JPH09135418A - Image recording device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To enable independent reproduction for each recording unit and to improve picture quality at the time of high-speed reproduction by making the recording unit correspondent to an encoding unit. SOLUTION: An input signal 10 of an image signal is divided into the encoding units by a blocking means 11, encoded based on a code amount control signal 18 by a block unit encoding means 13, outputted thereafter to a signal separating means 15 for recording as a block unit encoded signal 14 and outputted as a recording candidate encoded signal 16 after the irreducibly minimum code words required for recording are separated at this means 15. When the code amount exceeds a code amount recordable for the recording unit, at a code amount discriminating means 17, the code amount control signal 18 is outputted to the block unit encoding means 13 so as to control the code amount, but when the code amount is settled within a target code amount, this signal is outputted to an encoded signal recording means 20 as the encoded signal 19 for recording and recorded so that the encoding unit can be correspondent to the recording unit.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an apparatus for encoding and recording image signals.

[0002]

2. Description of the Related Art In recent years, MPEG1 and MPEG2 have been widely used as an international standard as a coding method for compressing a moving image. In both MPEG1 and MPEG2, motion compensation using orthogonal correlation between frames and orthogonal transformation (D
By using techniques such as CT) and variable length coding, a large amount of image data is compressed. Also,
MPEG2 is expected to be applied to a wide range of applications such as digital broadcasting, and can be encoded at a wide range of data rates. For this reason, applications such as recording on a VTR are also under study.

[0003]

However, MPE is generally used.
The code amount control in the case of using the G method coding is performed within a screen or a fixed number of plural screens (GO) including motion compensation coding.
P), the code amount varies from screen to screen, the code amount is not constant even within the screen, and the macro block (hereinafter referred to as MB) is the smallest unit that can be reconstructed by decoding. ), The code amount is not constant but varies.

For this reason, when it is attempted to record MPEG encoded data on the VTR, it is very difficult to make the recording position of the VTR correspond to the position of the encoded screen. When recording is performed in this way, it is impossible to reproduce the data necessary for obtaining sufficient image quality in high-speed reproduction.

Generally, when recording data on a medium such as a VTR tape, the recording portion of the medium is often recorded with a fixed-length block (sync block, SB) as one recording unit.

In view of the above point, the present invention, when recording image data encoded by a system such as MPEG, associates a recording unit with an encoding unit and performs encoding control within the range of the recording unit. The object is to solve the above problems.

[0007]

According to a first aspect of the present invention, an input signal which is a screen to be encoded is divided into blocks in units of encoding,
Blocking means for outputting as a block signal, block unit coding means for coding the block signal in a block unit by a coding method such as MPEG and outputting as a block unit coded signal, and the block unit coded signal From the above, a recording signal separating means for separating only the minimum necessary code for decoding in block units and outputting as a recording signal, and an encoded signal recording for recording the recording signal in a predetermined recording unit. And an image recording device including means.

According to a second aspect of the present invention, a block forming means for dividing an input signal, which is a screen to be encoded, into blocks by an encoding unit, and outputting as a block signal, and an orthogonal transformation by orthogonally transforming the block signal in block units. Orthogonal transform means for outputting a transform coefficient signal, quantizing means for quantizing the orthogonal transform coefficient signal using a quantization parameter and outputting a quantized signal, and variable length coding of the quantized signal And a variable length coding means for outputting as the above, and a code amount is calculated from the coded signal and compared with a target code amount. When the code amount of the coded signal is larger than the target code amount, the quantization parameter incrementing means An increment control signal, and in other cases, a code amount determination means for outputting the recording coded signal to the coded signal recording means, and the increment control signal. Based on this, a quantization parameter increment means for incrementing the pre-quantization parameter by a predetermined number and outputting it to the quantization means and the quantization parameter memory means, and storing the quantization parameter, and the quantization as the pre-quantization parameter. The image recording device comprises a quantization parameter memory means for outputting to the parameter increment means, and an encoded signal recording means for recording the recording encoded signal in a predetermined recording unit.

According to a third aspect of the present invention, a block forming means for dividing an input signal, which is a screen to be encoded, into blocks by an encoding unit, and outputting as a blocked signal, and an orthogonal transformation by orthogonally converting the blocked signal in a block unit. Orthogonal transform means for outputting a transform coefficient signal, quantizing means for quantizing the orthogonal transform coefficient signal using a quantization parameter and outputting a quantized signal, and variable length coding of the quantized signal Variable length coding means for outputting as, and calculating the code amount from the coded signal, outputting a control signal by comparing with the target code amount, and within the target code amount and closest to the target code amount,
Code amount determination means for outputting a coded signal for recording, and reference quantization parameter update for calculating and updating a reference quantization parameter serving as a reference for changing a quantization parameter based on the control signal and the number of times of encoding Means, a quantization parameter calculation means for calculating a quantization parameter from the reference quantization parameter and the pre-quantization parameter, and outputting the quantization parameter to the quantization means and the quantization parameter memory means, and the quantization parameter stored therein. An image recording comprising a quantization parameter memory means for outputting to the quantization parameter increment means as the pre-quantization parameter and an encoded signal recording means for recording the recording encoded signal in a predetermined recording unit. It is a device.

In a fourth aspect of the present invention, an input signal, which is a screen to be encoded, is divided into blocks in a coding unit and is output as a block signal, and the block signal is orthogonally transformed in a block unit. An orthogonal transform means for outputting the transform coefficient signal to the quantizing means, a quantized orthogonal transform coefficient signal by using a quantization parameter, and an AC component zero value replacing means and a first variable length coding means as the quantized signal. The quantizing means for outputting, the first variable-length coding means for variable-length coding the quantized signal and outputting it as the first coded signal to the code amount determining means, and the predetermined number of AC components in the quantized signal AC component zero-value replacing means for replacing the zero-value and outputting it as a zero-value replacing quantized signal to the second variable-length coding means, and variable-length coding the zero-value replacing quantized signal to obtain a second coded signal. The second variable-length coding means for outputting to the code amount determining means, the code amount of the first coded signal, the code amount of the second coded signal, and the target code amount, and compares Both the encoded signal and the second encoding are
When the code amount exceeds the target code amount, the control signal is output to the quantization parameter determining means, and when any of the code amounts falls within the signal target code amount, the coding having the maximum code amount within the target code amount range is performed. A code amount determination means for selecting a signal and outputting it as a recording coded signal, and a quantization parameter for recalculating a pre-quantization parameter based on the control signal and outputting it to the quantization means and a quantization parameter memory means. Determining means, quantization parameter memory means for storing the quantization parameter and outputting it as the pre-quantization parameter to the quantization parameter determining means, and a code for recording the recording coded signal in a predetermined recording unit. The image recording device includes an encoded signal recording unit.

According to a fifth aspect of the present invention, an input signal which is a screen to be coded is divided into blocks by a coding unit and is output as a blocking signal, and the blocking signal is orthogonally transformed in a block unit. Orthogonal transforming means for outputting the transform coefficient signal to the quantizing means, quantizing the orthogonal transform coefficient signal using a quantizing parameter, and outputting as a quantized signal to the AC component zero value replacing means and the variable length coding means. Quantizing means, a predetermined number of AC components in the quantized signal is replaced with a zero value, AC component zero value replacing means for outputting to the variable length coding means as a zero value replacement quantized signal,
Variable-length coding means for variable-length coding the quantized signal and outputting it as a coded signal to a code amount determination means and a memory means, and storing the coded signal, and shifting the time to code as a pre-coded signal. Memory means for outputting to the quantity determining means,
If the code amount of the coded signal is compared with the code amount of the pre-coded signal and the target code amount, and if the coded signal and the pre-coding both exceed the target code amount, the control signal Is output to the quantization parameter determining means, and when any of the code amounts falls within the signal target code amount, the coded signal having the maximum code amount in the target code amount range is selected and used as the recording code signal. A code amount determining means for outputting, a quantization parameter determining means for recalculating a pre-quantization parameter based on the control signal, and outputting the quantization parameter to the quantization means and a quantization parameter memory means, and storing the quantization parameter. ,
An image recording apparatus including a quantization parameter memory unit that outputs the pre-quantization parameter to the quantization parameter determination unit, and an encoded signal recording unit that records the recording encoded signal in a predetermined recording unit. Is.

A sixth aspect of the present invention is the same as the variable length coding,
In an encoding method that can also perform fixed-length encoding by adding an escape code, when recording data for each encoding unit in multiple recording units, divide the input signal, which is the screen to be encoded, into blocks by the encoding unit. A blocking means for outputting as a blocked signal; an orthogonal transform means for orthogonally transforming the blocked signal in block units to output an orthogonal transform coefficient signal; and a quantizing parameter for quantizing the orthogonal transform coefficient signal. Quantization means for outputting a quantized signal, coding method switching means for switching the quantized signal to fixed-length coding means or variable-length coding means on the basis of a switching control signal, and outputting the quantized signal Variable-length coding means for outputting a variable-length coded signal to the code amount determining means, and a fixed code by adding an escape code to the quantized signal. Fixed-length coding means for long-coding and outputting the fixed-length coded signal to the code amount judging means, and when the code amount of the variable-length coded signal exceeds the code amount of the recording unit, stop at a point where it does not exceed Fixed-length coding is performed and 1
After the fixed length coding is performed, a switching control signal is output to the coding method switching means so that the variable length coding is performed again, and the preceding variable length coding is performed more than the code amount of a predetermined number of recording units. If the sum of the code amounts of the encoding and the fixed length coding is larger, the quantization control signal is output to the quantization parameter determining means, and if the sum is smaller, the code amount determining means outputs the coded signal. A quantization parameter determining means for recalculating a pre-quantization parameter based on the quantization control signal and outputting it to the quantization means and a quantization parameter memory means; and storing the quantization parameter, the pre-quantization Quantization parameter memory means for outputting to the quantization parameter determination means as a parameter, and the last escape code existing within the code amount of one recording unit from the encoded signal. Is detected, the coded signal is divided before and after the escape code and sequentially output as a coded signal for recording, and a coded signal recording means for recording the coded signal for recording in a predetermined recording unit. The image recording device is composed of

[0013]

BEST MODE FOR CARRYING OUT THE INVENTION In the image recording apparatus of the first invention, the recording unit and the encoding unit are made to correspond to each other, and the position of the encoding unit on the screen and the recording position are made to coincide with each other, so that the code in MPEG It is possible to omit the start code of the coding unit and the code that indicates the position of the coding unit in the screen.Since only the code words of the quantization parameters and coefficients need to be recorded, the code amount can be reduced and the recording units can be independent. Playback and decryption are possible.

In the image recording apparatus of the second invention, in the coding in the coding unit, at the time of quantization, the code amount is evaluated from the one having the smaller quantization parameter, that is, the one having the larger code amount, By stopping the evaluation when the target code amount of one recording unit is reached and recording as it is, it is possible to perform encoding without saving the previous encoding state in the buffer, and to save the buffer memory. It saves money and reduces hardware.

In the image recording apparatus of the third invention, the quantization parameter can be efficiently determined even when the number of times of changing the quantization parameter is limited due to the limitation of hardware. When the quantization parameter is represented by 5 bits as in MPEG, it is possible to determine and record the quantization parameter that fits within the target value by a maximum of five evaluations, and a code generated in the process of determining the quantization parameter. It is possible to reduce the delay in conversion.

In the image recording apparatus of the fourth invention, the code amount is obtained by using the result of encoding with an appropriate quantization parameter and the result of encoding after substituting the zero value of the AC component with the same quantization parameter. By performing the control, it becomes possible to perform more precise code amount control as compared with the conventional code amount control using only the quantization parameter, and it is possible to efficiently fall within the target code amount range.

In the image recording apparatus of the fifth invention, in addition to the fourth invention, the result of encoding by an appropriate quantization parameter and a quantization parameter smaller than the quantization parameter and a zero value of the AC component. By using the result obtained by performing the encoding after the replacement, according to the fourth invention,
It is possible to perform more precise code amount control, and it is possible to efficiently fit within the target code amount range.

In the image recording apparatus of the sixth invention, M
Like PEG, when not only variable length coding but also fixed length coding by adding an escape code is possible, if the coding amount of one coding unit exceeds one recording unit, it is forced. By recording an escape code, recording a fixed-length code following the escape code in the next recording unit, and continuing encoding, it is possible to encode in multiple recording units. Since the code amount of can be used, the image quality can be improved.

FIG. 1 shows a block diagram of an image recording apparatus in an embodiment of the first invention. In the figure, 10 is an input signal,
11 is a block forming means, 12 is a block forming signal, 13 is a block unit encoding means, 14 is a block unit encoding signal, 15 is a recording signal separating means, 16 is a recording encoding signal, and 17 is a code amount determining means. , 18 is a code amount control signal, 19 is a coded signal, 20
Is a coded signal recording means.

The operation of the image recording apparatus of the present embodiment constructed as above will be described below. The input signal 10, which is an image signal, is divided into coding units by the blocking means 11. For example, if the encoding method is MPEG, the minimum unit of encoding that can reconstruct the screen at the time of decoding is a macro block, so that it is a rectangular block of 16 pixels vertically and horizontally. If the encoding unit is the macroblock unit, the luminance signal and the color difference signal at the same screen position can be reproduced, and therefore, the screen area having the size of the macroblock can be reproduced. Here, MP
The case of encoding with EG has been described as an example, but a block unit suitable for the encoding method may be selected.

The blocked signal 12 thus divided is
In the block unit coding means 13, the code amount control signal 18
Is coded based on. When MPEG is selected as the encoding method, the encoding is performed in units of macroblocks, but here, in order to enable independent decoding in units of macroblocks during reproduction, one macroblock is defined as one slice, It is necessary to perform only intra-picture coding as an intra macroblock, code under fixed conditions, and code a quantization parameter for each macroblock. This is an example of encoding by MPEG, but the encoding is not limited to MPEG as long as the encoding is performed under the condition that the screen can be reconfigured independently. After being encoded in this way, it is output to the recording signal separating means 15 as the block unit encoded signal 14.

In the recording signal separating means 15, the minimum codeword necessary for recording is selected and separated from the codeword string of the block unit coded signal 14, and the recording candidate coded signal 16 is selected. Output as. Usually, a codeword sequence encoded by a method such as MPEG is added with a codeword capable of separating encoding units during decoding and a codeword indicating an encoding condition. A slice start code for separating a slice layer, which is a coding unit of MPEG, a type of macroblock, that is, a code indicating whether or not intra-picture coding is performed, and an orthogonal transformation of a macroblock,
For example, a code indicating whether or not to process the input image as an interlaced image can be given. In the present invention, since the recording unit and the encoding unit are made to correspond to each other, a start code or the like for separating the encoding unit is not necessary at the time of recording, and the block can be reproduced independently in the encoding unit. Since the coding condition is fixed at the time of the unit coding means 13, a code indicating the coding condition is also unnecessary. That is, the codewords to be recorded are only the codewords necessary for independently reconstructing the screen in coding units. For example, in the case of MPEG, a code word of a quantization parameter used in quantization of an orthogonal transform coefficient in macroblock units, a code word of a quantized and encoded orthogonal transform coefficient, and an end of the orthogonal transform coefficient code word are added. It suffices to record only the EOB code that is recorded. FIG. 7 shows an example of the arrangement of data recorded in one recording unit. By separating the code words to be recorded in this way, in the case of MPEG, the code amount can be reduced by about 40 bits per macro block. Although an example of MPEG has been described here, the invention is not limited to MPEG.

The recording candidate coded signal 16 is coded by the code amount judging means 17
When the code amount of the recording candidate coded signal 16 exceeds the code amount recordable in one target recording unit, the code amount control signal 18 is output to the block unit coding means 13. Then, for example, in the coding conditions, the quantization parameter is changed to reduce the code amount and control is performed so that the target code amount is achieved. Thus, when the code amount of the recording candidate encoded signal 16 falls within the target code amount, the recording encoded signal
It is output to the encoded signal recording means 20 as 19.

In the coded signal recording means 20, the coded signal for recording 19 is recorded at a recording position on a recording medium such as a magnetic tape.
Recording is performed in the arrangement as shown in FIG. 7 so that the coding unit and the recording unit correspond to each other. As described above, according to the present embodiment, the recording unit and the encoding unit are made to correspond to each other, and the position of the encoding unit on the screen and the recording position are made to coincide with each other. , The code indicating the position of the coding unit on the screen can be omitted, and since only the code words of the quantization parameter and the coefficient need to be recorded, the code amount can be reduced, and independent reproduction and decoding can be performed in the recording unit. It will be possible.

FIG. 2 shows a block diagram of an image recording apparatus in an embodiment of the second invention. In the figure, 1001 is an input signal, 1002 is a blocking means, 1003 is a blocking signal, 10
04 is an orthogonal transformation means, 1005 is an orthogonal transformation coefficient, 1006 is a quantization means, 1007 is a quantized signal, 1008 is a variable length coding means, 10
09 is a coded signal, 1010 is a quantization parameter, 1011 is a quantization parameter memory means, 1012 is a pre-quantization parameter,
Reference numeral 1013 is a quantization parameter increment means, 1014 is an increment control signal, 1017 is a code amount determination means, 1018 is a recording coded signal, and 1019 is a coded signal recording means.

The operation of the image recording apparatus of the present embodiment constructed as above will be described below. The input signal 1001 which is an image signal is divided into coding units by the blocking means 1002. For example, if the encoding method is MPEG, the minimum unit of encoding that can reconstruct the screen at the time of decoding is a macro block, so that it is a rectangular block of 16 pixels vertically and horizontally. If the encoding unit is the macroblock unit, the luminance signal and the color difference signal at the same screen position can be reproduced, and therefore, the screen area having the size of the macroblock can be reproduced. Here, the case of encoding by MPEG is taken as an example, but a block unit suitable for the encoding method may be selected.

Blocked signal 1003 thus divided
Is orthogonally transformed by the orthogonal transformation means 1004 and outputs the orthogonal transformation coefficient 1005 to the quantization means 1006. As the orthogonal transform, discrete cosine transform is used in MPEG or JPEG, but the present invention is not limited to this, and other orthogonal transform such as Haar transform or Hadamard transform may be used according to the encoding method. Absent.

In the quantizing means 1006, the quantizing parameter 10
The orthogonal transform coefficient 1005 is quantized based on 10, the coefficient to be encoded is reduced, and the quantized signal 1007 is output.

The quantized signal 1007 is converted into variable length coding means 1008.
In, a codeword is assigned and is output as a coded signal 1009. Two-dimensional Huffman coding is often used as variable-length coding. For example, in MPEG,
Quantized orthogonal transform coefficient (quantized signal 1007 here)
Are sequentially arranged from the low frequency component close to the DC component to the high frequency component, and the number of consecutive same values of coefficient value (level) (run)
A method is adopted in which sets are two-dimensionally formed, and short Huffman code words are assigned to combinations of runs and levels that appear frequently.

The code amount of the coded signal 1009 is evaluated by the code amount determining means 1017. When the code amount of the coded signal 1009 exceeds the code amount recordable in one recording unit which is a target, the increment control signal 1014 is output to the quantization parameter increment means 1013, and when it falls within the target code amount, it is output to the encoded signal recording means 1019 as the recording encoded signal 1018.

Quantization parameter increment means 1013
Then, in response to the increment control signal 1014, a predetermined number is added to the pre-quantization parameter 1012 at the previous stage stored in the quantization parameter memory means 1011, for example, 1 is added to the quantization parameter used for the next quantization. 1010 is output to the quantization means 1006 and the quantization parameter memory means 1011.

Generally, the larger the value of the quantization parameter,
The generated code amount is small. In the present invention, the code amount is sequentially evaluated from the smaller quantization parameter to the larger quantization parameter, that is, from the larger code amount to the smaller quantization parameter, and the code amount control is controlled. For example, in the case of MPEG (in the case of linear quantization), there are 31 levels of quantization parameters from 1 to 31. When the present invention is implemented, control is performed such that the quantization parameter is sequentially increased from 1. If the control is performed such that the quantization parameter is decreased from the larger one, when the target code amount is exceeded, the change of the quantization parameter is stopped and the previous coding state is re-established.
Quantization needs to be redone. For this reason, it is necessary to re-execute the processing or to provide a buffer memory for storing the immediately preceding quantization state. On the other hand, in the present invention, since the quantization parameter is changed in a state where the code amount is large, that is, the target code amount is exceeded, the quantization parameter is determined when the code amount falls below the target code amount. , No need to redo quantization, nor buffer memory.

In the coded signal recording means 1019, the coded signal for recording 1018 corresponds to the recording position on the recording medium such as a magnetic tape by the coding unit and the recording unit in the arrangement as shown in FIG. To record.

As described above, according to the present embodiment, in the coding in the coding unit, at the time of quantization, the code amount is evaluated from the one having the smaller quantization parameter, that is, the one having the larger code amount. By stopping the evaluation when the target code amount of one recording unit is reached and recording it as it is, it is possible to perform encoding without storing the previous encoding state in the buffer, and the buffer memory And the hardware can be reduced.

FIG. 3 shows a block diagram of an image recording apparatus in an embodiment of the third invention. In the figure, 1001 is an input signal, 1002 is a blocking means, 1003 is a blocking signal, 10
04 is an orthogonal transformation means, 1005 is an orthogonal transformation coefficient, 1006 is a quantization means, 1007 is a quantized signal, 1008 is a variable length coding means, 10
09 is an encoded signal, 1020 is a quantization parameter, 1021 is a quantization parameter memory means, 1022 is a pre-quantization parameter,
1023 is a quantization parameter calculation means, 1024 is a reference quantization parameter, 1025 is a quantization reference parameter updating means, 1026
Is a control signal, 1027 is a code amount determination means, 1028 is a recording coded signal, and 1029 is a coded signal recording means.

The operation of the image recording apparatus of this embodiment having the above structure will be described below. The input signal 1001 which is an image signal is divided into coding units by the blocking means 1002. For example, if the encoding method is MPEG, the minimum unit of encoding that can reconstruct the screen at the time of decoding is a macro block, so that it is a rectangular block of 16 pixels vertically and horizontally. If the encoding unit is the macroblock unit, the luminance signal and the color difference signal at the same screen position can be reproduced, and therefore, the screen area having the size of the macroblock can be reproduced. Here, the case of encoding by MPEG is taken as an example, but a block unit suitable for the encoding method may be selected.

Blocked signal 1003 thus divided
Is orthogonally transformed by the orthogonal transformation means 1004 and outputs the orthogonal transformation coefficient 1005 to the quantization means 1006. As the orthogonal transform, discrete cosine transform is used in MPEG or JPEG, but the present invention is not limited to this, and other orthogonal transform such as Haar transform or Hadamard transform may be used according to the encoding method. Absent.

In the quantizing means 1006, the quantizing parameter 10
The orthogonal transform coefficient 1005 is quantized based on 20, the coefficient to be encoded is reduced, and the quantized signal 1007 is output.

The quantized signal 1007 is converted into variable length coding means 1008.
In, a codeword is assigned and is output as a coded signal 1009. Two-dimensional Huffman coding is often used as variable-length coding. For example, in MPEG,
Quantized orthogonal transform coefficient (quantized signal 1007 here)
Are sequentially arranged from the low frequency component close to the DC component to the high frequency component, and the number of consecutive same values of coefficient value (level) (run)
A method is adopted in which sets are two-dimensionally formed, and short Huffman code words are assigned to combinations of runs and levels that appear frequently.

The code amount of the coded signal 1009 is evaluated by the code amount determining means 1027. If the code amount of the coded signal 1009 exceeds the code amount recordable in the target recording unit, the control signal 1024 Quantization reference parameter updating means 1025
When it is within the target code amount, it is output to the coded signal recording means 1029 as the coded signal for recording 1028.

The quantizing reference parameter updating means 1025 stores the number of times of updating the quantizing reference parameter for determining the quantizing parameter based on the control signal 1024. Parameter 10
Update 24 and output. For example, when the quantization parameter is represented by 5 bits as in MPEG and the Nth update is performed, a value of 2 to the (5-N) th power is calculated, and based on the control signal 1024, If the previous code amount is larger than the target value, a value of 2 (5-N) power is output as the reference quantization parameter 1024, and if it is smaller than the target value, 2 (5-N) power of The value is converted into a negative value and output as the reference quantization parameter 1024.

In the quantization parameter calculation means 1023, the reference quantization parameter 1024 is added to the previous quantization parameter 1022 stored in the quantization parameter memory means 1021 and the quantization parameter 1020 used for the next quantization. Is output to the quantization means 1006 and the quantization parameter memory means 1021.

FIG. 8 shows the procedure for determining the quantization parameter. The figure shows an example of the case where the quantization parameter is expressed by 5 bits in the case of MPEG. Normally, in order to perform control to keep the target code amount, it is necessary to change the quantization parameter several times to determine an appropriate value. The present invention limits the quantization parameter due to hardware limitations. Even if the number of changes is limited, it is possible to efficiently provide a means for determining a quantization parameter. Referring to FIG. 8, the median value 16 of the maximum value 31 and the minimum value 1 of the quantization parameter is used for the first time. At this time, if the code amount is smaller than the target, 2 to the power of (5-2) is calculated in order to reduce the quantization parameter, and the reference quantization parameter becomes (-8), so 16-8 is calculated. Then 8
Is used as the second quantization parameter. Next, when the second encoding is performed and the code amount is smaller than the target,
To increase the quantization parameter, 2 (5-3)
Since the power is calculated and the reference quantization parameter is (+4), 8 + 4 is calculated and 12 is used as the third quantization parameter. In this way, when the calculation is carried out sequentially,
When the quantization parameter is represented by 5 bits, all the quantization parameters can be searched in 5 times.

However, when the target code amount is equal to or smaller than the target code amount at the fourth time and equal to or less than the target code amount at the fifth time, the optimum quantization parameter closer to the target code amount is obtained at the fourth time. Must be encoded using
In addition, when it is less than or equal to the target code amount at the fourth time, and it is necessary to re-encode with the quantization parameter at the fourth time even when the target code amount is exceeded at the fifth time, in the case of the quantization parameter represented by 5 bits. It is possible to determine the optimum quantization value with a maximum of 6 times.

In the coded signal recording means 1019, the coded signal for recording 1018 corresponds to the recording position on the recording medium such as a magnetic tape in the arrangement as shown in FIG. To record.

As described above, according to the present embodiment, it is possible to efficiently determine the quantization parameter even when the number of times the quantization parameter is changed is limited due to the hardware limitation. When the quantization parameter is represented by 5 bits as in MPEG, it is possible to determine and record the quantization parameter that fits within the target value by a maximum of five evaluations, and a code generated in the process of determining the quantization parameter. It is possible to reduce the delay in conversion.

FIG. 4 shows a block diagram of an image recording apparatus in an embodiment of the fourth invention. In the figure, 1001 is an input signal, 1002 is a blocking means, 1003 is a blocking signal, 10
04 is an orthogonal transformation means, 1005 is an orthogonal transformation coefficient, 1006 is a quantization means, 1007 is a quantized signal, 1108 is a first variable length coding means, 1109 is a first coded signal, 1030 is a quantization parameter,
1031 is a quantization parameter memory means, 1032 is a pre-quantization parameter, 1033 is a quantization parameter determination means, 1034 is a control signal, 1206 is an AC component zero value replacement means, 1207 is a zero value replacement quantized signal, and 1208 is a second. Variable length coding means, 1209 is a second coded signal, 1300 is a code amount determining means, 1301 is a recording coded signal, and 1302 is a coded signal recording means.

The operation of the image recording apparatus of the present embodiment configured as above will be described below. The input signal 1001 which is an image signal is divided into coding units by the blocking means 1002. For example, if the encoding method is MPEG, the minimum unit of encoding that can reconstruct the screen at the time of decoding is a macro block, so that it is a rectangular block of 16 pixels vertically and horizontally. If the encoding unit is the macroblock unit, the luminance signal and the color difference signal at the same screen position can be reproduced, and therefore, the screen area having the size of the macroblock can be reproduced. Here, the case of encoding by MPEG is taken as an example, but a block unit suitable for the encoding method may be selected.

Blocked signal 1003 thus divided
Is orthogonally transformed by the orthogonal transformation means 1004 and outputs the orthogonal transformation coefficient 1005 to the quantization means 1006. As the orthogonal transform, discrete cosine transform is used in MPEG or JPEG, but the present invention is not limited to this, and other orthogonal transform such as Haar transform or Hadamard transform may be used according to the encoding method. Absent.

In the quantizing means 1006, the quantizing parameter 10
The orthogonal transform coefficient 1005 is quantized based on 30 to reduce the coefficient to be coded, and the quantized signal 1007 is output to the first variable length coding means 1108 and the AC component zero value replacement means 1206.

The quantized signal 1007 is the first variable length coding means.
A codeword is assigned at 1108 and the first coded signal 11
It is output as 09. Two-dimensional Huffman coding is often used as variable-length coding. For example, in MPEG, quantized orthogonal transform coefficients (quantized signal 1007 in this case) are sequentially arranged from a low frequency component close to a direct current component to a high frequency component, and the number of consecutive coefficient values (levels) of the same value ( Runs that have a high appearance frequency,
A method of assigning a short Huffman codeword to the level combination is adopted.

In parallel, the quantized signal 1007 is replaced by the AC component zero value replacement means 1206 with a zero value when there is a non-zero value in the high frequency component of the AC component of the quantized signal 1007. As a result, it is possible to reduce the code amount when quantized with the same quantization parameter. As described above, by controlling the code amount by the AC component zero value replacement at the same time as controlling the code amount by the quantization parameter, more precise code amount control becomes possible. The signal thus replaced is output to the second variable length coding means 1208 as a zero-value replaced quantized signal 1207, variable length coded, and output to the code amount determination means 1300 as a second coded signal 1209.

First coded signal 1109 and second coded signal 1209
Means that the code amount is evaluated by the code amount determining means 1300, and if both code amounts exceed the recordable code amount in one target recording unit, the control signal 1034 is output to the quantization parameter determining means 1033. If either of them falls within the target code amount, the coded signal for recording 1301 is output to the coded signal recording means 1302. Further, when both code amounts are within the target code amount and are close to the target code amount, either one that is closest to the target code amount is selected, and the coded signal is recorded as the recording coded signal 1301. It is output to the means 1302.

The quantization parameter determining means 1033 receives the control signal 1034, adds a predetermined number to the pre-quantization parameter 1032 of the previous stage stored in the quantization parameter memory means 1031, for example, adds 1 and then It is output to the quantizing means 1006 and the quantizing parameter memory means 1031 as the quantizing parameter 1030 used for the quantizing.

In the coded signal recording means 1019, the coded signal for recording 1018 corresponds to the recording position on the recording medium such as a magnetic tape in the coding unit and the recording unit in the arrangement as shown in FIG. To record.

As described above, according to the present embodiment, the coding is performed using the result of the coding with the appropriate quantization parameter and the result of the coding after the zero value substitution of the AC component with the same quantization parameter. By performing the amount control, it becomes possible to perform more precise code amount control as compared with the conventional code amount control using only the quantization parameter, and it is possible to efficiently fall within the target code amount range.

FIG. 5 shows a block diagram of an image recording apparatus in an embodiment of the fifth invention. In the figure, 1001 is an input signal, 1002 is a blocking means, 1003 is a blocking signal, 10
04 is an orthogonal transformation means, 1005 is an orthogonal transformation coefficient, 1006 is a quantization means, 1007 is a quantized signal, 1408 is a variable length coding means, 14
09 is a coded signal, 1030 is a quantization parameter, 1031 is a quantization parameter memory means, 1032 is a pre-quantization parameter,
1033 is a quantization parameter determining means, 1034 is a control signal, 12
06 is an AC component zero value replacing means, 1207 is a zero value replacing quantized signal, 1208 is a second variable length coding means, 1209 is a second coded signal, 1401 is a code amount judging means, 1402 is a recording coded signal. , 1403 is an encoded signal recording means, 1404 is a memory means, 14
05 is a pre-encoded signal.

The difference between the configuration of the fourth embodiment and this embodiment is that the first variable-length coding means 1208 and the second variable-length coding means are made common and the variable-length coding means 1408 is provided. The memory means 1404 and the pre-encoded signal 1405 are added.

With respect to the operation of the image recording apparatus of the present embodiment configured as above, the operation common to the operation of the embodiment of the fourth invention is omitted, and the operation of the different constituents will be described below. Described in.

The quantized signal 1007 is the same in the AC component zero value replacing means 1206 by replacing the quantized signal 1007 with a non-zero value in the high frequency component of the AC component, by replacing it with a zero value. When quantization is performed with the quantization parameter, the code amount can be reduced.

In the present embodiment and the embodiment of the fourth invention, the AC component zero value replacement process is performed on the quantized signal, but this process is performed at the time of variable length coding. It is also possible. FIG. 9 is a block diagram showing an example of the AC component zero value replacement process.

In the figure, 101 is a quantized signal, 102 is a variable length coding means, 103 is a codeword, 104 is a codeword adding means, 105 is a first codeword string, 106 is a first codeword memory means, 107 is second
A code word string, 108 is a code amount determining means, 109 is a recording coded signal, 110 is a second code word memory means, and 111 is a third code word string.

The operation of the AC component zero value replacement device configured as described above will be described below. The quantized signal 101 is variable length coded by the variable length coding means 102 for each quantized coefficient and output as a code word 103. Codeword 103
Is added to the second codeword string 107, which is in the previous coding state, by the codeword adding means 104, and is output to the first codeword memory means 106 as the first codeword string 105.

In the first code word memory means 106, the first code word string 105 is stored, and at the same time when the next code word string is input, the stored code word string is set as the second code word string 107. It outputs to the code word adding means 104, the second code word memory means 110, and the code amount determining means 108.

In the second code word memory means 110, the second code word string 107 is stored, and at the same time when the next code word string is input, the stored code word string is set as the third code word string 111. It is output to the code amount determination means 108. At this time, at the same time, the first codeword memory means 106 and the second codeword memory means 1
The difference between the codeword strings stored in 10 and 10 is that the most recently encoded one codeword is not added to the codewords stored in the second codeword memory means 110.
That is, at the time when the coding is completed, the last codeword (usually a nonzero-value codeword of the high frequency component of the AC component)
The code word string from which is omitted is stored in the second code word memory means 110. In this way, the AC component zero value replacement process can be performed at the time of variable-length coding. This device is also applicable to the embodiment of the fourth invention.

Here, the difference between this embodiment and the embodiment of the fourth invention is that in the embodiment of the fourth invention, the AC component zero value replacement processing and the normal quantization processing are performed in parallel. In contrast to the configuration, the serially configured, previous quantized and coded state is stored in the memory means 1404, the pre-coded signal 1405 is output to the coding determination means 1401. . That is, the quantization parameter is alternately changed with time, and the AC component zero value replacement is performed. For example, in a time series, the quantization parameter is updated and the zero value is replaced so that the code amount becomes small as follows.

The first is the quantization parameter 1, the second is the quantization parameter 1 and zero value replacement, the third is the quantization parameter 2, and the fourth is the quantization parameter 2 and zero value replacement, 5
The third is the quantization parameter 3 ,. . . , (2 * q−1) th is the quantization parameter q, 2 * q is the quantization parameter q and zero value substitution ,. . . Becomes Further, in the present embodiment, they are configured in series, and the comparison between the first and the second, the second and the third is performed by the code amount comparing means 1401. In the case of the embodiment of the fourth invention, the comparison of the first and the second, the third and the fourth is performed, so that the embodiment is
Since there are many conditions to be compared, finer code amount control is possible.

As described above, according to this embodiment, in addition to the fourth aspect of the present invention, the result of encoding by an appropriate quantization parameter and a quantization parameter smaller than the quantization parameter and having zero AC component. By using the result obtained by performing the value substitution and then the encoding, it is possible to perform more precise code amount control from the fourth aspect, and it is possible to efficiently set the target code amount within the range.

FIG. 6 shows a block diagram of an image recording apparatus in an embodiment of the sixth invention. In the figure, 1001 is an input signal, 1002 is a blocking means, 1003 is a blocking signal, 10
04 is an orthogonal transformation means, 1005 is an orthogonal transformation coefficient, 1006 is a quantization means, 1007 is a quantized signal, 2101 is a variable length coding means, 21
02 is a variable length coded signal, 2103 is a coding method switching means, 2104 is a fixed length coding means, 2105 is a fixed length coded signal, 2106 is a switching control signal, 2010 is a quantization parameter, and 2011 is a quantization parameter memory. Means, 2012 is a pre-quantization parameter, 2013 is a quantization parameter determination means, 2014
Is a quantization control signal, 2015 is a code amount determination means, 2016 is a coded signal, 2017 is a codeword division means, 2018 is a recording coded signal, and 2019 is a coded signal recording means.

The operation of the image recording apparatus of the present embodiment constructed as described above will be described below. The input signal 1001 which is an image signal is divided into coding units by the blocking means 1002. For example, if the encoding method is MPEG, the minimum unit of encoding that can reconstruct the screen at the time of decoding is a macro block, so that it is a rectangular block of 16 pixels vertically and horizontally. If the encoding unit is the macroblock unit, the luminance signal and the color difference signal at the same screen position can be reproduced, and therefore, the screen area having the size of the macroblock can be reproduced. Here, the case of encoding by MPEG is taken as an example, but a block unit suitable for the encoding method may be selected.

Blocked signal 1003 thus divided
Is orthogonally transformed by the orthogonal transformation means 1004 and outputs the orthogonal transformation coefficient 1005 to the quantization means 1006. As the orthogonal transform, discrete cosine transform is used in MPEG or JPEG, but the present invention is not limited to this, and other orthogonal transform such as Haar transform or Hadamard transform may be used according to the encoding method. Absent.

In the quantizing means 1006, the quantization parameter 20
The orthogonal transform coefficient 1005 is quantized based on 10, the coefficient to be encoded is reduced, and the quantized signal 1007 is output.

The quantized signal 1007 is outputted to the fixed length coding means 2104 or the variable length coding means 2101 in the coding method switching means 2103 based on the switching signal 2106 outputted from the code amount judging means 2015. . Switching of this coding method, the fixed length coded signal 2105 and the variable length coded signal 2102 are input, the switching is determined in the code amount determination means 2015, and the switching control signal 2106 is sent to the coding method switching means 2103. Is output. At this time, the coding method switching criterion is variable length coding by the variable length coding means 2101 until just before the code amount exceeds the code amount for one recording unit, and when it exceeds, the fixed length coding. By means 2104,
It is fixed length coded. After that, the variable length coding means 2101 performs variable length coding until the recording of the next recording unit is completed. That is, only when the break of the recording unit is reached,
It is encoded by the fixed length encoding means 2104.

In this embodiment, like MPEG,
If fixed-length coding is possible by adding an escape code as well as variable-length coding, the purpose is to record one coding unit in multiple recording units. And fixed length coded signal
It is realized by using 2105 and as a break between recording units.

Further, the code amount determining means 1017 evaluates the entire code amount, and when the code amount of the encoded signal 1009 exceeds the target recording code amount, the quantization control signal 2014 is set to the quantization parameter determining means 2013. , And if it falls within the target code amount, the code word dividing means 20
Output to 17.

In the code word dividing means 2017, the escape code of the code word string having the code amount within the code amount of one recording unit and having the closest code amount is detected and divided from the encoded signal 2016, and the code for recording is recorded. The encoded signal 2018 is output to the encoded signal recording means 2019.

FIG. 10 shows an example in which a code word of one coding unit is recorded in a plurality of recording units, for example, two recording units. In the first recording unit of the two recording units, the escape code indicates the end of the code word. However, since there may be multiple escape codes in the same codeword, 1
The closest escape code within the code amount corresponding to the recording unit is detected. In the example of this figure, two recording units are used, but the present invention is not limited to this.

In this way, by recording one coding unit in a plurality of recording units, the number of recordable codes per one coding unit can be increased as compared with the case where one coding unit is recorded in one recording unit. Therefore, the image quality per one coding unit can be improved.

The quantization parameter determining means 2013 receives the quantization control signal 2014 and receives the quantization parameter memory means.
Pre-quantization parameter of the previous stage stored in 2011 2012
Is added to the quantization means 1006 and the quantization parameter memory means 2011 as a quantization parameter 2010 to be used for the next quantization.

In the coded signal recording means 2019, the coded signal for recording 2018 is arranged so that the coding unit and the recording unit correspond to the recording position on the recording medium such as the magnetic tape in the arrangement shown in FIG. To record.

As described above, according to this embodiment, MPE
In the case where not only variable-length coding but also fixed-length coding is possible by adding an escape code like G, when the code amount of one coding unit exceeds one recording unit, it is forced. The escape code is recorded in, the fixed length code following the escape code is recorded in the next recording unit, and encoding can be performed in a plurality of recording units by continuing the encoding. Since a large amount of code can be used, the image quality can be improved.

[0082]

EFFECTS OF THE INVENTION One recording unit or a fixed number of recording units are matched with a unit compression block (MPEG macro block) for encoding, and encoding is performed within the range of the recording capacity. The amount is fixed, and each recording unit can be independently reproduced. In particular, in high-speed reproduction, when reproduction is difficult in a continuous recording range, if reproduction is possible for each recording unit, the reproducible portion increases and the image quality during high-speed reproduction improves.

When correction is necessary after recording, if reproduction is performed in coding units and recording is performed at the same recording position after correction,
The correction can be made without affecting other recorded parts.

[Brief description of the drawings]

FIG. 1 is a block diagram of an image recording apparatus according to an embodiment of the first invention.

FIG. 2 is a block diagram of an image recording apparatus in an embodiment of the second invention.

FIG. 3 is a block diagram of an image recording apparatus in an embodiment of the third invention.

FIG. 4 is a block diagram of an image recording apparatus in an embodiment of the fourth invention.

FIG. 5 is a block diagram of an image recording apparatus in an embodiment of the fifth invention.

FIG. 6 is a block diagram of an image recording apparatus in an embodiment of the sixth invention.

FIG. 7 is an explanatory diagram related to an example of data arrangement on a recording medium when the embodiment of the first invention is used.

FIG. 8 is an explanatory diagram related to a procedure for determining a quantization parameter in the embodiment of the third invention.

FIG. 9 is a block diagram of an apparatus that realizes AC component zero value replacement processing.

FIG. 10 is an explanatory diagram related to an example of data arrangement on a recording medium when the embodiment of the sixth invention is used.

[Explanation of symbols]

 11 block forming means 13 block unit encoding means 15 recording signal separating means 17 code amount determining means 20 encoded signal recording means 102 variable length encoding means 104 codeword adding means 106 first codeword memory means 108 code amount determining means 110 Second code word memory means 1002 Blocking means 1004 Orthogonal transformation means 1006 Quantization means 1008 Variable length coding means 1017 Code amount determination means 1019 Coded signal recording means 1011 Quantization parameter memory means 1013 Quantization parameter increment means 1027 Code Quantity determination means 1029 Coded signal recording means 1021 Quantization parameter memory means 1023 Quantization parameter calculation means 1025 Reference quantization parameter update means 1108 First variable length coding means 1031 Quantization parameter memory means 1033 Quantization parameter determination means 1206 AC Zero component replacing means 1208 Second variable length coding means 1300 Code amount judging means 1302 Coding Code recording means 1401 Code amount determination means 1403 Coded signal recording means 1404 Memory means 1408 Variable length coding means 2011 Quantization parameter memory means 2013 Quantization parameter determination means 2015 Code amount determination means 2017 Codeword division means 2019 Coded signal recording Means 2101 Variable length coding means 2103 Coding method switching means 2104 Fixed length coding means

Claims (6)

[Claims] 1. A blocking means for dividing an input signal, which is a coded screen, into blocks in a coding unit and outputting the block signal as a block signal, and coding the block signal in block units by a coding method such as MPEG. A block unit coding means for outputting as a block unit coded signal, and a recording unit for separating from the block unit coded signal only a code necessary for decoding in block units and outputting as a recording signal An image recording apparatus comprising: a signal separation unit; and an encoded signal recording unit that records the recording signal in a predetermined recording unit. 2. Blocking means for dividing an input signal, which is a screen to be coded, into coding units and outputting as a blocking signal, and orthogonal transforming the blocking signal in block units to output orthogonal transform coefficient signals. An orthogonal transform unit that quantizes the orthogonal transform coefficient signal using a quantization parameter and outputs a quantized signal, and a variable length that quantizes the quantized signal and outputs the quantized signal as a coded signal. An encoding unit, a code amount is calculated from the encoded signal, compared with a target code amount, and when the code amount of the encoded signal is larger than the target code amount, an increment control signal is sent to the quantization parameter increment unit. A code amount determining means for outputting a coded signal for recording to the coded signal recording means in other cases, and based on the increment control signal, Quantization parameter increment means for incrementing a child parameter by a predetermined number and outputting it to the quantization means and quantization parameter memory means, and storing the quantization parameter, and the quantization parameter increment means as the pre-quantization parameter. An image recording device comprising: a quantization parameter memory means for outputting to a recording medium; and an encoded signal recording means for recording the recording encoded signal in a predetermined recording unit. 3. Blocking means for dividing an input signal, which is a screen to be coded, into coding units and outputting as a blocking signal, and orthogonal transforming the blocking signal in block units to output orthogonal transform coefficient signals. An orthogonal transform unit that quantizes the orthogonal transform coefficient signal using a quantization parameter and outputs a quantized signal, and a variable length that quantizes the quantized signal and outputs the quantized signal as a coded signal. Encoding means, calculates the code amount from the coded signal, outputs a control signal by comparing with the target code amount, if the closest to the target code amount within the target code amount, the recording coded signal Code amount determination means for outputting, and based on the control signal and the number of times of coding, a reference quantization parameter update for calculating and updating a reference quantization parameter serving as a reference for changing a quantization parameter. And a quantization parameter calculation means for calculating a quantization parameter from the reference quantization parameter and the pre-quantization parameter and outputting the quantization parameter to the quantization means and the quantization parameter memory means, and storing the quantization parameter. An image recording comprising a quantization parameter memory means for outputting to the quantization parameter increment means as the pre-quantization parameter and an encoded signal recording means for recording the recording encoded signal in a predetermined recording unit. apparatus. 4. A block forming means for dividing an input signal, which is a screen to be encoded, into blocks in a coding unit and outputting the block as a block signal, and an orthogonal transform coefficient signal for orthogonally transforming the block signal in block units. Orthogonal transformation means for outputting to the quantization means, the orthogonal transformation coefficient signal is quantized using a quantization parameter, and AC component zero value replacement means as the quantized signal;
Quantizing means for outputting to the variable length coding means; first variable length coding means for variable length coding the quantized signal and outputting it to the code amount judging means as a first coded signal; AC component zero-value replacing means for replacing a predetermined number of AC components in 0 with zero values and outputting to the second variable-length encoding means as a zero-value replacing quantized signal; and variable-length encoding the zero-value replacing quantized signal. , A second variable length coding means for outputting to the code amount determination means as a second coded signal, and a code amount of the first coded signal, a code amount of the second coded signal and a target code amount. Then, when both the first coded signal and the second coded exceed the target code amount, the control signal is output to the quantization parameter determining means,
When any one of the code amounts falls within the signal target code amount, a code amount determining unit that selects the coded signal having the maximum code amount in the range of the target code amount and outputs the coded signal for recording, the control signal A quantization parameter determining means for recalculating a pre-quantization parameter and outputting the quantization parameter to the quantization means and a quantization parameter memory means, and storing the quantization parameter, and the quantization parameter as the pre-quantization parameter. An image recording device comprising: a quantization parameter memory means for outputting to a determining means; and an encoded signal recording means for recording the recording encoded signal in a predetermined recording unit. 5. Blocking means for dividing an input signal, which is a screen to be coded, into coding units, and outputting as a blocking signal, and orthogonal transforming the blocking signal in block units to quantize orthogonal transform coefficient signals. An orthogonal transform means for outputting to the quantizing means, a quantizing means for quantizing the orthogonal transform coefficient signal using a quantization parameter, and outputting as a quantized signal to the AC component zero value replacing means and the variable length coding means, A predetermined number of AC components in the quantized signal are replaced with zero values, AC component zero value replacement means for outputting to the variable length coding means as a zero value replacement quantized signal, and the quantized signal is variable length coded, Variable length coding means for outputting to the code amount judging means and the memory means as a coded signal, and storing the coded signal, and shifting the time to the code amount judging means as a pre-coded signal Comparing the code amount of the encoded signal, the code amount of the pre-encoded signal and the target code amount, the encoded signal and the pre-encoding both exceed the target code amount. In this case, the control signal is output to the quantization parameter determining means, and if either code amount falls within the signal target code amount, the coded signal having the maximum code amount in the target code amount range is selected, A code amount determination means for outputting as a coded signal for recording, a recalculation of a pre-quantization parameter based on the control signal, and a quantization parameter determination means for outputting to the quantization means and a quantization parameter memory means, A quantization parameter memory unit that stores a quantization parameter and outputs it to the quantization parameter determination unit as the pre-quantization parameter, and records the encoded signal for recording in a predetermined recording unit. Image recording device comprising an encoding signal recording means. 6. In a coding system capable of fixed-length coding by adding an escape code at the same time as variable-length coding, when recording data for a coding unit in a plurality of recording units, a coded screen is displayed. A block dividing unit that divides the input signal into blocks by an encoding unit and outputs the block signal as a block signal; an orthogonal transform unit that orthogonally transforms the block signal in block units and outputs an orthogonal transform coefficient signal; Quantizing means for quantizing the coefficient signal using a quantization parameter and outputting the quantized signal; and outputting the quantized signal by switching to fixed length coding means or variable length coding means based on a switching control signal. Encoding method switching means, variable length encoding means for variable length encoding the quantized signal and outputting the variable length encoded signal to the code amount determining means, and the amount A fixed length coding means for adding a escape code to the coded signal to perform fixed length coding and outputting the fixed length coded signal to the code quantity determination means; and the code quantity of the variable length coded signal is the code quantity of a recording unit. When it exceeds the limit, the fixed length coding is stopped at the point where the limit is not exceeded, the fixed length coding is performed once, and then the switching control signal is output to the coding method switching means so that the variable length coding is performed again. If the sum of the code amounts of the preceding variable length coding and the fixed length coding is larger than the code amount of a predetermined number of recording units, the quantization control signal is sent to the quantization parameter determining means. A code amount determining means for outputting and outputting as a coded signal if small, and a quantum for recalculating a pre-quantization parameter based on the quantization control signal and outputting to the quantization means and a quantization parameter memory means. Conversion A parameter determining unit, a quantization parameter memory unit that stores the quantization parameter and outputs it as the pre-quantization parameter to the quantization parameter determining unit; and within a code amount for one recording unit from the encoded signal. A codeword dividing unit that detects the last existing escape code, divides the coded signal before and after the escape code, and sequentially outputs the coded signal for recording, and records the coded signal for recording in a predetermined recording unit. And an image recording device including a coded signal recording means for performing the same.