JPH05235778A - High efficiency coding method - Google Patents

Abstract

PURPOSE:To reduce deterioration in data due to a transmission error. CONSTITUTION:Input data Di from a terminal 101 are inputted to a prediction circuit 102, a subtractor circuit 103 and a remainder arithmetic operation circuit 106. The prediction circuit 102 outputs one preceding data Di-1. The subtractor circuit 103 subtracts the prediction data Di-1 from the data Di and outputs a prediction error Si. A category circuit 104 outputs a category number Ji according to the prediction error Si. A conversion circuit 105 outputs divisor data OUi or data Mi based on the category number Ji. The remainder arithmetic operation circuit 106 divides the data Di by the divisor data OUi and outputs the residue Ei. A coding circuit 109 codes the category number Ji, the coding circuit 110 obtains the bit number Mi of the residue data Ei based on the category number Ji and outputs a low-order Mi-bit of the residue data Ei in a bit serial form and multiplexes the bits onto the coded data and outputs the result from a terminal 108.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a high efficiency coding method for reducing the amount of information such as data obtained by sampling and quantizing analog signals such as video and audio.

[0002]

2. Description of the Related Art There are various types of high-efficiency coding, and there is also a combination of these types. At present, standardization of high-efficiency coding schemes for images and voices is being carried out, and standardization of still picture coding schemes is being carried out by the JPEG subordinate organization of the International Organization for Standardization (ISO).

As a conventional high-efficiency encoding method, the DCT method, which is a high-efficiency encoding method of JPEG, will be described as an example (reference: Journal of the Television Society Vol. 44, No. 2 (1990) pp15.
8 to 159).

The input signal is a sampled and quantized image signal,
That is, it is digital image data. Image data, which is a pixel array of raster scan, is divided into a rectangular area (referred to as a block) of 8 pixels in each of the horizontal and vertical directions of the screen and converted into data in block units.
This is called blocking. An 8th-order two-dimensional discrete cosine transform (hereinafter referred to as DCT) is performed for each block, and the obtained DCT coefficient is quantized at a predetermined quantization step Q determined for each coefficient (that is, divided by Q and rounded). ). The AC coefficient of the quantized DCT coefficient is two-dimensionally Huffman coded, and the DC coefficient of the quantized DCT coefficient is predictively coded.

The predictive coding method of the DC coefficient will be described. The input data is the DC coefficient of the quantized DCT coefficient, and the data Di (i = 0, 1, 2,
3, ...., i is the data number and is equal to the block number). In the predictive coding, a predictive value Pi is obtained using encoded input data, a predictive error Si that is a difference between the input data Di and the predictive value Pi is obtained, and the predictive error Si is encoded. In the prediction method, the previous input data is used as the predicted value in the previous value prediction.

A method of encoding the prediction error Si will be described. The prediction error Si is classified into a predetermined category according to its size, the number of the corresponding category is obtained, and this is Huffman-encoded. The category number corresponds to the upper bit information of the prediction error. For the prediction error, the lower bits are cut out by the number of bits L determined by the category number, and are output following the Huffman-coded category number. That is, the prediction error is divided into high-order bit information and low-order bit information and encoded. Note that if the lower L bits of the prediction error are cut out as they are, a sign that overlaps with a positive value and a negative value occurs, so if the prediction error is negative, 1 is subtracted in advance, and then the lower L bits are cut out.

In Huffman coding, a code having a short word length is assigned to data having a high occurrence probability, and a code having a long word length is assigned to data having a low occurrence probability. Is. Since the correlation between adjacent input data is high and the prediction error has a high probability of becoming a value near 0, by assigning a short code to the category number that represents the range in which the absolute value of the prediction error is small, highly efficient coding is possible. realizable.

[0008]

However, since predictive coding obtains a decoded value by accumulating the prediction errors, once a transmission error occurs, the influence of the error is accumulated, and thereafter, only an erroneous decoded output is obtained. There was a problem that there is no (error propagation).

[0009]

A high-efficiency coding method of the present invention uses a sampled and quantized signal as input data, obtains a predicted value of the input data, and calculates a difference between the input data and the predicted value. , A second step of classifying the prediction error according to its size and outputting a category number representing a corresponding category, and a predetermined step defining a range of the corresponding category. A third step of dividing the input data to obtain a remainder with a predetermined value larger than a difference between an upper limit value and a predetermined lower limit value as a divisor, and a fourth step of encoding and outputting the category number and the remainder. And the fourth step assigns the same codeword to the category number to which the maximum value of the prediction error belongs and the category number to which the minimum value belongs,
When the prediction error is classified into a category to which the maximum value or the minimum value belongs, the fourth step directly encodes and outputs the input data instead of the residual data. ..

[0010]

Since the high-efficiency encoding method of the present invention transmits the lower bit information of the input data by the above-mentioned configuration,
The error propagation does not always occur, and the transmission error resistance can be improved as compared with the conventional predictive coding method. Further, when the absolute value of the prediction error is large, the input data is directly transmitted, so that a correct decoded value can be obtained for the data after the direct transmission of the input data even if error propagation occurs.

[0011]

DESCRIPTION OF THE PREFERRED EMBODIMENTS First, the encoding method of the present invention will be described below by assigning a number to each processing step.

1. A prediction value Pi of the input data Di (an integer without a sign) is obtained, and this is subtracted from the input data Di to obtain a prediction error Si. Note that Di represents the i-th input data, and in the following, the symbol with the subscript i attached is Di.
It indicates that the data corresponds to.

2. The prediction error S is calculated using a predetermined classification table.
i is classified according to its size. A category number Ji representing the range to which the prediction error Si belongs is obtained. Therefore, if the upper and lower limits of the prediction error range indicated by the category number Ji are SXi and SNi, respectively,

[0014]

[Equation 1]

Is satisfied. Furthermore, in correspondence with the category number Ji one-to-one,

[0016]

[Equation 2]

Predetermined divisor data OUi satisfying the above is obtained. 3. The remainder Ei is obtained by dividing the input data Di by the divisor data OUi. That is,

[0018]

[Equation 3]

Is satisfied. However, Ni is a quotient. 4. The category number Ji and the remainder data Ei are encoded.

A specific example of the prediction error classification table is shown in (Table 1).

[0021]

[Table 1]

Table 1 shows not only the category number Ji and the corresponding prediction error range (SNi to SXi),
The word lengths Mi of the divisor data OUi and the remainder data Ei are shown in association with each other.

As has been made clear from the above description, the encoding method of the present invention encodes and transmits the category number, which is the upper bit information of the prediction error, and the residual data, which is the lower bit information of the input data. Is. When the input data Di is a signed integer, the main encoding can be easily performed by converting the code into an unsigned integer.

Next, this decoding method will be described. In the encoding method of the present invention, the category number Ji and the remainder data Ei
Are encoded and transmitted. To obtain the data Di, the divisor data OUi, the remainder data Ei, and the quotient data Ni are required as shown in (Equation 3). Category number Ji
Since there is a one-to-one correspondence between the divisor data OUi and the divisor data OUi, the category number-divisor data conversion table is prepared and the divisor data OUi can be obtained from the transmitted category number Ji. Since the surplus data Ei has been transmitted, the data Di can be obtained if the quotient data Ni is obtained.

A method of obtaining the quotient data Ni that is required therefor will be described. Substituting (Equation 3) into (Equation 1) to eliminate the prediction error Si, the following equation is obtained.

[0026]

[Equation 4]

Is obtained. Further, by substituting (Equation 3) into (Equation 4) and eliminating Di, the following equation is obtained.

[0028]

[Equation 5]

Is obtained. The predicted value Pi is obtained from the decoded data Di, and SXi and SNi are category numbers Ji.
Since there is a one-to-one correspondence with, a conversion table is created in advance, and by using this, the conversion table can be obtained from the category number Ji. Further, the quotient data Ni is an integer, and the difference (SXi-SNi) / OUi between the leftmost term and the rightmost term of (Equation 5) is less than 1 than (Equation 2), and thus satisfies (Equation 5). The quotient data Ni can be uniquely determined. Therefore, the following expression obtained by extracting the expression on the left side of (Equation 5)

[0030]

[Equation 6]

Find the smallest integer Ni that satisfies
The following expression obtained by extracting the expression on the right side of (Equation 5)

[0032]

[Equation 7]

It suffices to find the maximum integer Ni that satisfies the above condition. That is, the quotient data Ni can be obtained by using any of the formulas (5), (6), and (7). (Equation 7)
The method using is the simplest to process because it only needs to truncate the fractional part of the division result on the right side.

The quotient data Ni obtained from the above is given by (Equation 3)
To obtain the data Di, that is, the data Di can be decrypted.
The decoding methods are summarized below and numbered for each processing unit.

1. The encoded data is decoded to obtain the category number Ji and the remainder data Ei.

2. Data Dk already obtained by decoding (however, k
Is an integer smaller than i) and the predicted value Pi is obtained.

3. The divisor data OUi, the upper limit value SXi of the prediction error range or the lower limit value SNi of the prediction error range is obtained from the category number Ji.

4. The quotient data Ni is obtained using (Equation 7) or (Equation 6) or (Equation 5). 5. The data Di is obtained using (Equation 3).

Since the remainder data Ei is less than the divisor data OUi, its code length Mi is (log ₂ OUi) bits. To minimize this, that is, to improve the coding efficiency, the minimum value that satisfies (Equation 2) may be used as the divisor data OUi. Further, by making the divisor data OUi a power of 2, it is possible to realize a remainder operation and division with a very simple circuit, and since the code length Mi becomes an integer value, the remainder data Ei can be efficiently encoded in binary. That is,

[0040]

[Equation 8]

The divisor data, the upper limit value and the lower limit value of the prediction error may be set so as to satisfy the above. The prediction error classification table (Table 1) is created so as to satisfy (Equation 8) when the word length of the input data Di is 8 bits.

FIGS. 1A and 1B are block configuration diagrams of an encoding device and a decoding device in an embodiment to which the high efficiency encoding method and the decoding method of the present invention can be applied. The input data is a sampled and quantized analog video signal obtained by raster scanning an image. The input data is an 8-bit unsigned integer (0 to 255).

In FIG. 1A, 101 is an input terminal of data Di to be encoded, 102 is a prediction circuit for obtaining a prediction value Pi, and 103 is a prediction error Si obtained by subtracting the prediction value Pi from the input data Di. , 104 is a classification circuit which inputs the prediction error Si and outputs a category number Ji, 105 is a conversion circuit which obtains divisor data OUi from the category number Ji, and 106 is division of the data Di by the divisor data OUi. A remainder arithmetic circuit for obtaining the remainder data Ei, 107 is a first encoding circuit for encoding the category number Ji and the remainder data Ei to obtain encoded data Ci, 108 is an output terminal for the encoded data Ci, 1
Reference numeral 09 is a second coding circuit that codes the category number Ji to obtain coded data CJi, 110 is a third coding circuit that codes the remainder data Ei to obtain coded data CEi, and 111 is the code It is a multiplexing circuit that connects the encoded data CJi and the encoded data CEi to obtain the encoded data Ci.

In FIG. 1B, 112 is an input terminal for encoded data Ci, and 113 is the encoded data Ci.
Is a first decoding circuit for decoding the category number Ji and the remainder data Ei, 114 is a conversion circuit for obtaining the upper limit value SXi of the prediction error range of the category of the number Ji, and the divisor data OUi. 115 is the quotient data Ni. A quotient data calculation circuit 116 indicates the divisor data OUi and the quotient data Ni.
And a synthesis circuit for reproducing data Di from the remainder data Ei, 117 is an output terminal of the data Di, 118 is the already decoded data Dk (k is an integer smaller than i)
A prediction circuit that outputs the predicted value Pi of the data Di using
119 is a buffer memory for temporarily storing the coded data Ci from the terminal 112, and 120 is the coded data CJi multiplexed at the beginning of the coded data Ci obtained from the buffer memory 119 and decoded to obtain the category number Ji. A second decoding circuit for obtaining 121, a third decoding circuit for obtaining coded data CEi multiplexed in the remaining portion of the coded data Ci from the buffer memory 119 and decoding it to obtain remainder data Ei, and an adder circuit 122. , 123 is a subtraction circuit for subtracting the remainder data Ei from the output from the addition circuit 122,
Reference numeral 124 denotes the output from the subtraction circuit 123, which is the divisor data O
Divide by Ui, only the integer part of the obtained result is quotient data N
A division circuit for outputting as i, 125 is the divisor data O
A multiplication circuit that multiplies Ui and the quotient data Ni to obtain an offset Fi, and 126 is an addition circuit that adds the remainder data Ei and the offset Fi to obtain new decoded data Di.

The operation of the encoding apparatus and decoding apparatus of the present embodiment configured as above will be described below.

In the encoding device, the input data Di from the terminal 101 is input to the prediction circuit 102, the subtraction circuit 103, and the remainder calculation circuit 106. The prediction circuit 102 performs the previous value prediction, and holds the previous data D _i-1 1
It is composed of only one register and outputs a predicted value Pi = D _i-1 .

The subtraction circuit 103 subtracts the prediction value Pi from the data Di and outputs a prediction error Si. The classification circuit 104 receives the prediction error Si as an input and classifies the prediction error Si according to its size according to (Table 1), and outputs a category number Ji indicating a corresponding classification item. The conversion circuit 105 can be composed of a ROM (Read Only Memory), and according to (Table 1), the divisor data OUi or data M from the category number Ji is used.
Output i. The remainder calculation circuit 106 divides the data Di by the divisor data OUi and outputs the remainder Ei. In (Table 1), the divisor data OUi is set to 2 to the power of Mi. Therefore, the remainder calculation circuit 106 can be realized by a simple gate circuit that extracts only the lower Mi bits of the data Di. In this case, the conversion circuit 105 may output the data Mi instead of the divisor data OUi.

The encoding circuit 109 uses the category number Ji.
Is Huffman-encoded (a type of entropy encoding) and output in bit-serial format. This output is the coded data CJi. Since the probability of occurrence of a prediction error with a category number of Ji and the probability of occurrence of a prediction error with a category number of -Ji are almost the same, the same Huffman code is assigned to these two categories in this embodiment. The coded data CJi is obtained by adding a 1-bit flag G indicating that to the Huffman code.
When the category number is 0, the flag G is unnecessary. When the flag G = 0, the category number is positive and G = 1.
If, the category number is negative.

The encoding circuit 110 uses the category number Ji.
Then, the bit number Mi of the remainder data Ei shown in (Table 1) is obtained, and the lower Mi bits of the remainder data Ei are output in a bit serial format. This output is the encoded data CEi
Is. Only the lower Mi bits of the remainder data Ei are output because the remainder data Ei can be represented by Mi bits. The multiplexing circuit 111 outputs the coded data Ci obtained by connecting the coded data CEi after the coded data CJi from the terminal 118 in a bit serial format. The encoding of the data Di is realized by the above operation.

The maximum value 255 and the minimum value -2 of the prediction error Si
The category numbers Ji to which 55 belongs are 8 and -8, respectively.
Is. In this case (Table 1), the word length Mi of the remainder data is 7, and if this is combined with 1 bit of the flag G,
It is 8 bits, which is the same as the word length of the input data Di. Therefore, the transmission efficiency does not change even if the input data is transmitted instead of sending the flag G and the remainder data. However, the input data Di
In the present embodiment, when the absolute value of the category number is 8, the input data Di is replaced by the input data Di in place of the remainder data and the flag G as the exception processing, and the minimum required word length ( The word length is a necessary word length that can represent the dynamic range of input data, and is transmitted as it is (or may be encoded such as code conversion) in 8 bits in this embodiment.

In the decoding device, the coded data Ci from the terminal 112 is temporarily stored in the buffer memory 119.

First, the decoding circuit 120 uses the buffer memory 1
The coded data CJi while judging the code length from 19
Is read and decoded, and the codeword length L1 and the category number Ji of the read code are output. Buffer memory 119
Receives the codeword length L1 and receives the coded data CJ
The head position of the coded data CEi following i is obtained and set to the read pointer inside it.

Subsequently, the decoding circuit 121 is connected to the decoding circuit 120.
Of the remainder data Ei shown in (Table 1) is obtained from the category number Ji from the above, the coded data CEi of Mi bits is read from the buffer memory 119, data 0 is added to the upper order, and a bit parallel format is obtained. Remainder data Ei which is data is reproduced.

The buffer memory 119 receives the word length Mi from the decoding circuit 121 and receives the encoded data CEi.
The leading position of the next encoded data CJi following is obtained, and the read pointer is updated to prepare for the next data decoding.

The conversion circuit 114 can be composed of a ROM, for example, and outputs the upper limit value SXi of the prediction error range and the divisor data OUi shown in (Table 1) from the category number Ji.

The quotient data calculation circuit 115 receives the remainder data Ei, the SXi and the prediction value P from the prediction circuit 118.
Using i and i, the calculation shown on the right side of the equation (7) is performed, and the quotient data Ni that is the integer part is output.

The synthesizing circuit 116 receives the quotient data Ni, the divisor data OUi, and the remainder data Ei as input, reproduces the data Di by performing the calculation shown in the equation (Equation 3), and outputs it from the terminal 117.

The prediction circuit 118 has the same configuration as the prediction circuit 102 in the encoding device, and outputs the prediction value Pi with the data Di as an input.

By the above operation, the decoding of the data Di is realized. When the absolute value of the category number is 8, the flag G and the input data Di are directly transmitted instead of the remainder data Ei as an exceptional process, and thus are output as they are. In this case, since the predicted value Pi is not required for decoding, even if error propagation occurs, error propagation does not occur in the subsequent decoded output.

Next, the operation and effects of the present invention will be described with reference to specific data examples. It is assumed that the input data Di to be encoded by the encoding device is 46 and the previous input data D _{i-1 which} has already been encoded is 35. The prediction circuit 102 outputs a prediction value Pi = D _i-1 = 35.
Prediction error Si = 46-35 in the subtraction circuit 103
= 11 is obtained. From the prediction error Si = 11 in the classification circuit 104, the category number J according to (Table 1)
i = 4 is obtained.

In the conversion circuit 105, the category number Ji is a power of 2 according to (Table 1)
Ui or its exponent part data Mi = 3 is output. The remainder calculation circuit 106 divides the data Di into the divisor data OUi = 2 ^Mi
The remainder data Ei = 6 divided by 8 is output. Since the divisor data OUi is a power of 2, it is not necessary to perform normal division, and the remainder data Ei can be obtained by only extracting the lower Mi = 3 bits of the data Di.

In the encoding circuit 109, the category number Ji is Huffman encoded. Ji = 4 or Ji =
The Huffman code that represents -4 is "10" with a binary number of 3 bits.
1 "(in the following, the binary code is surrounded by""), the coded code CJi of the category number Ji = 4
Becomes "1010" and is output in a bit serial format. The 1-bit data "0" added at the end is a flag G indicating whether the category number is positive or negative, and indicates that the category number is positive.

In the encoding circuit 110, the lower data Mi = 3 bits of the remainder data Ei = 6 are output in the bit serial format to become the encoded data CEi "110".
In the multiplexing circuit 111, the encoded data CJi ″ 1
The encoded data CEi "110" is added after 010 "to obtain the encoded data Ci =" 1010110 ", and the left end (the most significant bit) is output from the terminal 118 in the bit serial format.

In the decoding device, the coded data from the terminal 112 is temporarily stored in the buffer memory 119 in the bit serial format. The decoding is completed up to the current data D _i-1 = 35, and from this, the encoded data Ci to the data Di
Shall be decrypted.

The decoding circuit 120 reads the encoded data Ci in the bit serial format from the memory address indicated by the pointer in the buffer memory 119. The decoding circuit 120 can detect that the code length LJ of the encoded data CJi is 4 bits at the time of reading from the first bit of the encoded data Ci to “101”, and if the Ji is not 0, an additional 1 bit is added. The flag G of is read. That is,
All 4-bit coded data CJi = "1010" is read. Since the flag G = 0 indicates that the category number Ji is positive, the decoding circuit 120 determines that the category number Ji is
i = 4 is output.

Subsequently, the decoding circuit 121 receives the above-mentioned decoded category number Ji = 4 and outputs M according to (Table 1).
The coded data CEi = “110” is obtained by reading the data for i = 3 bits from the buffer memory 119, the coded data CEi = “110” is converted into a parallel format, and the remainder data Ei = 6 is obtained by adding 0 to the upper bits. ,Output.
The pointer of the buffer memory 119 is updated to indicate the head memory address of the next encoded data C _{i + 1} .

According to (Table 1), the conversion circuit 114 uses the upper limit SXi = 15 of the prediction error range in the category of category number Ji = 4 and the divisor data OU which is a power of two.
The exponent part Mi of i is output.

The quotient data calculation circuit 115 includes the prediction circuit 11
8 predicted value Pi = D _i-1 = 35 and the upper limit value SXi
= 15, the remainder data Ei = 6 is subtracted and further divided by the divisor data OUi = 2 ^Mi = 8 to obtain and output the quotient data Ni = 5.

The synthesizing circuit 116 multiplies the quotient data Ni = 5 by the divisor data OUi = 2 ^Mi = 8 to obtain the offset Fi, and adds the remainder data Ei = 6 to this to complete the decoding data Di = 46. Is obtained and output from the terminal 117. This completes the decoding of the data Di.

By the way, since the divisor data OUi is a power of 2, it is not necessary for the division circuit 124 to perform a normal division, and the quotient data Ni is obtained by removing the lower Mi = 3 bits of the data Di, and the multiplication circuit 125 does not need to perform a normal multiplication, and an offset F which is a multiplication result can be obtained by adding 0 of Mi bits to the lower part of the quotient data Ni.
i is obtained. Further, in the adder 126, the remainder data Ei which is one input is 0 except for the lower Mi bits, and the offset Fi which is the other input has the lower Mi bits being 0, so that it is not necessary to perform normal addition.
It suffices to replace the lower Mi bits of the offset Fi with the lower Mi bits of the remainder data Ei. Therefore, the divider 1
24, the multiplier 125, and the adder 126 are collectively realized by an extremely simple circuit, that is, a circuit that replaces the lower Mi bits of the output of the adder 123 with the lower Mi bits of the remainder data Ei and outputs the decoded data Di. it can.

Next, consider decoding when the immediately preceding data D _i-1 = 36 has been decoded as 31 having a relatively small error due to a transmission error. Since the prediction value Pi = D _i−1 , error propagation always occurs in the conventional predictive coding.
However, when actually decoding the high-efficiency coding method of the present invention, a correct result of Di = 46 is obtained. That is, no error propagation occurs. This is the quotient data N using (Equation 7) or (Equation 5) or (Equation 6) for decryption.
This is because i is obtained because correct quotient data Ni can be obtained even if the prediction value Pi used in these equations has a certain range of error from the prediction value at the time of encoding. According to (Equation 7) D
When i = 46, it can be seen that correct quotient data Ni = 5 is obtained when the predicted value Pi is 31 or more and 38 or less.

As described above, according to this embodiment, since error propagation does not always occur, transmission error resistance can be greatly improved. Furthermore, since the input data Di is sent instead of the residual data Ei and the flag G in the category to which the maximum value and the minimum value of the prediction error belong, the transmission error resistance can be further enhanced.
Also, by making the divisor data a power of 2, the coding apparatus of the present invention can be realized with a circuit scale as small as that of the conventional predictive coding apparatus.

In the above embodiment, the remainder data was coded by cutting out the lower Mi bits as a simple binary code, but it may be converted into another code and output.

When the decoded value used for prediction has a considerably large error due to a transmission error, only the predicted value Pi with a large error can be obtained by the same prediction method as at the time of encoding, and a correct decoded value can no longer be obtained.

However, if the prediction value Pi 'with a small error is obtained by a prediction method using another decoded value that is not affected by the transmission error, that is, a prediction method different from that at the time of encoding, correct decoding is possible. is there. This is because the encoding method of the present invention uses high-order bit information (category number J
i) and lower bit information (remainder data Ei) of the data to be encoded are encoded and transmitted, and if the quotient data Ni which is the only unknown in (Equation 3) is obtained, the correct decoded data Di can be obtained. Because it is done. In particular, when the divisor data OUi is large, the existence range of the quotient data Ni is narrowed, which facilitates the determination.

In the above embodiment, reversible coding was performed. However, for example, when the prediction error is large, a lossy coding method is possible by rounding and transmitting the lower bits of the remainder data Ei. .. In this case, in order to match the prediction values in the encoding device and the decoding device, it is necessary to provide a local decoding device in the encoding device and create the prediction value from the decoded data. It is needless to say that it is possible to provide the local decoding device even in the lossless encoding as in the present embodiment.

The present invention is not limited to these embodiments, various prediction methods can be applied, and arithmetic coding or the like can be applied as an entropy coding method.

[0078]

As described above, the present invention is a high-efficiency coding method characterized by transmitting the upper bit information of the prediction error and the lower bit information of the input data, without lowering the coding efficiency. The transmission error tolerance can be greatly improved, and its practical effect is great.

[Brief description of drawings]

FIG. 1 is a block configuration diagram of an encoding device and a decoding device in an embodiment using the high efficiency encoding method and the decoding method of the present invention.

[Explanation of symbols]

 101 Input terminal of data Di to be encoded 102 Prediction circuit 103 Subtraction circuit 104 Classification circuit 105 Conversion circuit 106 Residue arithmetic circuit 107 Encoding circuit 108 Output terminal of encoded data Ci 112 Input terminal 113 of encoded data Ci Decoding circuit 114 Transformation Circuit 115 Quotient data calculation circuit 116 Compositing circuit 117 Output terminal of decoded data Di 118 Prediction circuit

Claims (3)

[Claims] 1. A first step of obtaining a prediction value of the input data by using a sampled and quantized signal as input data and obtaining a prediction error which is a difference between the input data and the prediction value, and the prediction. A second step of classifying the errors according to their magnitudes and outputting a category number representing a corresponding category; and a difference between a predetermined upper limit value and a predetermined lower limit value defining the range of the relevant category. The method further comprises: a third step of dividing the input data with a predetermined value as a divisor to obtain a remainder; and a fourth step of encoding and outputting the category number and the remainder, the fourth step comprising: The same codeword is assigned to the category number to which the maximum value of the prediction error belongs and the category number to which the minimum value belongs, and the prediction error is divided into the category to which the maximum value or the minimum value belongs. If similar, the fourth step directly encodes the input data instead of the residual data, and outputs the encoded data. 2. The third step sets a divisor to a value obtained by adding 1 to the difference between the upper limit value and the lower limit value, and sets the upper limit value and the lower limit value of each category so that the divisor becomes a power of 2. The high-efficiency coding method according to claim 1, wherein 3. The fourth step comprises the steps of entropy coding the category number and converting the remainder into a variable length code according to the size of the divisor. High efficiency coding method.