate as small as possible while maintaining sufficient quality. As one example, a high-efficiency coding apparatus using DCT (discrete cosine transform) and Huffman coding has been proposed. In such an encoding apparatus, data such as video and audio to be encoded is converted into coefficient data from a DC component to a high-order AC component by a discrete cosine transform (DCT) circuit or the like for each block. The data is quantized, and the quantized coefficient data is encoded by a variable length encoding circuit.

[0003] In DCT, a DC component to a high-order AC component of blocked data is converted into two-dimensional coefficient data. Then, the transform coefficients obtained by this transform are scanned (zigzag scan) and encoded as shown in FIG. Encoding is performed according to the order of scanning, and up to the last coefficient that becomes non-zero after quantization is taken as an encoding range.

[0004] Coefficient data within the coding range by the zigzag scan are arranged in one column as shown in FIG. 25, for example, and the coefficient data arranged in one column is converted into a Huffman code as described above in a variable length encoding circuit. It is encoded by a technique such as conversion. This Huffman coding technique is disclosed in
It is disclosed in JP-A-132530.

In FIG. 25, DC is coefficient data of a DC component (in this example, it is “168”), AC0
1 to AC33 are coefficient data of the AC component. The larger the number attached to the coefficient data of the AC component, the higher the AC component. As shown in the figure, the coefficient data DC33 of the higher-order AC component is converted from the coefficient data DC of the DC component.
Are sequentially encoded into codes C1 to Cn. As shown in FIG. 25, when each coefficient data is encoded,
Encoding is performed using consecutive “0” and subsequent values. As a word indicating the number of consecutive “0”,
The word "zero run" is used. For example, "0 run 0"
Indicates that the number of consecutive “0” (0 run) is “0”, that is, there is no “0”.

In FIG. 25, a coefficient AC2 of an AC component
0 is “0”, the following AC11 is “0”, the following AC02 is “0”, and the following AC03 is “4”. In this case, this is called “0 run 3 value 4”. Also,
Since the AC component coefficient AC01 is “120”, in this case, the expression is “0 run 0 value 120”.

Now, the case of encoding the coefficient data shown in FIG. 25 will be described with reference to FIG. FIG.
Is a table for giving codes according to 0 runs and the value following the last "0". For convenience of explanation, FIG. 26 shows only the value of the 0 run shown in FIG. 25 and the coefficient values following the 0 run. The table shown in FIG. 26 is created according to a well-known Huffman code creation procedure.

First, the AC component coefficient AC01 is “12”.
Since it is “0”, as shown in FIG. 26, it becomes “0 run 0, coefficient value“ 120 ””, so that the variable length code to be assigned is, for example, code C1, and the AC component coefficient AC10 is “50”.
Therefore, as shown in FIG. 26, “0 run 0, coefficient value“ 5
0 "", the variable length code to be assigned is, for example, code C2.

The following coefficient AC20 is "0" and the coefficient AC11 is
Is "0", coefficient AC02 is "0", coefficient AC03 is "4", that is, three "0" s are continuous, and the next non-zero "4" is arranged. As shown in FIG. 26, “0 run 3, coefficient value“ 4 ””, and therefore, the variable length code to be assigned is code C3. Similarly, the coefficient AC1
2 is “0”, the coefficient AC21 is “0”, and the coefficient AC30 is “3”. In this case, as shown in FIG.
Run 2 has a coefficient value of “3”, and the variable length code to be assigned is code C4.

Next, since the coefficient AC31 is "1", in this case "0 run 0, value" 1 "" as shown in FIG. 26, the variable length code to be assigned is code C5. Next, since the coefficient AC22 is “0”, the coefficient AC13 is “0”, and the coefficient AC23 is “1”, in this case, “0 run 2, value“ 1 ”” is obtained as shown in FIG. The code to be assigned is code C6. Since the coefficient AC32 is "1", it becomes "0 run 0, value" 1 "" as shown in FIG. 26. Therefore, the code to be assigned is code C7. Similarly, since the coefficient AC33 is "1", it is shown in FIG. Thus, “0 run 0, value“ 1 ”” is obtained, and the code to be assigned is code C8.

As described above, in the conventional encoding apparatus, the coefficient data obtained by the discrete cosine transform is scanned from the DC component to the high-order AC component by zigzag scanning to determine the encoding range of the coefficient data. The coefficient data in the range of encoding is arranged in one column in the order of zigzag scan, and the coefficient data arranged in one column is created according to the Huffman code creation procedure. The encoding is performed by referring to a table consisting of codes corresponding to non-zero values following the.

The applicant of the present invention first performs transform coding such as cosine transform on a two-dimensional image block composed of (n × n) pixels, and converts a DC component obtained by the transform coding into a predetermined bit. , And divides (n 2 −1) AC components into (m × m) (n> m) sub-blocks, and stores address information of sub-blocks having significant data in sub-block units. There has been proposed a data transmission apparatus in which the amount of transmission data can be controlled to be smaller than a predetermined target value by feedforward control by transmitting significant coefficient data in a transmitted sub-block (Japanese Patent Laid-Open No. Hei 2 (1994)). -226
886).

The applicant of the present invention performs transform coding such as cosine transform on a two-dimensional image block composed of (n × n) pixels, and converts a DC component obtained by the transform coding into a predetermined bit. , And the frequency of occurrence in a unit period of one field, one frame, etc. is detected with respect to the data of (n squared -1) AC components, and the frequency is converted into cumulative distribution data, and the cumulative frequency distribution is calculated. A data transmission apparatus has been proposed in which the amount of generated data in a unit period is controlled to be smaller than a target transmission data amount by referring to data (see Japanese Patent Application Laid-Open No. Hei 2-238787).

[0014]

SUMMARY OF THE INVENTION As described above, DC
In transform coding such as T, the inside of a block is scanned in a zigzag manner from a low-order coefficient to a high-order coefficient, and a continuous “0” and a subsequent number other than “0” are paired to uniquely determine a code. Had been assigned. Even in sub-band coding,
In general, a code that is uniquely determined is assigned to a set of consecutive “0” s in a band or between bands and a coefficient subsequent thereto.

In these methods, when the compression rate is high, "0" s are continuous, and encoding can be performed efficiently. However, when the compression rate is low, "0" decreases, so that encoding is performed efficiently. Can not do it. That is, in the above-described transform coding in which codes are uniquely assigned to coefficient data, there is a disadvantage that coding efficiency cannot be improved.

The present invention has been made to solve such a problem, and an object of the present invention is to propose an encoding apparatus capable of greatly improving encoding efficiency.

[0017]

According to the present invention, there is provided a method for encoding a digital signal, comprising the steps of:
Step (a) of converting to a data sequence that is continuous with non-zero coefficients
(B) obtaining, from the data sequence, a plurality of events consisting of a series of zero coefficients having a predetermined run length and at least one non-zero coefficient before or after the series of zero coefficients; (C) setting one code word for a plurality of events based on table data that gives a different code word to each event in accordance with each event.

Further, the encoding table generating method of the present invention comprises:
(A) converting a predetermined digital signal into a continuous data sequence consisting of continuous zero coefficients and non-zero coefficients; and, from the data sequence, a series of zero coefficients having a predetermined run length and a series of zero coefficients. Obtaining a plurality of events consisting of at least one non-zero coefficient connected before or after (b)
(C) detecting, for each of a plurality of events, the frequency of occurrence of each event appearing next to the event; and, in accordance with the frequency of occurrence of each event, changing a desired data sequence to the state of the previous data sequence. (D) of generating table data to be used when encoding according to the above and storing the table data in a memory.

The method of encoding a digital signal according to the present invention further comprises the step of: (a) converting a predetermined digital signal into a data string that is continuous with continuous zero coefficients and non-zero coefficients;
From the data sequence, a series of zero coefficients having a predetermined run length and at least one of the series of zero coefficients before or after the series of zero coefficients.
Obtaining a plurality of events having two non-zero coefficients (b)
(C) detecting, for each of the plurality of events, the frequency of occurrence of each event appearing next to the event, and a desired data string according to the frequency of occurrence of each event obtained in step (c). (D) generating table data to be used when encoding is performed in accordance with the state of the previous data string and storing the table data in a memory; and converting an arbitrary digital signal into continuous zero coefficients and non-zero coefficients. Step (e) of converting into a continuous data string, step (f) of obtaining each event from the data string obtained in step (e), and a plurality of events obtained in step (f) based on the table data. (G) determining one codeword.

The encoding apparatus according to the present invention further comprises conversion means 2 and 3 for converting a digital signal into a data sequence that is continuous with continuous zero coefficients and non-zero coefficients, and a series of zero coefficients having a predetermined run length. Storage means 4a for storing, for a plurality of events consisting of at least one non-zero coefficient before or after the series of zero coefficients, a plurality of table data representing code words to be given to the following events, respectively. And encoding means 4 which obtains a plurality of events from the data sequence and gives one code word to each of the plurality of events based on the table data.

The digital signal encoding method according to the present invention further comprises the steps of: (a) converting a predetermined digital signal into a data string that is continuous with continuous zero coefficients and non-zero coefficients;
(B) classifying the non-zero coefficients into a plurality of groups according to their values; and, from the data sequence, a series of zero coefficients having a predetermined run length, and at least one continuous series before or after the series of zero coefficients. (C) obtaining a plurality of events having non-zero coefficients, and for each event, for a plurality of events, based on table data giving a codeword for each group in the immediately preceding event, It comprises the step (d) of determining one code word each.

Further, in the digital signal encoding method of the present invention, in step (d), the order information in the group is further added to the code word in the step (d).

The method of encoding a digital signal according to the present invention further comprises the steps of: (a) converting a predetermined digital signal into a data string that is continuous with continuous zero coefficients and non-zero coefficients; (B) obtaining a plurality of events consisting of a zero coefficient having at least one non-zero coefficient before or after the continuation of the zero coefficient, and classifying the non-zero coefficients into a plurality of groups according to their values. (C), detecting the occurrence frequency of each event appearing next to each group (d), and changing the desired data sequence to the state of the previous data sequence according to the occurrence frequency of each event. (E) generating table data to be used when encoding according to the data and storing the data in a memory; and converting an arbitrary digital signal into a continuous data sequence of continuous zero coefficients and non-zero coefficients. (F), a step (g) for obtaining each event from the data sequence obtained in the step (f), and a plurality of events obtained in the step (g) based on the table data. (H) determining a codeword.

The encoding apparatus according to the present invention further comprises conversion means 2 and 3 for converting a digital signal into a continuous data sequence of continuous zero coefficients and non-zero coefficients, and a plurality of non-zero coefficients corresponding to the values. Classifying means 4 for classifying a plurality of events having a continuous run of zero coefficients having a predetermined run length and at least one non-zero coefficient connected before or after the run of zero coefficients. A storage means 4a for storing a plurality of table data representing codewords to be given to the next event in response to the plurality of events from a data string, And encoding means 4 for giving one code word each.

Further, the encoding apparatus of the present invention includes conversion means 2 and 3 for converting a digital signal into a data stream in a predetermined form, and a plurality of events in the data stream in a predetermined form.
And a plurality of codewords to be given to the next event
Formed by storing the table data storing means 4a, Day
Multiple events are obtained from the data sequence, and multiple events are
It comprises encoding means 4 for giving one code word to each of a number of events . The encoding method of the present invention
Conversion of a digital signal into a data stream of a predetermined format
Steps and multiple events in a given sequence of data
Represents the code word given to the next event
A storage step for storing a plurality of table data;
Obtain multiple events from a data string and
, A code that gives one codeword for each of multiple events
It consists of an encryption step.

[0026]

According to the digital signal encoding method of the present invention described above, in step (a), the digital signal is converted into a continuous data sequence consisting of continuous zero coefficients and non-zero coefficients. From the sequence, a plurality of events consisting of a series of zero coefficients having a predetermined run length and at least one non-zero coefficient before or after the series of zero coefficients are obtained. , One code word is set for a plurality of events based on table data that gives a different code word to each event.

According to the encoding table generating method of the present invention described above, in step (a), the predetermined digital signal is converted into a continuous data sequence of continuous zero coefficients and non-zero coefficients. )
Obtaining a plurality of events consisting of a series of zero coefficients having a predetermined run length and at least one non-zero coefficient before or after the series of zero coefficients, and in step (c), for each of the plurality of events, A table used to detect the occurrence frequency of each event appearing next to the event and to encode a desired data sequence in step (d) according to the occurrence frequency of each event according to the state of the previous data sequence Generate data and store it in memory.

According to the above-described digital signal encoding method of the present invention, in step (a), a predetermined digital signal is converted into a continuous data sequence of continuous zero coefficients and non-zero coefficients. In step (c), a plurality of events consisting of a series of zero coefficients having a predetermined run length and at least one non-zero coefficient before or after the series of zero coefficients are obtained. , The occurrence frequency of each event appearing next to the event is detected, and in step (d), a desired data string is replaced with the previous data string according to the occurrence frequency of each event obtained in step (c). Table data to be used for encoding according to the state is generated and stored in a memory, and in step (e), an arbitrary digital signal is linked with a continuous zero coefficient and a non-zero coefficient. In step (f), each event is obtained from the data string obtained in step (e), and in step (g), a plurality of events obtained in step (f) are obtained based on the table data. On the other hand, one codeword is determined.

According to the configuration of the encoding apparatus of the present invention described above, the conversion means 2 and 3 convert the digital signal into a continuous data sequence of continuous zero coefficients and non-zero coefficients, and a predetermined run length is obtained. For a plurality of events consisting of a series of zero coefficients and at least one non-zero coefficient preceding or following the series of zero coefficients, a plurality of table data representing a code word to be given to each of the following events is stored. Means 4a, and a plurality of events are obtained from the data sequence by the encoding means 4,
One codeword is assigned to each of a plurality of events based on the table data.

According to the digital signal encoding method of the present invention described above, in step (a), the predetermined digital signal is converted into a continuous data sequence consisting of continuous zero coefficients and non-zero coefficients. ), The non-zero coefficient is
According to the value, the data is classified into a plurality of groups. In step (c), the data sequence is divided into a series of zero coefficients having a predetermined run length and at least one non-zero coefficient connected before or after the series of zero coefficients. Multiple events,
In step (d), one codeword is determined for each of a plurality of events based on table data that provides a codeword for each group in the previous event for each event.

Further, according to the digital signal encoding method of the present invention described above, in step (d), the order information in the group is further added to the code word.

According to the above-described digital signal encoding method of the present invention, in step (a), a predetermined digital signal is converted into a continuous data sequence of continuous zero coefficients and non-zero coefficients. ), From the data sequence, a plurality of events consisting of a zero coefficient having a predetermined run length and at least one non-zero coefficient before or after the continuation of the zero coefficient are obtained. In step (c), the non-zero coefficient is obtained by: In step (d), the frequency of occurrence of each event that appears next is detected for each group, and in step (e), a desired frequency is determined according to the frequency of occurrence of each event. Table data to be used when encoding the data sequence according to the state of the previous data sequence is generated and stored in the memory. In step (f), an arbitrary digital signal is In the step (g), each event is obtained from the data string obtained in the step (f). In the step (h), based on the table data, Step (g)
One code word is determined for each of the plurality of events obtained in step (1).

Further, according to the configuration of the encoding apparatus of the present invention described above, the converting means 2 and 3 convert the digital signal into a continuous data sequence of continuous zero coefficients and non-zero coefficients, The zero coefficients are classified into a plurality of groups according to their values, and are classified into a plurality of events including a series of zero coefficients having a predetermined run length and at least one non-zero coefficient before or after the series of zero coefficients. On the other hand, a plurality of table data representing a code word to be given to the next event according to each group is stored in the storage unit 4a, and a plurality of events are obtained from the data sequence by the encoding unit 4 to obtain the table data. , One code word is assigned to each of a plurality of events.

Further, according to the configuration of the encoding apparatus of the present invention described above, the conversion means 2 and 3 convert the digital signal into a data stream of a predetermined form and convert the digital signal into a data stream of a predetermined form.
Given to multiple events that occur next
Serial multiple table data representing the codeword storage unit 4a
The encoding means 4 obtains a plurality of events from the data sequence,
One for each of multiple events based on table data
Gives the code word of The structure of the encoding method of the present invention described above is also described.
According to the configuration, a digital signal is converted into a data stream of a predetermined form.
Into multiple events in the data stream of the specified form.
On the other hand, a compound representing the code word given to the next event
Stores a number of table data and allows multiple events
And based on the table data, for multiple events,
Each gives one codeword.

[0035]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of a digital signal encoding method, an encoding table generating method, an encoding apparatus and an encoding method according to the present invention will be described below in detail with reference to FIGS.

First, an encoding method, an encoding device, and a high-efficiency encoding device to which the encoding method of the present invention is applied will be described with reference to FIG. FIG. 1A shows an encoder of the high-efficiency encoder, and FIG. 1B shows a decoder of the high-efficiency encoder.

First, the encoder will be described with reference to FIG. 1A. In FIG. 1A, 1 is a digital V
An input terminal to which image data or the like to be converted from an electronic device main body circuit such as a TR is supplied. Image data is supplied to the conversion circuit 2 via the input terminal 1.

The transform circuit 2 is, for example, a discrete cosine transform (DCT) circuit, and performs a discrete transform process on the image data supplied via the input terminal 1 for each block using only the cosine function. Next, coefficient data from a DC component to a higher-order AC component is obtained. As the orthogonal transform considered here, for example, a Fourier transform, a Hadamard transform,
The Karhunen-Loeve transform is conceivable, but the Fourier transform involves complex arithmetic and the configuration is complicated. Further, as a conversion method other than the orthogonal transformation, for example, a gradient transformation (slant transformation) or a Haar transformation can be considered. In this example,
A DCT that has a high-speed operation algorithm, can be realized by a one-chip LSI that enables real-time conversion of an image, and has excellent power concentration on low-frequency components that directly affects coding efficiency is used.

The image data is converted by the conversion circuit 2 into coefficient data and supplied to the quantization circuit 3. The quantization circuit 3 quantizes the coefficient data from the conversion circuit 2. The quantized coefficient data is supplied to the variable length coding circuit 4. The variable length coding circuit 4 scans (zigzag scans) the coefficient data from the quantization circuit 3 as described with reference to FIG. Then, the coefficient data within the coding range is arranged in one line as shown in FIG. 4, for example, and the coefficient data arranged in one line is encoded by the Huffman coding technique.

In this example, a plurality of tables corresponding to the condition series are prepared in order to encode coefficient data arranged in one column, and when a certain coefficient data is to be encoded, the coefficient before the coefficient data is set. A table corresponding to the data condition is selected, and encoding is performed with reference to the selected table. In the transform coding, since the coefficients after the transform have a correlation with each other, the coding efficiency can be greatly improved by performing the coding using the coefficients.

FIG. 2 is for creating such a table, and is a Huffman table generating apparatus to which the encoding table generating method of the present invention is applied.

In FIG. 2, reference numeral 11 denotes an input terminal to which video signals having the same contents are supplied, for example, in a predetermined cycle. An external video signal is supplied to the conversion circuit 12 via the input terminal 11. The conversion circuit 12 performs a conversion process such as DCT (Discrete Cosine Transform) on a video signal supplied from the outside via the input terminal 11, and supplies the converted video signal to the quantization circuit 13.

The quantization circuit 13 quantizes the conversion output from the conversion circuit 12, that is, the coefficient data, and supplies it to the Huffman coding circuit 14. The Huffman encoding circuit 14 performs Huffman encoding on the quantized coefficient data from the quantization circuit 13 with reference to a table stored in the memory 16, and supplies the total code length data to the quantization circuit 13. The total code length data is for changing the roughness of the quantization in the quantization circuit 13. Here, although not shown, the memory 16 is a read / write circuit and, for example, a DRAM (dynamic RAM) or an SRAM (static RAM).
And the like.

On the other hand, the coefficient data from the quantization circuit 13 is also supplied to the occurrence frequency detection circuit 15. The occurrence frequency detection circuit 15 detects the occurrence frequency of the coefficient data from the quantization circuit 13, generates a table for performing Huffman coding based on the detection result, and supplies the generated table data to the memory 16.

Next, the operation of the Huffman table generation device shown in FIG. 2 will be described. For example, a video signal of a predetermined length for table creation is repeatedly supplied to the input terminal 11. In addition, a table randomly selected as an initial value is previously loaded from the occurrence frequency detection circuit 15 into the memory 16.

The quantizing circuit 13 quantizes the table-producing video signal of a predetermined length from the input terminal 11 by, for example, the roughness by the initial control from the Huffman encoding circuit 14, and quantizes the coefficient data into the Huffman encoding circuit 14. And to the occurrence frequency detection circuit 15 respectively. The Huffman coding circuit 14 codes the coefficient data according to a table loaded as an initial value from the memory 16.

On the other hand, the occurrence frequency detection circuit 15 detects the occurrence frequency of the coefficient data from the quantization circuit 13, creates a new table based on the detection result, and loads the created new table into the memory 16. .

The condition series referred to here is “0 run, value (coefficient value)”. In addition, the detection of the occurrence frequency as referred to herein means finding the occurrence probability of a set of 0 run and value for each condition series. Then, using this occurrence probability, a set of codes for each condition sequence is created according to the Huffman method. As a well-known method, a symbol (a set of 0 run and value) is used.
Are arranged in ascending order of their occurrence probabilities, and “1” and “0” are arbitrarily assigned to the symbol having the smallest appearance probability and the symbol having the second smallest occurrence probability. The number of symbols is reduced by one by taking the occurrence probabilities of the determined symbols as the sum of the occurrence probabilities before merging, and this is considered as a new information source, and the symbols are rearranged again in descending order of occurrence probability. The process is repeated until the symbol becomes "1", and when the symbol becomes "1", "1" and "0" assigned to the symbol for each symbol are read out in reverse order, and this is read out for that symbol. Code word.

Generally, the absolute value of a coefficient in a DCT block tends to be large when one is large, the other coefficient is also large when one is small, and the other coefficient is also small when one is small. That is, the appearance frequency of the set of 0 run and value also changes depending on the condition series, and by using this to assign a variable length code, the entire code length can be further reduced.

When this new table is loaded into the memory 16, the Huffman encoding circuit 14 encodes the coefficient data from the quantization circuit 13 with reference to the table, and encodes the total code length data into the quantization circuit 13. To change the roughness of the quantization in the quantization circuit 13. In the same manner, encoding is performed based on the changed table with respect to the table creation data of a predetermined length, and the total code length data obtained by this encoding is supplied to the quantization circuit 13 and the quantization roughness is calculated. Is changed, the frequency of occurrence is detected, a new table is created, and encoding is performed using this table. By repeating this operation, a large number of encoding tables corresponding to the above-described condition series are created.

The created table data is stored in the variable length coding circuit 4 of the encoder of the high efficiency coding apparatus shown in FIG. 1A.
, For example, an EEPROM, a RAM with a backup function, a PROM, a one-time ROM, and the like. As a method of storing the data, a method of writing table data by a device for writing data to the memory 4a and mounting the memory 4a in which the table data is written on the variable length coding circuit 4 and a method of generating the Huffman table shown in FIG. To load the table data into the memory 4a of the variable-length coding circuit 4 from.

FIG. 3 is a configuration diagram showing a specific configuration example of the Huffman table generation device shown in FIG. 3, parts corresponding to those in FIG. 2 are denoted by the same reference numerals, and detailed description thereof will be omitted.

In the figure, reference numeral 17 denotes a control unit 17a and RO
The controller 17 is composed of M17b, and the ROM 17b of the controller 17 stores the table data as the above-described default value. For example, when the power is turned on or when a reset switch (not shown) is pressed, the control unit 17a supplies a switch control signal to the switch 18 and sets the movable contact 18c of the switch 18 to one fixed contact 1
8a.

The movable contact 18c of the switch 18 is connected to the input terminal of the Huffman encoding circuit 14, and the other fixed contact 18b of the switch 18 is connected to the output terminal of the memory 16. Therefore, the movable contact 18 of the switch 18c
c is connected to one fixed contact 18a, the ROM 17
b is loaded into the Huffman encoding circuit 14 via the switch 18, and when the movable contact 18c of the switch 18 is connected to the other fixed contact 18b, the table data is stored in the memory 16. The generated table data is stored in the Huffman encoding circuit 14.
Is loaded.

When the power is turned on or reset, the controller 17a of the controller 17 connects the movable contact 18c of the switch 18 to one of the fixed contacts 18a. It is supplied to the encoding circuit 14. The Huffman encoding circuit 14 encodes the coefficient data from the quantization circuit 13 with reference to the supplied table, and supplies the total code length data to the quantization circuit 13 so that the roughness of the quantization in the quantization circuit 13 is reduced. To change.

When the new table is loaded into the memory 16, the control unit 17a operates the movable contact 1 of the switch 18.
Since 8c is connected to the other fixed contact 18b, new table data from the memory 16 is supplied to the Huffman encoding circuit 14. The Huffman encoding circuit 14 encodes the coefficient data from the quantization circuit 13 with reference to the table and supplies the total code length data to the quantization circuit 13 to change the quantization roughness in the quantization circuit 13. Let it.
In the same manner, encoding is performed based on the changed table with respect to the table creation data of a predetermined length, and the total code length data obtained by this encoding is supplied to the quantization circuit 13 and the quantization roughness is calculated. Is changed, the frequency of occurrence is detected, a new table is created, and encoding is performed using this table. By repeating this operation, a large number of encoding tables corresponding to the above-described condition series are created. The created table data is
It is stored in the writable memory 4a of the variable length coding circuit 4 of the encoder of the high efficiency coding device shown in FIG. 1A.

As a condition considered when the table data is created as described above, a value of 0 run and a value of non-zero following it can be considered. As an example, a case will be described in which coefficient data arranged in one row shown in FIG. 4 is encoded with reference to a table corresponding to a condition series created as described above.

In FIG. 4, DC is coefficient data of a direct current component (“168” in the example of this figure), and AC 01 to A
C33 is the coefficient data of the AC component. The larger the number attached to the coefficient data of the AC component, the higher the AC component. As shown in this figure, the components from the DC component coefficient data DC to the higher-order AC component coefficient data AC33 are sequentially encoded into codes 1 to n. Further, as shown in FIG. 4, when encoding each coefficient data, as described above, encoding is performed in accordance with a continuous "0" and a value subsequent thereto. 24, the word "0 run" is used as the word indicating the number of consecutive "0". For example, when "0 run 0" is described, the number of consecutive "0" (0 run) is "0", that is, "0".
Indicates that there is no

In FIG. 4, the AC component coefficient AC22
Is "0", the subsequent AC13 is "0", and the subsequent AC23 is "1". In this case, "0 run 2, value 1" is obtained. Further, since the AC component coefficient AC01 is “120”, in this case, it is “0 run 0, value 120”.

Next, the coefficient data AC10 shown in FIG.
Will be described with reference to FIG. FIG. 5 shows a table of “0 run 0, sign of value“ 120 ”” which is a condition series of AC10. That is, as shown in FIG. 5, the table assigns a code corresponding to the value of the 0 run and the value of the coefficient data for each condition series. Here, the difference from the prior art is that, in this example, a large number of tables are created in accordance with the Huffman code creation procedure according to the condition series, and when encoding certain coefficient data, That is, a corresponding table is selected from many tables based on the condition of the coefficient data, and the coefficient data is encoded with reference to the selected table.

FIG. 6 shows a table created based on the condition series described with reference to FIG. 5 selected by a series of coefficient data before the coefficient data to be coded, and referring to the selected table. FIG. 5 is an explanatory diagram showing a case where the conversion is performed based on the example shown in FIG. 4.

As shown in FIGS. 6 and 4, first, since the coefficient AC01 having the value "120" is the head of the AC component, there is no condition series. Therefore, for the coefficient at the head of the AC component, the variable length code 1 is obtained from the table (the table shown in FIG. 26) in which the code is given to the current value, that is, the value of the 0 run and the value of the coefficient. To do.

Subsequently, the AC component coefficient AC10 is "50".
However, since the coefficient before this coefficient AC10 is AC01 and its value is “0 run 0, coefficient value“ 120 ””, a table having this as a conditional sequence is selected from a plurality of tables. The variable length code 2 is given to the value “50” of the coefficient AC10 with reference to the table.

The following coefficient AC20 is "10". The coefficient before this coefficient AC20 is AC10, and its value is "0 run 0, coefficient value" 50 "". A table as a condition series is selected, and a variable length code 3 is given to the value “10” of the coefficient AC20 with reference to the table.

The following coefficient AC11 is "20". The coefficient before this coefficient AC11 is AC20, and its value is "0 run 0, value" 10 "". A table to be a series is selected, and a variable length code 4 is given to the value “20” of the coefficient AC11 with reference to the table.

The following coefficient AC02 is "30". The coefficient before this coefficient AC02 is AC11, and its value is "0 run 0, value" 20 "". A table to be a series is selected, and a variable length code 5 is given to the value “30” of the coefficient AC02 with reference to the table.

The following coefficient AC03 is "41", but the coefficient before this coefficient AC03 is AC02, and its value is "0 run 0, value" 30 "". A table to be a series is selected, and a variable length code 6 is given to the value “41” of the coefficient AC03 with reference to the table.

The following coefficient AC12 is "10". The coefficient before this coefficient AC12 is AC03, and its value is "0 run 0, value" 41 "". A table to be a series is selected, and a variable length code 7 is given to the value “10” of the coefficient AC12 with reference to the table.

Next, the coefficient AC21 is "10",
Since the coefficient before this coefficient AC21 is AC12 and its value is “0 run 0, value“ 10 ””, a table having this as a conditional sequence is selected from a plurality of tables, and the table is referred to. The variable length code 8 is given to the value “10” of the coefficient AC21.

Next, although the coefficient AC30 is "3", the coefficient preceding this coefficient AC30 is AC21, whose value is "0 run 0, value" 10 "". Is selected as a condition series, and the variable length code 9 is set to the value “3” of the coefficient AC30 with reference to the table.
give.

Next, although the coefficient AC31 is "1", the coefficient preceding this coefficient AC31 is AC30, and its value is "0 run 0, value" 3 "". Is selected as a condition series, and the variable length code 10 is set to the value “1” of the coefficient AC31 with reference to the table.
give.

Next, the coefficient AC22 is "0", and the following coefficient A
C13 is "0" and the following coefficient AC23 is "1",
Since the coefficient before the coefficient AC22 is AC31, and its value is “0 run 0, value“ 1 ””, a table having this as a conditional sequence is selected from a plurality of tables, and the table is referred to and Coefficient AC22 value “0”, coefficient AC
The variable length code 11 is given to the value “0” of 13 and the value “1” of the coefficient AC23.

Next, the coefficient AC32 is "1", and the coefficients before this coefficient AC32 are AC22, AC13, AC13.
23, and its value is “0 run 0, value“ 1 ””. Therefore, a table having this as a condition series is selected from a plurality of tables, and the table is referred to the value “1” of the coefficient AC32. A variable length code 12 is provided.

Next, although the coefficient AC33 is "1", the coefficient preceding this coefficient AC33 is AC32, and its value is "0 run 0, value" 1 "". Is selected as the condition series, and the variable length code 13 is set to the value “1” of the coefficient AC33 with reference to the table.
give.

Now, the encoder for encoding as described above has been described. Next, the decoder of the high efficiency encoding apparatus will be described with reference to FIG. 1B.

In FIG. 1B, reference numeral 6 denotes an input terminal to which, for example, a transmission path (not shown) or transmission data from an output side of an electronic device (not shown) or reproduction / readout data is supplied. These various data are supplied to the variable length code decoding circuit 7. The variable length code decoding circuit 7 decodes the variable length code data supplied via the input terminal 6 to obtain quantized coefficient data, and supplies the coefficient data to the quantization restoration circuit 8.

The quantization restoration circuit 8 restores the coefficient data from the variable length code decoding circuit 7, obtains the coefficient data before quantization, and supplies the coefficient data to the inverse transformation circuit 9. The inverse transform circuit 9 performs the inverse transform such as the discrete cosine transform and the Hadamard transform described for the transform circuit 2, and outputs the original data obtained by performing the inverse transform via the output terminal 10 to the high efficiency encoding apparatus. Is supplied to the reproduction system of the electronic device equipped with.

When the variable-length code data is decoded by the variable-length code decoding circuit 7, that is, when "0 run, value" is coded as a conditional sequence, for example, the coding table is used as it is. A table for obtaining the original coefficient data from the code data (corresponding to the respective coefficient data to the code data, provided that the data indicating which table is referred to and coded is the coefficient data coded at the time of coding) Must be added to the variable-length code decoding circuit 7.

As described above, in the present embodiment, a table is created by detecting the frequency of occurrence of the coefficient data quantized by the quantization circuit 13 using the Huffman table generator, and Huffman coding is performed based on the created table. An operation of encoding by the circuit 14 and supplying the total code length data at that time to the quantization circuit 13 to control the roughness of the quantization is repeated, and a large number of tables based on the condition series (0 run, value) are obtained. When the created table is loaded into the memory 4a of the variable length encoding circuit 4 of the encoder of the high efficiency encoding device, and the coefficient data is encoded by the high efficiency encoding device, the coefficient data before the coefficient data is encoded. Since the table corresponding to the condition series is selected and the encoding is performed with reference to the selected table, the entire code length can be reduced.

In the above example, FIG. 2 or FIG.
Has been described in which a plurality of tables based on the condition series generated by the Huffman table generating apparatus shown in FIG. 1 are written or loaded into the memory 4a of the variable length coding circuit 4 of the high efficiency coding apparatus shown in FIG. 2 and subsequent circuits of the quantization circuit 13 shown in FIGS.
, The Huffman encoding circuit 14 and the occurrence frequency detection circuit 1
5, the memory 16, the Huffman encoding circuit 1 in the case of FIG.
4. The occurrence frequency detection circuit 15, the memory 16, the controller 17, and the switch 18 may be built in the variable length coding circuit 4. In that case, the memory 4a shown in FIG. 1 can be shared with the memory 16 shown in FIG. 2 or FIG.

Next, a case where a condition series (size) other than the above-described condition series “0 run, value” is used will be described with reference to FIGS. 7 to 23. The “size” is a value of the exponent of the power when the value of the coefficient is represented by a power of 2, and is used as an index when the coefficient is classified according to the value. However, the 2 9th power is 256, but the maximum value when representing this with 9 bits is 255. Therefore, when actually classifying by size, the exponent of the power + 1 when the value of the coefficient is represented by a power of 2 +
Becomes

FIG. 7 shows an example in which coefficient data is classified based on the concept called the size. As shown in FIG. 7, the coefficient value “0” is the size 0 (zero power) and the coefficient value “−”.
1 ”and the coefficient value“ 1 ”are size 1 (the first power),.
The coefficient value “−1023 to −512” and the coefficient value “5
12 to 1023 ”are size 10 (10th power)... Coefficient values“ −8191 to −4096 ”and“ 4096 to
8191 "is classified as size 13 (13th power).

In the method of arranging coefficient data in each size, coefficient data having larger values are arranged in order from the left in FIG. How to classify,
It is assumed that the table shown in FIG. 7 is provided in the Huffman encoding circuit 14 shown in FIGS. 2 and 3, for example. Also, the sizes 0 to 13 shown in the figure indicate exponents of the respective powers, and coefficient data having a value within the range is assigned to these sizes. For example, if the value of the coefficient data is “50”, the “50” has a power exponent of “6”, that is, a range of “64” in terms of size. Therefore, as shown in FIG. 7, the coefficient data having the value “50” has a size of 6.

In this example, the values of the coefficient data are classified by size, a number of tables corresponding to the sizes of the coefficient data are prepared, and when encoding the coefficient data, the size of the previous coefficient data is reduced. And a Huffman code is given to the coefficient data to be encoded by referring to the selected table. In the example described above, “0 run, value” is used as the condition series, but in this example, the size is used as the condition series. However, it is the same in that a table corresponding to the condition series of the previous coefficient data is selected, and encoding is performed with reference to the selected table.

Here, the operation of creating a table corresponding to a condition series (the size of the previous coefficient data) using the Huffman table generation device shown in FIGS. 2 and 3 will be described with reference to the flowchart of FIG. . For the leading coefficient of the AC component, the size value is obtained by using the table (see FIG. 7) for obtaining the size value from the coefficient values already described.

First, in step S1, "0" is substituted for a variable J indicating the value of the size of the previous coefficient which is a condition series. Then, control goes to a step S2. That is, the occurrence frequency detection circuit 15 shown in FIGS. 2 and 3 sets the variable J to “0”.

In step S2, it is determined whether or not the variable J ≦ the maximum value of the size of the coefficient immediately before the condition series. If “YES”, the process proceeds to step S3, where “N” is determined.
If "O", the table creation processing ends.

In step S3, for the subset whose previous size is J, the occurrence probability of the current size in the subset is determined. Then, control goes to a step S4.

In step S4, the sizes are checked in descending order of occurrence probability. Then, control goes to a step S5.

In step S5, the smallest occurrence probability and the next smallest occurrence probability are degenerated, and a new size is assigned. The occurrence probability is summed, and the sign bit “0” is assigned to the smaller one and the sign bit is assigned to the larger one. Add bit “1”. Then, control goes to a step S6.

In step S6, it is determined whether or not there is only one with probability 1. If "YES", the flow shifts to step S7, and if "NO", the flow shifts to step S4 again. Here, “there is only one with a probability of 1” is a state in which all have been reduced to one as a result of repeating the work of reducing the occurrence of the minimum occurrence probability and the next one to one. .

In step S7, the Huffman code is obtained under the condition that the size of the previous coefficient which is the condition series is the variable J. And step S
Move to 8. In other words, the Huffman code for each of the current coefficient sizes where the condition series is J can be obtained by reading the assigned bits of “1” and “0” in reverse during degeneration.

In step S8, 1 is added to the variable J.
Then, the process returns to step S2. That is, a code table for the next condition series is obtained.

In the above example, statistics of all data may be obtained by ignoring the condition series for the leading coefficient of the AC component.

Next, with reference to FIG. 4 again, encoding when the size of the condition series is described. In the following description, only the coefficient data shown in FIG.
Reference numerals 1 to 13 and arrows corresponding to the coefficient data and frames surrounding the values of the coefficient data are not referred to.

First, the coefficient AC01 which is the head of the AC component
Is "120" and the size is "7" from the table shown in FIG. By the way, as described in the example in which “0 run, value” described in one embodiment is a condition series,
This is the head of the AC component. Therefore, as shown in FIG. 23, the leading coefficient is assigned a variable length code by referring to a table consisting of the current coefficient size and the variable length code assigned according to the size. As shown in FIG. 23, this table is a table including the current coefficient size and the variable length code assigned to the size, and is a table for obtaining the variable length code from the first coefficient of the AC component. .

That is, the coefficient A which is the head of the AC component
Since C01 is "120", the size "7" is given from the table shown in FIG. 7, and since it is the head of the AC component, it corresponds to the current coefficient size "7" from the table shown in FIG. The obtained variable length code “111110” is given. Here, data indicating the number of “120” in the table shown in FIG. 7 is further added to this code.

Here, the data indicating the number is represented by 0
The first, second, and so on are counted from the 0th. The bit length of the data indicating the number is represented by the same bit length as the value of the size. For example, in the case of the coefficient corresponding to the size “7”, the bit length of the data indicating the number from the size “7” is 7 bits. Here, the fact that the order of the coefficient is counted from "0" means that the data indicating the order by the bit length of the numerical value indicated by the size can be represented for all the coefficients within the size range.

Now, the data indicating the number is actually described. The size of the value “120” of the coefficient AC01 is “7” and the 0th of the size “7” is “−127”. 120 "is -127, -126, ...
−64, 64, 65,... 120, and in this case, the 120th, that is, “11110”
00, the encoded output of the value “120” of the coefficient AC01 is “1111110 1111000”.

Subsequently, the AC component coefficient AC10 is set to "5
0 "is a, the size of the previous coefficient AC01 coefficient AC10 is"", the size from the table shown in FIG. 7" 6 "because it is, shown in FIG. 1 6, the size of the previous coefficient" 7 7 And a variable length code "01" corresponding to the current coefficient size "6" of the table is given.
Here, data indicating the number of “50” in the table shown in FIG. 7 is added to this code. “5
Since “0” has a size of “6” and the 0th of the size “6” is “−63”, “50” is −63, −62,.
...- 32, 32, 33, ..., 63, so in this case it is the 50th.

The subsequent coefficient AC20 is "10" and the size is "4" from the table shown in FIG.
Since the size of the coefficient AC10 before 20 is “6”,
1 5, select the table of the previous size of the coefficient "6", giving the variable length code "00" corresponding to the size "4" of the now coefficient of that table. Here, this code:
Further, in the table shown in FIG. 7, "10" adds data indicating the number. Since “10” has a size of “4” and the 0th of the size “4” is “−15”,
"10" is -15, -14, ... -8, 8, 9, ...
.., 15 in this case, which is the tenth.

The following coefficient AC11 is "20" and the size is "5" from the table shown in FIG.
Since the size of the coefficient AC20 before “11” is “4”,
1 3, select the table of the previous size of the coefficient "4" gives the variable length code "110" corresponding to the size "5" of the current coefficients of the table. Here, data indicating the number of “20” in the table shown in FIG. 7 is further added to this code. Since “20” has a size of “5” and the 0th of the size “5” is “−31”, “20” is −31, −30,.
Since it is counted as 16, 17,... 31, it is the 20th in this case.

The subsequent coefficient AC02 is "30" and the size is "5" from the table shown in FIG.
Since the size of the coefficient AC11 before “02” is “5”,
1 4, select the table of the previous size of the coefficient "5", giving the variable length code "10" corresponding to the size "5" of the current coefficients of the table. Here, this code:
Further, in the table shown in FIG. 7, "30" adds data indicating the number. “30” has a size of “5”
And the 0th of the size “5” is “−31”, so “30” is −31, −30,.
.., 31, so that in this case it is the 30th.

The following coefficient AC03 is "41" and the size is "6" from the table shown in FIG.
Since the size of the coefficient AC02 before “03” is “5”,
1 4, select the table of the previous size of the coefficient "5", giving the variable length code "111110" corresponding to the size "6" of the current coefficients of the table. Here, in the table shown in FIG.
Appends data indicating the number. Since “41” has a size of “6” and the 0th of the size “6” is “−63”, “63” is −63, −62,.
63, so that in this case it is the 41st.

The following coefficient AC12 is "10" and the size is "4" from the table shown in FIG.
Since the size of the coefficient AC03 before “12” is “6”,
1 5, select the table of the previous size of the coefficient "6", giving the variable length code "00" corresponding to the size "4" of the now coefficient of that table. Here, this code:
Further, in the table shown in FIG. 7, "10" adds data indicating the number. "10" means size "4"
And the 0th of the size “4” is “−15”, so “10” is “−15, −14,... -8, 8,
In this case, it is the tenth since it is counted as 9,...

Next, the coefficient AC21 is "10" and the size is "4" from the table shown in FIG.
Since the size of the previous coefficient AC12 of C21 is "4", 1 3, select the table of the previous size of the coefficient "4", corresponding to the size "4" of the now coefficient of the table A variable length code "10" is given. Here, in the table shown in FIG. 7, "10" is added to the code to indicate the number. In this case as well, it is the tenth as described above.

Next, the coefficient AC30 is "3" and the size is as shown in FIG.
Is "2" from the table shown in FIG.
Since the size of the previous coefficient AC21 0 is "4", 1 3, select the table of the previous size of the coefficient "4", corresponding to the size "2" of the now coefficient of the table A variable length code "011" is given. Here, this code:
Further, in the table shown in FIG. 7, "3" adds data indicating the number. Since “3” has a size of “2” and the 0th of the size “2” is “−3”,
"3" is counted as -3, -2, 2, 3,
In this case, it is the third.

Next, the coefficient AC31 is "1" and the size is "1" from the table shown in FIG.
Since the size of the coefficient AC30 before “31” is “2”,
1 1, select the table of the previous size of the coefficient "2", it gives the variable length code "00" corresponding to the size "1" now in the coefficient of that table. Here, this code:
Further, in the table shown in FIG. 7, "1" adds data indicating the number. Since “1” has a size of “1” and the 0th of the size “1” is “−1”,
Since "1" is counted as -1, 1 in this case, it is the first.

Next, the coefficient AC22 is "0" and the size is "0" from the table shown in FIG.
Since the size of the coefficient AC31 before “22” is “1”,
1 0, the size of the previous coefficient select table "1", giving the variable length code "00" corresponding to the size "0" of the current coefficients of the table. Here, in this case, data indicating what number “0” is is not added to this code. Since the size "0" has only "0" as a member and has no meaning itself, the data indicating the number is not added.

Next, the coefficient AC13 is "0" and the size is "0" from the table shown in FIG.
Since the size of the coefficient AC22 before “13” is “1”,
The table in which the previous coefficient size is “0” shown in FIG. 9 is selected, and a variable length code “00” corresponding to the current coefficient size “0” of the table is given. Also, as described above,
No data indicating the number of “0” is added to this code.

Next, the coefficient AC23 is "1" and the size is "1" from the table shown in FIG.
Since the size of the coefficient AC13 before “23” is “0”,
The table with the previous coefficient size “0” shown in FIG. 9 is selected, and the variable length code “01” corresponding to the current coefficient size “1” in the table is given. Here, in the table shown in FIG. 7, data indicating the number is added to this code. Since “1” has a size of “1” and the 0th of the size “1” is “−1”,
Since "1" is counted as -1, 1 in this case, it is the first.

Next, the coefficient AC32 is "1" and the size is "1" from the table shown in FIG.
Since the size of the coefficient AC23 before 32 is “1”,
A table with the previous coefficient size "1" shown in FIG. 10 is selected, and a variable length code "01" corresponding to the current coefficient size "1" in the table is given. Here, in the table shown in FIG. 7, data indicating the number is added to this code. The data indicating the number is "1" as described above.

Next, the coefficient AC33 is "1" and the size is "1" from the table shown in FIG.
Since the size of the coefficient AC32 before “33” is “1”,
A table with the previous coefficient size "1" shown in FIG. 10 is selected, and a variable length code "01" corresponding to the current coefficient size "1" in the table is given. Here, in the table shown in FIG. 7, data indicating the number is added to this code. The data indicating the number is "1" as described above.

On the other hand, when the variable-length code data is decoded by the variable-length code decoding circuit 17, that is, the condition series is set to a size, a table is selected based on the size of the previous coefficient, and the selected table is referred to. When decoding the encoded code data, as described above, data indicating the number of the coefficient data in the table shown in FIG. 7 is added to the code data. The tables shown in FIGS. 8 to 23 may be used as they are, or a large number of these tables may be compressed for decoding only to create a new table, and the created new table may be used.

In this encoding method, data indicating the number of the coefficient data in the table shown in FIG. 7 is added to the code data. Can be easily decoded simply by having the tables shown in FIGS. 7 to 23 or these compressed tables by referring to the data, and the value of the coefficient data is assigned to each code data in FIG. Since the data indicating the order of the table shown in (1) is added, even if the code is the same, the coefficient data of the code can be easily detected by detecting the data indicating the order added to the code. Can be obtained.

As described above, in this example, the frequency of occurrence of the coefficient data quantized by the quantization circuit 13 is detected to create a table in which the size is a conditional sequence, and based on the created table, the Huffman encoding circuit is used. 14, the total code length data at that time is supplied to the quantization circuit 13, and the operation of controlling the quantization roughness is repeated to create a large number of tables based on the condition series. The table is loaded into the memory 4a of the variable-length coding circuit 4 of the encoder of the high-efficiency coding apparatus, and when the coefficient data is coded by the high-efficiency coding apparatus, the table corresponds to the size of the coefficient data before the coefficient data. Since a table is selected and encoding is performed with reference to the selected table, the entire code length can be reduced. Further, in this example, data indicating the number of the size of the coefficient data in the size table is added to the encoded coefficient data, so that the encoded data can be decoded with a simple configuration and processing. can do.

The above embodiment is an example of the present invention.
It goes without saying that various other configurations can be adopted without departing from the spirit of the present invention.

[0118]

According to the above-described digital signal encoding method of the present invention, in step (a), the digital signal is converted into a continuous data sequence with continuous zero coefficients and non-zero coefficients, and step (b) In step (c), a plurality of events consisting of a sequence of zero coefficients having a predetermined run length and at least one non-zero coefficient before or after the sequence of zero coefficients are obtained from the data sequence. One code word is set for a plurality of events based on table data that gives a different code word to each event according to each event, so that correlation between coefficients can be used. This has the effect that the coding efficiency can be greatly improved.

Further, according to the encoding table generating method of the present invention described above, in step (a), the predetermined digital signal is converted into a data string that is continuous with continuous zero coefficients and non-zero coefficients. )
Obtaining a plurality of events consisting of a series of zero coefficients having a predetermined run length and at least one non-zero coefficient before or after the series of zero coefficients, and in step (c), for each of the plurality of events, A table used to detect the occurrence frequency of each event appearing next to the event and to encode a desired data sequence in step (d) according to the occurrence frequency of each event according to the state of the previous data sequence Since the data is generated and stored in the memory, it is possible to create table data using the correlation between the coefficients, and thus, when encoding is performed using the table data thus created, the code is generated. There is an effect that the conversion efficiency can be greatly improved.

Further, according to the digital signal encoding method of the present invention described above, in step (a), the predetermined digital signal is converted into a continuous data sequence of continuous zero coefficients and non-zero coefficients, and step (b) In step (c), a plurality of events consisting of a series of zero coefficients having a predetermined run length and at least one non-zero coefficient before or after the series of zero coefficients are obtained. , The occurrence frequency of each event appearing next to the event is detected, and in step (d), a desired data string is replaced with the previous data string according to the occurrence frequency of each event obtained in step (c). Table data to be used for encoding according to the state is generated and stored in a memory, and in step (e), an arbitrary digital signal is linked with a continuous zero coefficient and a non-zero coefficient. In step (f), each event is obtained from the data string obtained in step (e), and in step (g), a plurality of events obtained in step (f) are obtained based on the table data. On the other hand, since one codeword is determined, the correlation between coefficients can be used, which has the effect of greatly improving coding and decoding efficiency.

Further, according to the encoding apparatus of the present invention described above,
The conversion means converts the digital signal into a continuous data sequence of continuous zero coefficients and non-zero coefficients, and a series of zero coefficients having a predetermined run length and at least one non-continuous series of zero coefficients before or after the series of zero coefficients. For a plurality of events having zero coefficients, a plurality of table data representing code words to be given to each of the following events is stored in a storage means, and a plurality of events are obtained from a data string by an encoding means, and a table is obtained. Since one codeword is given to each of a plurality of events based on the data, the correlation between the coefficients can be used, and the coding efficiency can be greatly improved.

Further, according to the above-described digital signal encoding method of the present invention, in step (a), the predetermined digital signal is converted into a continuous data sequence of continuous zero coefficients and non-zero coefficients, and step (b) ), The non-zero coefficient is
According to the value, the data is classified into a plurality of groups. In step (c), the data sequence is divided into a series of zero coefficients having a predetermined run length and at least one non-zero coefficient connected before or after the series of zero coefficients. Multiple events,
In step (d), one codeword is determined for each of a plurality of events based on table data that provides a codeword for each group in the immediately preceding event for each event. , The correlation between the coefficients can be used, and thereby the coding efficiency can be greatly improved.

Further, according to the digital signal encoding method of the present invention described above, in step (d), the rank information in the group is further added to the code word. Efficiency can be improved.

Further, according to the above-described digital signal encoding method of the present invention, in step (a), the predetermined digital signal is converted into a continuous data sequence of continuous zero coefficients and non-zero coefficients, and step (b) ), From the data sequence, a plurality of events consisting of a zero coefficient having a predetermined run length and at least one non-zero coefficient before or after the continuation of the zero coefficient are obtained, and in step (c), the non-zero coefficient is obtained by: In step (d), the frequency of occurrence of each event that appears next is detected for each group, and in step (e), a desired frequency is determined according to the frequency of occurrence of each event. Table data to be used in encoding the data sequence according to the state of the preceding data sequence is generated and stored in the memory. In step (f), an arbitrary digital signal is In the step (g), each event is obtained from the data string obtained in the step (f). In the step (h), based on the table data, Step (g)
Since one code word is determined for each of the plurality of events obtained in the above, one code word is determined, so that correlation between coefficients can be used, thereby performing encoding and decoding. There is an effect that efficiency can be greatly improved.

Further, according to the encoding apparatus of the present invention described above,
The conversion means converts the digital signal into a continuous data sequence with continuous zero coefficients and non-zero coefficients, classifies the non-zero coefficients into a plurality of groups according to the values, and classifies a predetermined run length. For a plurality of events consisting of a series of zero coefficients and at least one non-zero coefficient before or after the series of zero coefficients, a code word to be given to an event that comes next according to each group. Since a plurality of table data are stored in the storage unit, a plurality of events are obtained from the data sequence by the encoding unit, and one code word is given to each of the plurality of events based on the table data. There is an effect that the correlation between the coefficients can be used, and thereby the coding efficiency can be greatly improved.

Further, according to the encoding apparatus of the present invention described above,
The conversion means converts the digital signal into a data stream of a predetermined form and responds to a plurality of events in the data stream of the predetermined form.
And a plurality of codewords to be given to the next event
Storing table data in a storage unit, de-in coding means
Obtain multiple events from the data sequence and, based on the table data,
Since one code word is given to each of multiple events,
There is an effect that the correlation between the preceding and following data can be used, and thereby the coding efficiency can be greatly improved.
According to the configuration of the encoding method of the present invention described above,
Signal is converted into a data stream in a predetermined format, and
Next to multiple events in the data string
Multiple table data representing codewords given to events
To obtain multiple events from the data sequence,
One codeword for each of multiple events based on data
To use the correlation between the data before and after.
Can greatly improve the coding efficiency
effective.

[Brief description of the drawings]

FIG. 1 is a configuration diagram of a high-efficiency encoding apparatus showing an embodiment of an encoding method, an encoding apparatus, and an encoding method of a digital signal according to the present invention.

FIG. 2 is a configuration diagram of a Huffman table generation device showing an embodiment of a coding table generation method of the present invention.

FIG. 3 is a configuration diagram of a Huffman table generation device showing another example of the encoding table generation method of the present invention.

FIG. 4 is an explanatory diagram for explaining a determination of a code for coefficient data for explaining an embodiment of a digital signal encoding method, an encoding device, and an encoding method according to the present invention.

FIG. 5 is an explanatory diagram for explaining a table corresponding to a condition series (0 run, value) used for describing an embodiment of a digital signal encoding method, an encoding device, and an encoding method of the present invention.

FIG. 6 is an explanatory diagram showing an example of a case where a code is determined based on a coefficient data value and a condition sequence for explaining an embodiment of a digital signal encoding method, an encoding apparatus, and an encoding method of the present invention. is there.

FIG. 7 is an explanatory diagram showing a table for determining the size of coefficient data for explaining an embodiment of the encoding table generation method of the present invention.

FIG. 8 is a flowchart for explaining an operation of creating a table corresponding to a condition series for explaining an embodiment of the encoding table generation method of the present invention.

FIG. 9 is an explanatory diagram showing an example of a table corresponding to a size “0” for explaining an embodiment of a digital signal encoding method, an encoding device, and an encoding method of the present invention.

FIG. 10 is an explanatory diagram showing an example of a table corresponding to a size “1” for describing an embodiment of a digital signal encoding method, an encoding device, and an encoding method of the present invention.

FIG. 11 is an explanatory diagram showing an example of a table corresponding to a size “2” for describing an embodiment of a digital signal encoding method, an encoding device, and an encoding method according to the present invention.

FIG. 12 is an explanatory diagram showing an example of a table corresponding to a size “3” for explaining an embodiment of a digital signal encoding method, an encoding device, and an encoding method of the present invention.

FIG. 13 is an explanatory diagram showing an example of a table corresponding to a size “4” for describing an embodiment of a digital signal encoding method, an encoding device, and an encoding method of the present invention.

FIG. 14 is an explanatory diagram showing an example of a table corresponding to a size “5” for explaining an embodiment of a digital signal encoding method, an encoding device, and an encoding method of the present invention.

FIG. 15 is an explanatory diagram showing an example of a table corresponding to a size “6” for explaining an embodiment of an encoding method, an encoding device, and an encoding method of a digital signal of the present invention.

FIG. 16 is an explanatory diagram showing an example of a table corresponding to a size “7” for explaining an embodiment of a digital signal encoding method, an encoding device, and an encoding method of the present invention.

FIG. 17 is an explanatory diagram showing an example of a table corresponding to a size “8” used for describing an embodiment of a digital signal encoding method, an encoding device, and an encoding method of the present invention.

FIG. 18 is an explanatory diagram showing an example of a table corresponding to a size “9” for explaining an embodiment of a digital signal encoding method, an encoding device, and an encoding method according to the present invention.

FIG. 19 is an explanatory diagram showing an example of a table corresponding to a size “10” for explaining an embodiment of a digital signal encoding method, an encoding device, and an encoding method of the present invention.

FIG. 20 is an explanatory diagram showing an example of a table corresponding to a size “11” used for describing an embodiment of a digital signal encoding method, an encoding device, and an encoding method according to the present invention.

FIG. 21 is an explanatory diagram showing an example of a table corresponding to a size “12” for explaining an embodiment of an encoding method, an encoding device, and an encoding method of a digital signal of the present invention.

FIG. 22 is an explanatory diagram showing an example of a table corresponding to a size “13” for explaining an embodiment of a digital signal encoding method, an encoding device, and an encoding method according to the present invention.

FIG. 23 is an explanatory diagram showing an example of a table for converting a leading coefficient of an AC component used for describing an embodiment of a digital signal encoding method, an encoding device, and an encoding method of the present invention.

FIG. 24 is an explanatory diagram of a discrete cosine transform scanning method for describing a conventional encoding device.

FIG. 25 is an explanatory diagram for explaining a code determination based on the number of consecutive 0s and a coefficient value following the last 0, which is used for describing a conventional encoding device.

FIG. 26 is an explanatory diagram for describing encoding of coefficient data for describing a conventional encoding device.

[Explanation of symbols]

 2,12 conversion circuit 3,13 quantization circuit 4 variable length coding circuit 4a, 16 memory 7 variable length code decoding circuit 8 quantization restoration circuit 9 inverse conversion circuit 14 Huffman coding circuit 15 occurrence frequency detection circuit 17 controller 17a Control unit 17b ROM 18 switch

──────────────────────────────────────────────────続 き Continuation of the front page (56) References JP-A-5-46358 (JP, A) Special Table Sho 61-500345 (JP, A) (58) Fields investigated (Int. Cl. ⁷ , DB name) H03M 7/40 H04N 1/41

Claims (10)

(57) [Claims] 1. A digital signal comprising: a continuous zero coefficient;
Step (a) of converting to a data sequence that is continuous with non-zero coefficients
(B) obtaining from the data sequence a plurality of events consisting of a series of zero coefficients having a predetermined run length and at least one non-zero coefficient preceding or following the series of zero coefficients; (C) setting one codeword for the plurality of events based on table data that gives a different codeword to each event in accordance with each preceding event. Encoding method of a digital signal. 2. A step (a) of converting a predetermined digital signal into a data string that is continuous with continuous zero coefficients and non-zero coefficients, and, from the data string, a series of zero coefficients having a predetermined run length. (B) obtaining a plurality of events consisting of at least one non-zero coefficient before or after the series of zero coefficients, and for each of the plurality of events, occurrence of each event appearing next to the event (C) detecting the frequency; generating table data used in encoding a desired data sequence according to the state of the previous data sequence according to the frequency of occurrence of each of the above events; (D) generating a coding table. 3. A step (a) of converting a predetermined digital signal into a data string which is continuous with continuous zero coefficients and non-zero coefficients, and from the data string, a sequence of zero coefficients having a predetermined run length. (B) obtaining a plurality of events consisting of at least one non-zero coefficient before or after the series of zero coefficients, and for each of the plurality of events, occurrence of each event appearing next to the event (C) detecting the frequency, and according to the occurrence frequency of each event obtained in the step (c),
(D) generating table data to be used in encoding a desired data sequence according to the state of the previous data sequence and storing the table data in a memory; A step (e) of converting the data string into a series of zero coefficients, a step (f) of obtaining each event from the data string obtained in the step (e), and a step (f) based on the table data. (G) determining one code word for the plurality of events obtained in (d). 4. The method of claim 1, wherein the digital signal is represented by a continuous zero coefficient,
Conversion means for converting the data sequence into a series of non-zero coefficients; a plurality of events consisting of a series of zero coefficients having a predetermined run length and at least one non-zero coefficient before or after the series of zero coefficients. On the other hand, storage means for storing a plurality of table data each representing a code word to be given to the next event, and obtaining the plurality of events from the data sequence, based on the table data, An encoding device comprising encoding means for giving one code word to each event. 5. A step (a) of converting a predetermined digital signal into a continuous data sequence of continuous zero coefficients and non-zero coefficients, and classifying the non-zero coefficients into a plurality of groups according to their values. And (b) obtaining, from the data sequence, a plurality of events consisting of a sequence of zero coefficients having a predetermined run length and at least one non-zero coefficient before or after the sequence of zero coefficients. c) and determining one codeword for each of the plurality of events based on table data that provides codewords for each group in the previous event for each event (d) ), A method for encoding a digital signal. 6. The digital signal encoding method according to claim 5, wherein in the step (d), rank information in the group is further added to the codeword. 7. A step (a) of converting a predetermined digital signal into a continuous data sequence consisting of continuous zero coefficients and non-zero coefficients, and from said data sequence, a zero coefficient having a predetermined run length; (B) obtaining a plurality of events consisting of at least one non-zero coefficient before or after the series of zero coefficients; and (c) classifying the non-zero coefficients into a plurality of groups according to their values. (D) detecting, for each of the groups, the frequency of occurrence of each of the following events; (E) generating table data to be used at the time of conversion and storing it in a memory; and (f) converting an arbitrary digital signal into a continuous data sequence of continuous zero coefficients and non-zero coefficients. , (G) obtaining each event from the data string obtained in the step (f); and codewords for each of the plurality of events obtained in the step (g) based on the table data. (H) determining a digital signal. 8. The method of claim 1, further comprising: converting the digital signal to a continuous zero coefficient;
Converting means for converting the data into a series of data with non-zero coefficients; classifying means for classifying the non-zero coefficients into a plurality of groups according to their values; continuation of zero coefficients having a predetermined run length; For a plurality of events consisting of at least one non-zero coefficient before or after a series of coefficients, a plurality of table data representing a code word to be given to an event that comes next according to each group is stored. Storage means, and encoding means for obtaining the plurality of events from the data sequence and giving one codeword to each of the plurality of events based on the table data. Encoding device. 9. A conversion means for converting a digital signal into a data stream in a predetermined form, and a plurality of events in the data stream in the predetermined form.
Storage means for storing a plurality of table data each representing a code word given to the next event; and obtaining the plurality of events from the data sequence .
For each event that is encoded,
By referring to the above table data,
An encoding device comprising encoding means for giving a word . 10. A conversion step of converting a digital signal into a data stream of a predetermined form, obtaining the plurality of events from the data stream, and
A different codeword is assigned to each event depending on the previous event.
Based on the obtained table data, for each obtained event,
The above table data according to the event encoded before
, An encoding step of giving one code word .