JPH0888854A - Moving picture encoding system - Google Patents

Abstract

PURPOSE: To facilitate editing operations such as the replacement of compression codes or the like without disturbing signals. CONSTITUTION: In this encoding system for compressing a required signal rate by obtaining reference pictures from pictures in the past and future in picture signals constituted of the plural pictures, detecting the motion vector and performing movement compensation, reference from the past pictures is approximately periodically =(a)} inhibited and it is realized by controlling a mode controller 230 by a GOP measurement device 250.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention compresses / compresses moving image signals.
The present invention relates to an encoding device, and more particularly to a moving image encoding system that enables editing of a compression encoded signal.

[0002]

2. Description of the Related Art In recent years, a high-efficiency compression / encoding system for reducing a required signal rate by removing the redundancy of an information source has been put into practical use. As a typical example of this, there is a so-called hybrid coding system in which spatial redundancy is removed from a moving image signal by orthogonal transformation such as DCT, and motion detection / compensation is performed to reduce temporal redundancy.

FIG. 8 shows an example of the structure of a hybrid coding device. A moving image signal is input to the input terminal 100.
The original image signal that has undergone processing such as filtering and noise reduction in the preprocessing circuit 110 is subtracted by the subtractor 120.
And to the motion detector 220. In addition, the preprocessing circuit 1
At 10, activity detection is performed to detect the complexity of the original image for determining the quantization step in the quantizer 140 described later. Further, in the pre-processing circuit 110, the scanning line order of the original image signal is converted into a scan (block scan) in units of, for example, horizontal 8 pixels and vertical 8 pixels (8 × 8 pixels) block (also a DCT processing block unit) and outputs it. To do.

Here, the coding mode in the hybrid coding device will be briefly described. First switch 210
And 200 are in the open state, in which the output signal of the subtractor 120 becomes equal to the input signal and is input to the DCT unit 130. The DCT unit 130 performs DCT processing (Discrete Co) on the block.
sine Transform). The DCT process transforms the spatial domain of horizontal 8 samples / vertical 8 lines forming a block into the frequency domain.
Here, as a general property of the image signal, it is known that the sample value signals randomly arranged in the block are concentrated in a specific frequency region by the DCT process. The quantizer 140 utilizes this property and the DCT unit 1
Of the 30 output signals, the minute components are truncated, and the signal components that are not truncated are also converted into discrete numerical values (the degree of discretization here is the activity signal and the generated code amount described later). Determined adaptively from the constraints).

Therefore, the number of signals required to express this block area can be reduced by the DCT processing and the quantization processing. In this way, for the signal of the input original image,
A mode in which compression is performed using only the spatial property in the image is called an intra coding mode.

Next, the operation in the coding mode in which the switches 220 and 230 are closed will be described. It is assumed that these switches 220 and 230 are closed by being controlled by the mode controller 230. Then, the output of the quantizer 140, for example, the intra-coded original image signal is decoded into a signal in the spatial domain by the inverse quantizer 150 and the DCT ⁻¹ unit 160 (inverse DCT processing). (Of course, the decoded image is irreversibly processed in the quantizer 140, and is not equal to the original image). Delay device 18
At 0, a signal delay corresponding to the time for forming one screen is performed, and the output signal is a motion compensator 190.
And the motion detector 220.

In the motion detector 220, the delay device 180 is used.
And the original image signal from the pre-processing circuit 110 are compared to detect the motion vector value for each block. That is, the motion vector value for the block arranged in the still region becomes zero, and for the block arranged in the moving region, the motion vector value corresponding to the motion direction and the distance thereof is detected. The motion vector value, which is the output signal of the motion detector 220, is input to the motion compensator 190 via the mode controller 230.
The motion compensator 190 moves the block area according to the motion vector value.

The motion-compensated signal is sent to the switch 21.
It is input to the subtractor 120 via 0, and the switch 200
And is input to the adder 170. The subtracter 120 subtracts the motion-compensated decoded image signal from the original image signal. Therefore, the output signal of the subtractor 120 is
The difference signal is obtained only for the area corresponding to the block for which the motion compensation is insufficient, and the output is zero for most areas. On the other hand, the adder 170
In the above, the reverse processing is performed, and the difference signal is added to the motion compensation signal, so that the output of the adder 170 is a decoded image signal close to the original image. Further, since it is expected that the difference signal itself of the subtractor 120 also includes spatial redundancy, DCT processing and quantization similar to those in the intra coding mode are performed.

As described above, when the switches 200 and 210 are closed, the correlation with the past image signal is utilized, and the temporal redundancy can be reduced. This is an encoding mode. To tell.

As a moving picture coding method based on the above-described hybrid coding method, MPEG (Moving Pictu) is used.
re Image Coding Experts Group) system has been proposed, and this MPEG system will be described.

In the hybrid coding system shown in FIG. 8, the inter coding mode is coded by utilizing the correlation with the past image. That is, information that has been compression-encoded once is decoded, a motion vector is detected between the decoded image and the input image, and motion compensation of the decoded image is performed based on this motion vector to create a predicted image signal.
Then, the compression is performed by obtaining the difference between the input image signal and the predicted image signal. On the other hand, in MPEG, the correlation with future images is also used. Therefore, in MPEG, an intra coding mode that uses only spatial redundancy within an image (this is I-Picture) and an inter coding mode that also uses temporal redundancy with past images (this is called P -Picture Predictive
-Picture) and the temporal redundancy between the past and future images are also used together (this is BP
image Bidirectionally-Pr
There is editive).

FIG. 9A shows an example of the order of arrangement of each picture performed by the MPEG system. Ixx is I-Picture, Pxx is P-Picture, B
xx represents B-Picture, and in this example,
Two B's are inserted between I or P pictures, and an I picture is inserted every 10 (frames).

Since the B picture is used for reducing the temporal redundancy with the future image, the input order of the original image and the order of the encoded image are different. That is, FIG.
As shown in (B), for B2 and B3, a temporal redundancy search using I1 and P4 is performed. (B2 and B
(3, I1 and P4 are called past and future reference images). Therefore, it is necessary to encode P4 prior to encoding B2 and B3. As shown in FIG. 9A, the encoding order is different from the original image order, that is, I1 and P4 are encoded. It is done prior to the encoding of B2 and B3.

In FIGS. 9A and 9B, P
Or in the B-picture, for the entire image area,
Although it is described that motion compensation is applied, in practice, a macroblock consisting of two blocks in the line direction and two in the vertical direction, that is, a total of four macroblocks is defined, and an intra code is defined for each macroblock. The I-macro block, the P-macro block that refers to the past image, and the B-macro block that refers to the past and future images are determined.

Here, for I pictures, the use of P and B macroblocks is prohibited, and thus all of them are composed of I macroblocks. In addition, I and P macroblocks are used in P pictures, and I, P, and B macroblocks can be used in B pictures. For example, assuming that a moving object is moving in an image in a B picture, a region hidden by a moving object is referred to by a future image to become a B macroblock, and the moving object itself is changed from the past. Can be referred to, and it becomes a P macroblock.

Further, with respect to the macroblock including both the moving object and the background, the motion compensation cannot be sufficiently performed, so that the I macroblock is selected. The selection of such a macro block is achieved by the following device configuration.

In FIG. 8, the motion detector 220 has a motion vector from a past image (hereinafter referred to as a forward motion vector) and a motion vector from a future image (hereinafter referred to as a backward motion vector) for a macroblock of interest. ) 2
Detect two motion vectors. Next, the difference value between the original image and the decoded image when motion compensation by the forward motion vector is performed is detected. Similarly, the difference value between the original image and the decoded image when the motion compensation is performed only by the backward motion vector, and the difference between the original image and the decoded image when the motion compensation is performed by the forward and backward bidirectional motion vectors. Detect the difference value. Also, a signal corresponding to the difference value between the original image and the decoded image when no motion vector is used, that is, when only the original image is used, is created. These two types of motion vector values (forward direction, backward direction) and four types of difference values are input to the mode controller 230.

Since the motion vector value and the like cannot be used for the I picture in the mode controller 230, the motion compensation value = 0 is set and a signal for turning off the switches 200 and 210 is output. For P pictures, for each macroblock, the difference value based on the forward motion vector is compared with the difference value signal when the motion vector is not used, the smaller one is selected, and the selected motion vector value Is output to the motion compensator 190. Similarly, for the B picture, all four types of difference values are compared, and the motion vector value having the smallest difference value is output.

With the above arrangement, macroblocks that can be used for each picture can be combined to select a macroblock having the smallest difference value. That is, it means that the macro block that is expected to have the highest compression efficiency is adaptively selected.

In the compression / encoding of a moving image signal according to the MPEG system, there is a hierarchy called GOP (Group of pictures) as a unit of the compression encoded signal. There is always one or more I pictures in a GOP. I-pictures are coded pictures that do not require reference from other pictures, while P- or B-pictures require reference pictures. In other words, it is necessary to start decoding from the I picture. (Even if decoding is started from a P or B picture, a sufficient reproduced image cannot be obtained because it is a differential image signal).

By providing the hierarchy of GOP in this way, there is a merit that it is easy for the decoder to find a sequence capable of starting decoding. In the encoder, the GOP is necessary as a unit for editing the code signal. At least one I picture is present in the code of the group by cutting out, adding or deleting the code signal in units of GOP, and at least other images are I pictures.
This is because it is created based on the picture.

However, in order to actually edit the code signal, it may be necessary to perform the editing in units of a plurality of GOPs. For example, for the first B picture B8 and B9 of GOP1, the last P picture P7 of the previous GOP.
May also be used as a reference image (FIG. 9).
(See (B)). A GOP that creates a reference image in such a manner that it crosses the boundary between GOPs is called an OPEN GOP.
On the other hand, the case where P7 is not referred to for B8 and B9 is called a CLOSED GOP. Generally, O
Since PEN GOP has many reference image selection branches, high compression efficiency can be expected.

However, when the encoding is performed by the OPEN GOP, it is assumed that the reference of P7 is permitted in the processing of B8 and B9, and the encoding of B17 and B18 of GOP2 is also permitted. Considering an operation of re-encoding GOP1 for this code signal and replacing the re-encoded signal with the original GOP1,
The following procedure is required.

First, encoding is started from GOP0 and I1
From P to P7, the encoding is performed in exactly the same state as the original encoding (this is an essential operation for GOP0, since the code signals are not exchanged). As a result, the last P7 of GOP0 can be reproduced. This P7 is used to start encoding GOP1,
Encode GOP1. Finally, GOP2 is encoded so that only B17 and B18 referred to by P16 can be changed.

By the above operation, GOP1 is re-encoded and the encoded signals can be exchanged (to be precise, it is necessary to exchange B8 included in GOP1 to B18 included in GOP2). Considering the coding of GOP0 here, for I1 to P7, a signal exactly the same as the original coded signal must be created again. When considering this feasibility by limiting it to the encoding device, the quantization parameter in the quantizer used for encoding, the motion vector value used in each macroblock, etc. are set at the time of the previous encoding. Must be identical to the one used. This is possible by storing all these signals in the previous encoding. In general, the hybrid encoder is realized by a digital circuit, so that all the signals necessary for reproducing the previous encoding are stored, and at the time of re-encoding, the signals may be encoded. This is in principle a possible operation, as long as there is no equipment failure. However, the condition that the original image signal input under this assumption is the same as the image signal in the previous encoding is further required. Even if the signals such as parameters required for encoding are the same, if the original image signal is different by encoding twice, P7
It is impossible to reproduce.

Generally, the amount of information of the original image signal before being compressed / encoded is enormous. For example, the image signal used in the current television signal or the like constitutes one frame with 720 horizontal samples and 480 vertical lines, and the frame rate is 1/30 second. The sample value of 1 pixel is 8 bits
The amount of information per second in the case of is about 83 Mbits. This is only for one type of signal, and in the case of a color image, three types of signals are required, and the amount of information varies depending on the format, but for example, the signal bandwidth of the color signal component is reduced to half that of the luminance signal. The amount of information in a so-called 4: 2: 2 format color image signal is about 16
It becomes 0MB / S. As a device that is currently realized as a device that can record and store a huge amount of information for a long time,
There is a digital VTR (VCR) using a magnetic tape. However, the reproduction of these magnetically recorded signals usually contains errors. (Of course, a noise-free recording / storing device such as a semiconductor memory is also possible, but it is unrealistic to handle a long-time original moving image signal).

Therefore, the assumption that the original images are exactly the same cannot be established by the above-described two-time encoding. That is, it becomes impossible to realize an editing operation of re-encoding a signal encoded by OPEN-GOP and replacing a part thereof.

Here, the necessity of the editing operation for exchanging a part as described above will be mentioned. In the hybrid coding method, there is a restriction on the information amount (generated code amount) of the coded signal. In this case, the signal bandwidth of the channel for transmitting or accumulating those compression codes is determined, and therefore, it is necessary to set the amount of generated codes to a band that matches this.

On the other hand, the complexity of the original image signal changes from moment to moment, and as a method of controlling the generated code amount against this change, the discretization step size (quantization size) in the quantizer 140 or For example, the coding algorithm selection algorithm in the mode controller 230 may be changed.

Regarding the control method of these parameters, there is no generalized algorithm, and the quality of the decoded image signal is greatly influenced by this control method (for example, considering the quantization processing). In this case, it can be understood that this irreversible processing determines the degree of distance between the original image and the decoded image). Therefore, at the present time, it is necessary to change the above parameters for the original image signal, perform encoding a plurality of times, and select the one having the best decoded image quality.

[0031]

In the hybrid coding system found in the conventional example, it is not possible to perform an editing operation for replacing a part of the compressed code with a re-encoded one. Of course, all GOPs are CLOSED-GOP
However, it is not possible to expect sufficient compression efficiency in this mode. SUMMARY OF THE INVENTION An object of the present invention is to provide a moving picture coding system which enables the above editing operation while keeping the compression efficiency relatively high.

[0032]

In order to achieve the above object, the present invention realizes a hybrid coder which periodically generates CLOSED-GOP, and a coding mode control unit is provided with an external or external device. The GOP selection signal instructed from the inside causes that GOP to be CLOSED-
The coding mode is selected so that GOP is set.

[0033]

By the above means, it is possible to easily replace a part of each GOP and obtain a signal convenient for editing.

[0034]

Embodiments of the present invention will be described below with reference to the drawings. FIG. 1 shows a first embodiment of the present invention. Units having the same functions as those of the hybrid encoding device in the conventional configuration are designated by the same reference numerals, and the description of the functions will be omitted.

In FIG. 1, the GOP is different from the conventional configuration.
A measuring instrument 250 is added. GOP measuring instrument 250
Receives a signal indicating the boundary of GOP from the mode controller 230 via the signal line 25b. GOP by this signal
GO from the initial value (= a) inside the measuring instrument 250
When the boundary of P is counted down and the number of boundaries is input a times, the count value becomes zero, and the CLOSED-GOP instruction signal is input to the mode controller 230 via the signal line 250a.

In the mode controller 230, the CLOS
When the ED-GOP instruction signal is received, the backward motion vector selection is prohibited for the two B pictures at the head of the GOP so that the reference picture is limited to the I picture that comes next.
By this operation, CLOSE without reference from the previous GOP
A D-GOP will be formed.

FIG. 7 shows GOP1 as CLOSED-GOP.
3 shows a relationship diagram of the reference image shown in FIG. As described in the conventional example, the compression efficiency is expected to decrease due to the selection of CLOSED-GOP. Therefore, the initial value a given to the GOP measuring device 250 is set relatively large. Usually 1G
The number of pictures forming the OP is set to about 10 to 30 frames. This is determined by the role of the GOP decoding device shown in the conventional example. That is,
In the decoding device, when the signal to be decoded is switched by turning on the power or changing the channel, GOP is used as a unit for reproducing a normal image. Therefore, if the number of pictures making up one GOP is increased, it takes a long time to obtain a normal image. On the other hand, if the number of 1-GOP constituent pictures is reduced, the time required to obtain a normal image will be shortened, but the frequency of occurrence of I-pictures that do not require other reference images will increase.

Since the hybrid encoding apparatus uses the temporal redundancy of the moving image signal to improve the compression efficiency, the increase in the I picture frequency means the decrease in efficiency.

From the above viewpoint, the number of pictures constituting one GOP is 10 to 30 frames, that is, I to some extent.
The time is set so that the frequency of pictures is reduced and a reproduced image can be obtained without a feeling of strangeness.

The reproduction time of 1 GOP depending on the number of pictures is about 1/3 second to 1 second. The unit of creating the moving image is created based on the concept of a scene. That is, one scene is composed of a series of related images, and one program is created by combining these scenes. Here, the length of one scene is determined by the program creator, but from experience, it is generally about 30 seconds. Further, it is expected that the complexity of the image will change depending on the scene. Therefore, it is appropriate that the editing operation such as replacement of the compressed code signal by re-encoding also takes this scene as a unit.

From the above, the set value of the initial value a in the GOP measuring device 250 is once for about 30 seconds CLOSED-
The value is such that GOP occurs. 1 GOP = 30
Picture = 1 second, a = 30
CLOSED-GOP occurs at a rate of once per second.

A series of GOPs delimited by CLOSED-GOP is referred to from the previous GOP, and a GO which will be described later.
Since the reference from P is not received, there is no restriction that the same original image signal has to be supplied in the two encodings, which is seen in the conventional example, and therefore the editing operation can be performed. In the above operation, CLOSED
-GOP is set accurately for each a, but this may be set for each substantially regular period (approximately a).

FIG. 2 shows a second embodiment of the present invention.
A reference macroblock measuring instrument 260 is added to the first embodiment shown in FIG.

Here, the mode controller 230 will be described again. First, the motion detector 220 detects two motions of a macroblock of interest: a motion vector from a past image (hereinafter referred to as a forward motion vector) and a motion vector from a future image (hereinafter referred to as a backward motion vector). Detect a vector. Next, the difference value between the original image and the decoded image when motion compensation by the forward motion vector is performed is detected. Similarly, the difference value between the original image and the decoded image when the motion compensation is performed only by the backward motion vector, and the difference between the original image and the decoded image when the motion compensation is performed by the forward and backward bidirectional motion vectors. Detect the difference value.
Also, a signal corresponding to the difference value between the original image and the decoded image when no motion vector is used, that is, when only the original image is used, is created. These two types of motion vector values (forward direction, backward direction) and four types of difference values are used as the mode controller 23.
Input to 0.

Since the motion vector value and the like cannot be used for the I picture in the mode controller 230, the motion compensation value is set to 0 and a signal for turning off the switches 200 and 210 is output. For P pictures, for each macroblock, the difference value based on the forward motion vector is compared with the difference value signal when the motion vector is not used, the smaller one is selected, and the selected motion vector value Is output to the motion compensator 190. Similarly, for the B picture, all four types of difference values are compared, and the motion vector value having the smallest difference value is output.

As a result, macroblocks that can be used for each picture unit can be combined to select a macroblock having the smallest difference value. That is, it is possible to adaptively select a macro block that is expected to have the highest compression efficiency.

For the reference macroblock measuring device 260, the signal indicating the GOP boundary from the mode controller 230, the difference value after the above-described motion compensation and the motion vector used for it are given to each macroblock. Signal line 26
0a. In the reference macroblock measuring device 260, the two B pictures at the head of the GOP are sequentially processed.
Square measures the number of macro blocks which adopted countercurrent motion vector. The measurement is performed for each image. The small number of macroblocks that use the forward motion vector means low correlation with the last image of the previous GOP.

[0048] Thus, supplying a signal for inhibiting the selection of the forward Direction motion vectors in the case of small consisting setting values that the number of macro blocks (= b) through the signal line 260b to the mode controller 230. That is, the adoption of the motion vector detected between the past reference image and the decoded image is stopped. In other words, the reference of the past reference image is stopped and the motion compensation is performed by another reference image.

By performing the above macroblock measurement, the method has a function of selecting CLOSED-GOP for a GOP boundary in which the selection of the forward motion vector is small to some extent. Therefore, the periodic CLOSED-GO shown in FIG. 1 is used.
It can be expected that the degree of reduction in compression efficiency will be less than the insertion of P. Of course, the forward motion vector prohibition signal is the signal line 2
It is also input to the GOP measuring device 250 via 60b at the same time, and the GOP measuring device 250 resets the counter to the initial value a by the input of this signal.

Therefore, the reference macroblock measuring device 260
If the time until the forward motion vector prohibition signal is obtained and the time until the next forward motion vector prohibition signal is obtained is longer than the time corresponding to the initial setting value a, CL is automatically set.
OSED-GOP will be inserted.

FIG. 3 further shows a third embodiment of the present invention. A scene change detector 270 is added to the embodiment of FIG. As described above, the length of one scene is empirically about 30 seconds, but this length varies depending on the original image signal. Normally, the image content changes significantly when the scene changes. Therefore, the scene change detector 270 can detect a scene change by inputting the input image and the signal obtained by delaying the input image by one frame (or one field) by the delay device 271 to the subtractor 272 to perform subtraction processing. . The output of the subtractor 272 is input to the averaging unit 273, and the averaging unit 273 measures the length of the detected scene change and outputs the average value. Scene change detector 270
The average scene length (= c) of the input image detected by is input to the GOP measuring device 250. By replacing this c with the initial setting value a, the optimum initial value is set according to the image signal. be able to.

FIG. 4 shows a fourth embodiment of the present invention. Compared to the embodiment of FIG. 1, the activity adaptive initial value unit 280
Has been added. As described in the conventional example, the preprocessing circuit 1
10 has an activity detector 112 for detecting the complexity of the image signal as a parameter for determining the quantization step size in the quantizer 140. The detected activity indicates that the degree of complexity of the image signal is large. Since the generated code amount is originally limited, the quantizer 140 performs an operation to increase the quantization step size as the activity increases, that is, to increase the degree of discretization and reduce the generated code amount.

Now, the activity adaptive initial value unit 280
Then, first, the activity detected for each screen is averaged by the averaging device 281, and the average value is non-linearly converted by the non-linear converting device 282, and the initial value a of the GOP measuring device 250 is changed to d. An example of the conversion characteristic of the non-linear converter 282 is shown in FIG. This provides the function of setting the GOP measurement initial value to a large value when the averaging activity is small, that is, the degree of image complexity is small. This is because originally when the activity is low, the coding is relatively easy and therefore the requirement for re-coding is expected to be low. The initial value of GOP measurement decreases as the activity increases. This is because it is expected that re-encoding will be performed due to an increase in activity, and the frequency of requests for exchanging compression codes will increase. If the activity further increases, the initial value setting value will increase. This is because, due to the increase in activity, the degree of difficulty in encoding is high, and the decrease in compression efficiency due to CLOSED-GOP is prevented as much as possible. It is expected that the demand for re-encoding will be the highest in this area, but since encoding difficulty is originally high, CLOSED should be given priority to compression efficiency.
-Adopts an algorithm that inserts a GOP. This means that the unit of re-encoding becomes long, but in the region where the coding difficulty is high, the re-encoding unit is lengthened to increase the changeable code amount and increase the degree of freedom of parameter change. be able to.

FIG. 5 shows a fifth embodiment of the present invention. FIG.
The activity converter 290 is added to the embodiment. As described above, CLOSED-GOP is expected to reduce the compression efficiency. In this embodiment, the GOP measuring device 25
From 0, the value of the GOP measurement counter is input to the activity measuring device 290. For example, as shown by the characteristics in FIG. 6B, the activity value is changed according to the measured GOP value and input to the quantizer 140. I have to.

For example, as shown in (a) of FIG.
When the OP measurement value is large, that is, CLOSED-GOP
When the elapsed time from the occurrence of is short, the conversion from the input activity to the output activity is performed small, that is, the quantization step size in the quantizer 140 is reduced, the generated code amount is increased, and the count value is decreased. , The control is performed such that the amount of generated code is reduced by increasing the conversion of the input-to-output activity.

According to such an operation, the CLOSED-G
In order to compensate for the minimal reduction in compression efficiency that is expected to occur due to OP, the generated code amount is suppressed in advance since the occurrence of CLOSED-GOP, and the suppressed code amount is distributed to a region corresponding to CLOSED-GOP. As a result, it is possible to reduce deterioration in image quality due to a decrease in compression efficiency. That is, the decrease in compression efficiency due to CLOSED-GOP is compensated by suppressing the generated code amount in advance.

As the conversion characteristic, (b) in FIG. 6 (B)
As described above, the effect can be expected even if the operation of suppressing the generated code amount is started when CLOSED-GOP approaches.
As described above, there are various application examples as long as the compression code can be edited by inserting the CLOSED-GOP at least periodically, without departing from the gist of the present invention.

Header information is added to the head of the GOP, and CLOSED- is included in this header information.
Identification information of GOP and OPEN-GOP is inserted. Therefore, in the editing process, the GOP may be extracted by referring to the identification information.

[0059]

As described above, according to the present invention, it is possible to facilitate the editing operation such as the replacement of the compression code even if the input image signals are different in a plurality of encodings.

[Brief description of drawings]

FIG. 1 is a diagram showing a first embodiment of the present invention.

FIG. 2 is a diagram showing a second embodiment of the present invention.

FIG. 3 is a diagram showing a third embodiment of the present invention.

FIG. 4 is a diagram showing a fourth embodiment of the present invention.

FIG. 5 is a diagram showing a fifth embodiment of the present invention.

FIG. 6 is a diagram showing an example of activity conversion characteristics used in an embodiment of the present invention.

FIG. 7 is a diagram shown for explaining an operation example of the apparatus of the present invention.

FIG. 8 is a diagram showing a basic configuration of a hybrid encoding device.

FIG. 9 is a diagram showing an example of an image processing procedure in the MPEG system.

[Description of Reference Signs] 110 ... Preprocessing circuit, 120 ... Subtractor, 130 ... DCT
, 140 ... Quantizer, 150 ... Inverse quantizer, 160 ...
DCT- ¹ device, 170 ... Adder, 180 ... Delay device, 190
... motion compensator, 200, 210 ... switch, 220 ... motion detector, 230 ... mode controller, 240 ... encoder,
250 ... GOP measuring device, 260 ... Reference macroblock measuring device, 270 ... Scene change detector, 280 ... Activity adaptive initial value device, 290 ... Activity converter.

─────────────────────────────────────────────────── ─── Continuation of the front page (51) Int.Cl. ⁶ Identification code Internal reference number FI Technical indication H04N 7/133 Z

Claims (6)

[Claims] 1. An input image signal is created by replacing image signals that are sequentially arranged in time in the original image sequence at a fixed rate during coding and compression processing, and uses past and future reference image signals. An image motion vector is detected, a predicted image signal is created from a decoded image signal of an image that has been subjected to compression processing based on the detected motion vector, and a difference signal between the input image signal and the predicted image signal is obtained. In the image encoding method for performing the compression process according to, the moving image encoding method is characterized in that the use of the above-mentioned past reference image signal is temporarily prohibited at almost regular intervals (approximately a). 2. A method of performing compression processing by obtaining a difference signal between the input image signal and the predicted image signal is a moving image in which one image is divided into a plurality of small areas and the compression processing is performed for each area. It is an image coding method, has a means for detecting the number of areas using a reference image from the past image signal as a reference image of the small area, and if the detection result is smaller than a natural number x, The moving picture coding system according to claim 1, further comprising means for prohibiting the use of the predicted image signal based on the reference image signal. 3. A means for measuring a scene change by calculating a difference between an original image signal and a past image signal, and calculating an average value of the time intervals of the scene change. The moving picture coding system according to the first aspect, characterized in that the moving picture coding system has a setting means for setting as a). 4. The setting means for setting the regular period (substantially a) has an activity detecting means for detecting the complexity of an image based on image components and the like, and an activity averaging means for the output of the detecting means. 2. The moving picture coding method according to claim 1, wherein the periodical period (substantially a) is calculated and set according to the averaged degree of activity. 5. The means for calculating the periodical period (substantially a) from the averaged activity calculates a large value in a small area of the averaged activity, and the averaged activity is calculated. With the increase of
5. The moving picture coding method according to claim 4, wherein the value of a is increased again when the averaged activity is increased. 6. A path for controlling the amount of generated code generated according to the information after compression according to the detection result of the activity detecting means, and the path is controlled prior to the prohibition at each regular period. The moving picture coding system according to claim 4, further comprising means for converting the activity detection result so as to reduce the generated code amount in advance.