JPH06232758A - Data compression circuit - Google Patents

Abstract

PURPOSE:To obtain a compression circuit in which data quantity is remarkably reduced. CONSTITUTION:Output data DH from an A/D converter 1 are fed to a detection means 21, in which a difference between adjacent data is detected. When the difference indicates a negative value, the difference is outputted in a form of 2's complement. Output data DH1 of the detection means 21 are fed to a code bit shift means 22. When the MSB of the data DH1 is logical 1, after the 2's complement is obtained, the result is shifted by one bit toward the high- order bit and a sign bit of logical 1 is added to the LSB. When the MSB of the data DH1 is logical 0, the result is shifted by one bit toward the high-order bit and a sign bit of logical 0 is added to the LSB. Output data DH2 of the shift means 22 are fed to a correction means 23, and when only middle datum among three consecutive data DH differs, the middle datum is set equal to the succeeding and preceding data substantially. Output data DH3 of the correction means 23 are fed to a common processing means 24, in which the data are classified into plural bit data based on valid bits and the result is subjected to common processing.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a data compression circuit suitable for use in a Holter system (long-time electrocardiographic recorder).

[0002]

2. Description of the Related Art Conventionally, it has been known to record an electrocardiographic waveform signal from an electrocardiographic measuring device on a magnetic tape or the like in the form of an analog signal. However, there are problems such as poor reproduction accuracy due to head characteristics and the like. Therefore, conventionally, in order to improve the reproducibility, it has been proposed to record an electrocardiographic waveform signal on a magnetic tape, a semiconductor memory or the like in the form of a digital signal.

[0003]

By the way, the Holter system, for example, continuously records electrocardiographic waveform signals for 24 hours, has a very large amount of recorded data, and requires a recording medium suitable for it. There was a problem such as becoming.

Therefore, an object of the present invention is to provide a data compression circuit capable of significantly reducing the amount of data.

[0005]

According to the present invention, there is provided adjacent data difference detecting means for detecting a difference between adjacent data of input data, and when the most significant bit of output data of the adjacent data difference detecting means is a logical one. Shifts one bit to the upper bit side after obtaining the two's complement, adds a code bit of logic 1 to the least significant bit, and when the most significant bit of the output data is logical 0, it is 1 to the upper bit side as it is. Code bit moving means for bit-shifting and adding a code bit of logical 0 to the least significant bit, and output data of this code bit moving means are classified into a plurality of bit data having different effective bits, and data common processing is performed. And a data commonizing means.

[0006]

In the present invention, the adjacent data difference detecting means 2
1 and the sign bit moving means 22 process the similar data so that the similar data becomes large, and then the data commonizing means 24 collects the similar data and makes the data common, so that the data amount is significantly reduced. obtain. Also,
Data similar to the compressed data is output from the data commonizing means 24, so that the Huffman method and LZW (La
The compression efficiency can be increased when data compression is performed using a compression method capable of 100% expansion such as the mple-Ziv-Welch) method.

[0007]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described below with reference to the drawings. This example is an example applied to a recording system of a Holter system for recording an electrocardiographic waveform signal with a digital VTR.

FIG. 1 is a block diagram showing the overall configuration of the Holter system. In the figure, an analog electrocardiographic waveform signal SH output from an electrocardiographic waveform measuring device (not shown)
Is supplied to the A / D converter 1, and the sampling frequency is 2
It is converted into a digital signal at 50 Hz and 10 bits per sample. The data of each sample is treated as 16-bit data in the following processing.

The electrocardiographic waveform data DH output from the A / D converter 1 is subjected to data compression processing by the data compression circuit 2 and then supplied to the digital VTR 3 for recording. The data compression circuit 2 is configured by including a microcomputer (hereinafter referred to as "microcomputer"), and the data compression processing is performed by software with the microcomputer.

FIG. 2 is a functional block diagram showing the configuration of the data compression circuit. In the figure, the electrocardiographic waveform data DH output from the A / D converter 1 is supplied to the adjacent data difference detecting means 21. The detection means 21 detects the difference between adjacent data of the data DH. In this case, when the difference is positive, the detection means 21 outputs it as it is, and when the difference is negative, it outputs it in 2's complement.

Next, the data DH1 output from the detecting means 21 is supplied to the sign bit moving means 22. In this moving means 22, the data DH1 (b15, b14, ...
16-bit data of b0) most significant bit (MSB) b1
It is processed as follows according to the logic state of 5.

That is, the most significant bit b of the data DH1
When 15 is a logic 1, the 2's complement of the data DH1 is obtained, and then one bit is shifted to the upper bit side and the code bit of the logic 1 is added to the least significant bit (LSB) b0.

For example, when the data DH1 is "FFFFH", the bit b15 is a logical 1 and the complement of 2 is "0001H", which is shifted to the upper bit side by 1 bit and then the bit b0 is a logical 1 code. When the bit is added, it becomes “0003H”. Here, “H” indicates hexadecimal display. The same applies to the following.

On the other hand, when the most significant bit b15 of the data DH1 is a logical 0, it is shifted by 1 bit to the upper bit side as it is and a code bit of logical 0 is added to the least significant bit b0.

For example, when the data DH1 is "0001H", the bit b15 is a logical 0, and when the sign bit of the logical 0 is added to the bit b0 after shifting 1 bit to the upper bit side, it becomes "0002H".

FIG. 3 shows the correspondence between the data DH1 output from the detecting means 21 and the data DH2 output from the moving means 22.

Next, the data DH2 output from the moving means 22 is supplied to the data correcting means 23. Correction means 2
3, when only the central data among the three consecutive data of the data DH output from the A / D converter 1 is different, the processing that substantially makes the central data the same as the preceding and subsequent data. Is performed.

For example, the data DH2 is 0002H 000.
When it is continuous with 3H or 0003H and 0002H, it is 0000H 0000H.

As described above, when the sampling frequency in the A / D converter 1 is set to 250 Hz, 125 which is half the sampling frequency is 125 according to the well-known sampling theorem.
Can be completely restored up to Hz. However, in the case of the electrocardiographic waveform, it is not considered that the frequency band up to 125 Hz is necessary, and 125 H is obtained by the above-mentioned processing by the correction means 23.
The frequency component of z can be removed, thereby increasing similar data.

The correction processing as described above is performed by the A / D
Of course, the data DH output from the converter 1 or the data DH1 output from the detection means 21 can be used. When the data DH output from the A / D converter 1 is processed, for example, 03F3H 03F4
When continuous with H 03F3H, 03F3H 03F3H0
It will be 3F3H. When the data DH1 output from the detection means 21 is processed, for example, FFFFH
When continuous with 0001H or 0001H FFF
When it continues to FH, it is 0000H 0000H.

The correction process by the correction means 23 is not always necessary and may be omitted.

Next, the data DH3 output from the correction means 23 is supplied to the data commonization means 24 and
Is processed.

The data DH3 is classified into three types of data of valid 4 bits, valid 8 bits, and valid 16 bits based on the valid bits. For example, "0004
"H", "0005H", etc. are effective 4-bit data,
“0080H”, “0090H”, etc. are valid 8-bit data, and “0100H”, “0200H”, etc. are valid 16 bits.
It is bit data. When this classification is performed, a header is placed before each type of data so that the content of each type can be understood.

Here, the header is, for example, bits b7 to b0.
1 byte data. The header has the following meanings, for example. That is, 2 of "b7b6"
The type of data is represented by bits. For example,
“00” indicates valid 4-bit data, “01” indicates valid 8-bit data, and “10” indicates valid 16-bit data. Also, the 6 bits of "b5 to b0" represent the number of chunks of bit data. The number is 01H to 3FH
(1 to 63) can be represented.

The effective 4-bit data is in 8-bit units. In this case, if the same data is not continuous, bits b15 to b8 are removed to obtain 8-bit data. When the same data continues, 8-bit data "b7-b0" is formed as follows. That is, the continuous number of the same data is represented by 4 bits of "b7 to b4". Here, 0H to FH (0 to 15) are shown, but it is considered that 0 to 15 represents 1 to 16. Also, 4-bit data is represented by 4 bits of "b3 to b0".

For example, 0004H 0005H 0005H
The data of 0005H 0005H 0005H 0006H becomes 04H 45H 06H.

The effective 8-bit data is in 8-bit units. In this case, bits b15 to b8 are removed and 8
It is treated as bit data.

For example, 0010H 0020H 0030H
The data of 0040H 0050H 0060H is 10H 2
It becomes 0H 30H 40H 50H 60H.

As described above, in the data commonizing means 24,
The processing of is performed. For example, 0004H 0005
H 0005H 0005H 0005H 0005H 0006
The data of H 0080H 0090H 0100H 0200H is 03H 04H 45H 06H 42H 80H 90H 82
It becomes H 0100H 0200H. Where 03H, 42
H and 82H are headers.

Next, the data DH4 output from the commonizing means 24 is lossless compression means 25.
Is supplied to. In the compression means 25, data compression processing is performed using a compression method capable of 100% expansion such as the conventionally known Huffman method or LZW (Lample-Ziv-Welch) method.
The data DH5 output from the compression means 25 is supplied to the digital VTR 3 as the output of the data compression circuit 2.

As described above, in this example, the detection means 2
1, the moving means 22 and the correcting means 23 are processed to increase the similar data, and then the commonizing means 2 is used.
Since the similar data are collected by 4 and the data is made common, the amount of data can be significantly reduced. Further, since similar data is output from the commonizing means 24 as compressed data, the Huffman method or LZ in the compressing means 25 is output.
The compression efficiency of data compression by a compression method such as the W (Lample-Ziv-Welch) method can be increased.

For example, when the output data DH of the A / D converter 1 is as shown in FIG. 4, the detecting means 21 and the moving means 2
2. Output data D of the correction means 23 and the commonization means 24
H1 to DH4 are as shown in FIGS. In this case, the input 768 bytes (see FIG. 4) becomes 191 bytes (see FIG. 8) by the processing of the detecting means 21 to the commonizing means 24, which means that 191 / 768≈1 / 4 compression is performed. It should be noted that FIGS. 4 to 8 and FIG. 12 to be described later do not include the symbol “H”, but they are displayed in hexadecimal.

Also, the electrocardiographic waveform signal has two channels, the sampling frequency is 250 Hz, one sample is 10 bits (actually, it is treated as 16 bits data), and the number of input data is 52838 in a Holter system operating for 24 hours.
When it is 4 bytes, the number of data after compressed by the data compression circuit 2 of this example is 59707 bytes, and the compression ratio is 8.8.
It became 5. However, this is the case where the lossless compression means 25 uses the LZW method.

In the reproducing system, as shown in FIG. 9, the reproduced data of the digital VTR 3 is decompressed by the data decompressing circuit 4 and then output as an analog signal by the D / A converter 5. The decompression processing in the data decompression circuit 4 is shown in FIG.
That is, the reverse processing to the processing by each means is performed. However, the reverse process of the correction means 23 is omitted.

FIG. 10 shows an electrocardiographic waveform obtained by the compression-processed electrocardiographic waveform signal of this example, which is almost the same as the electrocardiographic waveform obtained by the non-compressed electrocardiographic waveform signal shown in FIG. The compression process does not adversely affect the reproduction accuracy.

Further, as described above, the correction means 23 need not always be provided. 12 shows that the output data DH of the A / D converter 1 is as shown in FIG.
The output data DH4 of the data commonizing means 24 in the case where is not provided is shown. In this case, the input 768 bytes (see FIG. 4) is 23 by the processing of the detecting means 21, the sign bit moving means 22 and the data commonizing means 24.
9 bytes (see Fig. 12), 239 / 768≈1 /
It means that the compression of 3.2 has been done.

As described above, when the correction means 23 is not provided, the compression rate is lower than when the correction means 23 is provided. This is because the correction means 23 is not provided, so the commonization means 24
It is considered that similar data is reduced in the data DH3 input to the.

The above embodiment is applied to a Holter system for recording an electrocardiographic waveform signal, but the present invention has many other signals, for example, an electrocardiographic waveform signal, which has a small level change. It is suitable for application to data compression processing when recording a signal.

[0039]

According to the present invention, after the similar data is processed by the adjacent data difference detecting means and the sign bit moving means so that the similar data is increased, the similar data is collected by the data commonizing means. Since they are shared, the amount of data can be significantly reduced. Further, since similar data is output as the compressed data from the data unifying means, the compression efficiency can be increased when the data is further compressed by using the compression method capable of 100% expansion such as the Huffman method and the LZW method.

[Brief description of drawings]

FIG. 1 is a diagram showing an overall configuration of an embodiment of the present invention.

FIG. 2 is a functional block diagram showing a configuration of a data compression circuit.

FIG. 3 is a diagram showing a correspondence relationship between output data DH1 of adjacent data difference detecting means and output data DH2 of sign bit moving means.

FIG. 4 is a diagram showing an example of output data DH of an A / D converter.

FIG. 5 is a diagram showing output data DH1 of the adjacent data difference detection means.

FIG. 6 is a diagram showing output data DH2 of the sign bit moving means.

FIG. 7 is a diagram showing output data DH3 of the data correction means.

FIG. 8 is a diagram showing output data DH4 of the data commonizing means.

FIG. 9 is a diagram showing a configuration of a reproduction system.

FIG. 10 is a diagram showing an electrocardiographic waveform based on data subjected to compression processing.

FIG. 11 is a diagram showing an electrocardiographic waveform based on data that has not been compressed.

FIG. 12 is a diagram showing output data DH4 (when there is no data correction means) of the data commonization means.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 A / D converter 2 Data compression circuit 3 Digital VTR 21 Adjacent data difference detection means 22 Code bit moving means 23 Data correction means 24 Data sharing means 25 Lossless compression means

Claims (1)

[Claims] 1. Adjacent data difference detecting means for detecting a difference between adjacent data of input data, and when the most significant bit of the output data of the adjacent data difference detecting means is a logic 1, after obtaining a two's complement When 1 bit is shifted to the upper bit side and a code bit of logic 1 is added to the least significant bit, and when the most significant bit of the output data is a logical 0, 1 bit is shifted to the upper bit side as it is and to the least significant bit A code bit moving means for adding a code bit of logic 0 and a data commonizing means for classifying the output data of the code bit moving means into a plurality of bit data each having a different effective bit and performing a data commonizing process are provided. A data compression circuit characterized by the above.