JP2002027459A - Apparatus and method for encoding dynamic image as well as recording medium for recording image encoding program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To solve the problem of unsuitability for fast encoding of a large amount and a lengthy contents because a conventional encoding system starts primarily encoding after analyzing a feature of encoding an overall image to be encoded so that its encoding process is not executed in real time requiring at least twice as long a time as normal encoding with the result that the processing is not executed in real time. SOLUTION: A method for encoding a dynamic image comprises the steps of converting image feature information into occurrence probability, calculating the target encoding amount from the probability, then regulating the quantization step width in response to the target encoding amount, and variably bit rate encoding the input image using the quantized step width.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to high-efficiency video coding, and more particularly to a video coding apparatus, a video coding method, and a video which perform real-time coding while controlling the bit rate during coding. The present invention relates to a recording medium on which an image encoding program is recorded.

[0002]

2. Description of the Related Art FIG. 7 is a block diagram showing a conventional moving picture coding apparatus disclosed in, for example, Japanese Patent Application Laid-Open No. 6-141298. In FIG. , 3 an image encoding unit, 11 a video tape recorder (hereinafter referred to as VTR) for outputting an image signal, and 12 an inter-frame prediction signal output from the image signal output from the VTR 11 by an inter-image predictor 17. Subtract
A prediction subtractor 13 for outputting a prediction residual signal; and D, which performs a discrete cosine transform on the prediction residual signal output from the prediction subtractor 12
CT and 14 are quantizers for quantizing the prediction residual signal after transformation by the DCT 13 with a quantization step width designated by the encoding control unit 22.

[0005] Reference numeral 15 denotes a local decoder which performs inverse quantization and inverse discrete cosine transform of the prediction residual signal quantized by the quantizer 14 and outputs a reproduced prediction residual signal. Output inter-frame prediction signal and local decoder 1
5 is an adder for adding the playback prediction residual signal output from the adder 5 and outputting a playback image signal; and 17 is an adder for delaying the playback image signal output from the adder 16 by one or more frames to perform motion compensation. This is an inter-picture predictor that generates a prediction signal.

[0004] Reference numeral 18 denotes a variable-length encoder that performs variable-length encoding on the prediction residual signal quantized by the quantizer 14 and outputs encoded data, 19 denotes a selector, and 20 temporarily stores the encoded data. Buffer to be stored, 21 is a selector, 22
Is an encoding control unit that controls the quantization step width in accordance with the degree of sufficiency of the buffer 20.

[0005] Reference numeral 23 denotes a temporary code amount counter that counts the code amount generated per unit time and outputs the count value as a temporary transfer rate. Reference numeral 24 denotes a temporary transfer rate per unit time that is used again in actual encoding. A temporary transfer rate memory for storing, 25 determines a conversion characteristic from the temporary transfer rate to the target transfer rate and outputs a code amount controller for outputting the conversion characteristic information. Accordingly, the target transfer rate setting unit sets the target transfer rate from the temporary transfer rate per unit time output from the temporary transfer rate memory 24.

Next, the operation will be described. The image encoding processing unit 3 has the same configuration as an image encoding device of an image encoding method that is a standard method such as ITU-T.
The image signal input in the image encoding processing unit 3 is input to the prediction subtractor 12. In the prediction subtractor 12, the inter-frame prediction signal output from the inter-picture predictor 17 is subtracted from the image signal, and the subtraction result is output to the DCT 13 as a prediction residual signal.

[0007] The prediction residual signal is subjected to a discrete cosine transform by the DCT 13 and guided to the quantizer 14. In the quantizer 14, quantization is performed according to the quantization step width output from the encoding control unit 22. The quantized prediction residual signal is variable-length coded by the variable-length encoder 18 and output to the buffer 20. Since the encoded data input to the buffer 20 is variable-length encoded, the code generation amount always fluctuates. However, the fluctuation is absorbed by the buffer 20 and the encoded data is output at a constant transfer rate.

On the other hand, the local decoder 15 performs inverse quantization and inverse DCT of the quantized prediction residual signal. The code is decoded into a reproduction prediction residual signal, which is output to the adder 16. The adder 16 adds the inter-frame prediction signal output from the inter-picture predictor 17 and the reproduced prediction residual signal output from the local decoder 15 to provide a reproduced image signal to the inter-picture predictor 17 Is done. Inter-picture predictor 17
In, the reproduced image signal is delayed by one or more frames to perform motion compensation, and an inter-frame prediction signal is generated and supplied to the prediction subtractor 12 and the adder 16.

Conventionally, a moving picture coding method employs a constant bit rate (CBR) coding method and a variable bit rate (VBR) coding method by controlling the output bit rate of coded data.
They can be broadly classified into methods and methods. The former is mainly used for adapting to the limitation of transmission path capacity in communication applications. On the other hand, the latter is mainly used for storing encoded data in a file medium, and under the constraint that the total code amount for a predetermined time length is constant, the bit rate is adjusted according to the properties of the image to be encoded. It fluctuates.

In the case of the fixed bit rate method, the encoding control unit 22 in the image encoding processing unit 3 sets a quantization step width according to the buffer filling amount fed back and outputs the quantization step width to the quantizer 14. The quantization step width is coarse when a large number of codes are stored in the buffer 20, and fine when the buffer 20 is almost empty, and is controlled so that overflow or underflow of the buffer 20 in which data is extracted at a constant rate does not occur. Need to be done. Therefore, the allowable range in which the quantization step width can be varied according to the properties of the image to be encoded is narrow, and image quality is likely to locally deteriorate in time.

On the other hand, in the variable bit rate method, the average picture quality can be improved by changing the bit rate according to the properties of the image to be coded under the constraint that the total code amount for a predetermined time length is constant. That is. The conventional variable bit rate coding apparatus executes a processing procedure of executing some temporary coding and reflecting the result in variable bit rate coding control of actual coding.

Hereinafter, an example of the variable bit rate coding apparatus will be described with reference to FIG. The encoding process is performed twice for the same encoded image signal. The first is performed in provisional encoding for setting a target transfer rate for each unit time, and the second is performed in actual encoding (actual encoding). First, the provisional encoding unit 1 is for setting the target transfer rate for each unit time. VTR11
Are input to the prediction subtractor 12. In the conventional example, the same image signal is twice
, All of the image signals before encoding are VT
It must be recorded on a large-capacity image recording medium such as R11.

The basic encoding process is the same as the operation of the image encoding processing unit 3 described above, but the encoded data output from the variable length encoder 18 is input to the selector 19. Here, the selector 21 is switched to the A side, and a predetermined fixed value is output to the encoding control unit 22 as the buffer sufficiency. The fixed value is such that an image encoded by the fixed value has a necessary and sufficient image quality. In the first temporary encoding process, the output of the variable length encoder 18 is input to the temporary code amount counter 23 via the selector 19. The temporary code amount counter 23 counts the code amount generated for each unit time and outputs the result as a temporary transfer rate.

Next, the target transfer rate setting section 2 sets a target transfer rate from the provisional code amount for each unit time so that the sum of the code amounts of the moving image signal becomes a predetermined value.
The temporary transfer rate output from the temporary code amount counter 23 is
It is input to the code amount controller 25 and the temporary transfer rate memory 24. The temporary transfer rate memory 24 stores all the temporary transfer rates per unit time that are used again in the actual encoding. The code amount controller 25 is for determining the conversion characteristic from the temporary transfer rate to the target transfer rate. When the provisional encoding is completed, the conversion characteristic information output from the code amount controller 25 is input to the target transfer rate setter 26.

Next, in the actual encoding process, encoding is performed while controlling the amount of generated codes so as to match the target transfer rate per unit time. The same image as in the temporary encoding is output again from the VTR 11, and the same encoding as the temporary encoding is performed. At the time of actual encoding, the output of the variable length encoder 18 is applied to the buffer 20 via the selector 19. The buffer 20 absorbs short-period fluctuations of the code and outputs coded data. The rate at which codes are read from the buffer 20 is controlled by a value supplied from the target transfer rate setting unit 26 per unit time. Therefore, the code amount output from the buffer 20 changes every unit time. The target transfer rate setting unit 26 converts the temporary transfer rate input from the temporary transfer rate memory 24 per unit time according to the conversion characteristics input from the code amount controller 25, and sets the target transfer rate. .

On the other hand, information on the sufficiency of the buffer 20 is input to the encoding control unit 22 via the selector 21. In the encoding control unit 22, the quantization step width is controlled by the buffer sufficiency. In this control, since the fluctuation in a long cycle is absorbed by the variable transfer rate per unit time, only the local fluctuation is absorbed, and the possibility of overflow is reduced. Note that, unlike the above-described conventional example, a moving image encoding apparatus configured to connect both the temporary encoding unit and the actual encoding unit in series can perform real-time processing.

[0017]

Since the conventional moving picture coding apparatus is configured as described above, the main coding is started after analyzing the coding characteristics of the entire coded picture. Therefore, the encoding process becomes a non-real-time process, and it takes at least twice as long as normal encoding. That is, such an encoding system has a problem that it is not suitable for high-speed encoding of a large amount of long content because the processing is not real-time processing.

Further, in a moving picture coding apparatus having a configuration in which both the provisional encoding means and the actual encoding means are connected in series, at least a double or more is required for the provisional encoding processing and the real encoding processing in real time. However, there is a problem that the device performance is required, so that the device scale or the power consumption is significantly increased, and the feasibility is poor.

SUMMARY OF THE INVENTION The present invention has been made to solve the above-described problems, and is capable of performing variable bit rate encoding in real time without causing a significant increase in device size and power consumption. It is an object to obtain a recording medium in which an apparatus, a moving picture coding method, and a moving picture coding program are recorded.

[0020]

A moving picture coding apparatus according to the present invention converts the image feature information extracted by the feature extracting means into an occurrence probability, and calculates a target code amount from the occurrence probability. Calculating means, and encoding means for adjusting a quantization step width according to the target code amount calculated by the target code amount calculating means, and encoding the input image at a variable bit rate with the quantization step width. It is a thing.

In the moving picture coding apparatus according to the present invention, the coding means codes the input picture so that the total code amount in a predetermined time length matches a specified value.

In the moving picture coding apparatus according to the present invention, when the target code amount calculating means calculates the target code amount, the image characteristic information is corrected.

In the moving picture coding method according to the present invention, when the image characteristic information is converted into the occurrence probability and the target code amount is calculated from the occurrence probability, the quantization step width is adjusted according to the target code amount. , The input image is subjected to variable bit rate encoding with the quantization step width.

In the moving picture coding method according to the present invention, an input picture is coded so that the total code amount in a predetermined time length matches a specified value.

In the moving picture coding method according to the present invention, the image feature information is corrected when the target code amount is calculated.

The recording medium on which the moving picture encoding program according to the present invention is recorded converts the image feature information extracted by the feature extraction processing procedure into an occurrence probability and calculates a target code amount from the occurrence probability. Encoding processing for adjusting the quantization step width according to the amount calculation processing procedure and the target code amount calculated by the target code amount calculation processing procedure, and performing variable bit rate encoding on the input image with the quantization step width. The procedure is recorded.

A recording medium on which a moving image encoding program according to the present invention is recorded is such that an encoding processing procedure encodes an input image such that a total code amount in a predetermined time length matches a specified value. It is.

The recording medium on which the moving image encoding program according to the present invention is recorded is such that the image characteristic information is corrected when the target code amount calculation procedure calculates the target code amount.

[0029]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS One embodiment of the present invention will be described below. Embodiment 1 FIG. FIG. 1 is a block diagram showing a moving picture coding apparatus according to Embodiment 1 of the present invention.
Is an image feature information extraction unit (feature extraction means) for extracting image feature information from an input image, and 32 is an image feature information extraction unit 31
Is a feature information-occurrence probability conversion table that converts the image feature information extracted by the above into an occurrence probability and outputs an occurrence probability section, and 33 calculates a target code amount from the occurrence probability section output by the feature information-occurrence probability conversion table 32 This is a target code amount calculation unit. Note that the characteristic information-occurrence probability conversion table 32 and the target code amount calculation unit 33 constitute a target code amount calculation unit.

An image encoding unit (34) adjusts the quantization step width according to the target code amount calculated by the target code amount calculation unit 33, and encodes the input image at a variable bit rate with the quantization step width. The image encoding unit 34 is the same as the image encoding device of the image encoding system which is a standard system such as ITU-T, for example, like the image encoding processing unit 3 of FIG. It has a configuration. In addition, for example, the configuration is the same as that of the image coding apparatus of the vector quantization system. 35 is an encoding control unit that controls the quantization step width according to the target code amount calculated by the target code amount calculation unit 33, and 36 performs a quantization operation using the quantization step width specified by the encoding control unit 35. The quantizer to perform,
Reference numeral 37 denotes an image encoding processing unit that performs an encoding operation on a moving image by an image encoding method employing a code amount control unit based on quantization.

FIG. 2 is a flowchart showing a moving picture coding method according to the first embodiment of the present invention. In the first embodiment, a moving picture coding apparatus is composed of an image feature information extracting unit 31, a feature information-occurrence probability conversion table 32, a target code amount calculating unit 33, and an image coding unit 34, which are hardware. Although these are shown, these may be constituted by software instead of hardware, and the moving image encoding program as the software may be recorded on a computer-readable recording medium.

Next, the operation will be described. It is assumed that the input image has already been digitized. The digital input image is input to the image feature information extraction unit 31, and image features, that is, image feature information that quantitatively represents the difficulty of encoding, are calculated from the digital input image (step ST1).
The image feature information may be, for example, an inter-frame difference sum with a previous frame for one frame of the encoded image. Alternatively, the sum of the variances or the horizontal or vertical frequency high frequency component may be used.

FIG. 3 is an explanatory diagram showing the general relationship between the inter-frame difference of an image and the code amount when coding is performed in the fixed quantization step. As shown in FIG. 3, the generated code amount is large when the inter-frame difference as the image feature information is large, and the generated code amount is small when the inter-frame difference is small. Therefore, it can be said that the inter-frame difference, which is the image feature information, quantitatively indicates the difficulty of image coding.

Next, the characteristic information-occurrence probability conversion table 32
In, the image feature information calculated by the image feature information extraction unit 31 is converted into the occurrence probability, and an occurrence probability section is output (step ST2). The relationship between image feature information and its occurrence probability is measured in many image sequences, and a feature information-occurrence probability conversion table 32 is created.

FIG. 4 is an explanatory diagram showing the relationship between an inter-frame difference, which is an example of image feature information, and its occurrence probability. FIG.
As shown in the following, the relationship between image feature information (from 0 to the maximum value S ₁₀₀ ) and its occurrence probability is represented by the following function ρ (S).

(Equation 1) Note that the relationship of the above equation (1) does not hold only when the interframe difference is used as the image feature information, but holds when any kind of image feature information is used. Further, in the example shown in FIG.

(Equation 2)

FIG. 5 is an explanatory diagram showing a configuration example of the feature information-occurrence probability conversion table 32 at the time of the occurrence probability distribution of the image feature information shown in FIG. In FIG. 5, the image feature information becomes an address signal of the ROM, and an occurrence probability section is output as data output of the ROM. In addition, in the example of FIG. 5, the occurrence probability sections are 0 to 3, but the number of section breaks is not limited and can be arbitrarily configured.

In the target code amount calculation unit 33, the occurrence probability section from the feature information-occurrence probability conversion table 32 matches the predetermined value of the total code amount in a predetermined time and follows the difficulty of image coding. The target code amount is converted and output (step ST3). The sum of the products of the probabilities indicated by each occurrence probability section in all occurrence probability sections and the bit rate corresponding to the target code amount to be converted is made to match the average bit rate obtained from the default value of the total code amount. Things. Further, the target code amount is set based on the difficulty of encoding the original image indicated by the occurrence probability section.

At the time of the occurrence probability distribution of the image feature information shown in FIG. 4, the feature information-occurrence probability conversion table 32 shown in FIG.
An example in which an occurrence probability section is output according to the configuration example described above will be described below. The target code amount of each occurrence probability section is as follows.

(Equation 3) In this case, the conversion to the target code amount is set so that the following relationship is satisfied. Total code amount / predetermined time = (0.25 × target code amount 1 Mbp)
s) + (0.25 × target code amount 2 Mbps) + (0.2
5 × target code amount 3 Mbps) + (0.25 × target code amount 4 Mbps)

The encoding control section 35 counts the code amount actually encoded by the image encoding processing section 37 with the target code amount as a target, and performs a quantization step so that the generated code amount converges to the target. Adjust the width (step ST
4). The image encoding processing unit 37 encodes the input image with the variable bit rate with the quantization step width adjusted by the encoding control unit 35 (step ST5). Note that the image encoding unit 34 performs the image encoding of the image encoding method employing the code amount control unit by quantization except that the target code amount is instructed to be fed forward from the target code amount calculation unit 33. The same processing is performed with the same configuration as the device.

As described above, according to the first embodiment, the characteristic information indicating the difficulty of encoding an image is obtained, and a large amount of code is assigned to an image that is difficult to encode. Since a small code amount is assigned to an image according to its occurrence probability distribution, the variable bit rate control for improving the average image quality of the coded image while keeping the total code amount within a predetermined time length within a predetermined value is performed. can do.
Also, since feedforward control is performed based on image feature information indicating the difficulty of image encoding, image encoding processing can be performed only once in real time.

Embodiment 2 FIG. 6 is a block diagram showing a moving picture coding apparatus according to Embodiment 2 of the present invention. In the figure, the same reference numerals as those in FIG. 1 denote the same or corresponding parts, and a description thereof will be omitted. Reference numeral 38 denotes a target code amount calculation unit that calculates a target code amount from the occurrence probability section output by the characteristic information-occurrence probability conversion table 32, and outputs an error signal to the characteristic information correction conversion unit 39. Therefore, it is a feature information correction conversion unit that corrects image feature information. Note that the target code amount calculation unit 38 and the characteristic information correction conversion unit 39 constitute a target code amount calculation unit.

Next, the operation will be described. In the first embodiment, the image feature information of the input image is calculated, the image feature information is converted into the occurrence probability, and the target code amount is calculated from the occurrence probability. When the occurrence probability of the information is biased, an error in the occurrence probability of the image feature information may be corrected.

That is, when the occurrence probability of the image feature information of the input image is deviated from the probability distribution of the feature information-occurrence probability conversion table 32, the probability distribution of the target code amount output from the target code amount calculation unit 38 Also bias. The target code amount calculation unit 38 counts the target code amount in a certain period, compares the average bit rate in the period with an average bit rate obtained from a predetermined value of the total code amount, and calculates the average bit rate of the target code amount. The deviation is output as an error signal.

The characteristic information correction converter 39 receives the error signal output from the target code amount calculator 38 and corrects the image characteristic information with respect to the image characteristic information output from the image characteristic information extractor 31. Then, the correction feature information is output. In the case of an error signal indicating that the average bit rate of the target code amount exceeds the average bit rate obtained from the default value of the total code amount, the correction operation corrects the image feature information to the image feature information having the higher occurrence probability. Calculate. Conversely, in the case of an error signal indicating that the average bit rate of the target code amount is lower than the average bit rate obtained from the predetermined value of the total code amount, a correction operation is performed to correct the image feature information to image feature information having a lower occurrence probability. I do. Further, in the case of an error signal indicating a match between the average bit rate of the target code amount and the average bit rate obtained from the predetermined value of the total code amount, the correction operation maintains the current correction operation.

In the feature information-occurrence probability conversion table 32, the corrected feature information is converted into its occurrence probability, and an occurrence probability section is output. Since the corrected feature information is close to the occurrence probability distribution of the feature information-occurrence probability conversion table 32, an output of an appropriate occurrence probability section can be obtained. From the output of an appropriate occurrence probability section, the first embodiment
Similarly to the above, the target code amount calculation unit 38 can convert the output to a target code amount that matches the predetermined value of the total code amount in a predetermined time and that is in accordance with the difficulty of image coding and outputs the target code amount. The encoding control unit 35 counts the code amount actually encoded by the image encoding processing unit 37 with the target code amount as a target, and adjusts the quantization step width so that the generated code amount converges to the target. I do.

As described above, according to the second embodiment, even when the occurrence probability of the image feature information of the input image is biased, the error of the occurrence probability of the image feature information is corrected. Therefore, even when the occurrence probability of the image feature information of the input image is biased with respect to the probability distribution of the feature information-occurrence probability conversion table 32, the total code amount within a predetermined time length falls within a predetermined value, and And variable bit rate control for improving the average image quality of the encoded image. Also, since feedforward control is performed based on image feature information indicating the difficulty of image encoding, image encoding processing can be performed only once in real time.

[0047]

As described above, according to the present invention, the target code amount calculating means for converting the image characteristic information extracted by the characteristic extracting means into the occurrence probability, and calculating the target code amount from the occurrence probability, And a coding unit that adjusts a quantization step width according to the target code amount calculated by the target code amount calculation unit and performs variable bit rate coding on the input image with the quantization step width. Therefore, there is an effect that variable bit rate encoding can be performed in real time without significantly increasing the device scale and power consumption.

According to the present invention, since the encoding means encodes the input image so that the total code amount in the predetermined time length matches the specified value, the average image quality can be improved. effective.

According to the present invention, when the target code amount calculating means calculates the target code amount, the image characteristic information is corrected, so that the average image quality can be improved.

According to the present invention, when the image characteristic information is converted into the occurrence probability and the target code amount is calculated from the occurrence probability, the quantization step width is adjusted according to the target code amount, and the input image is converted. Since the variable bit rate encoding is performed with the quantization step width, the variable bit rate encoding can be performed in real time without increasing the device scale and power consumption accompanying a remarkable increase in the amount of processing operations. effective.

According to the present invention, since the input image is encoded so that the total code amount in the predetermined time length matches the specified value, the average image quality can be improved.

According to the present invention, since the image characteristic information is corrected when calculating the target code amount, there is an effect that the average image quality can be improved.

According to the present invention, the target code amount calculation processing procedure for converting the image feature information extracted by the characteristic extraction processing procedure into the occurrence probability and calculating the target code amount from the occurrence probability, and the target code amount calculation Since the quantization step width is adjusted in accordance with the target code amount calculated by the processing procedure, and the encoding processing procedure of performing variable bit rate encoding on the input image with the quantization step width is configured to be recorded,
There is an effect that variable bit rate encoding can be performed in real time without increasing the device scale and power consumption accompanying a remarkable increase in the amount of processing operations.

According to the present invention, the encoding processing procedure is configured to encode the input image such that the total code amount in the predetermined time length matches the specified value, so that the average image quality can be improved. There is an effect that can be done.

According to the present invention, when the target code amount calculation processing procedure calculates the target code amount, the image characteristic information is corrected so that the average image quality can be improved.

[Brief description of the drawings]

FIG. 1 is a configuration diagram showing a moving picture coding apparatus according to Embodiment 1 of the present invention.

FIG. 2 is a flowchart showing a moving picture coding method according to the first embodiment of the present invention.

FIG. 3 is an explanatory diagram showing a general relationship between an inter-frame difference of an image and a code amount when encoding is performed in a fixed quantization step.

FIG. 4 is an explanatory diagram showing a relationship between an inter-frame difference, which is an example of image feature information, and its occurrence probability.

FIG. 5 is an explanatory diagram showing a configuration example of a feature information-occurrence probability conversion table.

FIG. 6 is a configuration diagram illustrating a moving image encoding device according to a second embodiment of the present invention.

FIG. 7 is a configuration diagram showing a conventional moving picture encoding device.

[Explanation of symbols]

31 image feature information extraction unit (feature extraction unit), 32 feature information-occurrence probability conversion table (target code amount calculation unit), 33 target code amount calculation unit (target code amount calculation unit), 34 image encoding unit (coding Means), 35 encoding control unit, 36 quantizer, 37 image encoding processing unit,
38 target code amount calculation unit (target code amount calculation means), 39
Characteristic information correction conversion unit (target code amount calculation means).

Continued on the front page F-term (reference) 5C053 FA22 GB37 JA01 KA05 KA20 KA22 KA24 LA14 5C059 KK06 KK49 MA00 MA01 MA23 MD02 PP04 SS11 SS20 TA53 TB00 TC19 TD06 TD14 UA02 UA33 UA39 5J064 AA04 BA01 BC09 BC16 BC03 BC03 BC03 AA10 BB01 CC02 GG06 HH16

Claims (9)

[Claims] 1. A feature extracting means for extracting image feature information from an input image, a target code amount for converting the image feature information extracted by the feature extracting means into an occurrence probability, and calculating a target code amount from the occurrence probability. Calculation means, and coding means for adjusting the quantization step width according to the target code amount calculated by the target code amount calculation means, and performing variable bit rate coding on the input image with the quantization step width. Video encoding device. 2. The moving picture encoding apparatus according to claim 1, wherein said encoding means encodes the input image such that a total code amount in a predetermined time length matches a specified value. 3. The moving picture coding apparatus according to claim 1, wherein the target code amount calculating means corrects the image feature information when calculating the target code amount. 4. Extracting image feature information from an input image, converting the image feature information into an occurrence probability, and calculating a target code amount from the occurrence probability, the quantization step width is determined according to the target code amount. A moving picture coding method for adjusting the input picture at a variable bit rate with the quantization step width. 5. The moving picture coding method according to claim 4, wherein the input picture is coded so that a total code amount in a predetermined time length matches a specified value. 6. The moving picture coding method according to claim 4, wherein image characteristic information is corrected when calculating the target code amount. 7. A feature extraction process for extracting image feature information from an input image, and a target for converting the image feature information extracted by the feature extraction process into an occurrence probability and calculating a target code amount from the occurrence probability. The coding amount calculation processing procedure and the coding that adjusts the quantization step width according to the target code amount calculated by the target code amount calculation processing procedure, and encodes the input image with the variable bit rate using the quantization step width. A recording medium in which a moving image encoding program having a processing procedure is recorded. 8. The moving image encoding program according to claim 7, wherein the encoding processing procedure encodes the input image such that a total code amount in a predetermined time length matches a specified value. Recording medium. 9. The moving image encoding program according to claim 7, wherein the target code amount calculating process corrects image feature information when calculating the target code amount. recoding media.