JPH05236450A - Video information compressing and encoding device - Google Patents

Abstract

PURPOSE:To improve compressing and encoding effects as a whole without increasing circuit scale by encoding an AC low frequency part together with a DC while using the same encoder, and concentrating the generating distribution of an AC high frequency part. CONSTITUTION:Video information data are inputted to a terminal 1, discrete cosine-transformation is executed by an orthogonal transformer 8, and an AC coefficient high frequency part output 91 is encoded by a first encoder 21. On the other hand, a DC coefficient AC coefficient low frequency part output 92 is encoded by a second encoder 22. Thus, the compressing and encoding effects can be improved by executing an encoded output at the encoder 21 exclusive for the AC high frequency, encoding the AC low frequency part together with the DC at plural encoders and comparatively selecting the minimum code among respective outputs.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a coding device for compressing and coding digital video information data.

[0002]

2. Description of the Related Art FIG. 14 is a block diagram showing a conventional encoder. In the figure, 1 is input digital data, 8 is an orthogonal transformer connected to the input data 1, and is an AC system numerical sequence.
101 and the DC system numerical value 102 are output. Encoding device 201
Numerals 202 designate encoders, which respectively output 10
1 and 102, each having outputs 700a and 700b.

Next, the operation will be described. The input image digital data 1 is input in the form of blocks each having 8 × 8 vertical and horizontal pixels. Hereinafter, the data value in the i-th row and the j-th column will be represented by f (i, j). Also i, j
Is called spatial coordinates. In this way, 64 pieces of data are input to the orthogonal transformer as one block, and as a result, the orthogonally transformed data are output as the AC system numerical value sequence 101 and the DC system numerical value 102.

Here, the orthogonal transformation process will be additionally described by taking a discrete cosine transformer (hereinafter abbreviated as DCT) as an example. This conversion is performed by calculating the following equation.

[0005]

[Equation 1]

The conversion result is as shown in FIG.
The data in the u-th row and the v-th column, which are obtained as individual data blocks, are represented as F (u, v). Of the data blocks, the data of sequence 0 in the upper left corner in Fig. 15 is DC.
It is called a coefficient and is output from the output line 101. The sequences 1 to 63 are called AC coefficient values, and are output as a data string from the output line 102. The output sequence is shown by arrows in FIG. The correspondence with sequence numbers is shown in FIG.

The data after the orthogonal transformation are coded by the encoder 201.
, 202 to be encoded. Encoder 201, 2
02 Both are basically constructed by the structure shown in FIG.

Here, the configuration and operation of the encoding device shown in FIG. 17 will be described. The input to the encoder is
It is 100 and outputs code 700. The encoding device itself is composed of an encoding table 2b and an encoder 2a, which are connected to each other by connecting lines 2c and 2d. The input digital data 100 is read by the encoder 2a, the encoding table 2b is referred to by the input data value via the signal line 2d, the code is obtained from the signal line 2c, and the output operation is performed as the output 700.

In this case, the code table is set so that the bit length of the code data can be changed according to the occurrence probability of the input data 1. That is, the occurrence probability of the data Di is Pi
If the post-coding code length at that time is Li, if the input data Dj and Dk are set so that if Pj> Pk, then Lj <Lk, then the total post-coding data is smaller than the input data. It becomes possible to do.

Assuming that the input data is 0, 1 as shown in FIG.
There are four types, 0, 20, and 30, and the occurrence probability is 50% and 30 respectively.
%, 10%, 10% are assumed. In this case, the input data string is 0, 20, 0, 0, 30, 10, 0, 10,
If it is 0 or 10, the total number of bits required for input data is 6 bits x 10 words = 60 bits, but if the encoding is performed using the table shown in Fig. 18, for example, all 17 bits can be encoded. As a result, the total amount of data can be reduced.

[0011]

Since the conventional encoder is configured as described above, it is effective only when the occurrence probabilities of the input data are locally concentrated, and the variance of the input data values is reduced. There is a problem that the compression effect of the data amount is hard to be obtained when the size becomes large.

That is, if the occurrence probability of the input data is, for example, as shown in FIG. 19, the amount of coded data does not necessarily decrease. If the input numerical value is in the vicinity of the value a, it is possible to reduce the code amount, but at a distant portion, a long code is assigned as much as the occurrence probability is reduced, and if the occurrence probability is not so low, As a result, the code amount becomes large and the total code amount cannot be reduced.

2 to 4 show measured values of the DC coefficient and the AC coefficient obtained by performing DCT on an actual video signal as an example. 2 to 4, an overall view is shown in FIG. 2, and a scale difference near the value 0 is shown in FIGS. 3 and 4. According to this figure D
It can be seen that in the C data and the AC data, the one with the lower sequence number (the one with the sequence 2 in the figure) has a considerable variance in the occurrence probabilities, and it is understood that compression encoding is difficult.

The present invention has been made to solve the above problems, and an object of the present invention is to obtain an encoding device which can obtain a high data compression effect even when the dispersion of input data values is large.

[0015]

A coding device of a first invention according to the present application performs an orthogonal transformation on input data to obtain a DC signal.
In addition, the low-order data up to the low-order data of the AC section is used as A
The remaining high-order data of the C data is used as high-frequency part data to perform a block segmentation operation, and then an optimal encoder is passed through each block.

The encoding device of the second invention according to the present application is provided with a single encoder dedicated to the high frequency band data and a plurality of encoders as low frequency band encoders. The feature is that the sign of is selected and output.

The encoding apparatus of the third invention according to the present application is the second invention.
In the present invention, an encoder that performs encoding by changing the allocated bit length according to the absolute value of the input data value as the low band encoder, and performs encoding with variable length bits according to the occurrence probability of the input data. It is characterized by being configured with an encoder.

[0018]

According to the first aspect of the present invention, the DC value having a large variance value and the AC low-order part are grouped together as the low-pass part data as the data value of the DCT output, and the input having a wide occurrence probability for the low-pass part encoding. In addition, it is possible to install an encoder that is effective for the high-frequency part, and as a result, it is possible to collect data that has a significant deviation in the occurrence probability, and an encoder that is highly effective for its input can be installed. The total amount of codes in the high-frequency part can be suppressed to a low level, the compression effect of data in the high-frequency part can be dramatically improved, and the total code amount itself can be reduced. Further, as for the high-band encoder, the features of the input data become more prominent, so that the circuit scale can be suppressed to be small and the design can be facilitated.

In the second aspect of the present invention, further compression coding can be performed by adopting a configuration in which the low band portion data which is difficult to obtain the compression effect is encoded by a plurality of encoders and the smallest one is selectively output.

In the third aspect of the invention, in the second aspect of the invention, as a high frequency encoder, a code depending on an absolute value is effective for the characteristic that the occurrence probability in the vicinity of the value 0, which is particularly found in AC low-order data, is slightly high. By using an encoder having a variable bit length and a variable length encoder according to the occurrence probability, an encoding device having a high compression effect can be realized.

[0021]

EXAMPLES Example 1. An embodiment of the present invention will be described below with reference to the drawings. In FIG. 1, 1 is an input data line, and 8 is an orthogonal transformer. The low frequency band data string is output from the orthogonal transformer output 91 and input to the dedicated encoder 21. same
At 92, the low frequency band data is output and input to the encoder 22.
Encoding is performed by each encoder, and code data is output via each output line 7a, 7b.

Next, the operation will be described. Hereinafter, description will be given by taking an example of actual encoding of video data. 2 to 4 are DCT output, sequence 2 is an example of DC and AC low frequency distribution, and sequence 5 is an example of AC high frequency distribution.
It is a figure showing the occurrence distribution about. In this example, the DCT output has an 11-bit width, and the result for each sequence of the luminance component obtained by performing the orthogonal transformation 2700 times per screen is shown.

Consider, for example, the following three encoders as the encoders 21 and 22. The first coding method is a fixed length, that is, the input data is output as it is in 11 bits (hereinafter abbreviated as FC), and the second is a code length that is changed according to the occurrence probability of the input data (hereinafter abbreviated as VLC). It is assumed that the code length is changed according to the magnitude of the absolute value of the input data (hereinafter abbreviated as RC).

For VLC, consider the encoder that performs the encoding shown in FIG. 5, for example, based on the occurrence probability of sequence 2, and for RC, the encoder that performs the encoding shown in FIG. Specifically, in the case of VLC, a short code is assigned in order from the one with the highest occurrence frequency, and the identification code (5 bits in this example) and the input data 11 bits in total are used for the less frequent codes.
Assigned a bit. The state of code allocation is also shown in FIG. In case of RC, the input data is -16 ~ as shown in FIG.
In the case of +15, encoding is performed with 6 bits (identification bit 1 bit + data code 1 bit + data lower 4 bits, total 6 bits). In other cases, 12 bits (identification bit 1
It is assumed that encoding is performed with a total of 12 bits (bit + data 11 bits). Note that this system will be referred to as the RC5 system below. The case where the range from -32 to +31 is encoded with 7 bits and the range outside is encoded with 12 bits by the same method is defined as the RC6 system, and is shown in FIG. Similarly, RC4, RC
5 etc. can be defined.

FIG. 7 shows the code amount in the case of actually encoding the sequence 2 as an example of DC and AC low frequency data and the sequence 5 as an example of the same high frequency data in each system. From the figure, in sequence 2, the total code amount increases in the order of RC → VLC → FC, and in sequence 5, increases in the order of VLC → RC → FC. It is understood from FIG. 7 that the distribution in DC increases in the order of FC → RC → VLC because the distribution is almost uniform as shown in FIG.

Also from the above example, as a result of performing the DCT operation, the occurrence frequency distribution gradually concentrates in the vicinity of the numerical value 0 in the order of DC → AC low band → AC high band, and the compression coding effect is applied to each coding. It was shown that FC->RC-> VLC in the descending order of.

In the high AC range, the VLC system is very effective and has no practical problem, but it can be improved in the low range of DC and AC. That is, in sequence 2, RC is the most effective, and the difference in the effects of FC and RC is not so large for DC. Therefore, rather than the conventional example in which the AC low frequency band is treated the same as the AC high frequency band, the DC is more effective. For example, it is expected that the effect will be higher if the encoding is performed by RC.

The configuration of this embodiment is the same as that of the first embodiment. It is FIG. 1, and FIG. 6 shows the difference in the code amount from the conventional device. From the figure, it can be understood that the present invention is improved to 51,508 bits compared to the conventional code amount of 52,130 bits.

Example 2. As described above, the VLC method is very effective in the AC high frequency band and has no practical problem, but the AC low frequency band can be further compressed and encoded by the following method, for example.

FIG. 9 shows some R for sequence 2.
The result of encoding by the C method is shown. In this example, R
C5 is the smallest, but RC4 is when the distribution situation shows 0 concentration, and RC6 when the dispersion tendency is a little stronger.
Can be minimal.

Further, FIG. 10 shows the case where VLC and RC5 are adaptively switched to perform encoding. This is an encoding that combines VLC that is effective when the frequencies are concentrated and RC that is effective when the frequencies are slightly dispersed. Even if one bit is added to each code to identify them, it is the smallest in the above example. This indicates that the encoding was successful.

FIG. 11 shows an encoding device configured by this method.
Is. The codes in the high frequency band are the same as those in the first embodiment, but in the low frequency band, both encoders 22a and 22b perform coding and output each output.
31 and 32 are compared by the code amount comparator 4, and the one having the smaller result code amount is selected and output by the selector 6 through the switching line 5, and finally the low frequency code is output from 7b. In the example of the figure, VLC is used as the high frequency encoder 21, and R is used as the low frequency encoders 22a and 22b.
By arranging C and VLC, it is possible to suppress the code amount especially in the AC low frequency band to a low level, and as a result, it is possible to configure an encoding device that suppresses the total code amount to a low level.

Example 3. FIG. 12 shows a case where the DC component of the second embodiment is configured by a dedicated encoder. As described above, the effect of RC is high in the AC low frequency region, and more efficient compression was possible by using it in combination with VLC. However, since the occurrence distribution of DC is almost uniform, FC It is most effective to encode with. Therefore, a dedicated FC is used for DC, and the system of Example 2 is used for the AC high range and low range.

FIG. 13 shows the result of actual encoding by the encoder of this system. From this figure, it can be confirmed that the total code amount of 49,822 bits is the lowest code amount among the encoding devices described above.

It should be noted that, in this configuration, the device itself seems to be complicated because the encoder dedicated to DC is arranged as compared with the second embodiment, but in the case of FC, the input value is actually the code output as it is. The circuit scale does not increase. Also, since it is a dedicated encoder for DC values only, R in the AC low frequency band
It is not necessary to add an identification bit for identifying the C and VLC codes, and it can be easily realized from the configuration of the second embodiment.

[0036]

As described above, according to the first aspect of the present invention, the AC low frequency band is encoded together with DC using the same encoder,
Since the occurrence distribution in the AC high frequency band is centralized, the data in the high frequency band is characterized, the compression coding effect is enhanced as a whole, and because it is composed of two encoders as in the conventional system, the circuit scale is reduced. It is possible to realize an encoding device having a very high compression effect without increasing the number. Further, since the input of the encoder in the high frequency band of AC is characterized, the encoder itself can be easily designed.

Further, in the second invention, since a plurality of encoders are provided for low-frequency data and the minimum code is selectively output, the compression effect can be increased for the low-frequency part where the compression effect is difficult to obtain. It will be possible.

According to the third invention, in the second invention, by using the RC coder and the VLC coder as the coder which matches the distribution condition of the input data values such as the low frequency part,
It is possible to obtain a high compression effect.

[Brief description of drawings]

FIG. 1 is a block diagram showing a configuration of a compression encoding device according to the present invention.

FIG. 2 is a graph showing a distribution of DCT output values of video data.

FIG. 3 is a graph showing a distribution of DCT output values of video data.

FIG. 4 is a graph showing a distribution of DCT output values of video data.

FIG. 5 is an explanatory diagram of a VLC code.

FIG. 6 is an explanatory diagram of an RC code.

FIG. 7 is a comparison table of code amount results for each coding method for each sequence data.

FIG. 8 is a comparison table showing a difference in code amount between an example according to the present invention and a conventional example.

FIG. 9 is a code amount comparison table for each RC method of sequence 2 data.

FIG. 10 is a code amount comparison table of VLC / RC selective coding schemes.

FIG. 11 is a block diagram showing another embodiment of the compression encoding device according to the present invention.

FIG. 12 is a block diagram showing another embodiment of the compression encoding device according to the present invention.

FIG. 13 is a code amount result of the encoding device according to the third embodiment.

FIG. 14 is a block diagram showing a compression encoding device in a conventional example.

FIG. 15 is an explanatory diagram showing a DCT output.

FIG. 16 is an explanatory diagram showing DCT output sequence numbers.

FIG. 17 is an operation explanatory diagram of a VLC encoding device in a conventional example.

FIG. 18 is an explanatory diagram of a VLC code table.

FIG. 19 is a graph showing an example of input data in which a VLC encoder is difficult to obtain a compression effect.

[Explanation of symbols]

 1 Data input line 21 High frequency data encoder 22 Low frequency data encoder 22a Low frequency data encoder 22b Low frequency data encoder 23 DC data encoder 31 Encoder output 32 Encoder output 4 Code amount comparison Unit 5 Code selection line 6 Code selection line 7 Code output line 7a High band code output 7b Low band code output 7c DC code output 8 Orthogonal converter 91 Orthogonal converter High band data output 92 Orthogonal converter Low band data output 93 Orthogonal converter DC data output

─────────────────────────────────────────────────── ───

[Procedure amendment]

[Submission date] June 8, 1992

[Procedure Amendment 1]

[Document name to be amended] Statement

[Name of item to be corrected] Claim 1

[Correction method] Change

[Correction content]

[Procedure Amendment 2]

[Document name to be amended] Statement

[Correction target item name] 0024

[Correction method] Change

[Correction content]

For VLC, consider the encoder that performs the encoding shown in FIG. 5, for example, based on the occurrence probability of sequence 2, and for RC, the encoder that performs the encoding shown in FIG. Specifically, in the case of VLC, a code having a low occurrence frequency is assigned in order from a low occurrence frequency, and an identification code (5 bits in this example) and an input data 11 bits, 16 bits in total, are assigned to a low frequency code. .. State of code allocation is also shown in FIG 5. In the case of RC, the input data is-
In the case of 16 to +15, 6 bids (identification bid 1 bid +
It is assumed that encoding is performed with a data code of 1 bit + data lower 4 bits, that is, a total of 6 bits, and in other cases, encoding is performed with 12 bits (identification bit 1 bit + data 11 bits, 12 bits in total). did. Note that this system will be referred to as the RC5 system below. In the same way, -32
The case where the range of +31 is coded with 7 bits and the code outside the range is coded with 12 bits is defined as the RC6 system, and is shown in FIG.
Similarly, RC4 and RC5 can be defined in the same manner.

[Procedure 3]

[Document name to be amended] Statement

[Correction target item name] 0028

[Correction method] Change

[Correction content]

The configuration of this embodiment is the same as that of the first embodiment. It is FIG. 1, and FIG. 8 shows the difference in the code amount from the conventional device. From the figure, it can be understood that the present invention is improved to 51,508 bids as compared with the conventional code amount of 52,130 bids.

[Procedure amendment 4]

[Document name to be amended] Statement

[Correction target item name] 0031

[Correction method] Change

[Correction content]

Further, FIG. 10 shows a case where VLC and RC5 are adaptively switched to perform coding. This is an encoding that combines VLC that is effective when the frequencies are concentrated and RC that is effective when the frequencies are slightly dispersed, and even if one bit is added to each code to identify both, the minimum in the above example. This indicates that the encoding was successful.

[Procedure Amendment 5]

[Document name to be amended] Statement

[Correction target item name] Explanation of code

[Correction method] Change

[Correction content]

[Explanation of Codes] 1 data input line 21 high frequency data encoder 22 low frequency data encoder 22a low frequency data encoder 22b low frequency data encoder 23 DC data encoder 31 encoder output 32 encoder output 4 code amount comparator 5 code selection line 6 code selector 7 code output line 7a high frequency code output 7b low frequency code output 7c DC code output 8 orthogonal transformer 91 orthogonal transformer high frequency part data output 92 orthogonal transformer low Area data output 93 Orthogonal converter DC data output

Claims (3)

[Claims] 1. An encoding device for compressing and encoding digital video information data, and an orthogonal transformer for orthogonally transforming input data to transform the data into a low band part and a high band part and outputting the data. A first encoder connected to the high-frequency output of the above, and a second encoder connected to the same low-frequency output for performing encoding with a different encoding method from the first encoder. A video information compression encoding device characterized by: 2. An encoding device for compressing and encoding digital video information data, which comprises: an orthogonal transformer for orthogonally transforming input data, converting the data into a low band part and a high band part, and outputting the same; and A first encoder connected to the high-frequency output, a second encoder connected to the low-frequency output of the orthogonal transformer, and an encoding different from the second encoder connected to the same low-frequency output. A third encoder that performs encoding according to the method, and a comparison selector that has as inputs the second encoder output and the third encoder output, and selects and outputs the encoder output with the minimum amount of data after coding. And a video information compression coding apparatus characterized by comprising: 3. An encoder that performs variable-length bit length encoding with a variable assigned bit length according to the absolute value of the input data value, and a code that performs variable-length bit length encoding according to the occurrence probability of the input data. 3. The video information compression coding apparatus according to claim 2, further comprising: a converter as an orthogonal converter low-band data encoder.