JP3015001B2 - Huffman decoding device - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a Huffman decoding device for decoding an input Huffman code and outputting decoded data.

[0002]

2. Description of the Related Art Image data contains a very large amount of information. Therefore, it is not practical to process the image data as it is in terms of memory capacity and communication speed. Therefore, image data compression technology becomes important.

One of the international standards for image data compression technology is JPEG (Joint Photographic Expert Group). In JPEG, DCT (Discrete Cosine Transform) for lossy encoding and DPCM (Differen
rial PCM). Hereinafter, DCT image data compression will be described.

FIG. 3 is a block diagram showing the basic configuration of a system for executing image data compression and image data expansion in the DCT system.

On the encoding side, the DCT processing unit 100 converts the input original image data into a discrete cosine transform (hereinafter, DC
T) and outputs DCT coefficients. This D
In the CT processing, first, as shown in FIG. 4, image data is divided into a plurality of 8 × 8 pixel blocks. As shown in FIG. 5, one 8 × 8 pixel block includes 64 pieces of pixel data P _XY (X, Y = 0,..., 7). When two-dimensional DCT processing is performed on each of the divided 8 × 8 pixel blocks, 64 DCT coefficients S _UV (U, V = 0,..., 7)
Is obtained.

[0006] The DCT coefficient _S00 is called a DC coefficient, and the remaining 63 DCT coefficients are called AC coefficients. As shown in FIG. 5, a block subjected to the DCT processing includes more high-frequency horizontal frequency components from left to right, and includes more high-frequency vertical frequency components from top to bottom.

[0007] The quantization unit 200 of FIG. 3 refers to the quantization table 400 and outputs the D output from the DCT processing unit 100.
The quantization is performed on the CT coefficients, and the quantized DCT coefficients are output. The image quality and the amount of encoded information are controlled by the quantization. FIG. 7 shows the DCT output from the quantization unit 200.
An example of the coefficient is shown. In FIG. 6, "A", "B",
“C”, “D”, “E”, and “F” represent values other than “0”.

[0010] The Huffman coding apparatus 300 shown in FIG. 4 performs Huffman coding on the DCT coefficients output from the quantization unit 200 with reference to the coding table 500 and outputs compressed image data. In the coding of the DC coefficient, a difference value between the DC coefficient of the immediately preceding block and the DC coefficient of the current block is obtained, and a Huffman code is assigned to the difference value.

In the coding of AC coefficients, as shown in FIG. 7, first, the AC coefficients are arranged one-dimensionally by zigzag scanning. This one-dimensionally arranged AC coefficient is
Encoding is performed using a run length indicating the length of successive coefficients (ineffective coefficients) of “0” and values of coefficients (effective coefficients) other than “0”. Effective coefficients are grouped, and a group number is assigned to each effective coefficient. In the coding of AC coefficients, Huffman codes are assigned to combinations of run lengths and group numbers. As described above, the original image data is encoded into the compressed image data.

On the decoding side, the Huffman decoding device 600
Decodes the compressed image data into decoded data consisting of a run length and a group number by Huffman decoding,
A DCT coefficient quantized based on the decoded data is output. The inverse quantization unit 700 includes a quantization table 400
Inverse quantization is performed on the quantized DCT coefficient with reference to
Output the DCT coefficient. The inverse DCT processing unit 800
An inverse DCT process is performed on the T coefficient to output reproduced image data.

FIG. 8 is a block diagram showing an example of a conventional Huffman decoding device. The cueing processing unit 11 detects the head position of the Huffman code from the compressed image data, and sends the compressed image data of the number of bits corresponding to the maximum code length of the Huffman code from the detected head position to the address input terminal AD of the memory 12. Give as address signal.

The memory 12 has a storage capacity of 2 ^k words. Here, k represents the maximum code length of the Huffman code.
At each address in the memory 12, decoded data corresponding to the Huffman code represented by the address is stored. Each decoded data includes the above-described run length and group number.

For example, if the maximum code length k of the Huffman code is 1
6, the Huffman code “11111” having a length of 16 bits is used.
The decrypted data corresponding to 11111110101 ″ is
It is stored at the address “1111111111110101”. 15-bit length Huffman code "1111111"
The decoded data corresponding to “11000010” is stored at two addresses “1111111111000010X”, where X represents 0 and 1. The decoded data corresponding to the 2-bit length Huffman code “01” is , ¹⁴ addresses "01XXXXXXXXXXXXXXX"
XX ".

As described above, since 16-bit compressed image data corresponding to the maximum code length is given as an address signal to the memory 12, the decoded data corresponding to the Huffman code shorter than the maximum code length includes a plurality of pieces of decoded data. Must be stored at the address.

For example, when the compressed image data includes a 2-bit Huffman code “01”, the memory 12 stores
6-bit compressed image data "01 ..." is given as an address signal. As a result, the decoded data stored at the address “01...” Is read out, and the data output terminal D
Output from O. In this manner, the Huffman code included in the compressed image data is decoded.

[0016]

As described above, in the conventional Huffman decoding device, the maximum code length k of the Huffman code is k.
Is given to the memory 12 as an address signal, so that the memory 12 requires a storage capacity of 2 ^k words.

In this case, the decoded data corresponding to the Huffman code shorter than the maximum code length k is stored at a plurality of addresses. That is, it is necessary to store extra decoded data at a much larger number of addresses than the number of Huffman codes. Assuming that the number of Huffman codes is n, the utilization efficiency of the memory 12 is very low at n / ^2k .

As a result, the circuit scale of the Huffman decoding device becomes large, and it is difficult to increase the processing speed.

It is an object of the present invention to provide a Huffman decoding device that is reduced in size and speed of processing.

[0020]

Means for Solving the Problems and Effects of the Invention (1)
First invention A Huffman decoding device according to a first invention is a Huffman decoding device that decodes an input Huffman code and outputs decoded data, wherein a plurality of first storage units, a plurality of coincidence units are provided. The apparatus includes a detection unit, a second storage unit, an occurrence frequency generation unit, and a third storage unit.

[0021] The plurality of first storage means stores a predetermined number of Huffman codes among the plurality of Huffman codes. The plurality of coincidence detecting means are provided corresponding to the plurality of first storage means, and detect coincidence between the input Huffman code and the Huffman code stored in the corresponding first storage means.
The second storage means stores a predetermined number of decoded data corresponding to the predetermined number of Huffman codes, respectively, and responds to output signals of the plurality of coincidence detection means to output any one of the predetermined number of decoded data. Is output. The occurrence frequency generation means generates a corresponding occurrence frequency based on the input Huffman code. The third storage means stores the decoded data at an address indicated by the occurrence frequency of at least the remaining plurality of Huffman codes among the plurality of Huffman codes, and receives the occurrence frequency generated by the occurrence frequency generation means as an address signal, The decoded data is output from the address specified by the address signal.

[0022] In the Huffman decoding apparatus according to the present invention, a predetermined number of Huffman codes of the plurality of Huffman codes are stored in the plurality of first storage means. Further, a predetermined number of decoded data respectively corresponding to the predetermined number of Huffman codes are stored in the second storage means. Further, decoded data corresponding to at least the remaining plurality of Huffman codes among the plurality of Huffman codes is stored in the third storage unit.
Each decoded data is stored at the address indicated by the frequency of occurrence of the corresponding Huffman code.

A match between the input Huffman code and the Huffman code stored in the plurality of first storage means is detected by each of the plurality of match detection means. When a match between the input Huffman code and one of the plurality of Huffman codes stored in the first storage means is detected by the match detection means, the second Huffman code is output in response to the output signals of the plurality of match detection means. Of the predetermined number of pieces of decoded data stored in the storage means. In this case, the input Huffman code is decoded at high speed by the match detection by the match detection means and the output of the decoded data from the second storage means.

On the other hand, a corresponding occurrence frequency is generated by the occurrence frequency generation means based on the input Huffman code. The occurrence frequency generated by the occurrence frequency generation means is given to the third storage means as an address signal. If the input Huffman code and the Huffman code stored in the plurality of first storage means do not match, decoding is performed from the third storage means based on the occurrence frequency given as an address signal from the occurrence frequency generation means. Coded data is output.

As described above, the predetermined number of decoded data respectively corresponding to the predetermined number of Huffman codes are stored in the second storage means, so that the input Huffman code matches one of the predetermined number of Huffman codes. In this case, the corresponding decoded data is read from the second storage means at high speed.
If the input Huffman code does not match the predetermined number of Huffman codes, the frequency of occurrence of the input Huffman code is generated, and the corresponding decoded data is stored in the third storage unit based on the frequency of occurrence. Is read.

Since the Huffman code and the frequency of occurrence have a one-to-one correspondence, and the frequency of occurrence and the decoded data also have a one-to-one correspondence, the number of decoded data stored in the third storage means is large. Both have the same number as a plurality of Huffman codes. Therefore, the storage capacity required for the third storage means is reduced.

Therefore, a Huffman decoding device which is reduced in size and speeded up in processing can be obtained.

(2) Second invention The Huffman decoding device according to the second invention is the Huffman decoding device according to the first invention, wherein the predetermined number of Huffman codes is larger than the remaining Huffman codes. It has a high frequency of occurrence. As a result, a Huffman code that occurs frequently can be decoded at a high speed, so that the operation of the Huffman decoding device is further accelerated as a whole.

(3) Third invention A Huffman decoding device according to a third invention is the configuration of the Huffman decoding device according to the first or second invention, wherein the occurrence frequency generation means includes a code length of the Huffman code. Constant storage means for storing a constant set for each, a minimum code storage means for storing a minimum code for each code length of the Huffman code, and an input based on the minimum code for each code length stored in the minimum code storage means. Code length detecting means for detecting the code length of the detected Huffman code; constant selecting means for selecting any of the constants stored in the constant storage means based on the code length detected by the code length detecting means; Calculating means for calculating the occurrence frequency based on the constant selected by the means and the input Huffman code.

The frequency of occurrence of the Huffman code is obtained by subtracting a constant set for each code length from the Huffman code. The constant storage means stores constants set for each code length of the Huffman code. Further, the minimum code storage means stores the minimum code for each code length of the Huffman code. The code length of the input Huffman code is detected based on the minimum code for each code length stored in the minimum code storage unit, and any of the constants stored in the constant storage unit is determined based on the detected code length. Selected. Then, the occurrence frequency is calculated based on the selected constant and the input Huffman code.

(4) Fourth invention The Huffman coding apparatus according to the fourth invention is the same as the Huffman coding apparatus according to the first, second or third invention, except that the second and third storage means are provided. And decoding data selection means for selectively outputting the decoding data output from.

If the input Huffman code matches a predetermined number of Huffman codes, the decoded data output from the second storage means is selectively output, and the input Huffman code is converted to a predetermined number of Huffman codes. If they do not match, the decoded data output from the third storage means is selectively output.

[0033]

FIG. 1 is a block diagram showing a configuration of a Huffman decoding apparatus according to an embodiment of the present invention.

The Huffman decoding apparatus shown in FIG. 1 has a cue processing unit 1, an occurrence frequency generation unit 2, a memory 3, and i registers R
1 to Ri, i comparators C1 to Ci, a register 4 and a selector 5. Here, assuming that the number of Huffman codes is n, i has a relationship of 0 <i <n. In this embodiment, i = 20. In this embodiment, the maximum code length k of the Huffman code is 16 bits. Generally, the shorter the Huffman code, the higher the frequency of occurrence. For example, the Huffman codes with the highest frequency of occurrence from the top to the twentieth have a code length of 8 bits or less.

The cueing processing unit 1 detects the head position of each Huffman code from the input compressed image data, and supplies 16-bit compressed image data from the detected head position to the occurrence frequency generation unit 2 to detect the Huffman code. The 8-bit compressed image data from the head position is supplied to the comparators C1 to Ci.

The occurrence frequency generation unit 2 generates the occurrence frequency of the Huffman code included in the compressed image data provided from the cueing processing unit 1 by a method described later, and outputs the generated occurrence frequency to an address input terminal of the memory 3. It is given to AD as an address signal.

As the memory 3, a RAM (random access memory) or the like is used. At each address of the memory 3, decoded data corresponding to the Huffman code having the frequency of occurrence indicated by the address is stored. The decoded data includes a run length (the number of consecutive 0s) and a group number. The Huffman code and the occurrence frequency have a one-to-one correspondence, and the occurrence frequency and the decoded data have a one-to-one correspondence. Therefore, the memory 3 stores a maximum of n pieces of decoded data.

The frequency of occurrence is determined by the address input terminal A of the memory 3.
By giving it to D as an address signal, decoded data corresponding to the Huffman code of the occurrence frequency is output from the data output terminal DO.

The i registers R1 to Ri store i Huffman codes whose occurrence frequencies are from the highest to the i-th, respectively. Comparators C1 to Ci are connected to registers R1 to Ri.
Is provided corresponding to each of them. Comparators C1 to Ci
Registers the Huffman codes included in the compressed image data supplied from the cueing processing unit 1 in the corresponding registers R1 to R1.
Compare with the Huffman code stored in Ri. Each comparator C
1 to Ci output, for example, a high-level match signal when the Huffman code given from the cueing processing unit 1 matches the Huffman code stored in the corresponding register R1 to Ri, and when they do not match. For example, a low-level mismatch signal is output.

The register 4 has i storage areas M1 to Mi corresponding to the registers R1 to Ri. In the storage areas M1 to Mi of the register 4, the occurrence frequency is i
Decoded data corresponding to the first Huffman code is stored. Each decoded data includes a run length and a group number.

The selector 5 selectively outputs the decoded data output from the memory 3 or the decoded data output from the register 4.

In this embodiment, the registers R1 to Ri are the first
, The comparators C1 to Ci correspond to the coincidence detecting means, and the register 4 corresponds to the second storing means. Further, the occurrence frequency generation unit 2 corresponds to an occurrence frequency generation unit, the memory 3 corresponds to a third storage unit, and the selector 5 corresponds to a selection unit.

FIG. 2 is a block diagram showing the configuration of the occurrence frequency generation unit 2 in the Huffman decoding apparatus of FIG.

The occurrence frequency generation unit 2 includes a constant storage unit 21, a minimum code storage unit 22, a code length detection unit 23, a selector 24, and an adder 25. The following relational expression holds between the Huffman code and the occurrence frequency.

Frequency of occurrence = Huffman code-constant m The constant m is unique to the code length of the Huffman code and can be obtained in advance by calculation. Therefore, the code length of the input Huffman code can be detected, and the occurrence frequency can be calculated using the constant m corresponding to the detected code length.

The constant storage unit 21 shown in FIG. 2 stores constants m corresponding to code lengths of 1 bit to 16 bits. The constant storage unit 21 includes, for example, a register.

The minimum code storage unit 22 stores the minimum code for each code length of the Huffman code. That is, the minimum code storage unit 22 stores a total of 16 minimum codes from the minimum code of the Huffman code having a code length of 1 bit to the minimum code of the Huffman code having a code length of 16 bits. For example, a Huffman code having a code length of 4 bits is "1".
Assuming that there are three bits, “010”, “1011” and “1100”, “1010” is stored as the minimum code of the Huffman code having a code length of 4 bits in the minimum code storage unit 22. The unit 22 includes, for example, a register.

The code length detection unit 23 detects the code length of the input Huffman code by comparing the input Huffman code with the 16 Huffman codes output from the minimum code storage unit 22.

The selector 24 selects one of the 16 constants m output from the constant storage unit 21 based on the code length detected by the code length detection unit 23, and adds the selected constant m to an adder. 25 to one input terminal. Adder 2
The other input terminal 5 is provided with the input Huffman code.

The adder 25 calculates the occurrence frequency by subtracting the constant m from the input Huffman code, and supplies the calculated occurrence frequency to the address input terminal AD of the memory 3 as an address signal. Thereby, decoded data including the corresponding run length and group number is output from the data output terminal DO of the memory 3.

In this embodiment, the constant storage unit 21 corresponds to constant storage means, the minimum code storage unit 22 corresponds to minimum code storage means, and the code length detection unit 23 corresponds to code length detection means. Further, the selector 24 corresponds to constant selecting means, and the adder 25 corresponds to calculating means.

Next, the operation of the Huffman decoding apparatus shown in FIG. 1 will be described. The cueing processing unit 1 detects the head position of each Huffman code included in the compressed image data, gives 16-bit compressed image data from the detected head position to the occurrence frequency generation unit 2, and starts from the detected head position. The 8-bit compressed image data is supplied to i comparators C1 to Ci.

Each of the comparators C1 to Ci compares the Huffman code included in the compressed image data supplied from the cueing processing unit 1 with the Huffman code stored in the corresponding register R1 to Ri. When the Huffman code supplied from the cueing processing unit 1 matches one of the i Huffman codes stored in the registers R1 to Ri, the comparators C1 to Ci
For example, a high-level match signal is output from any of the comparators, and a low-level mismatch signal is output from the other comparators.

The output signals of the comparators C1 to Ci are supplied to the register 4 as address signals. As a result, the decoded data is output from the storage area corresponding to the comparator that has output the match signal among the storage areas M1 to Mi of the register 4.

In this case, the output of the decoded data by the comparators C1 to Ci and the registers R1 to Ri is performed in one cycle of the reference signal.

If the Huffman code included in the compressed image data supplied from the cueing processing unit 1 does not match any of the Huffman codes stored in the registers R1 to Ri, the Huffman code included in the compressed image data is used. The occurrence frequency is output from the occurrence frequency generation unit 2.

The occurrence frequency output from the occurrence frequency generation unit 2 is used as an address signal as an address signal at the address input terminal AD of the memory 3.
Given to. As a result, decoded data corresponding to the Huffman code having the occurrence frequency is output from the data output terminal DO of the memory 3.

In this case, the output of the decoded data by the occurrence frequency generator 2 and the memory 3 is performed in three cycles of the reference signal.

If the Huffman code supplied from the cueing processing unit 1 is a Huffman code having the highest i-th occurrence frequency, the selector 5 outputs the decoded data output from the register 4 and performs cueing. If the Huffman code given from the processing unit 1 is not a Huffman code having the highest i-th occurrence frequency, the decoded data output from the memory 3 is output.

Since the appearance probability of the Huffman code having the frequency of occurrence from the top to the twentieth is about 90% or more,
Approximately 90% of the Huffman code given from the cueing processing unit 1
Are decoded by one cycle of processing by the comparators C1 to Ci and the register 4. Therefore, the processing of the Huffman decoding device is speeded up as a whole.

The Huffman code and the frequency of occurrence are one-to-one.
, And the occurrence frequency and the decoded data have a one-to-one correspondence, so that the storage capacity required for the memory 3 is the same as the Huffman code, that is, a maximum of n words. Therefore, the Huffman decoding device is downsized.

In the above embodiment, among the plurality of Huffman codes, the Huffman codes whose occurrence frequencies are from the highest to the twentieth are stored in the registers R1 to Ri. The number is not limited to this, and any number of Huffman codes can be stored in the register.

Further, in the above embodiment, the decoded data corresponding to all the Huffman codes are stored in the memory 3, but they correspond to the Huffman codes except for the i Huffman codes stored in the registers R1 to Ri. The decoded data may be stored in the memory 3.

[Brief description of the drawings]

FIG. 1 is a block diagram illustrating a configuration of a Huffman decoding device according to an embodiment of the present invention.

FIG. 2 is a block diagram illustrating a configuration of an occurrence frequency generation unit included in the Huffman decoding device of FIG. 1;

FIG. 3 is a block diagram showing a basic configuration of a system for executing image data compression and image data expansion of a DCT method.

FIG. 4 is a diagram showing blocking of image data.

FIG. 5 is a diagram showing an 8 × 8 pixel block and a block subjected to DCT processing.

FIG. 6 is a diagram illustrating an example of quantized DCT coefficients.

FIG. 7 is a diagram for explaining zigzag scanning.

FIG. 8 is a block diagram illustrating an example of a conventional Huffman decoding device.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Start-up processing part 2 Occurrence frequency generation part 3 Memory 4 Register 5 Selector R1, Ri register C1, Ci Comparator 21 Constant storage part 22 Minimum code storage part 23 Code length detection part 24 Selector 25 Adder

Continuation of the front page (58) Field surveyed (Int.Cl. ⁷ , DB name) H03M 7/40

Claims (4)

(57) [Claims] An Huffman decoding apparatus for decoding an input Huffman code and outputting decoded data, wherein a plurality of first Huffman codes storing a predetermined number of Huffman codes among the plurality of Huffman codes is provided. And a plurality of coincidence units provided corresponding to the plurality of first storage units, each of which detects coincidence between the input Huffman code and the corresponding Huffman code stored in the corresponding first storage unit. Detecting means for storing a predetermined number of decoded data respectively corresponding to the predetermined number of Huffman codes, and outputting one of the predetermined number of decoded data in response to output signals of the plurality of coincidence detecting means A second storage unit that generates a corresponding occurrence frequency based on the input Huffman code; and a generation unit that generates at least the remaining plurality of Huffman codes among the plurality of Huffman codes. Third storage means for storing the decoded data at the address indicated by the frequency, receiving the occurrence frequency generated by the occurrence frequency generation means as an address signal, and outputting the decoded data from the address specified by the address signal; A Huffman decoding device comprising: 2. The Huffman decoding apparatus according to claim 1, wherein said predetermined number of Huffman codes has a higher frequency of occurrence than the remaining Huffman codes. 3. A constant frequency storage means for storing a constant set for each code length of the Huffman code, a minimum code storage means for storing a minimum code for each code length of the Huffman code, Code length detection means for detecting the code length of the input Huffman code based on the minimum code for each code length stored in the minimum code storage means; and the constant based on the code length detected by the code length detection means A constant selection means for selecting any of the constants stored in the storage means; and a calculation means for calculating an occurrence frequency based on the constant selected by the constant selection means and the input Huffman code. 3. The Huffman decoding device according to claim 1, wherein 4. The apparatus according to claim 1, further comprising: decoded data selecting means for selectively outputting decoded data output from said second and third storage means. Huffman decoding device.