JPS5883313A - Encoding method for digital picture data - Google Patents

Abstract

PURPOSE:To decrease the value of DSV regarding picture data by substituting a word in a natural binary code by a prescribed word, and inverting it into a complementary word. CONSTITUTION:A color video signal is supplied to an A-D converter 2, which converts the input into a word in an 8-bit natural binary code. This data word is supplied to a DSV control encoding circuit 3, which substitutes the data word by an 8-bit word rearranged by CDS. At the same time, this substituted word is inverted into a complementary word, word by word, by utilizing correlation. Consequently, DSV for picture data is reduced greatly.

Description

[Detailed Description of the Invention] A color video signal is sampled, and each sample is converted into white pits 11, for example, 8-bit digital data (1 word), recorded on a VTR, played back, and transmitted as this digital data. is being thought and done.

In this case, if one sample of 8-bit data is recorded as is on a VTR, for example, the binary levels of the recording signal @1'' and '0'' do not appear uniformly, so the recorded signal is However, since general magnetic head devices cannot reproduce the DC component during reproduction, the recording system is designed to prevent this DC component from being included in the recorded signal. k encoding is performed on the recording signal.

The encoding at the time of recording is D8 of the recording signal.
V (Digital 8um Variation
) This is a process to reduce the force as much as possible. Here, DSV is the value obtained by integrating the binary levels @1m and @O'' corresponding to +1 and -1, respectively, and this
SV has a value at any point in time or period. Then, for the continuous binary signal, D from the beginning.
When calculating the SV, if the DSV increases or decreases without limit, then ``the signal has a DC component, and if the DSV is bounded, it does not have a DC component.

Conventionally, as an encoding processing method to reduce the DSV of this 8-bit recording signal, so-called 8-9 conversion,
- Block code like 10 transform, Gu and M2 (Mo
An encoding method such as d...ed Mi 1lsr) Q:) is adopted. These encoding methods increase the recording bit rate relative to the source bit rate. In other words, the data length of 8 bits is
The DSV of the recording signal was made smaller by expanding it to bits, 10 bits, and 16 bits.

However, recently, as the amount of information in recorded signals increases, the source bit rate itself tends to increase. For this reason, the recording bit rate cannot be made very high, and it is required to be able to record at the same source bit rate.

A first object of the present invention is to provide a method that satisfies the above requirements regarding image data to be recorded.

That is, in this invention, the following two processes are performed to prevent the source bit rate of image data from increasing.

The first process is to obtain the DSV value of one word of n-bit data, that is, CD8 (Codsnror.
d Dlgitsl 8wn ) K Yotte Rearrangement is a process of replacing each word of the same<n bits in a one-to-one correspondence.

In this case, the replacement process i basically replaces a group of correlated image data words with a group of words with the same CD8 value.

The second process is to further replace the data words replaced based on the value of CD8 with correlated data words word by word or in units of multiple words that are complementary to each gator word.

By performing these two processes, it is possible to significantly reduce the DSV for image data, as will be described later, without increasing the recording bit rate relative to the source bit rate.

Tohiro, when recording ζ0 image data, one image data is divided into blocks of multiple samples for convenient image data processing during playback, and synchronization words and address words are added to this block, and further recording is performed. Media and transmission path OM
It is common practice, and also necessary, to add an error correction code in consideration of the nature of the error.

Therefore, these additional words are actually the D
Processing to reduce the SV is performed and added to the nine-stroke data word, but the DSV reduced by these additions becomes such that the processing becomes meaningless. Sometimes it got too big.

A second object of the invention is to avoid the above-mentioned drawbacks when adding additional words to image data words.

That is, in this invention, only those with a small CD8 are selected from a word group with a large number of unnecessary words as additional data, and each selected word is replaced with an additional word.

Hereinafter, one embodiment of the method of this invention will be described with reference to the drawings.

FIG. 1 shows a system diagram when the method of the present invention is applied to an apparatus for recording and reproducing a color video signal as a digital signal.

The main specifications in this case are as follows.

Signal recording method Luminance signal Y1 Color difference signal U=B-Y.

■, -R-Y component -ryp IJy/number of revolutions l; 4fsc, U, V;
2fsc where fac is the color subcarrier frequency.

Original number of samples: 768 numbers x 51H/segment (IH: 1 horizontal line, 1 sample) 8-bit sample, 1 field, 5 segments) /N'I'80 or higher.

In the figure, (1) is the input terminal for the color video signal, and the color video signal through this is supplied to the A-D converter (2).
The sampled value is converted to an 8-bit natural binary code.

The 8-bit parallel image data word from this A/D converter (2) is supplied to a DSV control (coating circuit).

In this encoding circuit (3), as described above, each 8-bit data or word is replaced with each 8-bit word rearranged by the CD8. In this case, the words to be replaced are stored, for example in a ROM, and are read out by specifying their address in a corresponding natural binary code so that each word can be replaced.

Figure 2 A, B, and C are 8-bit natural binary
It shows an example of the correspondence of words to be rearranged for all 256 words of the code. In the figure,
What is indicated as "1 element value" is the decimal representation of each word of the Nachetsuru binary code obtained from the A-reader converter (2), which corresponds to the address of the ROM.

Previously, what was displayed as "replacement value" is 1 of the word to be replaced.
It is displayed in 0 decimal notation, and the one displayed as "D8VccJ" is its binary code display.

As is clear from this second figure, natural binary
Basically, a group of words having the same value of CD8 is assigned as a replacement word for the digital data that has the same value as the code OWs. If the CD8s of correlated signals do not have the same value, a signal with a CD80 difference as small as possible is assigned.

Moreover, the median value with a high probability of occurrence as data is 1”128
For the vicinity of J4D, an 8-bit word with CD8 of 0 is allocated, and as the distance from the value ``128I'' increases, the value ``128J'' decreases to -2, -4. . . , in the direction in which the value r128J increases, the absolute value of CD8 is sequentially allocated to +2, +4, and so on.

Each word thus replaced in this encoding circuit (3) is supplied to an inversion processing circuit (4). In this inversion processing circuit (4), for example, a process is performed in which each word is replaced with a complementary word.

That is, in this inversion processing circuit (4), a certain word is output as is, and one word is output as @1''.
and ``0'' are replaced with complementary words that are completely inverted. For example, "Original value 5" in Figure 2A
3 Replacement value 49'' and CD8 is > -)'(D8VCC) If (00110001) is inverted, this is converted to (11001110). This value corresponds to the data word "Original value 202 Replacement value 206", and its CD8 is +2. That is, in this case, the complementary word of the data word of the encoding circuit (3) is a word whose positive/negative polarity of CD8 is opposite to that of the original O word.

In this way, from this circuit (4), the same word and the inverted word are obtained alternately. Considering that the input color video signal is a highly correlated signal, this inverted D of the output data word of the processing circuit (4)
It is easy to understand that if you calculate 8V, the value will be "0" in the direction of the station house.

The image data word from the inverter (4) is supplied to the recording processor (5).

In this Pousetuna (5), for example, the image data for one horizontal section is divided into four, and the image data divided into four is considered as one block, and the OIi image data for this one block is divided into four parts, and the OIi image data for one block is divided into four. CRCC (Cycle it)
Redundancy Check Code) is added, that is, the CRCC for 3 words is calculated for 1 block of both image data (for example, 204 words) k, and as shown in FIG. It is added to the last part of the image data.
・The same block synchronization signal is added to each block O9B at the position of 3 words of image data for one block.In this case, the block synchronization signal is 'D'. 8 Considering V, for example, [00110
001) (11001100) (”01110011)
Three codewords are used:

The address code is inserted between the block synchronization signal and the image data for six words, of which the first two words c[)a and CD are taken as the actual address. Of the two words, the first word CD3 is used as a head specification and segment specification address, and the second word CD3 is used as a head specification and segment specification address.
2 is a block designation address.

In this case, the block designation address is determined according to the number of the block within the segment. In this example, the number of 0 blocks in the l segment is 216, but this number is not the gist of this invention, so a description thereof will be omitted.

3rd word CDI and 4th word CDo of address data
is the CRCC for the first two words.

Further, the fifth word CW1 and the sixth word CW are codes obtained as follows.

And in this case, the block address word CD! teeth,
The value of the ordering of blocks within the segment is its tt@n
Instead, as shown in Fig. 4, the value of CD8 is assigned to "OJO" from the lowest number in the segment of the block.The value of CD8 is "0".
” after the end of the word, the CD80 value is “+2”.
Words that are j and words that are “-2” alternate Kp! 1
7a can be seen.

Similarly, when a word with the value of CD8 is "f sat 2" is completed, a word with the value of CDS of "+4" is completed in the same way,
-4 words are alternately assigned to each word.

In this way, the reason for alternately assigning positive and negative CD8 values is to ensure that the number of positive and negative values is equal, which contributes to a reduction in DSV as a whole. This is the ninth thing to do.

(6) is a ROM in which the above address word is stored.
The corresponding block address word cD2 is stored at the address of this ROM (6) indicated by the number (order) of the block within one segment.

ζ Then, the data in the format shown in Figure 3 after one block is supplied from the processor (5) to the parallel-to-serial converter (7) K, and the 8-bit parallel data is It is converted to serial data, and this is transmitted to a VTR (8
) K is supplied and recorded.

In this case, in the VTR (8), data for one field of the color video signal is recorded on the tape as a plurality of diagonal tracks.

When the VTR (8) is put into the playback state, the playback data is supplied to the serial-to-barrel converter (9), where the serial data is converted into an 8-bit parallel word-by-word data string. The data word of the converter (9) is a reproduction process! It is supplied to a.

In the reproducing processor αQ, when the address word CD of each block is converted by the ROM (Ill) into a word representing the number of the block within one segment, %KCRCC is used to detect and correct the error.

The output data from the Ryufuza α is supplied to the inversion processing circuit υ, and the image data that was converted into complementary words during recording is returned to the original word. This inversion processing circuit The output data of a is the DSV controller.
The D8V decoding circuit ← is supplied to the D8V converter 0, and a replacement that is completely opposite to that of the recording system encoding circuit (3) for the image data hood is performed. - Therefore, this circuit as has a ROM.

The word obtained from this decoding circuit aj is the original natural binary code, which is D-A
The signal is supplied to the converter I, where it is converted back to the original analog color video signal, and output to the output terminal α9.

In the present invention, as for image data, natural binary code words are stored on a CD.
A one-to-one replacement process is performed with the words rearranged based on the value of S, and an inversion process is performed to replace the replaced word with a complementary word word by word using the correlation of the image signal. By doing so, the DSV value of image data in the recording signal can be significantly reduced.

In this case, the DSV value can be reduced to some extent by focusing on the correlation of the input signals and inverting the words of the natural binary code word by word or word by word. However, in the case of natural binary codes, there are some parts where the DSV does not decrease even if the inversion process is performed, and the effect as achieved by the present invention cannot be obtained.

FIG. 4 is a diagram for comparing the CDi values of a code word and a code word that has been subjected to D8V control encoding, which is the first processing in this example. Figure 6 shows the change in to when the DSVO value at rNJ t from the sample value rOJ is calculated using natural binary code with N-2551.
FIG. 7 shows the changes when calculated using a code subjected to DSV control encoding processing.

From these figures, we can see that in natural binary code, s
Even among people whose CD5o values are correlated, there is a relatively large change, and the magnitude of the change is not constant.On the other hand, words encoded with D8V ConF roll are significantly different between people whose CD8 values are correlated. It can be seen that the values are the same.

And especially in the case of normal input Eirin signal, level i is "16" to r240J11 degree is used as data, r
Considering that the probability of appearance of data around ``x2g'' is high, it can be seen that the difference in the value of ``CD8'' becomes smaller among all words in the D8V control and t-detagged food.

As mentioned above, in the case of natural binary code, K as is clear from FIGS. 5 and 6.

The value of DS-V is increasing and decreasing. Therefore, there are some parts where it cannot be said that K D '8 V is reduced even if the inversion process is performed. For example, in Figure 5, the value "3"
If the inversion process is performed from the code with the value "1" to the code with the value "32", the DSV in this part becomes 8, which actually becomes larger.

On the other hand, in the case of image data encoded with D8V control according to the present invention, the CDS is almost equal for data whose DSV values are correlated as shown in FIGS. 5 and 7; If inversion processing is performed on certain data, the DSV will always tend to decrease.

Incidentally, as in the above example, when a color video signal is recorded on a VTR by forming multiple tracks per field, each segment is inverted in a natural binary code state. When, the D8V control I/H-D according to the present invention, the CD80 average value when Googled and further inversion processing, DS
V. The maximum value, the final value for the DSV1-kl component, is shown in FIG.

Note that this is a result only for image data,
The block synchronization signal, address signal and CRCC are the same without any addition.

As is clear from this Figure 8, regarding the image data,
According to the present invention, the final value of the DSV is reduced by eight compared to the final value of the DSV when the natural binary code is unchanged and inverted.

The above is a study on DSV with only one image data, but as mentioned above, in practice, in order to perform optimal processing for recording and reproducing to a VTR, image data is divided into blocks of multiple words and this synchronous message),
Add CRCC, address signals, etc. The maximum DSV value, etc. when added to the above 6 image data and recorded and played back without making any modifications to the added data is the 9th.
As shown in the top two rows of the figure.

In other words, compared to the case of only image data in FIG.
A significant increase in V was observed, especially in this IME! D8
It can be seen that if an address or the like is added to the image data that has been subjected to the V control two-coding and inversion processing, the effect of the processing performed on this image data is greatly diminished.

Incidentally, since it is sufficient to allocate a fixed code word to the synchronization signal, an increase in DSV should be prevented to some extent by considering the code word to be allocated.

Furthermore, since CRCC is a code that is generated completely randomly, the increase in DSV due to CRCC is not considered to be much of a problem.

Then, it is considered that the main cause of the increase in DSV is the address word.

Based on the above idea, in the above example, allocation processing was performed for block address word CD2 in order from the word with the smallest absolute value of CD8.

The bottom row of FIG. 9 shows the final value of DSv when this address word is processed and added together with a synchronization signal and CRCC to the image data processed according to the present invention. As is clear from this, if the address word is processed according to the present invention, the maximum value of DSν will decrease by about + and the final value will decrease by about 100% compared to the case where no processing is applied to this address etc. be.

As described above, according to the present invention, it is possible to significantly reduce the transmission image data 0D8V without increasing the transmission bit rate with respect to the source bit rate of the image data.
Since the data to be added is also subjected to special processing, it is dangerous to impair the effect of reducing O, D 8 V on the image data. ), DSV for the entire transmitted data
can be significantly reduced.

Note that the correspondence table for the image data shown in FIGS. 2A, B, and C and the correspondence table for the block address words shown in FIG. 4 are examples), and in any case, even if K>%/%, Appropriate rearrangement may be performed within the range of word groups whose CD8 values are the same.

Further, the inversion process for image data may not be performed for each word, but may be performed for each plurality of words as long as the data are correlated. Alternatively, focusing on the vertical correlation of the Eirin signal, inversion processing may be performed for each horizontal line.

In addition, in the above example, address words with the lowest absolute values of CD8 are assigned alternately to those with positive values and those with negative values, but all address words must appear. Therefore, if the number of positive value words and the number of negative value words are made equal, there is no need to alternately allocate positive and negative values.

Furthermore, since the address word has a smaller number of words than the image data word, the DSV can be reduced without changing the transmission bit rate by converting the 8-bit address word to 9 or 10 bits. For the address word you can use this 8-9.8-10
Block encoding processing such as conversion may also be performed. In this case, when selecting a 9-bit, 10-bit F word from an 8-bit word, the words are selected in order from the smallest CD8, as in the above example.

In addition, when forming horizontal parity and vertical parity and adding them to image data for error correction, after calculating these parities, 8-9 conversion and 8-9 conversion are applied to these codes. -10 conversion may also be performed.

However, the invention can be applied to all cases where digital signals are transmitted not by recording digital image data on a VTR, but by other transmission means that do not pass a direct current component.

[Brief explanation of the drawing]

Fig. 1 is a system diagram of an example of a device to which the ζ0 invention method is applied, Fig. 2 is a diagram showing an example of words replaced in image data, and Fig. 3 explains the format of one block of recording signal. Figure 4 is a diagram showing examples of words to be replaced for address data; Figures 5 to 7 are diagrams showing examples of words to be replaced for an image according to the present invention; FIGS. 8 and 9 are diagrams used to explain the present invention in detail. (2) A-D converter, (31)! DSV coy)
o-pyx encoding circuit, (4) is an inversion processing circuit, (5) is a recording processor, (6) is an address ROM
It is. Figure 1 Figure 8 (110 kekusuke) Figure 2 A yi-41@il DSVCCCDS 7c@
Value If value osvcc coso o
00000000 -815 +7
00010001 -4 16 1
8 00010010 -421 36 0
0100100 -4 22 40
00101000 -423 48 0011
0000 -4 24 65 111
000001 -425 66 010000
10 -1 26 68 0100
0100 -427 72 01001000
-II 28 80 01010
000 -435 160 10100000
-4 36 192 1100000
0-4 Figure 2 B! 61 232 11101000 0
Summer 62 240 11110000 0
171 9101011011 2
)72 93 01011101 2nd 255 255 11111111 8 5th figure 215 11111111 ShiMo5vc
c cos 00000010 00100000 01000000 00000110 00001100 00010001 00010010 00010.100 00011000 0010QOQI 00110000 01000001 010000 10 01000100 01001000 10000010 10010000 10100000 C

Claims (1)

[Claims] When digital image data words are divided into blocks into multiple words and additional data such as a block synchronization signal and address signal are added to the blocks and transmitted, the digital image data words are transmitted at a transmission bit rate higher than the source bit rate. At the same time, the additional data is rearranged in such a way that the DSV is optimized without causing any damage, and the above additional data is
A method of encoding digital image data in which only a small D8 is selected and the selected book is used.