JP2000134617A - Image encoding device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To stabilize a picture quality by suppressing the local or instantaneous fluctuation of a quantizing step while eliminating the difference of encoding distortion between parallel processing concerning a device for dividing an input image and parallelly compression-encoding them. SOLUTION: The input image is divided to two images by a picture dividing circuit 2, two divided image signals are parallelly compression-encoded through the applied quantizing step by MPEG2 encoders 3 and 4, and the code data of two channels outputted from the MPEG2 encoders 3 and 4 are multiplexed by a multiplexing means 17. The multiplexing means 18 generates one synthetic stream capable of decoding the entire input image from the code data of two channels, controls stuffing processing corresponding to the generated code amount of the synthetic stream and feeds the same quantizing step back to two MPEG2 encoders 3 and 4 so that the generated code amount of the synthetic stream can become the desired code amount.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

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

[0002]

2. Description of the Related Art A system combining DCT (Discrete Cosine Transform) and motion compensation is effective as a highly efficient compression coding technique for natural moving pictures.
The G2 system (ISO / IEC13818) has begun to be widely used. In recent years, a one-chip coded LSI corresponding to MP @ ML (main profile / main level) has also been developed, and has been relatively easily coded (for example, ISSCC97 [FP16.2]).

[0003] On the other hand, in the field of broadcasting, new broadcasting systems such as a progressive system and HDTV have been proposed with the aim of improving the image quality of images, and studies are being conducted on the realization of digital broadcasting using a compression encoding technique.

[0004]

However, high-quality video signals equivalent to H14L (high 14 level) or HL (high level) such as 525 progressive, 1125 interlace, and 750 progressive, which are proposed as new broadcasting systems, are not available. Since the number of effective pixels per unit time is large, it is impossible in terms of processing capacity to perform encoding with one MP @ ML encoding LSI.

Therefore, attempts have been made to parallelize processing by using a plurality of coding LSIs. However, in a parallel coding method in which the processing is divided spatially or temporally and coding is performed independently, the parallel processing is not performed. Adaptive bit allocation becomes difficult. For example, some coding LSIs perform stuffing processing on invalid data because the amount of generated code is extremely small, while other coding LSIs generate an extremely large amount of generated code and have to terminate encoding. Balance may occur. Furthermore, even if it is not an extreme case so far, it has been a problem that distortion is easily detected because a difference in encoding distortion between parallel processes makes a boundary of division more prominent, thereby deteriorating subjective image quality.

Accordingly, it is an object of the present invention to provide an image encoding device capable of achieving high image quality.

[0007]

In order to solve this problem, according to the present invention, first, an image encoding apparatus for dividing an input image into N (N is an integer of 2 or more) regions and performing compression encoding. , A screen dividing means for obtaining N divided images, N encoding means for compressively encoding image signals divided into N systems in a given quantization step, and N encoding means Multiplexing means for multiplexing N-channel code data output from the multiplexing unit to generate one combined stream capable of decoding the entire input image from the N-channel code data, thereby realizing parallel coding. In addition, in the multiplexing means, unnecessary stuffing is suppressed by controlling the stuffing process by the total generated code amount after multiplexing, that is, the generated code amount of the combined stream, and By constantly recalculating the quantization step such that the code amount becomes the desired code amount and performing code amount control to feed back the same quantization step to all the parallel encoding means, it is possible to reduce the amount of parallel processing. The bit distribution is automatically optimized to achieve high image quality.

Second, when the encoding means complies with MPEG2, the MPEG2 encoder does not cause the input buffer of the decoder to overflow or underflow, so that the VBV (Video Buffering Verifi
er) Buffer control needs to be performed. In the present invention, instead of performing VBV buffer control for each parallel coding unit, local or instantaneous quantization is determined by determining a quantization step in accordance with the VBV buffer sufficiency of the entire image that changes slowly. This makes it possible to stabilize the image quality by suppressing the fluctuation of the image formation step.

Third, if it is predicted that it will be difficult to suppress the code amount to a desired code amount even by controlling the code amount by updating the quantization step shown in the second means, the AC (AC) is determined by the sufficiency of the VBV buffer. By selectively encoding the coefficients according to the frequency, changing the weighting according to the I, P, B picture type, or changing the properties of the pre-filter to change the properties of the input image, This makes it possible to perform encoding in which distortion is hardly visually detected even in a complicated moving image in which a large amount of quantization distortion occurs.

More specifically, the code amount can be controlled by the control (VBV buffer control) shown by the second means. However, for example, when small pictures that are difficult to compress are continuously input, the second By adjusting the data amount by the third means in addition to the means described above, it is possible to suppress image quality deterioration. Third means (AC depends on whether the VBV buffer is full)
(AC) Selectively encode coefficients according to frequency, change weighting according to I, P, B picture type, or change characteristics of pre-filter to change properties of input image ) Is difficult to encode and VB
This is particularly effective when the V buffer is driven. The prediction can be easily made by checking the level of the VBV buffer (that is, the degree of sufficiency of the VBV buffer).

Next, a point at which the AC coefficient is selectively encoded according to the frequency according to the degree of sufficiency of the VBV buffer, or an AC coefficient to be encoded by N encoding means is selected according to the degree of sufficiency of the VBV buffer. Will be described. In DCT coding, data to be recorded and transmitted is decomposed into frequencies. The lower frequency component has a greater effect on the reproduction image quality, and the higher frequency component contains only fine information. Therefore, especially when the VBV buffer is driven in and encoding is difficult, the low-frequency components having relatively high importance can be encoded by sequentially cutting off the high-frequency components. That is, selective encoding means that high-frequency components are discarded with priority given to low-frequency components.

Next, a description will be given of how the weighting is changed according to the I, P, and B picture types. For example, V
If the BV buffer is not enough and is driven, I,
When weighting is performed in the order of P and B as the importance is high, I
A picture is coded with a larger code amount. Usually I,
An improvement in image quality can be obtained with respect to not weighting P and B.

Next, a description will be given of how the characteristics of the input image are changed by changing the characteristics of the pre-filter. As before, if the VBV buffer runs out of room,
Blur the input image by filtering. As a result, the high-frequency components are reduced, so that the image quality can be improved in the same manner as when the low-frequency components having relatively high importance are selectively encoded as in the above example.

Specifically, the multiplexing means includes N FIFO memories to which the outputs of the N encoding means are supplied, a switch for switching and outputting the outputs of the N FIFO memories, Calculate the number of bits of invalid data to be stuffed for each frame based on the N code amount counters for integrating the code amounts output from the N encoding means, and the sum of the integrated values of the N code amount counters. A stuffing control circuit, a stuffing circuit for inserting invalid data by the number of bits indicated by the stuffing control circuit into the output from the changeover switch, and a sum of integrated values of the N code amount counters indicating the number of invalid data bits to be stuffed. In order to control the generated code amount of the added final synthesized stream to the desired code amount, the optimum and only quantization step is calculated and N Comprising a quantization switch determining circuit that feeds back to the coding means. As a result, since the feedback loop of the code amount control is completed before the FIFO memory in spite of the parallel encoding, the feedback delay can be minimized, and fine and accurate code amount control can be realized.

[0015]

BEST MODE FOR CARRYING OUT THE INVENTION An embodiment of the present invention will be described below with reference to FIG. 1 by taking as an example a case where two MPEG-2 encoders of MP @ ML are arranged in parallel to encode a 525 progressive signal corresponding to MP @ H14L. This will be described with reference to FIG.

(First Embodiment) FIG. 1 is a block diagram showing an image coding apparatus according to a first embodiment of the present invention. In FIG. 1, reference numeral 1 denotes an input terminal for a 525 progressive signal. Reference numeral 2 denotes a screen dividing circuit as a screen dividing means for dividing a 525 progressive image into two equal parts vertically. Reference numerals 3 and 4 denote first and second MPEG2 encoders as parallelized encoding means, respectively, which compressively encode image signals divided into two systems in a given quantization step. Reference numerals 5 and 6 denote first and second code amount counters for accumulating the code amounts generated by the first and second MPEG2 encoders 3 and 4, respectively.

7 and 8 are the first and second MPEG2
First and second FIFO memories for temporarily storing code data generated by the encoders 3 and 4. 9 is FIFO
This is a memory output switch. Reference numeral 10 denotes a quantization step determination circuit that determines a quantization step for obtaining a desired code amount. A stuffing control circuit 11 calculates the number of bits of invalid data to be stuffed for each frame. A stuffing circuit 12 inserts invalid data by the number of bits specified by the stuffing control circuit 11 in order to prevent buffer underflow. Reference numeral 13 denotes a header processing circuit that rewrites a parameter of a header that needs to be changed by combining the streams. An output terminal 14 is a compressed stream of the 525 progressive signal that is finally synthesized. 15,16
Are adders, respectively.

In the above, the first and second code amount counters 5, 6, the first and second FIFO memories 7,
8, FIFO memory output changeover switch 9, quantization step determination circuit 10, stuffing control circuit 11,
The stuffing circuit 12 and the header processing circuit 13 constitute multiplexing means 17 and multiplex the two-channel code data output from the first and second MPEG2 encoders 3 and 4 to thereby input the two-channel code data. One composite stream capable of decoding the entire image is generated. Further, the multiplexing means 17 controls the stuffing process according to the generated code amount of the combined stream, and controls all the first and second MPEG2 encoders 3 and 4 so that the generated code amount of the combined stream becomes a desired code amount. The same quantization step is fed back.

The parameter rewriting in the header processing circuit 13 is different depending on the number of the N parallel encoding units, but for example, the combined stream must have an image size larger than that of the divided stream. . Therefore, the parameters related to the image size need to be rewritten.

The operation of the image coding apparatus configured as described above will be described below, including details of the operation of the multiplexing means 17.

FIG. 2 is a conceptual diagram showing the concept of converting a 525 progressive signal into two equal parts vertically and converting it into a pseudo interlace signal. The 525 progressive signal input from the input terminal 1 is, as shown in FIGS.
This is a progressive scan with 720 effective pixels × 480 lines and a frame rate of 59.94 Hz, and is a video signal equivalent to MP @ H14L in MPEG2. By dividing the screen vertically into two equal parts by the screen dividing circuit 2 and alternately extracting each line, as shown in FIGS. 2B and 2D, the effective number of effective pixels is 720 pixels × 120 pixels. It can be divided into interlaced scanning with a line and field rate of 119.88 Hz.

FIGS. 2 (a) and 2 (b) show that the lines are alternately extracted line by line. As a result, the progressive signal can be handled as an interlace signal. Black circles and white circles in FIGS. 2C and 2D indicate pixels, and black and white have the same relationship as the solid line and the broken line in FIGS. 2A and 2B. A broken line in which a circle is skewed indicates the direction of the time axis, and indicates the interval between adjacent frames / fields.

As described above, the number of effective pixels per unit time of the pseudo-interlaced signal generated by dividing the 525 progressive signal into upper and lower halves is MP ＠ ML in MPEG2.
Therefore, the encoding can be performed by using two MP @ ML MPEG2 encoders in parallel in terms of processing amount.

Here, the upper half image divided into two equal parts is encoded in parallel by the first MPEG2 encoder 3, and the lower half image is encoded in parallel by the second MPEG2 encoder 4. It is assumed that the data is written in the FIFO memory 7 and the second FIFO memory 8.

The first code amount counter 5 integrates the generated code amount of the upper half, and the second code amount counter 6 integrates the generated code amount of the lower half. However, the second code amount counter 6 does not add up the code amount of header data (for example, a sequence header, a GOP header, a picture header, etc.) in a layer higher than the slice header. Therefore, the total sum of the integrated values of the first and second code amount counters 5 and 6 becomes equal to the code amount before stuffing of the stream synthesized by reading the switch 9 of the FIFO memory output at the subsequent stage.

The stuffing control circuit 11 determines the number of bits of invalid data to be stuffed for each frame by a first and second code amount counter in order to prevent buffer underflow when the generated code amount of the combined stream is extremely small. It is calculated from the sum of the integrated values of 5 and 6. The stuffing circuit 12 inserts invalid data into the output from the changeover switch 9 by the number of bits specified by the stuffing control circuit 11.

The quantization step determination circuit 10 adds the number of stuffing bits (the number of invalid data bits to be stuffed) indicated by the stuffing control circuit 11 to the sum of the integrated values of the first and second code amount counters 5 and 6. The generated code amount of the final synthesized stream added is determined by the adder 1
5 and 16, the optimum and only quantization step is immediately calculated to control the generated code amount of the final combined stream to a desired code amount, and the first and second MPEG2 The same quantization step is fed back to the encoders 3 and 4.

Therefore, the parallelized MPEG2 encoders 3 and 4 all encode at the same quantization step at a certain time, and it is possible to suppress variations in encoding distortion between parallel processes. . The code data coded in parallel and written into the first and second FIFO memories 7 and 8 are alternately read out while being switched by the FIFO memory output changeover switch 9 each time both picture header codes are detected. Are combined into one stream.

FIG. 3 is a schematic diagram showing the operation of the changeover switch 9 for FIFO memory output. On the left side of FIG. 3, the sequence of data written to the first and second FIFO memories 7 and 8 and sequentially output, and the first and second FIFO memories are shown.
The reading order of data in the O memories 7 and 8 is shown. On the right side of FIG.
2 shows the arrangement of data output from. First FI
In the FO memory 7, a picture header and a slice header 0 to a slice header 14 are repeatedly arranged following the sequence header and the GOP header, and in the second FIFO memory 8, a picture header is arranged following the sequence header and the GOP header. And the slice headers 15 to 29 are repeatedly arranged.

At this time, the FIFO memory output changeover switch 9 discards the header data of the upper layer than the slice layer for the code data of the lower half image read from the second FIFO memory 8,
It can be attached so as to satisfy the syntax of MPEG2.

The stuffing control circuit 11 determines how many bits of invalid data should be inserted when the generated code amount of the synthesized stream is extremely small and the input buffer of the MPEG2 decoder may underflow. The quantization step determining circuit 1 calculates and instructs the stuffing circuit 12 to insert invalid data, and determines the quantization step from the total amount of generated codes.
The number of stuffing bits is also given to 0.

Here, the quantization step determining circuit 10, the stuffing control circuit 11, and the stuffing circuit 1
2 will be described. Quantization step decision circuit 1
0 determines the quantization step according to, for example, a function as shown in FIG. 6 in accordance with the VBV buffer sufficiency. The stuffing control circuit 11 determines whether stuffing is necessary and calculates the number of invalid data bits to be stuffed if necessary. The stuffing circuit 12 performs actual stuffing according to the result of the stuffing control circuit 11.

By synthesizing the streams as described above,
One header parameter that needs to be rewritten, such as an image vertical size in a sequence header, a VBV delay in a picture header, and a vertical position of a slice in a slice header, can be rewritten by the header processing circuit 13 and decoded as a 525 progressive signal. Complete the compressed stream.

With this configuration, the feedback loop of the code amount control is completed in front of the first and second FIFO memories 7 and 8 while the parallel coding is performed, and the feedback delay is minimized. Fine and accurate code amount control can be realized, and coding efficiency can be improved by suppressing unnecessary stuffing in units of parallel encoders and performing stuffing only on a synthesized stream. Furthermore, an optimal bit distribution is automatically achieved between parallel processes due to the parallel coding using the same quantization step. As a result, high image quality can be achieved.

(Second Embodiment) Hereinafter, a second embodiment of the present invention will be described.
The image encoding device according to the embodiment will be described with reference to the drawings.

With the same configuration as that of the first embodiment, the operation of the quantization step determining circuit 10 for achieving higher image quality coding, especially when the coding method is MPEG2, An example will be described.

FIG. 4 shows that the input frame order 41 is the encoding order 42
MPEG2 encoded output 43 rearranged and encoded
FIG. 6 is a time chart showing an increase / decrease 44 of the VBV buffer due to V.V. The amount of code consumed by decoding the frame is subtracted from the accumulated code amount, and the number of VBV buffers that repeats a saw-tooth change is increased / decreased 44. The gradient in the case of the fixed rate is constant, and the gradient increases as the rate increases.

In order to obtain the highest image quality within a range that does not cause the input buffer of the decoder to fail, the increase / decrease 44 of the sufficiency of the VBV buffer is controlled as if it always lies between the upper limit value and the lower limit value while forcibly as much as possible. It is important not to limit the amount of generated code. Therefore, the VBV buffer sufficiency 45 that changes relatively slowly is defined as the sum of the difference between the actual generated code amount and the target code amount.
Are used to determine the quantization step.

Here, the relationship between the actual generated code amount, the target code amount, and the upper and lower limits of the increase / decrease 44 of the VBV buffer will be described. The target code amount is the average bit rate of 1
Although they are equally divided per frame, they do not actually match exactly as they are. Therefore, although the integrated value of the actual code amount generated in the past becomes substantially equivalent to the degree of sufficiency of the VBV buffer, it is necessary to keep the integrated value from exceeding the upper limit and the lower limit.

FIG. 5 is a block diagram showing a quantization step determining circuit of an image coding apparatus according to a second embodiment of the present invention. In FIG. 5, reference numeral 51 denotes an integration circuit for obtaining the degree of sufficiency of the VBV buffer from the sum of the difference between the generated code amount and the target code amount. Reference numeral 52 denotes a quantization step determination function for determining a quantization step from the sufficiency of the VBV buffer.

As in the image encoding apparatus shown in FIG.
In the case of parallel encoding, the difference between the generated code amount of two macroblocks (referred to as MB in the figure) and the target code amount for two macroblocks that are processed at the same time is integrated. A macro block is a unit of a unit when a screen is divided into blocks and encoded. The code amount per macroblock is obtained by dividing the code amount of one frame composed of a plurality of macroblocks by the number of macroblocks. When processing is performed in parallel, there are two macroblocks of interest at the same time, and therefore, consideration is made in units of two macroblocks.

The quantization step can be determined directly from the VBV buffer sufficiency using a function. For example, FIG.
When a quantization step decision function as shown in FIG.
By setting the sensitivity to be low in a range in which the degree of sufficiency of the BV buffer is close to the target value, it is possible to suppress a sudden change in the quantization step due to the characteristics of a local image and achieve image quality stabilization, while achieving VBV. Where the buffer sufficiency is extremely large, the sensitivity is set high to prevent overflow. As described above, it is possible to arbitrarily set the image quality and the response time in the code amount control by this function form.

In the above embodiment, the 525 progressive signal is divided into two equal parts at the top and bottom, and two MP @ ML
The above description has been made with reference to an example in which encoding is performed by an MPEG2 encoder. However, in the case where HDTV signals such as 1125 interlace and 750 progressive are encoded in parallel,
Even when the processing speed of the MPEG2 encoders is even slower, it can be similarly implemented by setting the dividing method and the number of parallel processing to the optimum (any value of 3 or more) and changing the combining method.

(Third Embodiment) On the basis of extension of the second embodiment, the first MPEG2 encoder 3 and the second MPEG2 encoder 4 perform quantum operations determined according to the VBV buffer sufficiency. As the conversion step becomes larger, the high frequency AC coefficient is not coded, and only the low frequency AC coefficient is selectively coded.
It is possible to prevent the buffer from overflowing and to suppress image quality deterioration.

(Fourth Embodiment) In the first and second MPEG2 encoders 3 and 4 of the second embodiment, the larger the quantization step determined according to the VBV buffer sufficiency, the more (I Target code amount of picture)> (P
(Target code amount of picture)> (Target code amount of B picture)
Thus, by making a difference depending on the picture type, it is possible to prevent the overflow of the VBV buffer and suppress the image quality deterioration.

(Fifth Embodiment) In the first and second MPEG2 encoders 3 and 4 of the second embodiment, the larger the quantization step determined according to the VBV buffer sufficiency, the larger the input image By changing the characteristics of the low-pass filter that limits the pass band of the input image to remove the high-frequency components of the input image, it is possible to prevent overflow and suppress image quality degradation.

[0047]

As described above, according to the image coding apparatus of the first aspect of the present invention, unnecessary stuffing is not performed in units of parallel coding, and stuffing is performed only on a synthesized stream. This can improve the coding efficiency. Further, since the optimal bit distribution is automatically performed between the parallel processes because the parallel encoding is performed using the same quantization step, high image quality can be achieved.

According to the image encoding apparatus of the second aspect,
When MPEG2 is used as the compression encoding method, the quantization step is changed in accordance with the VBV buffer sufficiency of the synthesized stream of the entire screen, which changes gradually, thereby suppressing local and instantaneous fluctuations of the quantization step and improving image quality. Can be stabilized.

According to the image encoding apparatus of the third, fourth or fifth aspect, A encoding is performed in accordance with the degree of sufficiency of the VBV buffer.
By selecting the C coefficient, changing the bit allocation (weighting) according to the I, P, and B picture types, or changing the characteristics of the pre-filter, even a complicated moving image in which a large amount of quantization distortion occurs. Encoding can be performed so that distortion is hardly detected visually.

According to the image coding apparatus of the sixth aspect,
Since the feedback loop of the code amount control is completed in front of the FIFO memory in spite of parallel coding, the feedback delay can be minimized, and fine and accurate code amount control can be realized.

[Brief description of the drawings]

FIG. 1 is a block diagram illustrating a configuration of an image encoding device according to a first embodiment of the present invention.

FIG. 2 is a conceptual diagram illustrating upper and lower equal division of a 525 progressive signal according to the first embodiment of the present invention.

FIG. 3 is a schematic diagram illustrating an operation of a changeover switch for FIFO memory output according to the first embodiment of the present invention.

FIG. 4 is a time chart showing increase / decrease of a VBV buffer according to a second embodiment of the present invention.

FIG. 5 is a block diagram illustrating a configuration of a quantization step determination circuit according to a second embodiment of the present invention.

FIG. 6 is a schematic diagram illustrating an example of a quantization step determination function according to the second embodiment of the present invention.

[Explanation of symbols]

 Reference Signs List 1 input terminal 2 screen division circuit 3 first MPEG2 encoder 4 second MPEG2 encoder 5 first code amount counter 6 first code amount counter 7 second FIFO memory 8 second FIFO memory 9 FIFO memory output Changeover switch 10 Quantization step decision circuit 11 Stuffing control circuit 12 Stuffing circuit 13 Header processing circuit 14 Output terminal 15 Adder 16 Adder

 ──────────────────────────────────────────────────続 き Continued on front page F term (reference) 5B047 AA30 EA07 5B057 AA20 CG02 CG03 CH04 5C059 KK11 KK13 PP04 RB12 SS01 SS11 TA41 TA45 TA46 TB07 TC18 TD07 TD11 UA31 UA34 5C078 AA04 BA57 CA00 DA01 DA07

Claims (6)

[Claims] 1. An image encoding apparatus which divides an input image into N (N is an integer of 2 or more) areas and performs compression encoding, comprising: a screen dividing means for obtaining N divided images; N encoding means for compressively encoding image signals divided into systems in parallel at a given quantization step, and multiplexing of N-channel code data output from the N encoding means Multiplexing means for generating one composite stream capable of decoding the entire input image from the N-channel code data, wherein the multiplexing means controls a stuffing process according to a generated code amount of the composite stream, The same quantization step is fed back to all of the N encoding means so that the generated code amount of the synthesized stream becomes a desired code amount. Encoding device. 2. N number of encoding means are MPEG2
An encoder, wherein the multiplexing unit performs VBV buffer control on the entire image based on the generated code amount of the combined stream to be generated, and feeds back the feedback to the N encoding units according to the VBV buffer fullness of the entire image. 2. The image coding apparatus according to claim 1, wherein the step is determined. 3. A multiplexing unit, which performs encoding by N encoding units in accordance with the degree of sufficiency of a VBV buffer of an entire image.
3. The image coding apparatus according to claim 2, wherein a C coefficient is selected. 4. The multiplexing means changes weights of I pictures, P pictures, and B pictures to be coded by N coding means according to the VBV buffer fullness of the entire image. Item 3. The image encoding device according to Item 2. 5. The image encoding apparatus according to claim 2, wherein the multiplexing means performs an adaptive filtering process on the entire input image according to the degree of sufficiency of the VBV buffer of the entire image. 6. The multiplexing means includes: N FIFO memories to which outputs of N encoding means are supplied; and N FIFO memories.
A changeover switch for switching and outputting the output of the FO memory; an N code amount counter for integrating the code amounts of the outputs of the N encoding means; and a total sum of integrated values of the N code amount counters. A stuffing control circuit for calculating the number of bits of invalid data to be stuffed for each frame, a stuffing circuit for inserting invalid data into the output from the changeover switch by the number of bits specified by the stuffing control circuit, and Calculate the optimal and only quantization step to control the generated code amount of the final combined stream to the desired code amount by adding the bit number of the invalid data to be stuffed to the sum total of the integrated value of the code amount counter. 2. The image code according to claim 1, further comprising a quantization switch determination circuit that feeds back to said N encoding means. Apparatus.