CN105915227A - Adaptive mixed data lossless compression system - Google Patents

Abstract

The invention relates to an adaptive mixed data lossless compression system, which is designed to improve the effectiveness and reliability of coding. The present invention includes: a statistical analysis module, which analyzes the statistical characteristics of the signal to be coded and compares it with the set threshold value to realize the adaptive switching between the linear predictive Golomb coding method and the adaptive Golomb coding method; if the linear predictive Golomb coding method is applicable, Then the linear prediction module performs linear prediction on the coded signal to obtain the residual signal, maps the residual signal to a positive integer, and the statistical judgment module determines the Golomb coding parameters according to the mapped residual signal and performs Golomb coding; if adaptive Golomb coding is applicable method, the statistical judgment module directly maps the signal to be coded, and performs Golomb coding after determining the Golomb coding parameters according to the mapped data. The present invention compresses and encodes the signal by adaptively adopting different encoding modes and encoding parameters by analyzing the digital statistical characteristics of each frame signal, thereby improving the effectiveness and reliability of compression.

Description

Adaptive Hybrid Data Lossless Compression System

technical field

The invention relates to a signal compression technology, in particular to an adaptive hybrid data lossless compression system.

Background technique

With the development and application of multimedia technology, the information that people obtain includes text, image, video, audio and various animation media. The amount of data that needs to be processed and transmitted to transmit this information is very large, which restricts the storage and transmission of multimedia information and hinders information acquisition and transmission. Therefore, data compression technology and mass storage technology are extremely important for the development of multimedia technology. In recent years, with the emergence of large-capacity storage devices and the increase of network transmission bandwidth, a physical guarantee is provided for lossless compression, thereby promoting its development. The main content of lossless compression includes framing, prediction and lossless coding.

Most of the signals encountered in practical applications are not stationary signals, and for random signals, such as the compression processing of intermittent mutation signals, the encoding method of fixed code length will occupy more bits than that of variable length code encoding , reducing compression performance.

In view of the above-mentioned defects, the designer is actively researching and innovating in order to create an adaptive hybrid data lossless compression system to make it more industrially applicable.

Contents of the invention

To solve the above technical problems, the purpose of the present invention is to provide an adaptive hybrid data lossless compression system that improves the effectiveness and reliability of compression.

The self-adaptive hybrid data lossless compression system of the present invention includes:

Central controller, said central controller includes statistical analysis, linear prediction module, statistical judgment module, Golomb coding module;

The statistical analysis module compares the signal analysis statistical characteristics of the signal to be encoded with the set threshold value to determine whether the signal to be encoded needs to be predicted,

If prediction is required, the statistical analysis module outputs the signal to be encoded to the linear prediction module; the linear prediction module performs linear prediction on the signal to be encoded to obtain a residual signal, and outputs the residual signal to the statistical decision module; the statistical judgment module obtains Golomb encoding parameters based on the residual signal, outputs the Golomb encoding parameters to the Golomb encoding module, and the Golomb encoding module encodes the residual signal;

If prediction is not required, then the statistical analysis module outputs the signal to be encoded to the statistical judgment module; the statistical judgment module obtains the Golomb coding parameters based on the signal to be coded, and outputs the Golomb coding parameters to the Golomb coding module, the The Golomb encoding module encodes the signal to be encoded.

Further, the central controller also includes a prediction decision module, an inverse linear prediction module, and a Golomb decoding module;

The prediction judgment module judges whether the coded code stream has been linearly predicted,

If there is no linear prediction, the coded code stream is output to the Golomb decoding module, and the signal to be encoded is recovered through the processing of the Golomb decoding module;

If linear prediction is performed, the encoded code stream is output to the inverse linear prediction module for processing, and then output to the Golomb decoding module, and the signal to be encoded is recovered through the Golomb decoding module.

Further, the statistical analysis module includes a prediction coefficient calculation unit and a comparison unit,

The prediction coefficient calculation unit is used to run the Levinson-Durbin algorithm to solve the prediction coefficient of the two-order linear prediction of the signal to be encoded and output the autocorrelation coefficient;

The comparison unit is used to obtain the minimum prediction error power of the second-order prediction of the autocorrelation coefficient output by the prediction coefficient calculation unit and in the Yule-Walker equation

If the minimum prediction error power *β<autocorrelation coefficient Then the comparison unit outputs the signal to be encoded to the linear prediction module;

If the minimum prediction error power *β≥autocorrelation coefficient Then the comparison unit outputs the signal to be encoded to the statistical judgment module;

Among them, β is a parameter determined in advance according to the signal characteristics.

Further, the statistical judgment module includes a first judgment unit, a second judgment unit, and a parameter output unit, the first judgment unit receives a residual signal, maps the residual signal to a positive integer, and calculates a mean mean; The second decision unit receives the signal to be encoded, maps the signal to be encoded into a positive integer, and calculates a mean value;

The parameter output unit, if mean<m ₁ , then the parameter output unit outputs the selection parameter b=m ₁ /2 to the Golomb encoding module; if m ₁ ≤ mean<m ₂ , then the parameter output unit outputs the selection parameter b= m ₁ to the Golomb coding module; if m ₂ ≤ mean<m ₃ , the parameter output unit outputs the selection parameter b=m ₂ to the Golomb coding module; if mean≥m ₃ , the parameter output unit outputs the selection parameter b = m ₃ to the Golomb encoding module.

Further, the linear prediction module includes a residual calculation unit, which is used to obtain the prediction coefficient, predict the predicted value of the signal to be encoded based on the P previous signals to be encoded, and calculate the interpolation between the digital data and the predicted value of the digital data to obtain the prediction residual.

Preferably, the dedicated chip of the central processing unit uses a DSP as the main processor, an ARM chip as the main processor chip, or an FPGA as the main processor.

Further, it also includes an analog/digital signal input circuit, an analog/digital signal output circuit, a code stream output interface, and a code stream input interface respectively connected to the central controller. If the signal to be encoded is an analog signal, through the analog/digital signal The signal input circuit is input to the central processing unit after performing AD conversion, and if the signal to be encoded is a digital signal, it is directly input to the central processing unit;

The Golomb coding module processes and generates a coded code stream, and the coded code stream is output through the coded code stream output interface; the coded code stream is input to the central processing unit through the code stream input interface, and is processed and restored by the Golomb coded module Output the signal to be encoded, if it is necessary to restore the analog signal, the signal to be encoded is output through the analog/digital signal output interface.

By means of the above solution, the present invention has at least the following advantages:

The present invention first preprocesses the signal to improve the operation speed. Through the test and analysis of the signal data, the signal of each frame is selectively predicted in the linear prediction module. 2nd order linear prediction is used.

According to the digital characteristics of each frame signal, the linear prediction module selects different selection parameters for Golomb coding, so as to realize the adaptive performance of compression coding. In the lossless coding module, Golomb coding, which is different from Huffman coding, is used for entropy coding. After testing, it can be improved to improve the effectiveness and reliability of lossless compression coding.

Using Golomb coding as the main coding method, the signal is divided into frames, the digital characteristics of each frame signal are analyzed, and different selection parameters are used for Golomb coding to realize the adaptive performance of compression coding. The autocorrelation coefficient of each frame is calculated, and the two-order linear prediction coefficient is obtained by the Levinson-Durbin algorithm. Introduce linear prediction, choose whether to predict or not on the basis of analyzing the signal frame, improve the compression efficiency, judge whether the data frame needs linear prediction according to the judgment threshold, and use Golomb coding with adaptive selection parameters for the predicted frame and non-predicted frame respectively, with flexibility Strong, high compression ratio.

The above description is only an overview of the technical solutions of the present invention. In order to understand the technical means of the present invention more clearly and implement them according to the contents of the description, the preferred embodiments of the present invention and accompanying drawings are described in detail below.

Description of drawings

Fig. 1 is the flow chart of the self-adaptive mixed data lossless compression system of the present invention;

Fig. 2 is a structural block diagram of the self-adaptive hybrid data lossless compression system of the present invention.

detailed description

The specific implementation manners of the present invention will be further described in detail below in conjunction with the accompanying drawings and embodiments. The following examples are used to illustrate the present invention, but are not intended to limit the scope of the present invention.

The present invention uses a variable length code—Golomb code to compress and code the signal. Golomb coding can make the average code length of the positive integer data flow subject to geometric distribution the shortest, use shorter codes for smaller numbers, and use longer codes for larger numbers, and can directly give the best variable-length code.

The Golomb code of positive integer n is composed of "prefix code + tail code"

1. prefix code is the unary code word of q+1 position, and

$\mathrm{<span\; class="notranslate">\; q</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; I</span>}\mathrm{<span\; class="notranslate">\; N</span>}\mathrm{<span\; class="notranslate">\; T</span>}<span\; class="notranslate">\; (</span>\frac{\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; b</span>}}<span\; class="notranslate">\; )</span>$

②The tail code is the remainder of (n-1)/b

r=n-1-qb

binary encoding.

Among them, b=2 ^m is the selection parameter of Golomb code.

Each frame of data needs to analyze the digital characteristics of the signal frame, and select appropriate selection parameters according to the digital characteristics to realize adaptive coding. Through repeated experimental analysis, the present invention provides four Golomb codes with selected parameters, the values of b points are 4, 8, 16, and 32 respectively, and the maximum number of bits of the Golomb code tail code is 5. But for a data frame with a larger value, when the value of b is 32, the value of q will increase correspondingly, the prefix code will occupy a larger number of bits, and the code length will become larger.

Therefore, the present invention also introduces a linear prediction technology to selectively predict all data frames. That is, the data frame with a small value is not predicted, but directly encoded; for the data frame with a large value, the second-order linear prediction only needs to encode the prediction error, which reduces the number of bits occupied by the encoding.

Example 1

Referring to Fig. 1, the self-adaptive hybrid data lossless compression system described in a preferred embodiment of the present invention, the self-adaptive hybrid data lossless compression system includes: a central controller, the central controller includes statistical analysis, linear Prediction module, statistical judgment module, Golomb coding module;

The statistical analysis module compares the signal analysis statistical characteristics of the signal to be encoded with the set threshold value to determine whether the signal to be encoded needs to be predicted,

If prediction is required, the statistical analysis module outputs the signal to be encoded to the linear prediction module; the linear prediction module performs linear prediction on the signal to be encoded to obtain a residual signal, and outputs the residual signal to the statistical decision module; the statistical judgment module obtains Golomb encoding parameters based on the residual signal, outputs the Golomb encoding parameters to the Golomb encoding module, and the Golomb encoding module encodes the residual signal;

If prediction is not required, then the statistical analysis module outputs the signal to be encoded to the statistical judgment module; the statistical judgment module obtains the Golomb coding parameters based on the signal to be coded, and outputs the Golomb coding parameters to the Golomb coding module, the The Golomb encoding module encodes the signal to be encoded.

The central controller also includes a prediction and decision module, an inverse linear prediction module, and a Golomb decoding module;

The prediction judgment module judges whether the coded code stream has been linearly predicted,

If there is no linear prediction, the coded code stream is output to the Golomb decoding module, and the signal to be encoded is recovered through the processing of the Golomb decoding module;

If linear prediction is performed, the encoded code stream is output to the inverse linear prediction module for processing, and then output to the Golomb decoding module, and the signal to be encoded is recovered through the Golomb decoding module.

The central controller is respectively connected with the analog/digital signal input circuit, the analog/digital signal output circuit, the code stream output interface, and the code stream input interface. If the signal to be encoded is an analog signal, the AD conversion is performed through the analog/digital signal input circuit Then input to the central processing unit, if the signal to be encoded is the signal to be encoded, then directly input to the central processing unit;

The Golomb coding module processes and generates a coded code stream, and the coded code stream is output through the coded code stream output interface; the coded code stream is input to the central processing unit through the code stream input interface, and is processed and restored by the Golomb coded module Output the signal to be encoded, and the signal to be encoded is output through the analog/digital signal output interface.

In this embodiment, the lossless compression is roughly implemented by three parts. The first step is the frame division operation, which provides editability, which is an important and necessary feature of most signal compression algorithms to be encoded. Intra-frame decorrelation follows. The redundant information contained in the adjacent samples is removed by the linear predictor, and the linear predictor is selectively applied to the frame data to generate a prediction error sequence. The parameters of the lossless coded predictor together with the prediction error represent the frame signal. Finally there is entropy coding. Entropy coding is lossless coding, the redundancy of the error signal itself. No information is lost during this process. After analysis, this embodiment adopts Golomb codes to perform entropy coding on each frame signal.

In practical applications, the signal to be compressed is usually an analog signal, so it is necessary to convert the analog signal into a digital signal through A/D sampling, compress and encode the digital signal, and output the coded stream.

Firstly, the signal is preprocessed to improve the operation speed. Through the test and analysis of the signal data, the second-order linear prediction is adopted. The autocorrelation coefficient of each frame is calculated, and the two-order linear prediction coefficient is obtained by the Levinson-Durbin algorithm. Determine whether the data frame needs linear prediction according to the decision threshold.

If the data frame needs to be predicted, the subsequent signal pair is linearly predicted according to the sample values of the first two sampling points of the signal, and then the residual signal is obtained.

If the data frame does not require linear prediction, directly map the data frame, perform data analysis on the mapped data, select the appropriate parameter value b of Golomb coding, and then perform Golomb coding. The data frame also includes two parts, the flag bit and the Golomb code stream. There are 3 flag bits, which store the prediction flag bit and the selection parameter value b.

In the self-adaptive mixed data lossless compression system of this embodiment, if the data frame needs to be predicted, the subsequent signal pair is linearly predicted according to the sampling values of the first two sampling points of the signal, and then the residual signal is obtained. Considering that Golomb coding encodes positive integers, it is necessary to map the residual signal to a positive integer and perform Golomb coding on the residual signal. The data frame prediction flag is 1 bit, the first two sampling points are respectively coded in 16-bit binary, and the linear prediction coefficients are coded in 13-bit binary respectively. There are four cases for selecting parameter b, so two bits are occupied. Therefore, the encoding of the data frame has two parts, namely the flag bit and the Golomb code stream. There are 61 flag bits, which are used to store prediction flag bits, prediction coefficients, 2 sampling point values, and the selection parameter b of Golomb encoding. The Golomb code stream stores the Golomb encoded codewords of all residual signals.

If the data frame does not require linear prediction, directly map the data frame, perform data analysis on the mapped data, select the appropriate parameter value b of Golomb coding, and then perform Golomb coding. The data frame also includes two parts, the flag bit and the Golomb code stream. There are 3 flag bits, which store the prediction flag bit and the selection parameter value b.

In this embodiment, the decoding process is a process of decoding the input encoded code stream and recovering the signal to be encoded. Because each frame of the input code stream includes flag bits and coded code streams, it is necessary to first determine the prediction flag bits in the flag bits.

The decoding process is the process of decoding the input encoded code stream and recovering the signal to be encoded. Because each frame of the input code stream includes flag bits and coded code streams, it is necessary to first determine the prediction flag bits in the flag bits.

If the frame is a prediction frame, it is necessary to extract two prediction coefficients of the second-order prediction and two sampling point values, the selection coefficient b of Golomb coding, and then decode the coded stream to recover the signal to be coded.

If the frame is not a predictive frame, it only needs to extract the selection coefficient b of Golomb coding to decode the code stream and restore the signal to be coded.

Example 2

In the adaptive mixed data lossless compression system described in this embodiment, on the basis of Embodiment 1, the linear prediction module, including a prediction coefficient calculation unit, runs the Levinson-Durbin Levinson-Durbin algorithm to calculate each frame The linear prediction coefficient of the signal to be encoded, specifically including:

The Yule-Walker equation for p-order linear prediction is as follows:

in, is the autocorrelation coefficient;

There are p+1 equations in the Yule-Walker equation,

When k=0,1,2,...,p When known, solve a _pk [k=1, 2, ..., p] and Sample p+1 unknown quantities, where, a _pk prediction coefficient, is the minimum error power;

Autocorrelation Coefficient in Yule-Walker Equation for Linear Prediction Calculating the minimum forecast error power of the second-order forecast based on the Levinson-Durbin algorithm

The recursive formula of Levinson-Durbin algorithm is

where [k=1, 2, . . . , p];

For two-stage prediction, the value of k is 1 and 2, and the prediction coefficients are a ₁₁ and a ₂₂

a _ki = a _{k-1, i} + a _kk a _{k-1, ki} , i=1, 2, ..., k-1

From this, the prediction coefficients a ₁₁ and a ₂₂ of the two-order linear prediction are obtained.

The linear prediction module also includes a comparison unit, and the prediction coefficient calculation unit outputs the autocorrelation coefficient in the Yule-Walker equation of the linear prediction Minimum Forecast Error Power for 2nd Order Forecast to said result output unit;

The result output unit obtains the minimum prediction error power of the second-order prediction of the autocorrelation coefficient output by the prediction coefficient calculation unit and in the Yule-Walker equation

If the minimum prediction error power *β<autocorrelation coefficient Then the comparison unit outputs the signal to be encoded to the linear prediction module;

If the minimum prediction error power *β≥autocorrelation coefficient Then the comparison unit outputs the signal to be encoded to the statistical judgment module;

Among them, β is a parameter determined in advance according to the signal characteristics.

The adaptive module includes a prediction residual calculation unit, and the specific calculation formula of the prediction residual ε (n) is:

$\mathrm{<span\; class="notranslate">\; \&epsiv;</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; the\; s</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; -</span>\stackrel{<span\; class="notranslate">\; ^</span>}{\mathrm{<span\; class="notranslate">\; the\; s</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; )</span>$

Among them, s(n) is the data to be encoded, is the predicted value of the data to be encoded.

The predicted value of the data to be encoded It is predicted by using the past p data to be encoded s(n), where

$\stackrel{<span\; class="notranslate">\; ^</span>}{\mathrm{<span\; class="notranslate">\; the\; s</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}}^{\mathrm{<span\; class="notranslate">\; p</span>}}{\mathrm{<span\; class="notranslate">\; a</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}\mathrm{<span\; class="notranslate">\; i</span>}}\mathrm{<span\; class="notranslate">\; the\; s</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; )</span>$

Where a _ii is the linear prediction coefficient.

The linear prediction in this embodiment selectively predicts all data frames. That is, the data frame with a small value is not predicted, but directly encoded; for the data frame with a large value, the second-order linear prediction only needs to encode the prediction error, which reduces the number of bits occupied by the encoding.

In this embodiment, it also includes an analog/digital signal input circuit, an analog/digital signal output circuit, a code stream output interface, and a code stream input interface respectively connected to the central controller. If the signal to be encoded is an analog signal, the /The digital signal input circuit carries out AD conversion and then input to the central processing unit, if the signal to be coded is a digital signal, then directly input to the central processing unit;

The Golomb coding module processes and generates a coded code stream, and the coded code stream is output through the coded code stream output interface; the coded code stream is input to the central processing unit through the code stream input interface, and is processed and restored by the Golomb coded module Output the signal to be encoded, if it is necessary to restore the analog signal, the signal to be encoded is output through the analog/digital signal output interface.

In each of the above embodiments, the statistical judgment module includes a first judgment unit, a second judgment unit, and a parameter output unit, and the first judgment unit receives a residual signal, maps the residual signal to a positive integer, and calculates mean mean; the second decision unit receives the signal to be encoded, maps the signal to be encoded into a positive integer, and calculates the mean mean;

The first decision unit performs data mapping on the prediction residual of each frame to be coded, specifically including: replacing the coded value c of the prediction residual of each frame to be coded with the mapping value d:

Judging whether the coded value c in each frame to be coded prediction residual is less than 0,

If the coded value c is less than 0, then the mapping value d=2c;

If the coded value c is greater than or equal to 0, then the mapping value d=2c+1;

The second decision unit performs data mapping on the prediction residual of each frame to be coded, specifically including: replacing the coded value c of the prediction residual of each frame to be coded with the mapping value d:

Judging whether the coded value c in the prediction residual of each frame to be coded is less than 0,

If the coded value c is less than 0, then the mapping value d=2c;

If the coded value c is greater than or equal to 0, then the mapping value d=2c+1;

Perform data mapping on each frame of the signal to be encoded, specifically including: replacing the encoded value e of each frame of the signal to be encoded with the mapping value f:

Judging whether the coded value e in the prediction residual of each frame to be coded is less than 0,

If the coded value e is less than 0, then the mapping value f=2e;

If the coded value e is greater than or equal to 0, then the mapped value f=2e+1.

The parameter output unit is used to perform data analysis on the signal to be coded or the prediction residual to obtain a mean value;

If prediction is required, the parameter output unit obtains the selected parameters by calculating and analyzing the mean value of the residual signal output by the first decision unit, specifically including,

If the residual signal mean<m ₁ , then the parameter output unit outputs the selection parameter b=m _1/2 to the Golomb coding module;

If m ₁ ≤ residual signal mean<m ₂ , then the parameter output unit outputs the selection parameter b=m ₁ to the Golomb encoding module;

If m ₂ ≤ residual signal mean<m ₃ , then the parameter output unit outputs the selection parameter b=m ₂ to the Golomb encoding module;

If the residual signal mean≥m ₃ , the parameter output unit outputs the selection parameter b=m ₃ to the Golomb encoding module.

If prediction is not required, then the parameter output unit obtains the selected parameters by calculating and analyzing the mean value of the signal to be encoded output by the second decision unit, specifically including,

If the signal to be encoded mean<m ₁ , then the parameter output unit outputs the selection parameter b=m _1/2 to the Golomb encoding module;

If m ₁ ≤ signal to be encoded mean<m ₂ , then the parameter output unit outputs the selection parameter b=m ₁ to the Golomb encoding module;

If m ₂ ≤ signal to be encoded mean<m ₃ , then the parameter output unit outputs the selection parameter b=m ₂ to the Golomb encoding module;

If the signal to be encoded is mean≥m ₃ , the parameter output unit outputs the selection parameter b=m ₃ to the Golomb encoding module.

The values of m ₁ , m ₂ , and m ₃ are determined according to the mean value of the residual or the mean value of the coded data.

The above is only a preferred embodiment of the present invention, and is not intended to limit the present invention. It should be pointed out that for those of ordinary skill in the art, some improvements can be made without departing from the technical principle of the present invention. and modifications, these improvements and modifications should also be considered as the protection scope of the present invention.

Claims (7)

1. An adaptive hybrid lossless data compression system, comprising: the central controller comprises a statistical analysis module, a linear prediction module, a statistical decision module and a Golomb coding module;

the statistical analysis module compares the signal analysis statistical characteristics of the signal to be coded with a set threshold value to judge whether the signal to be coded needs to be predicted or not,

if prediction is needed, the statistical analysis module outputs a signal to be coded to the linear prediction module; the linear prediction module performs linear prediction on a signal to be coded to obtain a residual signal, and outputs the residual signal to the statistical decision module; the statistical decision module obtains a Golomb coding parameter based on the residual signal, outputs the Golomb coding parameter to the Golomb coding module, and the Golomb coding module codes the residual signal;

if the prediction is not needed, the statistical analysis module outputs a signal to be coded to the statistical judgment module; the statistical decision module obtains a Golomb coding parameter based on a signal to be coded, outputs the Golomb coding parameter to the Golomb coding module, and codes the signal to be coded by the Golomb coding module.

2. The adaptive hybrid lossless data compression system of claim 1, wherein the central controller further comprises a prediction decision module, an inverse linear prediction module, a Golomb decoding module;

a prediction judgment module for judging whether the code stream is subjected to linear prediction,

if no linear prediction exists, the coding code stream is output to a Golomb decoding module, and a signal to be coded is recovered through processing of the Golomb decoding module;

if linear prediction is carried out, the coding code stream is output to an inverse linear prediction module for processing, then output to a Golomb decoding module, and a signal to be coded is recovered through processing of the Golomb decoding module.

3. The adaptive hybrid data lossless compression system according to claim 1, wherein the statistical analysis module includes a prediction coefficient calculation unit and a comparison unit, the prediction coefficient calculation unit is configured to run a levinson-durbin algorithm to solve a prediction coefficient of two-order linear prediction of the signal to be encoded and output an autocorrelation coefficient;

the comparison unit is used for obtaining the minimum prediction error power of the autocorrelation coefficient 2-order prediction output by the prediction coefficient calculation unitAnd in the Euler-Watk Yule-Walker equation

If the minimum prediction error powerThe comparison unit outputs a signal to be coded to the linear prediction module;

if the minimum prediction error powerThe comparison unit outputs a signal to be coded to the statistical decision module;

where β is a parameter predetermined according to the signal characteristics.

4. The adaptive hybrid data lossless compression system according to claim 1, wherein the statistical decision module includes a first decision unit, a second decision unit, and a parameter output unit, the first decision unit receives a residual signal, maps the residual signal to a positive integer, and calculates a mean; the second decision unit receives a signal to be coded, maps the signal to be coded into a positive integer and calculates a mean value mean;

the parameter output unit, if mean<m₁If the parameter output unit outputs the selection parameter b ═ m₁2 to the Golomb coding module; if m₁≤mean＜m₂If the parameter output unit outputs the selection parameter b ═ m₁To the Golomb encoding module; if m₂≤mean<m₃If the parameter output unit outputs the selection parameter b ═ m₂To the Golomb encoding module; if mean is greater than or equal to m₃If the parameter output unit outputs the selection parameter b ═ m₃To the Golomb encoding module.

5. The adaptive hybrid lossless data compression system as claimed in claim 1, wherein the linear prediction module includes a residual calculation unit for obtaining prediction coefficients, predicting a predicted value of the signal to be encoded based on P preceding signals to be encoded, and calculating an interpolation of the digital data and the predicted value of the digital data to obtain a prediction residual.

6. The adaptive hybrid lossless data compression system of claim 1, wherein the dedicated chip of the central processing unit has a DSP as the main processor, an ARM chip as the main processor chip, or an FPGA as the main processor.

7. The adaptive hybrid data lossless compression system according to claim 2, further comprising an analog/digital signal input circuit, an analog/digital signal output circuit, a code stream output interface, and a code stream input interface, which are respectively connected to the central controller, wherein if the signal to be encoded is an analog signal, the signal is input to the central processor after being subjected to AD conversion by the analog/digital signal input circuit, and if the signal to be encoded is a digital signal, the signal is directly input to the central processor;

the Golomb coding module processes and generates a coding code stream, and the coding code stream is output through the coding code stream output interface; and the coding code stream is input into a central processing unit through a code stream input interface, is processed by the Golomb coding module to recover a signal to be coded, and if analog signals need to be recovered, the signal to be coded is output through an analog/digital signal output interface.