JP3509346B2 - Encoding device 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 quantizing device and a quantizing method for quantizing a residual between an input signal such as a digital audio signal and a digital image signal and a prediction value generated from the input signal.

[0002]

2. Description of the Related Art Predictive coding is known as coding for compressing the amount of transmission information such as digital audio signals and digital image signals. For example, one-dimensional DPCM
Forms the difference (residual) between the input sample value and the predicted value in the time direction, and the two-dimensional DPCM forms the residual difference between the input sample value and the predicted value in the spatial direction. Since the digital information signal has a correlation in the time direction and the spatial direction, the residual error is concentrated near 0. Therefore, the residual signal can be quantized with a smaller number of bits than the original number of quantized bits, whereby the amount of information can be compressed. Further, the amount of information can be further compressed by performing the variable length coding by utilizing the concentration of the distribution of the residual signal.

Generally, in the frequency distribution of the residual signal, the values are concentrated near 0. Therefore, as a quantizer for the residual signal, the quantization step width near 0 is made fine and the level is Non-linear quantizers are known that make the quantization step width coarser as the size increases. A conventional quantizing device including this non-linear quantizing targets all levels of a residual signal that can be generated as a quantizing object. For example, when one sample (one pixel) of the digital image signal is quantized to 8 bits, the residual signal may have a value of (-255 to +255). In the conventional quantizer, this entire range is targeted for quantization.

[0004]

Here, in the conventional quantizer, since the quantized value is represented by the bit representation by the binary number, the number of quantization steps has been limited to an even number. When the number of quantization steps is limited to an even number, and when the difference data is quantized, if the design is made to have 0 as a decoded value, the quantization does not become positive and negative symmetry. Alternatively, even if the number of quantization steps is 2n-1, it is wasteful because the number of quantization steps of 2n must be used.

When the number of quantization bits is changed from 2 bits to 3 bits, the number of quantization steps is 4.
To 8 As in this example, the degree of change in the number of quantization steps with respect to the number of quantization bits is large. This means that when the number of quantization bits is varied for the purpose of controlling the transmission data amount to be approximately constant, the change in the quantization step width with respect to the change in the number of bits becomes large, and the change in the image quality of the restored image becomes large. there were.

Accordingly, the purpose of this invention, it quantization step number is either odd / even, also, Oyo quantizer capable of suppressing the degree of change in the number of quantization steps
And providing a quantization method .

[0007]

In order to achieve the above-mentioned object, the invention of claim 1 is a code for quantizing an input digital signal and performing a variable length coding on the quantized digital signal. One of a plurality of variable length coding methods based on the number of quantization steps, and a quantization code generation means for generating a quantization code by quantizing a digital signal with a predetermined number of quantization steps And a variable length coding means for performing variable length coding on the quantized code.

According to the present invention of claim 8 , the residual signal between the input digital signal and the prediction value is quantized, and the quantized residual signal is subjected to variable length coding. Quantize the signal with a predetermined number of quantization steps to generate a quantization code, and determine one of a plurality of variable length coding methods based on the number of quantization steps, And a variable length coding means for performing variable length coding on the coded code.

According to a ninth aspect of the present invention, an image signal of the second layer in which the number of pixels is smaller than that of the image signal of the first layer is formed, and the image signal of the first layer is converted from the image signal of the second layer. In a coding device that performs prediction, generates a difference value between a first layer image signal and a predicted value, and transmits the difference value in the second layer image signal, the difference value and the second layer image signal are first
And a second quantizing device, and the first and second quantizing devices generate a quantized code by quantizing the input digital signal with an arbitrary number of quantization steps. An encoding device having means and a quantizer for generating a quantized value by associating a quantized code with a predetermined variable length code.

According to a tenth aspect of the present invention, an image signal of the second layer in which the number of pixels is smaller than that of the image signal of the first layer is formed, and the image signal of the first layer is converted from the image signal of the second layer. In a coding device that performs prediction, generates a difference value between a first-layer image signal and a predicted value, and transmits the difference value in the second-layer image signal, a second-layer image signal is generated using the first-layer image signal. A forming means for forming an image signal of a hierarchy, a predicting means for generating a predicted value by predicting a value of the first image signal from an image signal of the second hierarchy; and an image signal of the first hierarchy and the predicted value. A first quantizing means for generating a difference value, a first quantizing means for quantizing the difference value, and a second quantizing means for quantizing the image signal of the second layer; Quantizes the difference value with an arbitrary number of quantization steps to generate a quantized code. To generate a quantized value by making it correspond to a variable-length code determined in advance, and the second quantizing means quantizes the image signal of the second layer with an arbitrary number of quantization steps to generate a quantized code. An encoding device is characterized in that a quantized value is generated by associating a quantized code with a predetermined variable length code.

The present invention can make the number of quantization steps substantially odd. Therefore, when the difference data (residual signal) is quantized, the quantization becomes positive and negative symmetry even if it is designed to have 0 as a decoded value. When the number of quantization steps is 2n-1, the quantization can be performed using the optimal number of quantization steps.

[0012]

BEST MODE FOR CARRYING OUT THE INVENTION An embodiment of the present invention will be described below with reference to the drawings. Here, an example of a predictive coding encoder that generates a residual signal is shown in FIG. Figure 1A
At, for example, a digital image signal is supplied to the input terminal indicated by 11. The digital image signal is supplied to the prediction unit 12 and the subtractor 13, and the subtractor 13 subtracts the prediction value generated by the prediction unit 12 from each pixel value, and a residual signal is generated at its output.

This residual signal is supplied to the quantizer 15 via the blocking circuit 14 and quantized with a smaller number of quantization bits than the original number of quantization bits. This quantizer 15
The present invention is applied to, and an example thereof is shown in FIG. The quantization unit 15 generates a variable-length coded quantized value and side information. Side information is output terminal 17
Taken out. Practically, as shown in FIG. 1B, the output of the subtractor 13 is supplied to the local decoder 12 ′, and the decoded output from the local decoder 12 ′ is supplied to the subtractor 13.

A decoder corresponding to the encoder shown in FIG. 1 is shown in FIG. Input terminal 2 with quantized value and side information
1 and 22 respectively. The quantized value and the side information are supplied to the inverse quantization unit 24, and the quantized value is converted into a quantized decoded value (representative value).

The quantized decoded value from the inverse quantizer 24 is supplied to the adder 25. The output of the adder 25 is the output terminal 27.
And is supplied to the prediction unit 26. Prediction unit 2
The predicted value is obtained from 6, and this predicted value is supplied to the adder 25. At the time of the encoder, when the refresh sample value itself is periodically inserted to prevent the accumulation of errors, the adder 25 does not perform the addition operation,
The quantized decoded value is led to the output terminal 27 as it is.

FIG. 3 is for explaining an example of prediction and shows a part of one screen. Each of a to h indicates a locally decoded pixel value, and each of A to P indicates a pixel value before being encoded. The predicted value A ′ for the pixel value A is formed using the locally decoded pixel values in the vicinity. For example, A '= 4
c-3 (b-f), A '= f + c-d, etc.
The predicted value of is formed. The prediction value for the pixel value B is also calculated using the locally decoded value by the same calculation.

Then, the predicted value, for example, A'is subtracted from the true pixel value, for example, A (that is, Δa = A-A ') to form the residual Δa. The blocking circuit 14 divides the residual into blocks of 4 pixels × 4 pixels as shown in FIG. When handling a digital audio signal, a prediction value in the time direction is formed and a block of a one-dimensional residual signal is formed.

Here, FIG. 4 shows an embodiment of the detailed configuration of the quantizer 15. Subtractor 1 via input terminal 31
The residual signal from 3 is supplied to the quantizer 15. The quantizer 15 includes a ROM 32, a quantizer 34, and an activity detection circuit 33. As described above, the residual signal is divided into blocks of 4 pixels × 4 pixels, and the blocked data is supplied to the activity detection circuit 33 and the quantizer 34.

Dynamic range is used as an example of the activity determination method in the activity detection circuit 33. In addition to the dynamic range, the sum of absolute differences with respect to the average value, the absolute value of standard deviation, and the like can be used as activities.

Then, the activity detecting circuit 33 determines the quantization characteristic based on the detected activity. In this example, the number of quantization steps is determined, and a control signal indicating the number of quantization steps is supplied to the quantizer 34 and the ROM 32. Quantizer 34
Quantizes the data of the blocked residual signal with this number of quantization steps. Further, the ID indicating the number of quantization steps is taken out from the output terminal 17 as side information from the activity detection circuit 33. The quantization code and ID from the quantizer 34 are supplied to the ROM 32 as an address, and the variable length code is output from the ROM 32 to the output terminal 1
6 is taken out.

As described above, the quantizer 15 includes the quantizer 34.
By combining the ROM 32 with the ROM 32, it is possible to use an odd number of quantization steps, unlike the conventional even number of quantization steps. That is,
What is transmitted or recorded is a variable-length encoded quantized value, and the number of quantized steps in the preceding stage can be arbitrarily selected. By properly designing the variable length code conversion table stored in the ROM 32, an increase in the amount of transmission information can be suppressed. An example of the code conversion table will be described below. This example
For simplification of description, this is an example when the number of quantization steps is 3. FIG 5 is (quantization code from the quantizer 34) input values (- 255～ + 255) and shows the relationship between the output value to be read (-255～ + 255) from the ROM 32. In this way, by setting the number of quantization steps to an odd number (3), the representative value of one quantization code is always 0, and the quantization step width Δ can be arranged symmetrically in positive and negative directions.

The activity detecting circuit 33 detects the maximum value MAX and the minimum value MIN of the level distribution of the residual signal of the block, and calculates the dynamic range DR (= MAX-MIN) as described above. When the detected dynamic range DR is 0 to 63, the number of quantization steps is 1. When DR is 64 to 128, the number of quantization steps is 3, and when DR is 129 to 191, the number of quantization steps is 5, and DR is 192.
When the number is up to 255, the number of quantization steps is 7. The quantization step width Δ is determined based on this odd number of quantization steps.

The ROM 32 stores a correspondence table of the number of quantization steps, a quantization code, and a Huffman code as an example of variable length coding. This Huffman code is transmitted. As shown in FIG. 6, when the number of quantization steps is 1, the quantization code is only 1 and the Huffman code is 0. Further, when the number of quantization steps is 3, the quantization code is 1, 2, 3; when the quantization code is 1, the Huffman code is 10, and when the quantization code is 2, the Huffman code is 0, When the quantization code is 3, the Huffman code is 11. And
When the number of quantization steps is 5, the quantization code is 1, 2,
When the quantization code is 1, the Huffman code is 1100, when the quantization code is 2, the Huffman code is 100, and when the quantization code is 3, the Huffman code is 0, If the code is 4, the Huffman code is 101 and the quantization code is 5.
In this case, the Huffman code is 1101.

Further, when the number of quantization steps is 7, the quantization code is 1, 2, 3, 4, 5, 6, 7, and when the quantization code is 1, the Huffman code is 00100 and the quantization code is If is 2, the Huffman code is 000
And when the quantization code is 3, the Huffman code is 0
When the quantization code is 4, the Huffman code is 10, when the quantization code is 5, the Huffman code is 11, and when the quantization code is 6, the Huffman code is 0011 and the quantization code is 7. In this case, the Huffman code is 00101. In this way, a small number of bits is assigned to a quantized code corresponding to a 0 level with a high occurrence frequency, and a large number of bits is assigned to a quantized code with a low occurrence frequency. The number of bits can be compressed. Note that the above-described selection of the number of quantization steps and the variable length coding table are examples, and it is also possible to use other quantization characteristics and variable length coding.

The present invention can be applied not only to the predictive coding encoder shown in FIG. 1 but also to a quantizing device for a hierarchical coding encoder as described below. The hierarchical coding apparatus described here can prevent an increase in the number of pixels to be coded by performing prediction between layers and using a simple arithmetic expression for inter-layer data.

This hierarchical coding method will be described with reference to FIG. FIG. 7 shows, as an example, a schematic diagram between four layers, where the first layer is the lowest layer (original image) and the fourth layer is the highest layer. For example, when averaging spatially corresponding lower layer data of 4 pixels is adopted as the upper layer data generation method, the upper layer data is M,
If the lower layer pixel values are x ₀ , x ₁ , x ₂ , and x ₃ , the number of transmission pixels does not increase, and the number of pixels may be 4 pixels.

That is, by using M, x ₀ , x ₁ and x ₂ , the non-transmission pixel x is obtained by a simple arithmetic expression: x ₃ = 4 · M− (x ₀ + x ₁ + x ₂ ) ... (1) ₃ can be easily restored. Each layer data is 4 in the lower layer.
It is generated by pixel averaging. Therefore, for example, it is possible to restore all the data by the equation (1) without transmitting the shaded data in the figure.

Next, FIG. 8 shows a configuration example of five layers of hierarchical data by averaging. It is assumed that the first layer is the resolution level of the input image. The first layer has a block size (1
X1) has the data structure. The second layer data is the first
It is generated by averaging four pixels of hierarchical data. In this example, the second layer data X ₂ (0) is generated by the average value of the first layer data X ₁ (0) to X ₁ (3). X
_2nd layer data X ₂ (1) to X adjacent to ₂ (0)
₂ (3) is similarly generated by averaging four pixels of the first layer data. This second layer has a block size (1/2 ×
1/2) data structure.

Further, the third hierarchical data is generated by averaging four pixels of the spatially corresponding second hierarchical data. Similar to the above, this third layer has a block size (1/4 ×
It has a data structure of 1/4). Similarly, the data of the fourth layer is also controlled by the data of the third layer, and the data structure thereof has a block size (1/8 × 1/8).
Finally, the fifth layer data X ₅ (0), which is the highest layer
Are generated by averaging the fourth layer data X ₄ (0) to X ₄ (3). The data structure of the fifth layer has a block size (1/16 × 1/16).

By applying the class classification adaptive prediction to the upper layer data with respect to the hierarchical structure data in which the increase in the number of pixels to be encoded is prevented, the lower layer data is predicted,
A configuration example for reducing the signal power by generating the difference between the lower layer data and its predicted value (that is, the residual signal) will be described with reference to the block diagram shown in FIG. This Figure 9
The structural example at the encoder side of hierarchical encoding is shown. Input terminal 7
The first layer data d0 is supplied to the averaging circuit 72 and the subtractor 76 as input image data d0 via 1. The first layer data is the image data of the original resolution.

The supplied input pixel data d0 is subjected to ¼ averaging processing by the 2 pixel × 2 pixel block shown in FIG. 8 in the averaging circuit 72 to generate hierarchical data d1. This hierarchical data d1 corresponds to the second hierarchical data shown in FIG. The generated hierarchical data d1 is supplied to the averaging circuit 73 and the subtractor 77.

The averaging circuit 73 is applied to the hierarchical data d1.
Then, the same processing as that of the averaging circuit 72 described above is performed to generate the hierarchical data d2. This hierarchical data d2 corresponds to the third hierarchical data. Generated hierarchical data d2
Is supplied to the averaging circuit 74 and the subtractor 78. Similarly, the averaging circuit 74 also performs the same processing on the hierarchical data d2 as the above-described averaging circuits 72 and 73 to generate hierarchical data d3. This hierarchical data d3
Corresponds to the fourth layer data. The generated hierarchical data d3 is supplied to the averaging circuit 75 and the subtractor 79. Further, in the averaging circuit 75, similarly, the hierarchical data d3
Is subjected to the same processing as that of the averaging circuits 72, 73 and 74 described above, and hierarchical data d4 is generated. This hierarchical data d4 corresponds to the fifth hierarchical data. The generated hierarchical data d4 is supplied to the quantizer 84.

Then, prediction is performed between layers for these five hierarchical data. First, in the fifth layer,
Quantization processing for compression is executed in the quantization unit 84. The output data d21 of the quantizer 84 is supplied to the inverse quantizer 88. The output of the quantizer 84 is taken out to the output terminal 106 as data of the fifth layer. The output data d16 of the inverse quantization unit 88 to which the encoded data d21 is supplied is supplied to the class classification adaptive prediction circuit 92.

In the class classification adaptive prediction circuit 92, a prediction process is performed using the data d16 to generate a prediction value d12 of the fourth hierarchical data, and this prediction value d12 is used as the subtractor 7
9 is supplied. In the subtractor 79, the averaging circuit 74
The difference value between the hierarchical data d3 supplied from the above and the predicted value d12 is obtained, and the difference value d8 is supplied to the quantization unit 83.

The quantizing unit 83, to which the difference value d8 is supplied, carries out the compression process in the same manner as the quantizing unit 84. The output data of the quantizer 83 is supplied to the calculator 96 and the inverse quantizer 87. In this computing unit 96, 4 pixels to 1
A process of thinning out pixels is performed. The data d20 output from the arithmetic unit 96 is taken out to the output terminal 105 as the output data of the fourth layer.

The output data d15 of the dequantizer 87 is supplied to the class classification adaptive prediction circuit 91. In the class classification adaptive prediction circuit 91, a prediction process is performed using the data d15, a predicted value d11 of the third hierarchical data is generated, and the predicted value d11 is supplied to the subtractor 78. In the subtractor 78, the difference value between the data d2 supplied from the averaging circuit 3 and the predicted value d11 is obtained, and the difference value d7 is supplied to the quantization unit 82.

The output data of the quantizer 82 to which the difference value d7 is supplied is supplied to the calculator 95 and the inverse quantizer 86. The arithmetic unit 95 performs a process of thinning out one pixel from four pixels. The third hierarchical data d19 output from the calculator 95 is taken out to the output terminal 104.

The output data d14 of the dequantization unit 86 to which the encoded data is supplied from the quantization unit 82 is supplied to the class classification adaptive prediction circuit 90. Class classification adaptive prediction circuit 9
At 0, the prediction process is performed using the data d14, the predicted value d10 of the second hierarchical data is generated, and is supplied to the subtractor 77. In the subtractor 77, a difference value between the data d1 supplied from the averaging circuit 72 and the predicted value d10 is obtained, and the difference value d6 is supplied to the quantizer 81.

The output data of the quantizer 81 to which the difference value d6 is supplied is supplied to the calculator 94 and the inverse quantizer 85. In this arithmetic unit 94, a process of thinning out one pixel from four pixels is performed. The second hierarchical data d18 output from the arithmetic unit 94 is taken out to the output terminal 103.

The output data d13 of the inverse quantizer 85 to which the encoded data is supplied from the quantizer 81 is supplied to the class classification adaptive prediction circuit 89. Class classification adaptive prediction circuit 8
In 9, the prediction process is performed using the data d13, the predicted value d9 of the first layer data is generated, and the predicted value d9 is supplied to the subtractor 76. The subtractor 76 obtains a difference value between the input pixel data d0 supplied from the input terminal 71 and the predicted value d9, and the difference value d5 is supplied to the quantizer 80.

The output data of the quantizer 80 to which the difference value d5 is supplied is supplied to the arithmetic unit 93. This calculator 93
In, the process of thinning out one pixel from four pixels is performed. The first layer data d17 output from the arithmetic unit 93 is taken out to the output terminal 102. The quantizers 80 to 84 described above
Each has the structure as shown in FIG. 4 described above. That is, the number of quantization steps is an odd number, and a ROM for variable length coding is provided. Inverse quantizer 85-88
Converts the variable-length quantized value into a quantized code, and the quantized code into a representative value.

Class classification adaptive prediction circuits 89, 90, 9
1 and 92 classify the pixel of the lower hierarchy to be predicted based on the level distribution of a plurality of pixels spatially nearby (included in the upper hierarchy). Then, a table of prediction coefficients or prediction values for each class, which is obtained by learning in advance, is stored in the memory, and a plurality of prediction coefficients or one prediction value corresponding to the class is read from the memory. The prediction value is used as it is, and the prediction coefficient is generated by linear linear combination with a plurality of pixels. Such class classification adaptive prediction is disclosed in, for example, Japanese Patent Application No. 4-155719 proposed by the present applicant.

Next, FIG. 10 shows an example of the configuration on the decoder side of hierarchical encoding corresponding to the above encoder. The respective hierarchical data generated on the encoder side are supplied to the input terminals 131, 132, 133, 134 and 135 as d30 to d34, respectively. Then, the data of each layer is supplied to the inverse quantizers 146, 147, 148, 149, 150, respectively.

First, the fifth layer input data d34 is subjected to a decoding process corresponding to the quantization applied by the encoder in the inverse quantizer 150, and becomes the image data d39,
It is supplied to the class classification adaptive prediction circuit 162 and the calculator 158. Further, the image data d39 is taken out from the output terminal 167 as an image output of the fifth layer.

In the class classification adaptive prediction circuit 162, the fourth
The class classification adaptive prediction is performed on the image data of the hierarchy, and the prediction value d47 of the fourth hierarchy data is generated. Data d38 from the inverse quantization unit 149 (that is, difference value)
And the predicted value d47 are added by the adder 154. The image data d43 is supplied from the adder 154 to the arithmetic unit 158, and the arithmetic unit 158 executes the operation of Expression (1), and the image data d39 and the image data d43 supplied from the inverse quantization unit 150 are included in the fourth layer. All pixel values of are restored. In this calculator 158, the restored all pixel values are supplied to the class classification adaptive prediction circuit 161 and the calculator 157 as image data d51. Also, the image data d51
Is taken out from the output terminal 166 as an output of the fourth layer.

In the class classification adaptive prediction circuit 161, similarly to the above, the class classification adaptive prediction is performed on the image data of the third layer, and the predicted value d46 of the third layer data is generated. The data d37 from the inverse quantizer 148 and the predicted value d
46 and 46 are added by the adder 153. The image data d42 is supplied from the adder 153 to the calculator 157, and the calculator 1
In 57, the calculation of Expression (1) is executed, and all the pixel values of the third layer are restored from the image data d51 and the image data d42 supplied from the calculator 158. In this computing unit 157, the all pixel values restored are image classification data d50, and the class classification adaptive prediction circuit 160 and computing unit 156.
Is supplied to. The image data d50 is taken out from the output terminal 165 as the output of the third layer.

Further, in the class classification adaptive prediction circuit 160, the class classification adaptive prediction is performed on the image data of the second layer in the same manner as described above, and the prediction value d45 of the second layer data is obtained.
Is generated. The data d36 from the inverse quantization unit 147 and the predicted value d45 are added by the adder 152. Adder 1
The image data d41 is supplied from 52 to the calculator 156,
The arithmetic unit 156 executes the arithmetic operation of the equation (1), and the image data d50 and the image data d50 supplied from the arithmetic unit 157.
From 41, all pixel values of the second layer are restored. In this computing unit 156, the restored all pixel values are the image data d
As 49, it is supplied to the class classification adaptive prediction circuit 159 and the calculator 155. The image data d49 is the second
The output of the hierarchy is taken out from the output terminal 164.

Further, in the class classification adaptive prediction circuit 159, the class classification adaptive prediction is performed on the image data of the first layer in the same manner as described above, and the prediction value d44 of the first layer data is obtained.
Is generated. The data d35 from the inverse quantization unit 146 and the predicted value d44 are added by the adder 151. Adder 1
The image data d40 is supplied from 51 to the calculator 155,
The arithmetic unit 155 executes the arithmetic operation of the equation (1), and the image data d49 and the image data d supplied from the arithmetic unit 156.
From 40, all pixel values of the first layer are restored. In this computing unit 155, the restored all pixel values are the image data d
As 48, the output of the first layer is taken out from the output terminal 163. Thus, in hierarchical coding in which the number of pixels to be coded is prevented from increasing, it is possible to improve coding efficiency by introducing the class classification adaptive prediction.

As a concrete application example of the above-mentioned hierarchical coding system, when a database of still image of high-definition television is constructed, the lowest layer data, that is, the first layer (original image) data is reproduction data of high-definition resolution. Yes, the second layer data becomes the reproduction data of the standard resolution, and the highest layer data, that is, the fifth layer data,
This is low-resolution playback data for high-speed search.

When compression coding is used for the purpose of reducing the amount of information, the reproduced image data obtained by the decoding device does not always match the input original image data, but it is visually deteriorated. Can be made undetectable. Further, the average value is not limited to the simple average value, but a weighted average value may be formed.

The present invention can also be applied to the quantization of the residual signal generated by the predictive coding other than the predictive coding described above. The present invention can also be applied to a system having a buffering configuration that controls the amount of generated data by controlling the quantization step width Δ.

[0052]

According to the present invention, an odd number of quantization steps can be used as the number of quantization steps,
When the difference data (residual signal) is quantized, when the design is made to have 0 as a decoded value, the quantization characteristic becomes positive and negative symmetry. Further, according to the present invention, the degree of freedom in designing the quantizer is increased and the coding efficiency can be improved.

[Brief description of drawings]

FIG. 1 is a block diagram of a predictive coding encoder to which the present invention can be applied.

FIG. 2 is a block diagram of a predictive coding decoder to which the present invention can be applied.

FIG. 3 is a schematic diagram for explaining an example of predictive coding.

FIG. 4 is a block diagram of an embodiment of a quantization device of the present invention.

FIG. 5 is a schematic diagram used for explaining the number of quantization steps.

FIG. 6 is a schematic diagram used to explain variable-length coding according to the present invention.

FIG. 7 is a schematic diagram used to describe an example of hierarchical encoding to which the present invention can be applied.

FIG. 8 is a schematic diagram used to describe an example of hierarchical encoding.

[Fig. 9] Fig. 9 is a block diagram illustrating an example of a configuration on the encoding side of hierarchical encoding.

[Fig. 10] Fig. 10 is a block diagram illustrating an example of a configuration on the decoding side of hierarchical encoding.

[Explanation of symbols]

12 Predictor 13 Subtractor 15 Quantizer

   ─────────────────────────────────────────────────── ─── Continued front page    (72) Inventor Kunio Kawaguchi               6-735 Kita-Shinagawa, Shinagawa-ku, Tokyo               Knee Co., Ltd.                (56) References JP-A-4-314266 (JP, A)                 Edited by Hiroshi Yasuda, 1.2.2 DCT method               Expression algorithm, multimedia code               International Standardization, Maruzen Publishing Co., Ltd.,               June 30, 1991, p19,22,23,90

Claims (21)

(57) [Claims] 1. Quantizing an input digital signal,
In a coding device for performing variable length coding on the quantized digital signal, quantization code generating means for quantizing the digital signal with a predetermined number of quantization steps to generate a quantization code, A code having variable length coding means for determining one of a plurality of variable length coding methods based on the number of quantization steps and performing variable length coding on the quantized code. Device. 2. The predetermined number of quantization steps is an odd number.
The code according to claim 1, which is a step of
Device. 3. A memory means for storing a code conversion table for each number of quantization steps, in which the supplied quantization code and the variable length code are associated with each other, wherein the quantization means is the memory means. The encoding device according to claim 1, wherein the quantized code and the variable length code are associated with each other. 4. A detecting local properties of the digital signal, characterized in that it comprises means for selecting a pre SL amount coca number of steps based on the local characteristics, codes of claim 1 Device. 5. The variable-length code is assigned a small number of bits for a quantized code with a high occurrence frequency and a large number of bits for a quantized code with a low occurrence frequency. Coding device according to claim 1, characterized in that Wherein said quantization code generation means, and generating a quantization code by arranging the quantization step width on the positive and negative symmetrical encoding apparatus according to claim 1. 7. The encoding apparatus according to claim 1, wherein a representative value of one quantization code is 0. 8. An encoding device for quantizing a residual signal between an input digital signal and a predicted value and performing variable length coding on the quantized residual signal, wherein the residual signal is and quantization code generation means for generating a quantized code, based on the number of quantization steps determining one of a plurality of variable length coding method are quantized by a predetermined number of quantization steps, the quantization An encoding device, comprising: variable length encoding means for performing variable length encoding on a code. 9. An image signal of the second layer in which the number of pixels is smaller than that of the image signal of the first layer is formed, the image signal of the first layer is predicted from the image signal of the second layer, and the image signal of the first layer is predicted. 1
The difference value between the image signal of the hierarchy and the predicted value is generated, and the second value is generated.
In an encoding device for transmitting the difference value in an image signal of a layer, the difference value and the image signal of the second layer are supplied to first and second quantizers, and the first and second And a quantizing device for quantizing the input digital signal with an arbitrary number of quantizing steps to generate a quantizing code, and the quantizing code corresponds to a predetermined variable length code. And a quantizing unit for generating a quantized value. 10. An image signal of a second layer in which the number of pixels is smaller than that of an image signal of the first layer is formed, the image signal of the first layer is predicted from the image signal of the second layer, and the image signal of the first layer is predicted. In a coding device that generates a difference value between a one-layer image signal and a predicted value and transmits the difference value in the second-layer image signal, the second-layer image signal is generated using the first-layer image signal. Forming means for forming an image signal of a hierarchy; prediction means for generating the predicted value by predicting a value of the first image signal from the image signal of the second hierarchy; and an image signal of the first hierarchy It has a generating means for generating a difference value from the predicted value, a first quantizing means for quantizing the difference value, and a second quantizing means for quantizing the image signal of the second layer. , The first quantizing means measures the difference value by an arbitrary number of quantization steps. Generate a quantized code, generate a quantized value by associating the quantized code with a predetermined variable-length code, and the second quantizing means generates the quantized value of the image signal of the second layer. An encoding apparatus, which quantizes an arbitrary number of quantization steps to generate a quantized code, and generates a quantized value by making the quantized code correspond to a predetermined variable length code. 11. The encoding apparatus according to claim 10, wherein the quantization step includes an odd step. 12. The first to second quantizing means holds a table storing a value of a variable length code corresponding to the number of quantization steps and the quantization code. The encoding device according to claim 10. 13. The encoding device further includes an inverse quantizer that inversely quantizes the quantized image signal of the second layer, and the predictor that is inversely quantized by the inverse quantizer. The encoding apparatus according to claim 1, wherein the prediction value is generated by using the image signal of the second layer that has been encoded. 14. The encoding apparatus according to claim 10, wherein the forming unit forms the image signal of the second layer by obtaining an average value of the image signals of the first layer. 15. The variable-length code is assigned a small number of bits for a quantized code with a high occurrence frequency,
11. The encoding device according to claim 10, wherein a large number of bits are assigned to a quantized code having a low occurrence frequency. 16. The encoding apparatus according to claim 10, wherein the quantization code generation means generates a quantization code by arranging the quantization step widths in positive and negative symmetry. 17. The encoding apparatus according to claim 10, wherein a representative value of one quantization code is 0. 18. Quantizing an input digital signal
The variable length of the quantized digital signal.
In a coding method for performing coding, the digital signal is quantized by a predetermined number of quantization steps.
To generate a quantized code.
And a plurality of variable length coding methods based on the number of quantization steps
Determine one of the modalities and
A variable length coding step for performing variable length coding.
And an encoding method characterized by: 19. The input digital signal and predicted value
Quantized the residual signal of
In the coding method for performing variable length coding, the residual signal is quantized by a predetermined number of quantization steps.
A quantization code generation step for generating a quantization code, and a plurality of variable length coding methods based on the number of quantization steps
Determine one of the modalities and
A variable length coding step for performing variable length coding.
And an encoding method characterized by: 20. The number of pixels is smaller than that of the image signal of the first layer.
A reduced second layer image signal is formed, and the second layer image signal of the second layer is formed.
Predicting the image signal of the first layer from the image signal,
A difference value between the image signal of one layer and the predicted value is generated,
Encoding method for transmitting the difference value in a two-layer image signal
In the method, the difference value and the image signal of the second layer are first and
A second quantizer, wherein the first and second quantizers are the input digitizers.
Quantize the Tal signal by quantizing it with an arbitrary number of quantization steps
A quantization code generating step for generating a code, and said quantity
Make the child code correspond to a predetermined variable length code.
And a quantization step to generate a quantized value by
An encoding method characterized by the above. 21. The number of pixels is smaller than that of the image signal of the first layer.
A reduced second layer image signal is formed, and the second layer image signal of the second layer is formed.
Predicting the image signal of the first layer from the image signal,
A difference value between the image signal of one layer and the predicted value is generated,
Encoding method for transmitting the difference value in a two-layer image signal
Method, using the image signal of the first layer, the image of the second layer
A forming step of forming a signal, and calculating the value of the first image signal from the second layer image signal.
A prediction step of generating the prediction value by performing prediction, and a difference value between the image signal of the first layer and the prediction value
Generating step, a first quantizing step for quantizing the difference value, and a second quantizing step for quantizing the image signal of the second layer.
And the first quantizing step converts the difference value into an arbitrary quantum value.
Quantized by the number of quantization steps to generate a quantized code,
Make the quantization code correspond to a predetermined variable length code
To generate a quantized value, and the second quantizing step is performed by the second layer image signal.
Is quantized by an arbitrary number of quantization steps, and the quantization code is
A variable length code that is generated and that has the predetermined quantization code.
Is characterized by generating a quantized value by corresponding to
The encoding method.