TW201318354A - A guaranteeing-quality signal compression method and system without the use of inverse process - Google Patents

Abstract

The present invention discloses a guaranteeing-quality signal compression method and system without the use of inverse process. The system comprises a time domain-transform domain converter, a quantizer, and a lossless compressor. The invention can be used to realize guaranteeing-quality data compression without the complex inverse process, which has the better compression efficiency and simpler error control process.

Description

Quality assurance signal compression method and system thereof without reverse processing

The invention relates to a signal compression method and a system thereof, in particular to a signal compression method and a system thereof which do not use reverse processing, and replace the traditional inverse conversion process by using a simplified distortion index (SPRD1 and SPRD2), thereby simplifying Distortion control process and the cost savings of hardware.

For signal-type information, Transform coding utilizes the irrelevance of coefficients in the transform domain, and proposes a data compression method that maintains the concept of features, which leads to a great leap forward in data compression technology and performance. Wavelet-based data compression methods have the best compression performance in conversion coding, so they are also the most famous and widely used. However, the conversion coding is a distortion compression method, which must be based on the distortion measurement. Otherwise, unrestricted compression may result in unrecognizable results. Therefore, a distortion index is needed for objective comparison. Distortion indicators and compression ratios are two contradictory parameters, and distortion compression must seek a compromise between the two. Recently, due to the great progress of communication and storage hardware technology, the cost of construction of communication hardware has been effectively reduced, and the quality of reconstruction has been gradually taken seriously. This demand has driven the quality assurance mechanism to maintain high reconstruction quality and become a must-have function of the data compression system. An electrocardiogram is a type of medical signal that is particularly demanding on quality assurance mechanisms.

The heart is a three-dimensional organization. The electrocardiogram measured at any point on the surface of the body is the sum of the electrical signals of the heart tissue. Therefore, a single-lead ECG can only provide a rough heartbeat record and cannot provide useful information for the diagnosis of heart disease. A typical clinical requirement is to record a 12-lead one-day electrocardiogram (such as the Holter system). If you want to establish a detailed dynamic electrical signal pattern for each atrium and ventricle, you must deploy more leads.

Care for heart disease usually requires long-term and immediate observation. Using a miniaturized portable sensor with wireless transmission or direct recording is the most convenient and feasible mobile monitoring mode. This mode requires an instant multi-lead ECG compression system with quality assurance and low power. The quality assurance mechanism must measure the degree of distortion by means of a distortion measurement indicator. In the compression of ECG data, the percentage root-mean-square difference (PRD) is usually used as the distortion index. The values are defined as follows:

among them And S _j are the reconstructed and original signals respectively, and N is the data length of the signal. The above formula indicates that the calculation of the PRD value requires the use of the reconstructed signal. However, the compression step is performed in the conversion domain, which means that the conversion coding compression method must go through the full reverse conversion process to obtain the PRD value. Wavelet-based electrocardiogram (ECG) data compression system under the set partitioning in hierarchical trees (SPIHT) architecture requires three inverse conversion processes; ie, reverse SPIHT, inversion Quantifying and inverting wavelet transforms, the implementation cost of both hardware and software is high, and it is not economical.

The data compression system with quality assurance mechanism has two parts: hardware and software. Figure 1 shows the distortion measurement loop of a traditional Wavelet-based Electrocardiogram (ECG) data compression system, where QF is represented as a quantization scale control parameter. The hardware cost required for this distortion-controlled loop architecture is approximately twice that of a simple encoding process (ie, wavelet transform, quantization processing, and distortion-free encoding). Therefore, the key to simplifying the hardware process architecture is to avoid the use of reconstructed data.

For the sake of the job, the inventors of the present invention carefully developed and developed a signal compression method and system that can realize the quality assurance mechanism without using reverse processing. The hardware cost of the new method is only the traditional system architecture (as shown in the figure). 1) Half of the hardware cost. The invention helps to eliminate complicated operation steps such as inverting SPIHT, inversion quantization and inverse wavelet transform, and obtains a simplified PRD calculation flow. In addition, the present invention can be further developed into a portable multi-lead electrocardiograph recording device.

The object of the present invention is to provide a quality assurance signal compression system that does not use reverse processing, which does not use a signal compression method of complicated reverse processing steps, and thus has a better compression efficiency and a lower complexity distortion control flow.

Another object of the present invention is to provide a quality assurance signal compression method that does not use reverse processing, and can replace the reverse conversion process required in the conventional method by using only the simplified distortion index (SPRD1 and SPRD2), thereby saving half of the hardware cost. .

In accordance with the purpose of the present invention, a quality assurance signal compression system that does not use reverse processing is provided, comprising: a time domain-conversion domain converter, a quantization processor, a conversion domain distortion indicator calculator, and a distortionless compressor. The time domain-conversion domain converter is configured to convert the time domain signal to the conversion domain to obtain a conversion coefficient and a rounding error of a conversion coefficient; the quantization processor is configured to process the conversion coefficient to obtain a quantization coefficient and a Quantization error; and the conversion domain distortion index calculator is used to calculate a rounding error of the conversion coefficient plus a simplified distortion index value (SPRD1) of the quantization error; and the distortionless compressor is to perform distortion without distortion coding.

According to a feature of the signal compression system of the present invention, wherein the reduced distortion index value (SPRD1) falls within the allowable error range, the distortionless compressor can perform distortionless encoding on the quantized coefficient; otherwise, the quantization process The device needs to perform quantization processing again to obtain a new quantization error, and calculates the new quantization error plus the simplified distortion index value (SPRD1) of the rounding error of the conversion coefficient via the conversion domain distortion index calculator.

According to a feature of the signal compression system of the present invention, wherein the reduced distortion index value (SPRD1) ≧ 1%, the simplified distortion index value (SPRD1) can replace the root mean square (PRD) of the difference value.

According to the purpose of the present invention, a quality assurance signal compression system that does not use reverse processing is further provided, which includes: a time domain-conversion domain converter, a quantization processor, a conversion domain distortion index calculator, and a distortionless compressor. The time domain-conversion domain converter is configured to convert the time domain signal to the conversion domain to obtain a conversion coefficient; the quantization processor is configured to process the conversion coefficient to obtain a quantization coefficient and a quantization error; and the conversion domain The distortion index calculator is used to calculate a simplified distortion index value (SPRD2) of the quantization error; and the distortionless compressor is to perform distortionless encoding on the quantized coefficient.

According to a feature of the signal compression system of the present invention, wherein the reduced distortion index value (SPRD2) falls within the allowable error range, the distortionless compressor can perform distortionless encoding on the quantized coefficient; otherwise, the quantization process The device needs to be re-quantized to obtain a new quantization error, and the simplified distortion index value (SPRD2) of the new quantization error is calculated via the conversion domain distortion index calculator.

According to a feature of the signal compression system of the present invention, wherein the reduced distortion index value (SPRD2) ≧ 2%, the simplified distortion index value (SPRD2) can replace the root mean square (PRD) of the difference value.

According to another object of the present invention, a quality assurance signal compression method that does not use reverse processing mainly includes the following steps: converting a time domain signal to a conversion domain by using a time domain-conversion domain converter to obtain a conversion coefficient and a conversion. The rounding error of the coefficient; processing the conversion coefficient by a quantization processor to obtain a quantized coefficient and a quantization error; calculating a rounding error of the conversion coefficient by using a conversion domain distortion index calculator plus a simplified distortion index value of the quantization error (SPRD1); and the distortion coefficient is encoded without distortion by a distortionless compressor.

According to a feature of the signal compression method of the present invention, wherein the simplified distortion index value (SPRD1) ≧ 1%, the simplified distortion index value (SPRD1) can replace the root mean square (PRD) of the difference value.

According to another object of the present invention, a quality assurance signal compression method that does not use reverse processing mainly includes the following steps: converting a time domain signal to a conversion domain by using a time domain-conversion domain converter to obtain a conversion coefficient; The processor processes the conversion coefficient to obtain a quantized coefficient and a quantization error; calculates a reduced distortion index value (SPRD2) of the quantization error by using a conversion domain distortion index calculator; and performs the quantized coefficient in a distortionless compressor No distortion coding.

According to one feature of the signal compression method of the present invention, wherein the simplified distortion index value (SPRD2) ≧ 2%, the simplified distortion index value (SPRD2) can replace the root mean square percentage (PRD) of the difference value.

The present invention replaces conventional inverse conversion steps such as inverting SPIHT, inverting quantization, and inverting wavelet conversion with a simplified distortion indicator. In addition, the distortion index calculated by the signal compression system and method of the present invention can completely replace the root mean square percentage (PRD) of the traditional distortion index-difference value.

The effect of the invention:

1. Can be used to instantly record multi-path ECG compression data with quality assurance features.

2. Highly capable of preserving clinical diagnostic information.

3. The design of the system is considered to be hardware-based, and can be further developed into a portable multi-lead ECG recording device.

4. With the lowest hardware cost and the lowest power consumption.

The above and other objects, features, and advantages of the present invention will become more apparent and understood.

While the invention may be embodied in various forms, the embodiments illustrated in the drawings It is not intended to limit the invention to the particular embodiments illustrated and/or described.

FIG. 2 shows a quality assurance signal compression system 100 that does not use reverse processing according to the present invention. Referring now to FIG. 2a, the method includes: a time domain-conversion domain converter 101, a quantization processor 102, and a conversion domain distortion. The indicator calculator 103 and a distortionless compressor 104. The time domain-conversion domain converter 101 is configured to convert the time domain signal to the conversion domain to obtain a conversion coefficient 105 and a rounding error 108 of a conversion coefficient, and the time domain-conversion domain converter 101 is a wavelet converter. A one-dimensional non-returning wavelet converter and a reversible one-dimensional non-returning discrete periodic wavelet converter. The quantization processor 102 of the present invention is configured to process the conversion coefficient 105 to obtain a quantization coefficient 106 and a quantization error 107, and the quantization processor 102 is a quantization processor 102 that can obtain an approximate linear distortion data compression effect. . Next, the distortionless compressor 104 performs distortionless encoding on the quantized coefficients 106. The distortionless compressor 104 performs signal distortion-free compression using one of a discrete wavelet coding algorithm and a Set Partitioning in Hierarchical Trees (SPIHT). In the first embodiment of the present invention, the conversion domain distortion index calculator 103 is used to calculate the rounding error 108 of the conversion coefficient and the simplified distortion index value (SPRD1) of the quantization error 107. It should be noted that when the simplified distortion index value (SPRD1) falls within the allowable error range, the distortion-free compressor 104 can perform the distortion-free encoding on the quantized coefficient 106; otherwise, the quantization processor 102 needs to perform the quantization again. The processing obtains a new quantization error 107, and calculates the new quantization error 107 plus the reduced distortion index value (SPRD1) of the rounding error 108 of the conversion coefficient via the conversion domain distortion index calculator 103. When the simplified distortion index value (SPRD1) is %1%, the simplified distortion index value (SPRD1) can replace the root mean square (PRD) of the difference value.

Referring now to FIG. 2b, the second embodiment of the present invention differs from the first embodiment only in that the time domain-conversion domain converter 101 is configured to convert the time domain signal to the conversion domain to obtain a conversion coefficient 105, and The conversion domain distortion index calculator 103 is only used to calculate the simplified distortion index value (SPRD2) of the quantization error 107. When the simplified distortion index value (SPRD2) falls within the allowable error range, the distortion-free compressor 104 can perform the distortion-free encoding on the quantized coefficient 106; otherwise, the quantization processor 102 needs to perform the quantization process again to obtain a The new quantization error 107 is calculated, and the simplified distortion index value (SPRD2) of the new quantization error 107 is calculated via the conversion domain distortion index calculator 103. When the simplified distortion index value (SPRD2) ≧ 2%, the simplified distortion index value (SPRD2) can replace the root mean square (PRD) of the difference value.

3 is a quality assurance signal compression method 200 of the present invention that does not use reverse processing. Referring now to FIG. 3a, which is an embodiment of the present invention, and mainly includes the following steps: using a time domain-conversion domain The converter 101 converts the time domain signal to the conversion domain to obtain a conversion coefficient 105 and a rounding error 108 of the conversion coefficient; then, the quantization coefficient 102 is processed by a quantization processor 102 to obtain a quantized coefficient 106 and a quantization error 107; And using a conversion domain distortion index calculator 103 to calculate the rounding error 108 of the conversion coefficient plus the simplified distortion index value (SPRD1) of the quantization error 107; finally, the quantization coefficient 106 is performed by a distortionless compressor 104. Distortion coding. In addition, the present invention also proposes another quality assurance signal compression method 200 that does not use reverse processing. Referring now to FIG. 3b, the difference from the above steps is only that the time domain signal is converted by the time domain-conversion domain converter 101. The conversion domain is obtained to obtain a conversion coefficient 105, and the conversion domain distortion index calculator 103 calculates only the simplified distortion index value (SPRD2) of the quantization error 107. In an embodiment, an electrocardiogram original signal is converted by the dimensional non-recursive wavelet transform of the time domain-conversion domain converter 101 to obtain the conversion coefficient 105 and generate a rounding error 108 (fre) of the conversion coefficient. Next, the conversion coefficient 105 (wavelet coefficient) is quantized by the quantization processor 102. This step will result in a truncation error (te _j ), which is te _j shown in Fig. 3; this truncation error (te _j ) multiplied by the quantization scale (determined via the quantization scale control parameter (QF)) is the quantization error 107 ( Qe'). Next, the conversion domain distortion index calculator 103 performs an operation of the error control loop to determine whether the distortion degree falls within the allowable range. Taking the simplified distortion index value (SPRD1) as an example, when the distortion falls within the allowable error range, the distortion-free compressor 104 can perform the distortion-free encoding on the quantized coefficient 106; when the distortion falls outside the allowable range At this time, a quantization scale adjustment is performed, and a new quantization scale is obtained through a quantization scale control parameter (QF), and then the simplified distortion index value (SPRD1) is recalculated; in short, the quantization processor 102 needs to be re-re- The quantization process is performed to obtain a new quantization error 107, and the new quantization error 107 is added via the conversion domain distortion index calculator 103 to add a simplified distortion index value (SPRD1) of the rounding error 108 of the conversion coefficient, and the process is as follows. The figure shows. Similarly, when the conversion domain distortion index calculator 103 used only calculates the simplified distortion index value (SPRD2) of the quantization error 107, when the distortion degree falls outside the allowable range, the quantization value is also repeated. The simplified distortion index value of the quantization error 107 is re-calculated by a quantization scale control parameter (QF) until the simplified distortion index value (SPRD2) falls within the allowable range. In short, the quantization processor 102 needs to perform the quantization process again to obtain a new quantization error 107, and calculates the simplified distortion index value (SPRD2) of the new quantization error 107 via the conversion domain distortion index calculator 103. As shown in Figure 3b.

Among them, the simplified distortion index values (SPRD1 and SPRD2) of the present invention are as follows:

d ^* and d ^{* T} are respectively a vector and a transpose vector of the conversion coefficient 105 of the time domain signal to the conversion domain converted by the time domain-conversion domain converter 101, and qe' and qe ^'T are vectors and rotations of the quantization error 107, respectively. The vector is set and fre is the rounding error 108 of the conversion factor.

It should be noted that the compression method of the present invention can obtain a compression curve that is very linear and independent of the signal type. In order to get a simplified root mean square (PRD) calculation process for differential values (ie to exclude the reverse SPIHT step):

1. SPIHT encoding must be performed separately from quantization to get the best compression.

2. Wavelet transform can also be operated in an integer manner (without the use of more complicated lifting operations), and the same result as the real number operation is obtained and the quantization error 107 is more easily controlled.

Taking compressed ECG data as an example, when actually compressing 48 arrhythmia signals, randomly select 11 signals for training to obtain quantitative scale control parameters (QF). The quantized scale control parameters are then used to compress 48 signals, the results of which are shown in Figures 4 and 5, respectively. 4 and FIG. 5 respectively show the curves of the quantization scale control parameter (QF) and the simplified distortion index values SPRD1 and SPRD2 of the present invention, that is, 48 distortion curves. Wherein, when the range of the quantization scale control parameter is between 0 and 30, the percentage of the simplified distortion index values SPRD1 and SPRD2 is between 0 and 25%. It should be noted that SPRD1 and SPRD2 are key technologies for the root mean square (PRD) of simplified difference values. It can be clearly seen that the near-linear distortion performance is obtained for each signal.

Table 1 shows the actual compression effect for a given target distortion and allowable variation ( PRD _T ± ε , ε = 5% and 10%, where PRD _T is the given target distortion, ε is the allowable range of variation). When ε = 5%, the conversion domain distortion index calculator 103 takes about 2 loop times on average, and the smaller the PRD _{T is} , the smaller the number of loops required by the conversion domain distortion index calculator 103 is.

In the operation of the compression method of the present invention, when 48 electrocardiogram signals are used and the ratios of SPRD1/PRD and SPRD2/PRD are measured, the actual experimental results are shown in FIG. 6. Figure 6 shows that SPRD2 can replace PRD when PRD ≧ 2%, and SPRD1 can replace PRD when PRD ≧ 1%. It should be noted that PRD≦9% is a clinically acceptable area (PRD≦2% is a distortion invisible area), and the results of Fig. 6 fully show that SPRD1 and SPRD2 can completely replace PRD clinically. In addition, the advantage of using SPRD1 and SPRD2 is that the complicated reverse conversion process such as reverse SPIHT, inverse quantization and inverse wavelet conversion of the quality control loop in Fig. 1 can be omitted, which means that half of the hardware cost can be saved.

The functions and advantages of the present invention are:

1. Can be used to instantly record multi-path ECG compression data with quality assurance features.

2. Highly capable of preserving clinical diagnostic information.

3. The design of the system is considered to be hardware-based, and can be further developed into a portable multi-lead ECG recording device.

4. With the lowest hardware cost and the lowest power consumption.

While the present invention has been described in its preferred embodiments, it is not intended to limit the scope of the invention, and various modifications and changes can be made without departing from the spirit and scope of the invention. As explained above, various modifications and variations can be made without departing from the spirit of the invention. Therefore, the scope of the invention is defined by the scope of the appended claims.

100. . . Quality assurance signal compression method without reverse processing

101. . . Time domain-conversion domain converter

102. . . Quantization processor

103. . . Conversion domain distortion indicator calculator

104. . . Distortionless compressor

105. . . Conversion factor

106. . . Quantization coefficient

107. . . Quantization error

108. . . Rounding error of conversion factor

200. . . Quality assurance signal compression system without reverse processing

The above and other objects, features, and advantages of the present invention will become more apparent from the <RTIgt;

Figure 1 shows the distortion measurement loop of a conventional wavelet-converted electrocardiogram data compression system;

2a shows a quality assurance signal compression system (with SPRD1 as an example) that does not use reverse processing according to the present invention;

Figure 2b shows a quality assurance signal compression system (with SPRD2 as an example) that does not use reverse processing according to the present invention;

Figure 3a shows a quality assurance signal compression method (with SPRD1 as an example) that does not use reverse processing according to the present invention;

FIG. 3b shows a quality assurance signal compression method (for example, SPRD2) that does not use reverse processing according to the present invention;

4 is a schematic diagram showing the quantization scale control parameter (QF) and the SPRD1 curve of the present invention;

Figure 5 is a schematic diagram showing the quantitative scale control parameter (QF) and SPRD2 curves of the present invention;

Fig. 6 is a view showing a comparison of SPRD1, SPRD2 and PRD of the present invention.

100. . . Quality assurance signal compression method without reverse processing

101. . . Time domain-conversion domain converter

102. . . Quantization processor

103. . . Conversion domain distortion indicator calculator

104. . . Distortionless compressor

105. . . Conversion factor

106. . . Quantization coefficient

107. . . Quantization error

108. . . Rounding error of conversion factor

Claims (32)

A quality assurance signal compression system that does not use reverse processing, comprising: a time domain-conversion domain converter for converting a time domain signal to a conversion domain to obtain a conversion coefficient and a rounding error of a conversion coefficient; a quantization processor, And a conversion domain distortion index calculator for calculating a rounding error of the conversion coefficient plus a simplified distortion index value (SPRD1) of the quantization error; and a A distortionless compressor that encodes the quantized coefficients without distortion. The signal compression system according to claim 1, wherein the time domain-conversion domain converter is a wavelet converter, a one-dimensional non-returning wavelet converter and a reversible one-dimensional non-retroactive discrete One of the periodic wavelet converters. The signal compression system of claim 1, wherein the quantization processor is a quantization processor that obtains an approximate linear distortion data compression effect. The signal compression system of claim 1, wherein the distortionless compressor performs signal compression using a discrete wavelet coding algorithm. The signal compression system according to claim 4, wherein the distortionless compressor uses a Set Partitioning in Hierarchical Trees (SPIHT) for signal distortion-free compression. The signal compression system of claim 1, wherein the simplified distortion index value (SPRD1) is expressed as: d ^* and d ^{* T} are the vector of the conversion coefficient and the transpose vector, respectively, qe ' and qe ' ^T are the vector of the quantization error and the transposition vector, respectively, and fre is the rounding error of the conversion coefficient. The signal compression system of claim 1, wherein the reduced distortion index value (SPRD1) falls within a tolerance range, and the distortion-free compressor can perform distortion-free coding of the quantized coefficient; The quantization processor needs to perform quantization processing again to obtain a new quantization error, and calculates the new quantization error plus the simplified distortion index value (SPRD1) of the rounding error of the conversion coefficient via the conversion domain distortion index calculator. For example, in the signal compression system described in claim 1, wherein the simplified distortion index value (SPRD1) ≧ 1%, the simplified distortion index value (SPRD1) can replace the root mean square percentage of the difference value (PRD). ). A quality assurance signal compression system that does not use reverse processing, comprising: a time domain-conversion domain converter for converting a time domain signal to a conversion domain to obtain a conversion coefficient; and a quantization processor for processing the conversion coefficient A quantization coefficient and a quantization error are obtained; a conversion domain distortion index calculator calculates only the simplified distortion index value (SPRD2) of the quantization error; and a distortionless compressor, which performs distortionless coding. The signal compression system according to claim 9, wherein the time domain-conversion domain converter is a wavelet converter, a one-dimensional non-returning wavelet converter and a reversible one-dimensional non-retroactive discrete One of the periodic wavelet converters. The signal compression system of claim 9, wherein the quantization processor is a quantization processor that obtains an approximate linear distortion data compression effect. The signal compression system of claim 9, wherein the distortionless compressor performs signal compression using a discrete wavelet coding algorithm. The signal compression system of claim 12, wherein the distortionless compressor performs distortionless compression using a Set Partitioning in Hierarchical Trees (SPIHT). The signal compression system of claim 9, wherein the reduced distortion index value (SPRD2) falls within a tolerance range, and the distortion-free compressor can perform distortion-free coding of the quantized coefficient; The quantization processor needs to perform quantization processing again to obtain a new quantization error, and calculates a simplified distortion index value (SPRD2) of the new quantization error via the conversion domain distortion index calculator. For example, in the signal compression system described in claim 9, wherein the simplified distortion index value (SPRD2) ≧ 2%, the simplified distortion index value (SPRD2) can replace the root mean square percentage of the difference value (PRD). ). The signal compression system of claim 9, wherein the simplified distortion index value (SPRD2) is expressed as: d ^* and d ^{* T} are the vector of the conversion coefficient and the transpose vector, respectively, and qe ' and qe ' ^T are the vector of the quantization error and the transpose vector, respectively. A quality assurance signal compression method that does not use reverse processing, which mainly comprises the following steps: converting a time domain signal to a conversion domain by using a time domain-conversion domain converter to obtain a conversion coefficient and a rounding error of a conversion coefficient; Processing, by the processor, the conversion coefficient to obtain a quantized coefficient and a quantization error; calculating a rounding error of the conversion coefficient by adding a conversion domain distortion index calculator plus a simplified distortion index value (SPRD1) of the quantization error; The quantized coefficients are encoded without distortion using a distortionless compressor. The signal compression method of claim 17, further comprising the step of: calculating whether the simplified distortion index value (SPRD1) falls within a tolerance range, and if so, the distortionless compressor can The quantization coefficient is subjected to distortion-free coding; otherwise, the quantization processor needs to perform quantization processing again to obtain a new quantization error, and calculates a new quantization error plus a rounding error of the conversion coefficient via the conversion domain distortion index calculator. Distortion indicator value (SPRD1). The signal compression method according to claim 17, wherein the simplified distortion index value (SPRD1) is expressed as: d ^* and d ^{* T} are the vector of the conversion coefficient and the transpose vector, respectively, qe ' and qe ' ^T are the vector of the quantization error and the transposition vector, respectively, and fre is the rounding error of the conversion coefficient. For example, in the signal compression method described in claim 17, wherein the simplified distortion index value (SPRD1) ≧ 1%, the simplified distortion index value (SPRD1) can replace the root mean square percentage of the difference value (PRD). ). The signal compression method according to claim 17, wherein the time domain-conversion domain converter is a wavelet converter, a one-dimensional non-returning wavelet converter and a reversible one-dimensional non-retroactive discrete One of the periodic wavelet converters. The signal compression method according to claim 17, wherein the quantization processor is a quantization processor that can obtain an approximate linear distortion data compression effect. The signal compression method according to claim 17, wherein the distortionless compressor performs signal compression using a discrete wavelet coding algorithm. The signal compression method according to claim 23, wherein the distortionless compressor performs distortionless compression using a Set Partitioning in Hierarchical Trees (SPIHT). A quality assurance signal compression method that does not use reverse processing, which mainly includes the following steps: converting a time domain signal to a conversion domain by using a time domain-conversion domain converter to obtain a conversion coefficient; and processing the conversion coefficient by using a quantization processor, Obtaining a quantized coefficient and a quantization error; calculating a simplified distortion index value (SPRD2) of the quantization error by using a conversion domain distortion index calculator; and performing distortionless encoding on the quantized coefficient by a distortionless compressor. The signal compression method according to claim 25, further comprising the step of: calculating whether the simplified distortion index value (SPRD2) falls within an allowable error range, and if so, the distortionless compressor can The quantized coefficient is subjected to distortion-free coding; otherwise, the quantization processor needs to perform quantization processing again to obtain a new quantization error, and calculates a simplified distortion index value (SPRD2) of the new quantization error via the conversion domain distortion index calculator. For example, in the signal compression method described in claim 25, wherein the simplified distortion index value (SPRD2) ≧ 2%, the simplified distortion index value (SPRD2) can replace the root mean square percentage of the difference value (PRD). ). The signal compression method according to claim 25, wherein the simplified distortion index value (SPRD2) is expressed as: d ^* and d ^{* T} are the vector of the conversion coefficient and the transpose vector, respectively, and qe ' and qe ' ^T are the vector of the quantization error and the transpose vector, respectively. The signal compression method according to claim 25, wherein the time domain-conversion domain converter is a wavelet converter, a one-dimensional non-returning wavelet converter and a reversible one-dimensional non-retroactive discrete One of the periodic wavelet converters. The signal compression method according to claim 25, wherein the quantization processor is a quantization processor that can obtain an approximate linear distortion data compression effect. The signal compression method according to claim 25, wherein the distortionless compressor performs signal compression using a discrete wavelet coding algorithm. The signal compression method according to claim 31, wherein the distortionless compressor uses a Set Partitioning in Hierarchical Trees (SPIHT) for signal distortion-free compression.