JP2002171488A - Image signal coding method and image signal recorder - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a VTR that imbeds independent image information for high-speed reproduction and can update a picture at a high-speed even in the case of high-speed reproduction and to provide its coding method. SOLUTION: Variable length codes of AC components among variable length codes of high-speed reproduction purpose image data encoded by predetermined variable length codes are replaced with new variable length codes with a smaller data quantity. Thus, the data quantity per one high-speed reproduction purpose image can be reduced so as to increase the number of images to be reproduced at high-speed reproduction.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image signal encoding method for converting a variable length code of a digitally compressed image signal, and an image signal recording apparatus for recording a digitally compressed image signal on a magnetic tape. The present invention relates to a configuration for suitably recording image data for high-speed reproduction independent of normal reproduction data.

[0002]

2. Description of the Related Art Digital broadcasting represented by digital satellite broadcasting is becoming widespread. Also, a VTR that records and reproduces these digital broadcasts as digital signals has been proposed.

In digital broadcasting, image information is compressed and transmitted using digital image compression technology in order to effectively use the band of radio waves. Digital image compression techniques include MPEG (Moving Picture Exp).
ert Group). In these digital image compressions, the data amount is reduced by utilizing the correlation between the images of the preceding and succeeding frames.

In a VTR for recording a digitally compressed image, it is difficult to perform high-speed reproduction such as fast forward and rewind.
This is because the digital image compression utilizes the correlation between frames, so that an original image cannot be restored only by a signal partially reproduced from the magnetic tape during high-speed reproduction.

In order to solve this problem,
Japanese Patent Laid-Open No. 8-46 discloses a VTR that records only an intra-frame coded image that does not use frame correlation at a predetermined position on a magnetic tape scanned by a rotating magnetic head during high-speed reproduction.
913 and JP-A-8-98140. This makes it possible to reproduce the data of the intra-frame coded image from the magnetic tape and restore the image during high-speed reproduction.

A VTR for compressing data at a higher compression ratio and recording at a predetermined position on a magnetic tape separately from a normal digitally compressed image is disclosed in Japanese Patent Laid-Open No. Hei 5-2764.
No. 91 publication. In this VTR, an input video signal is converted into a digital compressed signal and recorded. In addition to normal compression, data for high-speed reproduction is generated by a high compression ratio circuit that compresses data at a higher compression ratio. And it is arranged at a predetermined position on the magnetic tape. This ensures that the image is correctly reproduced even during high-speed reproduction.

[0007]

The above-mentioned VTR has the following problems. JP-A-5-2765
The VTR described in Japanese Patent Publication No. 91 has a high compression ratio circuit for performing high compression separately from a compression circuit used for normal reproduction. However, for example, a digital video signal transmitted by digital broadcasting is a signal that has already been compressed. The compression ratio is determined by the broadcasting station,
The TR cannot control the compression ratio. Further, when re-compressing, it is necessary to return the compressed signal to an uncompressed video signal once by a decoder, or to generate high-compression data by a high-compression-ratio circuit. That is, in order to record a digitally compressed signal that has been compressed in advance, such as a digital broadcast, the VTR requires a built-in decoder for decoding the compressed signal. Therefore, it is not suitable for a VTR that records a signal that has been compressed in advance, such as digital broadcasting.

In the VTRs of Japanese Patent Application Laid-Open Nos. 8-46913 and 8-98140, only intra-frame coded images are extracted from an input stream.
Recorded as high-speed playback data. However, the data amount of the intra-frame coded image is determined when image compression is performed in the broadcasting station, and cannot be controlled on the VTR side.

Generally, an intra-coded image tends to have a larger data amount than an inter-coded image. Since the amount of data that can be reproduced during high-speed reproduction is very small, if the intra-frame encoded image is used as it is as data for high-speed reproduction, the number of images reproduced during high-speed reproduction will be very limited. There is a disadvantage that the speed of screen updating during high-speed playback is reduced.

[0010] An object of the present invention is to provide a VTR in which independent image information is embedded for high-speed reproduction, which can perform high-speed screen updating even when high-speed reproduction is performed, and an encoding method therefor. Is to do.

[0011]

An image signal encoding method according to the present invention decomposes an image signal into a plurality of spatial frequency components by a discrete cosine transform, and the spatial frequency components are predetermined for each frequency. Quantize using the quantization coefficient,
Each of the quantized spatial frequency components is converted to image data encoded with a predetermined variable length code using a new variable length code. Therefore, of the predetermined variable length codes, the variable length code of the AC component is converted into a new variable length code having a smaller data amount, and the variable length code of the image data is converted by the new variable length code. The replacement reduces the amount of image data.

Further, the image signal recording apparatus of the present invention decomposes an image signal into a plurality of spatial frequency components by discrete cosine transform, and converts each spatial frequency component using a predetermined quantization coefficient for each frequency. Image data extraction for quantizing, inputting image data obtained by encoding each of the quantized spatial frequency components with a predetermined variable length code, and extracting high-speed reproduction data from the input image data Means for converting the variable length code of the AC component among the predetermined variable length codes into a new variable length code having a smaller data amount, and converting the variable length code of the high-speed reproduction data into the new variable length code. Image data encoding means for reducing the data amount of the high-speed reproduction data by replacing with a long code; and a predetermined position on the magnetic tape to be scanned by the rotating magnetic head during high-speed reproduction. A structure having an image data recording means for recording by placing the reduced for high-speed playback data has been in the data volume.

[0013]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, a recording apparatus using a data reduction method according to the present invention will be described by taking a digital VTR as an example. The image compression method will be described as being based on the MPEG method.

FIG. 1 shows a digital VTR according to the present invention.
It is a block diagram of. 101 is an input terminal to which recording data is input, 102 is an image extraction circuit for extracting an in-frame compressed image from the recording data, 103 is an image memory for storing an image, and 104 is a packet detection circuit for detecting a packet having a specified PID 105, a CPU for processing image data, 106, a ROM for storing a CPU program and a variable length code (VLC) conversion table, 107, a data arrangement circuit for arranging data in a track, 108, a recording signal processing circuit, 109 Is a rotating drum, and 110 is a rotating magnetic head. Recording data is input to 101 from an external device such as a digital broadcast tuner in a packet format. This input packet includes video data, audio data, time information, and the like.

As described above, in order to perform recording corresponding to high-speed reproduction, it is necessary to extract a compressed image (I picture) within a frame from input data and record it at a predetermined position on a magnetic tape. There is. First of all,
An intra-frame compressed image is extracted from the input data.

FIG. 2 shows the format of a packet input from the input terminal 101. 201 is a synchronization byte, 202 is P
An ID 203 indicates a data area. Each packet is 188 bytes of data starting with a synchronization byte 201. The sync byte has a value of 47 in hexadecimal. In the packet,
PID20 for identifying the data attribute of the packet
2 is included. Various types of data such as video, audio, and time information are mixed in the input data. In order to identify which data each packet contains, a PID is added to each data. By identifying the PID in the packet, each packet can be separated.

Which PID each data corresponds to is determined by a program association table (PA).
T) and the respective packets of the program map table (PMT). To extract only video information from an input packet, PAT and PM
Analyze T and check the PID of the packet to which the video data is sent. For this purpose, first, the CPU reads the contents of the PAT using the packet detection circuit.

FIG. 3 shows a block diagram of the packet detection circuit. 301 is an input terminal to which a packet is input, 302 is a PID identification circuit for identifying the PID of the input packet, 30
3 is a memory for storing input packets, 304 is a CPU
This is a communication terminal for communicating with the device. Packet detection circuit 10
No. 4 compares the PID specified by the CPU 105 with the PID of the input packet, and stores the contents of the packet in the memory 303 only when they match. If the PIDs do not match, the contents of the packet are not written to memory. The PID matches, and the packet contents are stored in the memory 303.
Is notified to the CPU 105 via the communication terminal 304 that the packet has been detected.

First, the CPU 105 causes the packet detection circuit 104 to detect PAT. Since the PID of the PAT is defined as 0, the PI
Instructs to detect a packet with D = 0. The packet detection circuit 104 waits for a packet having a PID of 0 to be input, and when the packet is input, fetches the content and stores it in the memory 303. If the PAT packet is detected by the packet detection circuit 104, the CP
U105 reads the contents from the memory 303 and analyzes them to obtain the PID of the PMT.

Subsequently, the CPU 105 gives the PID of the PMT obtained by analyzing the PAT to the packet detection circuit 104, and instructs to detect a PMT packet. The packet detection circuit 104 waits until a packet corresponding to the PMT is input, and stores the contents in the memory 303. When the PMT is detected, the CPU 105 analyzes the contents of the PMT and obtains the PID of the video packet. Through the above processing, the PID corresponding to the video data packet is determined. The image extraction circuit 102 extracts the intra-frame compressed image from the obtained PID of the video packet and the start and end patterns of the intra-frame compressed image.

FIG. 4 shows the structure of image data. Image data is compressed in a unit called GOP (Group of Pictures). The GOP always includes an intra-frame compressed image (I picture), and a forward prediction coded image (P picture) based on this image.
And a bidirectional predictive coded image (B picture) are generated. At the head of the GOP is a GOP header, which indicates the start position of the GOP. Following the GOP header, picture data for 15 frames follows. Immediately before a GOP, a sequence header is usually added. The sequence header includes basic information of the image such as the number of pixels of the image and the frame frequency, and is used for image restoration in the GOP.

Each picture has a picture start code (PSC) indicating the start of the picture, a temporal reference (TR) indicating the picture display order, and a picture coding type (P) indicating the picture structure.
CT), followed by picture data.

Here, the picture start code is 3
It has a length of 2 bits and is a value of 00000100 in hexadecimal. The temporal reference has a length of 10 bits and indicates the position of each picture in the GOP. Picture coding type is 3 bits, compressed image in frame (I picture)
1, 2 for a forward prediction coded image (P picture), and 3 for a bidirectional prediction coded image (B picture).

FIG. 5 shows a block diagram of the image extracting circuit. Reference numeral 501 denotes an input terminal to which a packet is input, 502 denotes a PID filter that passes only packets of a specified PID, 503 denotes a communication terminal for performing settings from the CPU to the PID filter and the memory control circuit, and 504 denotes a PID.
A buffer memory for temporarily storing the packets that have passed through the filter; 505, a start code detection circuit for detecting whether a start code of an intra-frame compressed image is present in an input packet; 506, an end code of an intra-frame compressed image in an input packet An end code detection circuit for detecting whether or not there is, a memory control circuit 507 for controlling data input / output of the buffer memory, and an output terminal 508 for outputting data to the image memory.

When performing image extraction, the PID of the video packet is
Is set to The PID filter 502 allows only a packet having a designated PID to pass from among the input packets, and writes the packet into the buffer memory 504. Packets having a PID different from the video packet do not include video information, and are therefore removed by the PID filter 502.

The start code detection circuit 505 constantly monitors whether or not the input packet has the start code of the intra-frame compressed image. Similarly, whether or not there is an end code of the in-frame compressed image is constantly monitored by the end code detection circuit 506. Here, the start code of the intra-frame compressed image may be the sequence start code (000001B6), and the end code of the intra-frame compressed image may be the start code of the bidirectional predictive coded image (B picture).

The memory control circuit 507 includes the CPU 105
When the extraction of the compressed image in the frame is instructed, the process waits until the start code detection circuit 505 detects the start code of the compressed image in the frame. Packets input before the start code is detected are not output to the image memory 103 but are discarded from the buffer memory 504.

When the start code detection circuit 505 detects the start code of an intra-frame compressed image, it means that the intra-frame compressed image is included after that packet. The memory control circuit 507 controls reading from the buffer memory 504 and writing to the image memory 103 via the output terminal 508.

Until the end code detection circuit 506 detects the end code of the intra-frame compressed image, the packet data written in the buffer memory 504 is read out by the memory control circuit 507 and output to the output terminal 508.
Are sequentially output to the image memory 103 via the.

When the end code detecting circuit 506 detects the end code of the intra-frame compressed image, the end of the intra-frame compressed image is detected at that position. The reading from the 504 is stopped, and the CPU 105 is notified that the detection of the compressed image in the frame is completed.

With the above processing, the intra-frame compressed image is stored in the image memory 103. When the image extraction circuit 102 notifies the CPU 105 that the in-frame compressed image has been stored in the image memory 103, the CPU 1
05 starts the data amount reduction processing of the image data.

First, the CPU 105 determines from the packet
Extract picture data only. Image memory 103
Is stored in a packet format as shown in FIG. 2, and only the data area 203 effective as image data is extracted from the packet. A data reduction process is performed on the extracted image data.

Here, the image compression algorithm of the MPEG system will be briefly described. In the MPEG system, a compressed image in a frame is divided into macroblocks composed of 16 × 16 pixels. As shown in FIG. 6, the data in each macroblock includes luminance (Y) and two color differences (Cb, Cb).
r) and each is processed independently. The luminance information is divided into four blocks of 8 × 8 pixels. The color difference is obtained by performing sub-sampling and thinning out 16 × 16 pixels to 8 × 8 pixels.

The data of each block is decomposed into 8 × 8 spatial frequency components by a discrete cosine transform (DCT). The spatial frequency component includes an AC component (AC component) and a DC component (DC component) having no AC component. A
For the C component, each coefficient is encoded as it is, and for the DC component, the difference from the DC component of the previous block is encoded. Thereby, the coefficient value of the DC component can be significantly reduced. In addition, in the DC component and the AC component,
The assignment of variable length codes is also different. The AC component is characterized in that the intensity decreases as it goes to a higher frequency range, and the AC component is less visually noticeable.

The spatial frequency components decomposed by the DCT are each divided by a quantization table as shown in FIG. 7 and quantization is performed. At the time of quantization, a larger quantization coefficient is used for a higher frequency component. In other words, this means that the higher the frequency component, the coarser the quantization, and the lower the level of the higher frequency component, the lower the value becomes.

The quantized frequency components are arranged in order from a low-frequency component to a high-frequency component in the order shown in FIG. 8, and are encoded by a variable length code. Here, the appearance frequency of 0 tends to increase as going to the high frequency component. Therefore, the coefficient 0 is not coded independently,
Is used as the number of RUNs, and the coding is performed together with the coefficient value (level) of the coefficient continuing to the coefficient 0, thereby improving the coding efficiency. In addition, all subsequent high frequency components are 0.
, An EOB (end of block) code indicating that the valid data has ended is inserted, and subsequent 0s are omitted. Here, the EOB code is 1 in binary.
Set to 0.

FIG. 9 shows an example of a variable length code for encoding the RUN value and the coefficient value of the frequency component. Variable length code (VL
S in C) represents a sign bit, and 1 when the coefficient value is negative.
And For example, if the RUN value is 0 and the coefficient value is 3, the code is coded as 001010 in binary. Similarly,
If the RUN value is 0 and the coefficient value is -3, it is encoded as 001011.

The variable length codes are not set for all combinations of the number of RUNs and coefficient values that appear, but are limited to those with a high frequency of appearance. For infrequently occurring combinations, add an escape code followed by
By directly describing the number of RUNs and coefficient values, all combinations of coefficient values are supported. This suppresses an increase in the types of variable length codes.

As can be seen from the assignment of the variable length code (VLC), the shorter the AC component coefficient value, the shorter the code is assigned. That is, the smaller the coefficient value, the smaller the image data amount. In the data reduction method according to the present invention, all the coefficient values of the AC component are set to 1/2.
To reduce the amount of data.

Here, a method of converting a variable length code (VLC) will be described. FIG. 10 shows the structure of the image data. The image data is divided into 16 vertical pixels, each of which is a unit called a slice. At the beginning of the slice,
A slice start code is added so that the slice start position can be easily determined. Each slice is divided into 16 horizontal pixels to form a macroblock. A macro block is composed of 16 pixels in length and width.

As shown in FIG.
Luminance information (Y1 to Y4) and two color difference information (C
b, Cr). The 8 × 8 signal strength in each block is converted into spatial frequency strength by DCT, and quantization and coding with a variable length code are performed. Therefore, in order to convert the variable length code, the slice start code is searched, and for each macroblock in the slice, the change of the variable length code of the luminance information block (Y1 to Y4) and the color difference information block (Cb, Cr) is changed. Should be performed. The conversion of the variable length code is performed according to the variable length code conversion table in FIG.

FIG. 11 shows an example of a variable length code conversion table. The left side shows the coefficient before conversion and the coefficient value after conversion. On the right side, a variable length code (VLC) corresponding to each coefficient value is shown, and the state of conversion of the variable length code is expressed.

Although FIG. 11 shows only the coefficients having the number of RUNs of 0, the conversion tables corresponding to the coefficients of each AC component are similarly created for the variable length codes of the other coefficients. A conversion table is created so that each coefficient becomes 1/2. Here, if the odd coefficient is １ ／,
A fraction of 0.5 occurs, but the decimals are rounded off. That is, when the coefficient is 1, the value is 0.5, but the value after the decimal point is rounded up to 1. 1/2 when coefficient is 2
And set it to 1. In the following, conversion tables are prepared in advance so that the coefficients become 1/2 for all variable length codes. Using this table, variable-length codes are converted for all coefficients of the AC component in the block.

Since the combination of the number of RUNs and the coefficient value to which no variable length code is assigned is encoded by using an escape code, the coefficient value is extracted and converted. Of course, if there is a variable length code corresponding to the combination of the converted coefficient value and the number of RUNs,
The encoding may be performed using a variable length code. If EOB appears in the data in the block, the subsequent data is set to 0.
Therefore, the conversion process for the block is terminated,
EOB may be added.

FIG. 12 shows how the data in the block is converted. The DC component of the data in the block is not changed. All the AC components are converted according to the variable length code conversion table in FIG. Also, EOB
Is added as it is.

As described above, since the variable length code is assigned so that the data length becomes shorter as the coefficient becomes smaller, all the coefficient values are reduced to 、, so that almost all the variable length codes are , The data length can be shortened,
The data amount of the whole intra-frame compressed image is also small.

With the above processing, the conversion of the variable length code to the AC component is performed. By the way, the high-frequency AC component does not visually affect much even if its coefficient is changed. Therefore, it is possible to further reduce the data amount by performing the conversion of the variable length code and limiting the number of the variable length codes.

FIG. 13 shows an example in which the number of AC components is limited to three. The variable length code of the DC component in the data in the block is not changed. Only the first three AC components are converted according to the variable length code conversion table in FIG. All AC components after the fourth are deleted, and E
Add OB. As a result, it is possible to greatly reduce the data amount.

By the way, when the coefficient of the AC component is changed,
The DCT coefficient changes, and the original image changes. An actual DCT coefficient is obtained by multiplying a coefficient value encoded by a variable length code by a quantization coefficient represented by a quantization table. Therefore, when the coefficient value encoded by the variable length code is halved, if the value of the corresponding quantization table is doubled, the DCT coefficient does not change as a result. In the data compression method according to the present invention, the coefficient value of the AC component is
Therefore, the coefficient values of the corresponding quantization tables are all doubled.

FIG. 14 shows an example of conversion of coefficients in the quantization table. In the quantization table, 8 × 8 coefficients for the DCT spatial frequency are set. Here, the upper left quantization coefficient is a DC component that does not include an AC component. Since the coefficient is not changed for the DC component, the quantization coefficient is not changed. The coefficients are doubled for all other AC components.

Here, the quantization table is usually included in the sequence header. In the image extraction circuit, the sequence header is extracted up to the head of the bidirectional predictive coded image (B picture), so that the sequence header can be analyzed.

First, the sequence header is analyzed,
It is determined whether a quantization table is included. If the quantization table is included, all coefficients are doubled except for the coefficient corresponding to the DC component of the quantization coefficient. When the quantization table is not included in the sequence header, a predetermined default quantization table is used. Therefore, the coefficient corresponding to the DC component is determined based on the default quantization table defined by the MPEG standard. Is doubled and inserted into the sequence header. As a result, the coefficients of the quantization table are changed. As described above, by changing the variable length code C and changing the coefficients of the quantization table corresponding to the change, the data amount of the image data can be reduced without substantially deteriorating the image quality.

The intra-frame coded image whose data amount has been reduced as described above is converted into a packet structure as shown in FIG. 2 again, and sent to the data arrangement circuit 107 as a high-speed reproduction packet. At this time, the program association table and the program map table are also stored in the high-speed reproduction data area together with the packet of the intra-frame encoded image data whose data amount has been reduced so that the reproduced image is immediately reproduced at the time of high-speed reproduction. At the same time. Since the high-speed reproduction data does not include audio data or program information, the contents of the program map table need to be changed.

The data arrangement circuit 107 arranges the input packet and the packet for high-speed reproduction at predetermined packet positions. The recording position of the data for high-speed reproduction is a position corresponding to the position on the magnetic tape scanned by the rotating magnetic head during high-speed reproduction.

Here, the structure of the track on the magnetic tape will be described. FIG. 15 shows a track structure for one track. 1501 and 1509 are margins at both ends of the tape, 1502 is a preamble of the subcode, 1503
Is a subcode recording area, 1504 is a subcode postamble, 1505 is a gap, 1506 is a main data preamble, 1507 is a main data recording area,
Reference numeral 1508 denotes a postamble of main data.

Each track has a main data recording area 1
A subcode recording area 1503 is prepared along with 507. The subcode area 1503 is used for recording recording format information, program information, and the like. Preambles 1502 and 1 at the beginning of each recording area
506, followed by postambles 1504 and 150
8 is provided. By providing a gap 1505 between the subcode area 1503 and the main data area 1507, each area can be independently rewritten.

FIG. 16 shows the structure of a sync block in the main data area. 1601 is a synchronization byte, 1602 is I
D0, 1603 is ID1, 1604 is ID parity, 1
605 is main data and 1606 is internal parity. The main data recording area 1607 is composed of 336 sync blocks. Each sync block is 1
It is composed of 11 bytes of data.
2-byte synchronization byte 1601, 1-byte ID0 (1
602) and ID1 (1603), 1-byte ID parity 1604, 99-byte data area 1605, 8
This is the internal parity (C1 parity) 1606 of the byte.

The synchronization byte 1601 is used to determine a break between blocks, and the position of a sync block can be determined from ID0 and ID1. The ID parity 1604 can be used to confirm whether these IDs are correct. Since the sync byte 1601 and the ID (1602 and 1603) are written in each sync block, even when the rotating magnetic head reproduces over a plurality of tracks as in the case of high-speed reproduction, the sync block 1601 and the ID are written. Information can be reproduced correctly.

The main data 1605 is an area for recording packet data. 8 for each main data
Even if an error occurs during reproduction by adding the internal parity 1606 of the byte, error correction can be performed using these parities, and correct data can be obtained.

FIG. 17 shows details of the main data 1605 of the sync block. 1701 is the main header, 1
702 is additional data, 1703 is a packet header, 1
704 is a packet. The 188-byte MPEG packet is divided into two sync blocks and recorded as shown in FIG. Sync block main data 17
05 has a recording area of 99 bytes, of which the first 2 bytes are used as a main header 1701 indicating the content of the sync block. One byte is an area 1702 for recording additional data. In addition, the MPEG packet is added with 4-byte time information 1703 to be 192 bytes of data and divided into two sync blocks.

The main header 1701 defines the format of data recorded in the packet. For example,
The normal reproduction data may be set to 0, the high-speed reproduction data may be set to 2, and the packet in which nothing is recorded may be set to 1. In addition, information such as a code for identifying whether the packet is the first half or the second half of the data and a number of times the packet is used for high-speed reproduction are recorded.

During high-speed reproduction, the rotating magnetic head 110 scans across a plurality of tracks on the magnetic tape. An external parity (C2 parity) is added to the data for normal reproduction separately from the internal parity (C1 parity) of each sync block. This is so that error correction can be performed even when error correction cannot be performed on a sync block basis or when a sync block cannot be obtained.

Even at the time of high-speed reproduction, it is necessary to improve the error correction capability by using such an external parity. However, since normal external parity cannot be used for data reproduced during high-speed reproduction, a dedicated parity is added to high-speed reproduction data. Specifically, as shown in FIG. 18, ten external parities are added to a sync block of 102 words. The 112 high-speed playback sync blocks configured in this way are
It is arranged at the position 1902 where the rotating magnetic head scans during high-speed reproduction.

FIG. 19 shows the scanning position and data arrangement of the rotating magnetic head in 12-time speed reproduction. 19
01 is a scanning locus of the rotating magnetic head at the time of high-speed reproduction,
Reference numeral 1902 denotes data reproduced at the time of 12 × speed reproduction.
Normally, in a VTR, azimuth recording is performed. Therefore, in consideration of each of a positive azimuth head and a negative azimuth head, data reproduction is performed at a cycle of 24 tracks in 12 × speed reproduction. In order to reproduce data at the time of high-speed reproduction, it is necessary to arrange data in advance at these positions where the rotating magnetic head scans.

The data for high-speed reproduction is the data shown in FIG.
Since these blocks are composed of 12 sync blocks, 14 blocks are arranged in each area 1902.
At this time, high-speed reproduction data may be repeatedly recorded a plurality of times in consideration of track bending and tracking deviation in each VTR. On the other hand, normal recording data is recorded so as to avoid the recording position 1902 of the high-speed reproduction data. Through the above processing, high-speed reproduction data can be embedded at a predetermined position.

Next, recording signal processing is performed to record these data on a magnetic tape. FIG. 20 shows the data format of the recording signal. Sync block main data 1
Reference numeral 605 has 99 bytes, and these 306 words collectively constitute main data for one track. These data are divided into three, and a 10-byte external parity (C2 parity) is added to each stream. Therefore, a total of 30 words of external parity are generated, and 3
It is recorded in sync blocks No. 06 to No. 335. An 8-word internal parity (C1 parity) is added to the data of each sync block.

Furthermore, as shown in FIG. 16, each sync block has a sync byte 1601, ID0 (160
2) and ID1 (1603), ID parity 1104
, And a predetermined demodulation process is performed.
Supply 10 The rotating magnetic head 110 records a signal input from the recording signal processing unit 108 on a magnetic tape (not shown). As described above, a signal input from the input terminal 101 can be recorded at a predetermined position as a data for high-speed reproduction of a compressed image in a frame whose data amount has been reduced.

In the above embodiment, a method of doubling the quantization step has been described as an example, but the present invention is not limited to this. Requantization is possible at any step,
A quantization table and a variable-length code conversion table may be created accordingly. Of course, it is also possible to prepare a plurality of tables so as to correspond to a plurality of quantization steps and to switch between them. Further, it is also possible to dynamically change the quantization step according to the image quality and data amount of the original image.

It is also possible to make the quantization step of the coefficient corresponding to the low-frequency spatial frequency component fine, and to coarsen the quantization step of the coefficient corresponding to the high-frequency spatial frequency component. Further, the variable length code corresponding to only the high frequency spatial frequency component may be converted without changing the variable length code corresponding to the low frequency spatial frequency component. In these cases, the quantization table is also changed according to the variable-length code conversion table used.

Although the variable-length code table is prepared in advance on the ROM, the variable-length code table may be obtained by calculation at the time of initialization and stored in the RAM.

[0071]

According to the image signal encoding method of the present invention, the data amount is reduced by converting a variable length code representing an AC component of a spatial frequency component and changing a coefficient used for quantizing the AC component among the quantized coefficients. Since the data amount is reduced, the data amount reduction processing is easy, and the data amount can be reduced by program processing using a microprocessor.
Further, image quality deterioration can be suppressed as compared with the conventional method of reducing the AC component.

Further, in the image signal recording apparatus according to the present invention, since the image data to be independently recorded for high-speed reproduction is recorded with the data amount reduced by the above-mentioned image signal encoding method, the image data is reproduced during high-speed reproduction. The number of images to be updated is large, and the screen can be updated at high speed.

[Brief description of the drawings]

FIG. 1 is a block diagram of a digital recording apparatus according to the present invention.

FIG. 2 is a configuration diagram of an MPEG packet.

FIG. 3 is a block diagram of a packet detection unit.

FIG. 4 is a diagram showing a GOP structure of an MPEG image.

FIG. 5 is a block diagram of an image extraction unit.

FIG. 6 is a configuration diagram of a macro block.

FIG. 7 shows an example of a quantization table.

FIG. 8 is an arrangement order of spatial frequency information.

FIG. 9 shows an example of a variable length code.

FIG. 10 is a diagram showing a block structure of an MPEG image.

FIG. 11 shows an example of a conversion table of a variable length code.

FIG. 12 shows a method 1 for reducing spatial frequency information.

FIG. 13 is a spatial frequency information reduction method 2

FIG. 14 is a method of changing a quantization table.

FIG. 15 is a configuration of a truck.

FIG. 16 shows a configuration of a sync block.

FIG. 17 shows a method of arranging packets in a sync block.

FIG. 18 shows a packet structure of data for high-speed reproduction.

FIG. 19 shows a recording pattern of data for high-speed reproduction.

FIG. 20: Packet structure of main data

[Explanation of symbols]

101 input terminal, 102 image extraction circuit, 103 image memory, 104 packet detection circuit, 105 CP
U, 106 ROM, 107 data arrangement circuit, 108
... Recording signal processing circuit, 109 ... Rotating drum, 110 ... Rotating magnetic head, 201 ... Sync byte, 202 ... PID,
203: data area, 301: input terminal, 302: PI
D identification circuit, 303: memory, 304: communication terminal, 5
01 ... input terminal, 502 ... PID filter, 503 ...
Communication terminal 504 buffer memory 505 start code detection circuit 506 end code detection circuit 507
Memory control circuit, 508 output terminal, 1501 margin, 1502 preamble, 1503 subcode recording area, 1504 postamble, 1505 gap, 1506 preamble, 1507 main data recording area, 1508 postamble, 1509 ... margin, 1601 ... synchronization byte, 1602 ... ID0, 1
603 ID1, 1604 ID parity, 1605
Main data, 1606: Internal parity, 1701: Main header, 1702: Additional data, 1703: Packet header, 1704: Packet, 1901: Head trajectory at high speed reproduction, 1902 Reproduction data area at high speed reproduction

──────────────────────────────────────────────────続 き Continued on the front page (51) Int.Cl. ⁷ Identification symbol FI Theme coat ゛ (Reference) H04N 7/30 H04N 7/133 Z F-term (Reference) 5C018 JA01 JC04 LA01 5C053 FA20 FA22 GB01 GB04 GB06 GB07 GB08 GB22 GB26 GB32 GB37 HA24 KA01 KA05 KA11 KA24 LA07 5C059 KK00 MA00 MA23 MC11 MC33 MC34 ME01 ME17 PP05 PP06 PP07 PP16 RB02 RC24 RF05 SS17 UA02 UA11 UA38 5J064 AA03 BA09 BA13 BA16 BC01 BC02 BC16 BD03

Claims (10)

[Claims] An image signal is decomposed into a plurality of spatial frequency components by a discrete cosine transform, and each of the spatial frequency components is quantized using a predetermined quantization coefficient for each frequency. An image signal encoding method in which each spatial frequency component is converted by using a new variable length code with respect to image data encoded by a predetermined variable length code, wherein the predetermined variable length code is Of these, the variable length code of the AC component is converted into a new variable length code having a smaller data amount, and the variable length code of the image data is replaced with the new variable length code, thereby reducing the data amount of the image data. A video signal encoding method. 2. The image signal encoding method according to claim 1, wherein the conversion of the variable length code is performed based on a quantized value of the variable length code before the conversion and a variable length code after the conversion. A conversion is performed so that the quantized value becomes 1 / N times (N> 1), and a new quantization coefficient corresponding to the conversion is set by multiplying the predetermined quantization coefficient by N. An image signal encoding method comprising: 3. The image signal encoding method according to claim 1, wherein said variable-length code conversion comprises deleting a variable-length code for a high-frequency AC component. Method. 4. The image signal encoding method according to claim 1, wherein the conversion of the variable-length code includes a correspondence between the variable-length code before conversion and the variable-length code after conversion. An image signal encoding method characterized by using a table shown in FIG. 5. The image signal encoding method according to claim 1, wherein the image data converted by using the new variable length code is further converted to the new variable length code. Decoding based on the predetermined quantization coefficient to obtain each spatial frequency component, and restoring the image signal by inverse transform of the discrete cosine transform. 6. The image signal encoding method according to claim 2, wherein the image data converted using the new variable length code is further decoded based on the predetermined variable length code, Converting a quantized coefficient of an AC component among predetermined quantized coefficients into a new quantized coefficient by multiplying by N, and obtaining each spatial frequency component by using the new quantized coefficient; An image signal encoding method characterized by restoring an image signal by inverse transform of the transform. 7. An image signal recording apparatus for recording an image signal on a running magnetic tape by using a rotating magnetic head, wherein the image signal is decomposed into a plurality of spatial frequency components by discrete cosine transform, and each of the spatial frequency components is decomposed. Is quantized using a predetermined quantization coefficient for each frequency, and the quantized spatial frequency components are input as image data encoded with a predetermined variable-length code. Image data extracting means for extracting high-speed reproduction data from the extracted image data, and converting the AC component variable length code into a new variable length code having a smaller data amount among the predetermined variable length codes. Image data encoding means for reducing the data amount of the high-speed reproduction data by replacing the variable-length code of the high-speed reproduction data with the new variable-length code; Image data recording means for arranging and recording the high-speed reproduction data with the reduced data amount at a predetermined position on the magnetic tape scanned by the rotary magnetic head. apparatus. 8. The image signal recording apparatus according to claim 7, wherein the image data encoding means converts a quantized value indicated by the variable length code before conversion when converting the variable length code.
An image signal recording apparatus for performing conversion so that a quantized value indicated by a converted variable length code becomes 1 / N times (N> 1). 9. The image signal recording apparatus according to claim 7, wherein said image data encoding means deletes a variable length code for a high frequency AC component when converting the variable length code. An image signal recording device characterized by the above-mentioned. 10. The image signal recording apparatus according to claim 7, wherein said image data encoding means converts the variable-length code before conversion into a variable-length code when converting the variable-length code. An image signal recording apparatus which performs the processing using a table indicating correspondence with a variable length code to be described later.