JPH05308328A - Data compression and demodulation system - Google Patents

Abstract

PURPOSE:To make intermediate compressed data not exceed actual data bit length and obtain large compression effect by comparing prediction data from a predictor with current data and obtaining the frequency of prediction up to the coincidence between both data in the form of compressed data. CONSTITUTION:In a register 1, 1st input data is inputted and stored. The predictor 5 sends a 1st predicted value in a prediction data group 7 to a comparator 6. The comparator 6 compares the predicted value with the 1st input data stored in the register 1. When they do not match each other, a counter 8 counts up by one and the 1st predicted value is removed from the prediction data group 7. Then the predictor 5 predicts and outputs a new predicted value from the data group in the updated prediction data group 7 and a comparator 6 makes a 2nd comparison. Then this comparing operation is repeated until the new predicted value matches the input data value in the register 1. When a coincidence is obtained, a gate 9 is opened and the counted value of the counter 8, i.e., intermediate data is sent to a Huffman encoder 10 and transmitted as encoded compressed data.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a data compression / demodulation system in which data is compressed and transmitted or recorded as compressed data, and the compressed data is demodulated and reproduced.

[0002]

2. Description of the Related Art The DPCM method, which is currently known as a method for compressing image data, predicts target pixel data from pixel data prior to the target pixel data, and predicts the predicted data value and the actual pixel data. This is a method in which the difference from the data value is compressed as new data using a Huffman code or the like, and then transmitted or recorded, and at the time of reproduction, the compressed data is demodulated.

[0003]

By the way, the above DPC
The M method predicts the current data based on the past data as described above, and it goes without saying that the performance of the predictor influences the compression rate, but the magnitude relationship between the predicted data value and the actual data value is large or small. Therefore, there is a possibility that the difference data value of these data values may take both positive and negative values, and it is necessary to manage the positive and negative states, which causes a reduction in the compression rate.

Further, considering the case where the predicted data value of the predictor is extremely different from the actual data value, the bit length of these difference data will be twice as long as the original source data. This leads to an increase in quantity.

The present invention has been made in view of the above problems, and solves the above problems, and even if the predicted data value of the predictor is extremely different from the actual data value, the intermediate compressed data is An object of the present invention is to provide a data compression / demodulation method that is highly efficient and has a large compression effect without exceeding the data bit length.

[0006]

In order to achieve the above object, a data compression / demodulation system according to the present invention compresses data and transmits or records it as compressed data, and demodulates and reproduces the compressed data. In the data compression / demodulation method, register means for storing past data, storage means for storing a prediction data group for predicting current data,
Prediction means for predicting current data from the prediction data group based on past data stored in the register means, selecting this, and outputting the selected data as prediction data, and this prediction means The prediction data is deleted from the prediction data group and the prediction is repeated until the output prediction data matches the current data, and the number-of-times information based on the number of times of the prediction until the prediction data matches the current data is displayed. In a data compression / demodulation system that comprises a data compression unit for converting compressed data, and / or compresses the data, transmits or records the compressed data, and demodulates and reproduces the compressed data. Register means for storing, storage means for storing a predicted data group for predicting the present data, and past memory stored in the register means. A prediction unit that predicts current data from the prediction data group based on the data, selects the selected data, and outputs the selected data as prediction data; and the prediction unit based on the number-of-times information of the compressed data. The data demodulation means for deleting the prediction data according to the above from the prediction data group, repeating the prediction, and outputting the prediction data obtained by this prediction as output data.

[0007]

According to the present invention, at the time of data compression, the data compression means deletes the prediction data from the prediction data group and repeats the prediction until the prediction data output by the prediction means matches the current data. Number of times based on the number of times of the prediction until the predicted data matches the current data is transmitted or recorded as compressed data, and at the time of data demodulation, the data demodulating means is based on the number of times of the compressed data. The prediction data by the prediction means is deleted from the prediction data group and the prediction is repeated, and the prediction data obtained by this prediction is output as output data.

[0008]

Embodiments of the present invention will be described below with reference to the drawings.

FIGS. 1 and 2 are block diagrams showing the configuration of a data compression / demodulation apparatus adopting a data compression / demodulation system according to an embodiment of the present invention.
FIG. 2 shows a data compression unit in the demodulation device, and FIG. 2 shows a data demodulation unit.

First, the data compression section will be described with reference to FIG.

The register 1 is a register for inputting and storing input data such as pixels from an external device or the like (not shown). The input data accumulated in the register 1 is input to the register 2 based on a signal from a comparator 6 described later.

The register 2 is a register capable of storing as much input data as is necessary for prediction in the predictor 5 described later, and the selector 4 is switched according to the number of input data stored in the register 2. The timing and the initial value 3 are changed. That is, for example, when predicting the current value from the past 10 pieces of input data, it is impossible to make a prediction when the number of the input data is 10 or less. Using the initial value 3, when the input data reaches 10 or more, the register 2 is selected by the selector 4. However, for the sake of simplification, the description will proceed here on the assumption that consecutive N (predictable number) past data are accumulated in the register 2 in advance.

The predictor 5 is a data group stored in the prediction data group 7 based on the consecutive N past data stored in the register 2 selected by the selector 4 or the initial value 3. A more predicted value is selected and sent to the comparator 6. As described above, the selector 4 initially selects the initial value 3,
Therefore, the predictor 5 first makes a prediction based on the initial value 3. Then, after the number of input data in the register 2 reaches a predetermined number, the selector 4 selects the register 2, and the predictor 5 also inputs the continuous N pieces (predictable) input to the register 2. It is designed to make predictions based on a large number of past data.

The comparator 6 has a function of comparing the predicted value predicted by the predictor 5 with the input data stored in the register 1. When no match is found in this comparison, the counter 8 is incremented by "1" and the prediction data group 7 is controlled so as to remove the prediction value from the prediction data group 7. There is.

When the above-mentioned coincidence is found in the comparator 6, the gate 9 is opened and the count value in the counter 8 is sent to the Huffman encoder 10. Further, the count value in the counter 8 and the data group in the predicted data group 7 are initialized, new input data is stored in the register 1, and the register 2 is stored in the register 1. The first input data is stored.

The Huffman encoder 10 is an encoder for performing variable length coding using a Huffman code. The count value of the count 8 input from the count 8 is variable length coded as compressed data. It is adapted to transmit to a transmission line (not shown).

Next, the operation of the data compression section will be described.

When the input data (first input data) of the first pixel or the like is input to the register 1, the register 1
Stores this input data. At this time, the selector 4 selects the initial value 3 side, and the predictor 5 uses the prediction value of the first pixel or the like from the prediction data group 7 based on this to make a comparison with the comparator 6.
To send to.

The comparator 6 compares the predicted value obtained by the predictor 5 with the first input data stored in the register 1, and if they do not match, the counter 8 outputs "". The value is incremented by "1" and the predicted value is removed from the predicted data group 7. Thereafter, the predictor 5 predicts and outputs a new predicted value from the updated data group in the predicted data group 7, and the comparator 6 performs the second comparison. Then, the comparison operation by the comparator 6 is repeated until a new predicted value and the input data value in the register 1 coincide with each other in the comparator 6.

After the above-mentioned coincidence is found in the comparator 6, the gate 9 is opened, the count value in the counter 8 is sent to the Huffman encoder 10 and is Huffman-encoded, and the compressed data is transmitted through the transmission line (see FIG. (Not shown).

Thereafter, the count value in the counter 8 and the data group in the predicted data group 7 are initialized, the second input data is stored in the register 1, and the register 2 is further stored in the register 2. The first input data stored in the register 1 is input.

In summary, the above operation is repeated N + 1 times so that the register 2 has 1
Nth data is stored and N + 1 is stored in the register 1.
The data of item th will be stored. At this time, the selector 4 is switched to the register 2 side, and thereafter,
The predictor 5 predicts and outputs a predicted value from the N pieces of data, and repeats the above-described operation to output intermediate data corresponding to the number of times it takes to sequentially predict the hit, that is, the count value of the count 8. The Huffman encoder 10 performs Huffman coding and sends the compressed data.

Next, the data demodulation section will be described with reference to FIG. As in the case of the description of the data compression unit, the register 20 in FIG. 2 stores N consecutive past data, and the predictor 17 makes a prediction based on the data in the register 20. As described below.

The register 11 is a register for inputting and storing the compressed data output from the data compression section.

The Huffman decoder 12 is a decoder having the same coding function as that of the Huffman encoder 10, and demodulates the compressed data stored in the register 11 using a Huffman code, and the demodulated data. Is set as intermediate data in the counter 13.

The predictor 17 selects the consecutive N stored in the register 20 selected by the selector 16.
A predicted value is selected from the data group accumulated in the predicted data group 19 based on the past data or the initial value 15 and is transmitted to the gate 18. Like the selector 4 in the data compression unit, the selector 16 first selects the initial value 15 and guides the initial value 15 to the predictor 17, and after the second prediction. Selects the register 20 and the register 20
The continuous N past data input to the above are led to the predictor 17.

The comparator 14 detects the output value of the counter 13 when the predictor 17 outputs the predicted value,
It has a function of controlling opening / closing of the gate 18 by this output value. Then, the output value of the counter 13 is "0".
When the preset value is “0”, the gate 18 is opened and the predicted value from the predictor 17 is completely demodulated as an output data to an external device (not shown).
And the predicted value is stored in the register 20. At this time, the predicted data group 19 is initialized.

On the other hand, the output value of the counter 13 is "0".
If not, the comparator 14 closes the gate 18, the counter 13 is decremented by 1, and the predicted value is deleted from the predicted data group 19.

Next, the operation of the data demodulation section will be described.

When the first compressed data (first compressed data) is input to the register 11 from the data compression unit, the register 11 stores the first compressed data and sends it to the Huffman decoder 12. The Huffman decoder 12
, Demodulates the first compressed data and sends it to the counter 13 as intermediate data. Then, the intermediate data is preset in the counter 13.

At this time, the selector 16 selects the initial value 15, and the predictor 17 selects and outputs the predicted value from the predicted data group 19 based on the initial value 15. At the time of this output, if the output value of the counter 13 is "0", the predicted value passes through the gate 18,
The completely demodulated output data is sent to an external device or the like and is stored in the register 20. At the same time, the prediction data group 19 is initialized.

On the other hand, the predictor 17 uses the prediction data group 19
When selecting and outputting the predicted value from the
When the output of 3 is not "0", the counter 13 is decremented and the predicted value is deleted from the predicted data group 19. After that, the predictor 17 makes the next prediction, and this prediction is repeated until the output of the counter 13 becomes "0", and the completely demodulated data is transmitted.

After demodulating one data in this way, the register 11 stores the next compressed data (second compressed data).
Is input, and the above-described prediction / demodulation operation is repeated N times. Then, when N pieces of completely demodulated data are stored in the register 20 and the N + 1th piece of data is input, the selector 16 is switched to the register 20 side and this data is used for subsequent prediction. Will be done.

That is, the data demodulation unit includes a decoder having the same coding function as the Huffman encoder in the data compression unit, performs Huffman demodulation on the input compressed data, and corresponds to the number of repetitions for prediction. Get the value of the intermediate data. Prediction is performed based on the source data that has already been completely demodulated, and the predicted value is excluded from the source data group to be predicted until the number of repetitions corresponding to the intermediate data reaches about 1 less repeat. Then, when the number of times indicated by the intermediate data is reached, the predicted value is used as source data that can be completely demodulated for subsequent data demodulation.

In the data compression / demodulation method of this embodiment, initial values are set in both the compression section and the demodulation section, and the first data is predicted from the initial values. It is also possible to use only the first input data as it is as intermediate data or first compressed data without setting the initial value.

[0036]

As described above, according to the present invention, even if the predicted data value of the predictor is extremely different from the current data value, the compressed data does not exceed the bit length of the current data and has a high value. It is possible to provide a data compression / demodulation method that is efficient and has a large compression effect.

[Brief description of drawings]

FIG. 1 is a block diagram showing the configuration of a data compression unit in a data compression / demodulation apparatus that employs a data compression / demodulation system that is an embodiment of the present invention.

FIG. 2 is a block diagram showing a configuration of a data demodulation unit in a data compression / demodulation apparatus that employs the data compression / demodulation method of the above embodiment.

[Description of Codes] 1, 2 ... Register 3 ... Initial value 4 ... Selector 5 ... Predictor 6 ... Comparator 7 ... Predicted data group 8 ... Counter 9 ... Gate 10 ... Huffman encoder

Claims (3)

[Claims] 1. A data compression / demodulation system in which data is compressed and transmitted or recorded as compressed data, and the compressed data is demodulated and reproduced, and register means for storing past data and current data are predicted. For predicting the present data from the above-mentioned prediction data group based on the past data stored in the register means and the storage means for storing the prediction data group, and selecting the selected data as the prediction data. And a prediction unit that outputs the prediction data and the prediction data output from the prediction unit matches the current data, deletes the prediction data from the prediction data group and repeats the prediction, and the prediction data matches the current data. Data compression / demodulation, which comprises: Formula. 2. A data compression / demodulation system in which data is compressed and transmitted or recorded as compressed data, and the compressed data is demodulated and reproduced, and register means for storing past data and current data are predicted. For predicting the present data from the above-mentioned prediction data group based on the past data stored in the register means and the storage means for storing the prediction data group, and selecting the selected data as the prediction data. And a prediction unit that outputs the prediction data obtained by this prediction based on the number information of the compressed data and deleting the prediction data by the prediction unit from the prediction data group and repeating the prediction. A data compression / demodulation system characterized by comprising demodulation means. 3. The compressed data is variable length coded information based on the number of predictions until the prediction data matches the current data.
Data compression / demodulation method described.