JPH03270388A - Data compressing device - Google Patents

Abstract

PURPOSE:To easily perform processing fast by predicting the encoding quantity of a standard quantization step or standard band-limit value from a past activity value obtained by the filtering process and absolute value addition of data. CONSTITUTION:Data which is inputted and stored temporarily in a frame memory 11 and then read out is filtered by an operator processor 12 and inputted to an absolute value sum calculating circuit 13, whose arithmetic output (activity value) is inputted to a standard encoding quantity value converter 14. The standard encoding quantity value converter 14 consists of a ROM, etc., and is stored with the mean value of encoding quantities corresponding to a specific activity value, converts the activity value inputted from the absolute value sum calculating circuit 13 into a corresponding encoding quantity predicted value, thereby outputting it to a quantization step converter 15 and a band-limit value converter 16. Consequently, easy and fast processing become possible.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application J] The present invention relates to a data compression device suitable for use, for example, when compressing a color still image and recording it on a predetermined recording medium.

[Prior Art 1] When recording image data on a recording medium such as a magnetic disk, the data is compressed for efficient recording. For this compression, for example, image data is divided into blocks each having NXN (or NXM) pixels, and each block is orthogonally transformed. Orthogonal variation 15! ! The resulting data is roughly quantized with a predetermined quantization step, and then zero-run length encoded or Huffman encoded. Compressing data in this way effectively compresses the data, but the amount of code varies depending on the image.

Therefore, conventionally, control has been performed to keep the code amount constant as follows.

The first method is to calculate the amount of data actually quantized at a predetermined quantization step, and then adjust the number of quantization steps according to the calculation result so that the amount of data reaches the desired value. This method is to change it and redo the quantization.

The second method focuses on the fact that the coefficients of the data after orthogonal transformation have a predetermined relationship with the code amount, and calculates the sum of squares of the coefficients for each block. Each block is divided into, for example, four classes according to its size, and more bits are allocated to blocks in classes with larger amounts of data, and fewer bits are allocated to blocks in smaller classes. be.

[Problems to be Solved by the Invention] However, in the first method described above, the process of actually calculating the code amount of quantized data must be repeated at least twice, so high-speed processing is difficult. be.

Furthermore, in the second method described above, not only it is necessary to perform orthogonal transformation processing, but also information indicating the class must be added, which increases the amount of code and requires complicated processing.

Furthermore, in either method, it is not possible to allocate the amount of code between multiple fields or multiple frames.

The present invention was made in view of this situation, and enables simple and high-speed processing.

Furthermore, it is possible to adjust the allocation of code amount over multiple fields or multiple frames.

[Means for Solving the Problems 1] The data compression device according to claim 1 is a data compression device that divides input data into blocks each having a predetermined number of pixels, performs orthogonal transformation, quantizes, and further encodes the data. , a first processing means for filtering input data;
A calculation means that performs an operation including at least addition on the data processed by the first processing means, and converts the activity value outputted by the calculation means into a corresponding predicted code amount value in a standard quantization step or standard band limit value. a standard code amount value converting means for setting a target value of the data amount; a target value setting means for setting a target value of the data amount; and a target value set by the target value setting means and a code amount predicted value outputted from the standard code amount value converting means. a set value output means for comparing and outputting a quantization step value or a band limit value corresponding to the difference; The second processing means for limiting, the past activity value output by the standard code amount value conversion means, the predicted code amount value at the standard quantization step or standard band limit value predicted from the past activity value, and the past activity. and a storage means for storing at least one of the past code amount values encoded with the quantization step value or the band limit value set corresponding to the value, and the standard code amount value converting means is provided with a storage means for storing at least one of the past code amount values encoded with the quantization step value or the band limit value set corresponding to the value. The data compression device according to claim 2, wherein the data compression device performs the operation of converting the activity value into the predicted code amount value using the value obtained by dividing the input image data into blocks each having a predetermined number of pixels; In a data compression device that performs orthogonal transformation, quantization, and further encoding, there is a code amount prediction unit that predicts the code amount of input image data, and a code amount prediction unit that predicts the code amount of input image data, and a code amount prediction unit that predicts the code amount of the image data according to the output of the code amount prediction unit. processing means for limiting or quantizing the image data; encoding means for encoding the image data processed by the processing means; a plurality of blocks as one group, a plurality of groups as one segment; It is characterized by comprising a control means for controlling the code amount prediction means so that the amount is constant.

[Operation] In the data compression device according to claim 1, the data is filtered, and the code amount at the standard quantization step or the standard band limit value is determined from the past activity value obtained by, for example, adding absolute values. is predicted. Then, the number of quantization steps or the band limit value is adjusted according to the prediction result.

Therefore, simple and high-speed processing becomes possible.

Furthermore, in the data compression apparatus according to the second aspect, a plurality of fields or a plurality of frames are made into one segment, and the amount of code in each segment is adjusted.

Therefore, even in an encoding algorithm that is completed with multiple frames, yields, or multiple frames, it is possible to allocate an appropriate amount of code.

[Embodiment] FIG. 1 is a block diagram showing the configuration of an embodiment of a data compression device of the present invention.

Of the input image data, intra-frame predicted data is directly generated, and inter-frame predicted data is generated via the motion compensation and difference data generator 6, respectively, to code quantum i1'!
It is input to II unit 1 (code amount prediction means). The code amount predictor 1 predicts the code amount from the input data, sets the band limit value and the number of quantization steps corresponding to the predicted value, and performs quantization with the Bri filter 2 (second processing means or processing means). output to the device 4 (second processing means or processing means).

The BRI filter 2 band-limits the data supplied from the code amount predictor 1 in accordance with the band limit value set by the code amount predictor 1, and performs D CT (Discrete
(cosine transfer) circuit 3. The data processed by the DCT circuit 3 is sent to a quantizer 4
and is quantized by the number of quantization steps set by the code amount predictor 1. The quantized data is input to an encoder 5 (encoding means) and encoded.

A buffer memory 7 (storage means) stores the output of the code amount predictor 1 and the output of the encoder 5, and supplies necessary data to the code amount predictor 1 under the control of the CPU 8 (control means).

The code amount predictor 1 is configured as shown in FIG. 2, for example.

The input image data is temporarily stored in the frame memory 11. The data read from the frame memory 11 is filtered by the operator processor 12 (first processing means) and then input to the absolute value sum calculation circuit 13 (calculation means) where it is calculated.

The output (activity value) of the absolute value sum calculation circuit 13 is
The signal is input to the standard code amount line converter 14 (standard code amount value conversion means), and is converted into a code amount predicted value at the standard quantization step and standard band limit value.

This code amount predicted value is calculated by the quantization step converter 15 (setting value output means) and the band limit value converter 16 (setting value output means).
is output to.

A target value for the amount of data to be set is input to the quantization step converter 15 and the band limit value converter 16 from the CPU 8 via the control data value transmitting circuit 17 (target value setting means). The quantization step converter 15 and the band limit value converter 16 compare the code amount predicted value inputted from the standard code amount line converter 14 and the target value inputted from the control data value transmission circuit 17. A quantization step value and a band limit value corresponding to the error are output to the quantizer 4 and the Buri filter 2, respectively.

Next, its operation will be explained.

Input image data is input to the frame memory 11, and image data for one frame is temporarily stored therein. The data stored in frame memory 11 is read out, provided to operator processor 12, and filtered.

This filter processing will be explained with reference to FIG. 3.

FIGS. 3(a) to 3(C) each represent nine (3×3) coefficients of a bypass filter, a low-pass filter, or a Laplacian when each process is performed.

3×3 pixel data of a predetermined area is read out from the frame memory 11, and these data and 3×3 pixel data shown in FIG.
A factor of 3 is multiplied between the corresponding positions. The nine pieces of data obtained as a result of this multiplication are further added.

Next, the area read out from the frame memory 11 is shifted, for example, by one pixel to the right, and the 3×
Similar processing is performed for pixel data No. 3. Thereafter, similar processing is performed on all data for one frame.

At this time, the amount by which the area is sequentially shifted is not 1 pixel but 2 pixels.
It can also be a pixel, three pixels, etc.

The data thus filtered is input to the absolute value sum calculation circuit 13. The absolute value sum calculation circuit 13 receives the 3X data from the operator processor 12 as described above.
The absolute value of one piece of data obtained every time 3 pixel data is filtered is calculated, and the absolute values are added for one frame. This added value (sum of absolute values) is input to the standard code amount line converter 14.

The absolute value sum (activity value) output by the absolute value sum calculation circuit 13 approximately corresponds to the code amount, as shown in FIG. That is, when a predetermined activity value is given, the code amount of the image data is represented by the hatched area in FIG. The standard code amount line converter 14 is constituted by, for example, a ROM, and stores the average value of the code amount corresponding to a predetermined activity value, as shown by the solid line in FIG. Therefore, the standard code amount line converter 14 converts the activity value inputted from the absolute value sum calculation circuit 13 into a corresponding code amount predicted value, and outputs it to the quantization step converter 15 and the band limit value converter 16. do.

The operation of the standard code amount line converter 14 will be explained in more detail using mathematical expressions as follows.

The standard code amount line converter 14 is configured to band-limit the data of a large number of images with various patterns using a preset standard band limit value (A = O), and to perform a standard quantization step (F = The code amount obtained when quantizing in step 50) is calculated and stored as dera ortho data so that the activity value and the code amount correspond statistically to one to one.

When encoding past data or data of the first frame or field that does not yet exist, the following calculations are performed.

B ID (50, O): ACT X k here BI
D(50,O) is the standard quantization step (F=50),
In addition, the predicted value of the code amount at the standard band limit value (A = 0), ACT is an activity value, and k is a constant. That is, the activity value inputted from the absolute value sum calculation circuit 13 is multiplied by the constant k to calculate the code amount predicted value BID(
50,0).

Furthermore, when encoding data of the second and subsequent frames or fields in which past data exists, the following calculations are performed.

BID(50,0)=ACTX(BID-'(F,
8)/TGT)X (BID-” (50,0)/AC
T-') Here, BID-'(50,0) is the code amount predicted value at the past (previous time in this example) standard quantization step and standard band limit value, BID-'(
F, A) is the code amount value at the past (previous) actually used quantization step (F) and the band limit value (A);
ACT-1 is the past (previous) activity value, TGT
represent the target code amount, respectively.

The past code amount predicted value BI D-' (50, 0) and the activity value ACT'''1 are converted to the standard code data value converter 14.
The data is outputted to the buffer memory 7 and stored therein. Further, the past code amount value BID'''' (F, A) is output from the encoder 5 to the buffer memory 7 and is stored therein.

The values stored in the buffer memory 7 are supplied from the buffer memory 7 to the standard code electric value converter 14.

The TGT is supplied from the control data value transmitting circuit 17 to the standard code electric value converter 14.

In the above formula, B I D-”(F, A)/T G T
is the tendency of the error between the actual code amount and the target code amount, and
BI D-'(50, O)/ACT-'' is the slope in the hatched area in Figure 4 that has the strongest correlation from the time of prediction, and is expressed by the following code. This provides feedback to reduce errors.

The control data value transmitting circuit 17 outputs a preset target value of data amount to the quantization step converter 15 and the band limit value converter 16. The quantization step converter 15 and the band limit value converter 16 are also configured by, for example, a ROM, and each stores a corresponding number of quantization steps and a band limit value appropriate for obtaining a predetermined amount of code. . Therefore,
The quantization step converter 15 and the band limit value converter 16 are
The code amount prediction value input from the code electric value converter 14 and the target value input from the control data value transmitting circuit 17 are compared, and the set target value is obtained in accordance with the error. The appropriate number of quantization steps and band limit value are read out and output to the quantizer 4 and the Buri filter 2.

For example, as shown in Fig. 5, the Buri filter 2 is a filter with five types of characteristics in which the band limit value (cutoff frequency) sequentially changes in steps of 1716 in the range from 15/6 to 1/16 of the reference value. , and selects a filter with characteristics corresponding to the band limit value input from the band limit value converter 16. Data band-limited by the selected filter is supplied to the DCT circuit 3.

Alternatively, the brifilter 2 can also be configured as shown in FIG.

In this embodiment, the Brifilter 2 includes a filter 21, multipliers 22, 23, and an adder 24. The filter 21 has a reference value of 1, as shown in FIG.
It has a band limit value of 72.

After the input data in Table 1 is band-limited by the filter 21, it is input to the multiplier 22.
6 are multiplied by each coefficient.

Furthermore, the input data that has not passed through the filter 21 is sent to the multiplier 23.
16/16 to ○ corresponding to the level.
/16 are multiplied by each coefficient. The outputs of multipliers 22 and 23 are added by adder 24 and output to OCT circuit 3.

Even in this case, the same operation as shown in FIG. 5 can be performed.

The DCT circuit 3 divides the input data into blocks and orthogonally transforms the data. The orthogonally transformed data is input to the quantizer 4.

The quantizer 4 quantizes the input data by the number of quantization steps set by the quantization step converter 15 and outputs it to the encoder 5. The larger the number of quantization steps,
The amount of code becomes smaller. The encoder 5 performs run-length encoding or Huffman encoding on the quantized data. Alternatively, both run-length encoding and Huffman encoding are performed.

In the above, the input image data is input as is to the code amount predictor 1 and processed, but after motion compensation is performed in the motion compensation and difference data generator 6, the reference image data and the frame are It is also possible to generate a difference with the inter-predicted image data and input this difference data to the code amount predictor 1.

At this time, the CPU 8 makes each circuit and means process a plurality of frames or fields as one segment, as shown in FIG. In the case of this example, one frame consists of one frame encoded within the frame (intra) and three frames (first to third ink) motion-compensated for the difference data between the intra and intra frames. It is considered a segment.

For example, as shown in FIG. 9, if the code amount of the entire segment is 150 Kbits, the code amount of additional information such as motion vector information in the past segment is 30 Kbits.
Kb it s, the CPU 8 controls each circuit and means, and predicts that the additional information of the current segment will have the same amount of code, and changes it from 150 Kb to 30 Kb.
120 Kbits, which is obtained by subtracting i t s, is allocated to the code amount of data of each frame.

Then, the CPU 8 calculates the ratio of the first to third inks output by the absolute value sum calculation circuit 13 and the intra activity value, and allocates 120 Kbits in accordance with the ratio. For example, if these ratios are 1: l: l: 3
, 20Kb i for the first to third inks
t s and 60 Kbit s for intra, respectively. These values are output to the control data value transmitting circuit 17, whereby the first to third inters are respectively encoded with a target code amount of 20 Kbits.

Further, the CPU 8 connects the encoder 5 to the encoder 5 via the buffer memory 7.
The amount of code actually executed is monitored. Then, the error between the target value and the actual code amount is accumulated, and for the last encoded intra, the remaining code amount is output to the control data value transmitting circuit 17 as the target code amount. For example, the first
The actual code amount of each of the third to third inks is 25 Kbi.
ts.

When the data is 20Kbits and 25Kbits, the intra target code amount is 50Kbits.

Alternatively, as shown in FIG. 10, the target code amount may be set by applying a predetermined weight to the ratio of activity values. For example, when the ratio of activity values is 1: 1: 1: 3, if you uniformly multiply by 0.5 for ink and uniformly multiply by 1.5 for intra, the first to third 10Kbits to the internet
However, 90Kbits are allocated to each intranet.

In the above, 4 frames were considered as 1 segment, but
A plurality of blocks may constitute one group, and a plurality of groups may constitute one segment. In this case, one segment may be within one field or one frame, or may span multiple fields or multiple frames.

In addition, in the above, the sum of absolute values of the outputs of the operator processor 2 is calculated, but the sum of the square values of each output (sum of squares) or the average of the sum of square values @( The root mean square value) may also be calculated.

Furthermore, in the above description, both the band limit value and the number of quantization steps are controlled, but either one of them may be controlled.

[Effects of the Invention] As described above, according to the data compression device according to claim 1, since the code amount is predicted using past activity values, the statistical prediction can be performed quickly and easily. It is possible to predict the code amount while suppressing large variations.

Further, according to the data compression apparatus according to the second aspect, since data is processed in units of segments, the amount of code of a segment composed of a plurality of frames or a plurality of fields can be controlled to a constant value.

[Brief explanation of drawings]

FIG. 1 is a block diagram showing the configuration of an embodiment of a data compression device of the present invention, FIG. 2 is a block diagram showing the configuration of an embodiment of the code amount predictor in the embodiment of FIG. 1, and FIG. (a
) to (c) are diagrams explaining the operation of filter processing in the embodiment of FIG. 2, FIG. 4 is a diagram explaining the characteristics of the standard code electric value converter in the embodiment of FIG. 2, and FIG. FIG. 6 is a block diagram showing the configuration of an embodiment of the Buri filter shown in FIG. 1, and FIG.
This figure is a diagram for explaining the characteristics of the filter shown in FIG. 6, and FIGS. 8 to 10 are diagrams for explaining the processing of interframe prediction data. 1... Code amount predictor, 2... Buri filter, 3...
・DCT circuit, 4... quantizer, 5... encoder, 6
...Motion compensation and difference data generator, 7.Buffer memory, 8.CPU, 11.Frame memory, 12.Operator processor, 13.Absolute value sum calculation circuit, 14. ...Standard sign electric value converter, 15...
Quantization step converter, 16...band limit value converter,
17.--Control data immersion circuit 43, 21.. Filter, 22.23.. Multiplier, 24.. Adder.

Claims (2)

[Claims] (1) In a data compression device that divides input data into blocks of predetermined pixels, performs orthogonal transformation, quantizes, and further encodes the input data, a first processing means that performs filter processing on the input data; Regarding the data processed by the first processing means,
a calculation means for performing an operation including at least addition; a standard code amount value conversion means for converting an activity value output by the calculation means into a corresponding code amount predicted value in a standard quantization step or standard band limit value; and data amount. a target value setting means for setting a target value of , and comparing the target value set by the target value setting means and the code amount predicted value outputted from the standard code amount value converting means, and responding to the difference. a set value output means for outputting a quantization step value or a band limit value, and a second set value output means for quantizing or band limiting the data in accordance with the quantization step value or band limit value output by the set value output means. processing means, the past activity value outputted by the standard code amount value conversion means, the code amount predicted value in the standard quantization step or standard band limit value predicted from the past activity value, and the past storage means for storing at least one of the past code amount values encoded using the quantization step value or the band limit value set corresponding to the activity value; A data compression device, characterized in that the activity value is converted into the predicted code amount value using a value stored in a storage means. (2) In a data compression device that divides input image data into blocks of predetermined pixels, performs orthogonal transformation, quantizes, and further encodes the data, code amount prediction predicts the code amount of the input image data. means, processing means for limiting the band or quantizing the image data in accordance with the output of the code amount prediction means, and encoding for encoding the image data processed by the processing means. and control means for controlling the code amount prediction means so that a plurality of the blocks are treated as one group, a plurality of the groups are treated as one segment, and the code amount of each segment is constant. data compression device.