A material spectrum signal acquisition and detection system

Abstract

The present invention provides a material spectral signal acquisition and detection system, including a spectral signal acquisition unit, a signal processing unit, a control unit and a display unit. The spectral signal acquisition unit includes a high-sensitivity spectrometer, a sensor and a light source. The high-sensitivity spectrometer is used to capture the spectral signal of the material. The sensor is a photoelectric sensor, which is used to detect changes in the spectral signal and supplement the data of the spectrometer. The light source emits light of a specific wavelength through a sample or a substance to be tested. The present invention has an efficient noise removal capability. The present invention adopts a multi-scale wavelet transform denoising algorithm, which can effectively distinguish between noise and useful information in the signal, and accurately remove noise in different frequency bands through multi-level processing. This algorithm can adaptively adjust parameters to avoid signal loss problems that may occur in traditional denoising methods, thereby ensuring the integrity and accuracy of the spectral signal.

Description

Technical Field

[0001] The present invention relates to the technical field of spectrum detection, and in particular to a material spectrum signal acquisition and detection system. Background Art

[0002] With the development of science and technology and industry, the spectral detection technology of substances has been widely used in many fields, including chemical analysis, environmental monitoring, biomedicine, food safety, etc. Spectral signal acquisition and analysis technology is based on the principle of interaction between matter and electromagnetic waves. By analyzing the absorption, reflection, emission and other characteristics of substances to light of specific wavelengths, it can identify important information such as the composition, structure and concentration of substances. The application range of spectral technology is very wide, especially in quality control, rapid detection and remote monitoring. However, in the actual spectral signal acquisition and detection process, there are many technical difficulties, which restrict the performance and accuracy of the spectral signal processing system.

[0003] When processing spectral signals, the presence of noise is inevitable. Common noise includes high-frequency noise (random noise) and low-frequency noise (background interference). Traditional filtering denoising methods (such as low-pass filtering) are prone to lose the details of effective spectral signals. Wavelet transform is a multi-resolution analysis tool suitable for processing non-stationary signals. It can decompose spectral signals at different scales and achieve adaptive denoising. Summary of the invention

[0004] The present invention aims to provide a material spectrum signal acquisition and detection system, which separates high-frequency noise from low-frequency effective signals in spectrum signals through wavelet decomposition, and uses adaptive wavelet basis and threshold denoising technology to accurately remove noise and retain effective signals. The algorithm can adjust denoising parameters according to the local characteristics of the signal, avoiding the information loss problem caused by traditional denoising methods.

[0005] To achieve the above object, the present invention proposes the following technical solution: a material spectrum signal acquisition and detection system, comprising:

[0006] A spectral signal acquisition unit, the spectral signal acquisition unit includes a high-sensitivity spectrometer, a sensor and a light source, the high-sensitivity spectrometer is used to capture the spectral signal of the substance, the sensor is a photoelectric sensor, the photoelectric sensor is used to detect the change of the spectral signal and supplement the data of the spectrometer, the light source emits light of a specific wavelength through the sample or the substance to be tested;

[0007] A signal processing unit, the signal processing unit includes an analog signal converter, a processor, a memory module, a data filtering module and a signal denoising module. The analog signal converter converts the spectral signal from an analog signal into a digital signal. The processor is used to process the signal and execute data filtering, denoising and feature extraction algorithms. The memory module is used to temporarily store the collected spectral data and processing results, and supports caching and calculations during signal processing. The data filtering module has a built-in hardware module or a microprocessor to implement low-pass, high-pass or band-pass filtering of the collected signal. The signal denoising module uses a digital signal processing algorithm to remove noise and retain valid signals.

[0008] The control unit includes a controller, a touch screen interface and a communication interface. The controller is used to adjust various system parameters. The touch screen interface is used for user input and system settings, allowing the user to set acquisition parameters and view system status. The communication interface is used to exchange data and transmit control signals with other units.

[0009] The display unit includes a display screen and a graphics processing unit. The display screen is used to display the spectral signal processing results and provide a visual interface. Users can view real-time spectral data and analysis results through the display screen. The graphics processing unit is used to accelerate graphics rendering and improve system response speed.

[0010] Furthermore, in the present invention, the high-sensitivity spectrometer adopts a high-resolution spectral detector, which is one of a photodiode array and a spectral fiber detector, so as to capture the spectral signal of the substance.

[0011] Furthermore, in the present invention, the light source is one of a laser, a xenon lamp or an LED.

[0012] Furthermore, in the present invention, the processor uses a multi-scale wavelet transform denoising algorithm for denoising, and the multi-scale wavelet transform denoising algorithm includes the following steps:

[0013] Step 1: Perform wavelet decomposition on the original spectral signal and use a multi-scale wavelet transform method to separate the high-frequency noise and low-frequency useful signal in the signal;

[0014] Step 2: By selecting the best wavelet basis and adjusting the parameters of the wavelet basis according to the local characteristics of the signal, adaptive wavelet denoising is performed. Signal noise removal at each scale is performed using hard threshold or soft threshold methods;

[0015] Step 3: In the wavelet reconstruction phase, an adaptive algorithm is used to reconstruct the signal to remove noise and retain the spectral features.

[0016] Furthermore, in the present invention, the specific process of step 1 is as follows:

[0017] Wavelet transform decomposes the spectral signal x(t) into low-frequency approximate components A j and high frequency detail component D j ,in j Represents the number of layers of wavelet decomposition, and the decomposition formula is as follows:

[0018] Continuous wavelet transform,

[0019] in a is the scale factor, b is the time translation, and ψ is the mother wavelet function;

[0020] Discrete Wavelet Transform,

[0021] where h[n] is the low-pass filter coefficient used to approximate component A j , g[n] is the high-pass filter coefficient for the detail component D j , A j and D j are the low-frequency component and high-frequency component of the jth layer respectively;

[0022] The input spectral signal x(t) is decomposed layer by layer through wavelet transform to separate the detail components D at multiple scales j and approximate component A j , high frequency component D j Contains noise information, while the low-frequency component A j Contains the main signal information.

[0023] Furthermore, in the present invention, the specific process of step 2 is as follows:

[0024] In the decomposed high frequency component D j In the above example, the threshold denoising method is used to remove noise. The soft threshold method and the hard threshold method are used. The hard threshold formula directly sets the coefficients less than the threshold to zero:

[0025]

[0026] The soft threshold formula reduces the coefficients that exceed the threshold:

[0027]

[0028] Where T j is the threshold value, which can be calculated by the adaptive threshold method:

[0029] σ j is the high frequency component D j The standard deviation of is estimated as: N is the signal length;

[0030] Then process the high-frequency components D of each scale layer by layer j , remove noise by soft threshold or hard threshold method, and adaptively adjust the threshold T j , in order to ensure the removal effect of noise of different scales.

[0031] Furthermore, in the present invention, the specific process of step 3 is as follows:

[0032] For the high frequency components after denoising and the original low-frequency component A j Perform wavelet reconstruction to restore the spectral signal after denoising The wavelet reconstruction formula is as follows:

[0033] Where g[n] and h[n] are the high-pass and low-pass reconstruction filter coefficients, respectively;

[0034] Starting from the finest layer, the signal is reconstructed layer by layer, and the denoised detail components are With low frequency component A j Synthesize layer by layer to finally get the denoised spectral signal

[0035] Beneficial effects: The technical solution of this application has the following technical effects:

[0036] 1. The present invention has efficient noise removal capability. The present invention adopts a multi-scale wavelet transform denoising algorithm, which can effectively distinguish noise from useful information in the signal and accurately remove noise in different frequency bands through multi-level processing. This algorithm can adaptively adjust parameters to avoid signal loss problems that may occur in traditional denoising methods, thereby ensuring the integrity and accuracy of spectral signals.

[0037] 2. The multi-channel signal processing capability of the present invention. The signal processing system of the present invention supports the simultaneous acquisition and processing of multi-channel spectral signals, and can make full use of the correlation and complementary information between different channels. The multi-channel signals are weighted and fused through the attention mechanism in deep learning, which enhances the effective information of the signal and suppresses the interference of redundant or irrelevant information, thereby improving the feature extraction accuracy of the substance and the recognition ability of the system.

[0038] 3. The present invention has the adaptability and flexibility of signal processing. Based on the wavelet transform, the present invention proposes an adaptive denoising strategy, which can automatically adjust the algorithm parameters according to the local characteristics of the spectral signal to optimize the denoising effect. This adaptability ensures that the system can flexibly respond to spectral data under different substances or environmental conditions, and has strong robustness and stability.

[0039] 4. The present invention has efficient real-time processing and convenient operation. It uses a high-performance processor and a dedicated hardware acceleration module to ensure that the system can process large-scale spectral data in real time and support rapid response. Users can conveniently set acquisition parameters and view system status through the touch screen interface, and monitor signal processing results and analysis outputs in real time through the display unit. The introduction of the graphics processing unit improves the speed and display effect of graphics rendering, making data display more intuitive and clear.

[0040] 5. The system structure of the present invention is compact and has strong scalability. The system design of the present invention is compact, the functional modules are clearly divided, and the modules are connected through efficient communication interfaces. The system can be expanded as needed to support more complex spectral data acquisition and processing tasks. Whether in laboratory environments or in industrial applications, it can provide stable performance and high work efficiency.

[0041] It should be appreciated that all combinations of the foregoing concepts, as well as additional concepts described in greater detail below, may be considered to be part of the inventive subject matter of the present disclosure, provided such concepts are not mutually inconsistent.

[0042] The foregoing and other aspects, embodiments and features of the present invention can be more fully understood from the following description in conjunction with the accompanying drawings. Other additional aspects of the present invention, such as the features and / or beneficial effects of the exemplary embodiments, will be apparent from the following description or learned from the practice of the specific embodiments according to the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] The drawings are not intended to be drawn to scale. In the drawings, each identical or nearly identical component shown in various figures may be represented by the same reference numeral. For clarity, not every component is labeled in every figure. Embodiments of various aspects of the present invention will now be described by way of example and with reference to the accompanying drawings, in which:

[0044] Figure 1 It is a schematic diagram of the structure of the present invention. DETAILED DESCRIPTION

[0045] In order to better understand the technical content of the present invention, specific embodiments are cited and described as follows in conjunction with the accompanying drawings. Various aspects of the present invention are described in this disclosure with reference to the accompanying drawings, in which many illustrative embodiments are shown. The embodiments of the present disclosure are not necessarily defined to include all aspects of the present invention. It should be understood that the various concepts and embodiments introduced above, as well as those described in more detail below, can be implemented in any of many ways, because the concepts and embodiments disclosed in the present invention are not limited to any implementation. In addition, some aspects disclosed in the present invention can be used alone, or in any appropriate combination with other aspects disclosed in the present invention.

[0046] A material spectrum signal acquisition and detection system includes a spectrum signal acquisition unit, a signal processing unit, a control unit and a display unit. The spectrum signal acquisition unit includes a high-sensitivity spectrometer, a sensor and a light source. The high-sensitivity spectrometer is used to capture the spectrum signal of the material. The sensor is a photoelectric sensor. The photoelectric sensor is used to detect the change of the spectrum signal and supplement the data of the spectrometer. The light source emits light of a specific wavelength and passes through a sample or a substance to be tested. The high-sensitivity spectrometer uses a high-resolution spectrum detector. The spectrum detector is one of a photodiode array and a spectrum fiber detector to capture the spectrum signal of the material and can accurately measure the spectrum signals within different wavelength ranges. The spectrum signal acquisition unit can effectively avoid common signal interference problems and ensure the stability and accuracy of the signal. The light source is one of a laser, a xenon lamp or an LED. A suitable light source is selected according to the characteristics of the material. The light source emits light of a specific wavelength and passes through the sample or the substance to be tested. The light reflected, transmitted or emitted by the material is transmitted to the spectrum detector through an optical fiber or a lens. The photoelectric detector inside the spectrometer captures the reflected or emitted light and converts it into an electrical signal. The spectrum signal is converted into a digital signal via an analog signal converter for processing by the signal processing unit.

[0047] The signal processing unit includes an analog signal converter, a processor, a memory module, a data filtering module and a signal denoising module. The analog signal converter converts the spectral signal from an analog signal to a digital signal. The processor is used to process the signal to perform data filtering, denoising and feature extraction algorithms. The memory module is used to temporarily store the collected spectral data and processing results, and supports caching and calculations during signal processing. The data filtering module has a built-in hardware module or uses a microprocessor to implement low-pass, high-pass or band-pass filtering of the collected signal. The signal denoising module uses a digital signal processing algorithm to remove noise and retain valid signals. The digitized spectral signal is transmitted from the analog signal converter to the microprocessor or FPGA. The processor controls the data filtering module and the denoising module through programming to perform real-time signal processing. The feature extraction module analyzes the signal and extracts key information related to the material composition and structure according to the characteristics of the spectral data.

[0048] The control unit includes a controller, a touch screen interface and a communication interface. The controller is used to adjust various system parameters. The touch screen interface is used for user input and system settings, allowing the user to set acquisition parameters and view system status. The communication interface is used to exchange data and control signal transmission with other units. The controller exchanges data with the spectral signal acquisition unit and the signal processing unit through the communication interface. The control unit receives user input through buttons or touch screen interfaces, and converts it into corresponding control signals to adjust system operating parameters.

[0049] The display unit includes a display screen and a graphics processing unit. The display screen is used to display the spectral signal processing results and provide a visual interface. Users can view real-time spectral data and analysis results through the display screen. The graphics processing unit is used to accelerate graphics rendering and improve system response speed. The display unit is connected to the control unit or the signal processing unit through an interface to obtain and display the processed data. Users can adjust the display content and view detailed information through the touch screen operation interface.

[0050] The processor uses a multi-scale wavelet transform denoising algorithm for denoising. The innovation of this method is that it combines multi-scale wavelet transform with an adaptive reconstruction algorithm, thereby better retaining the effective components of the signal and avoiding the problem of information loss in traditional denoising methods. The algorithm can respond more flexibly to different types of noise. The multi-scale wavelet transform denoising algorithm includes the following steps:

[0051] Step 1: Perform wavelet decomposition on the original spectral signal and use a multi-scale wavelet transform method to separate the high-frequency noise and low-frequency useful signal in the signal;

[0052] Step 2: By selecting the best wavelet basis and adjusting the parameters of the wavelet basis according to the local characteristics of the signal, adaptive wavelet denoising is performed. Signal noise removal at each scale is performed using hard threshold or soft threshold methods;

[0053] Step 3: In the wavelet reconstruction phase, an adaptive algorithm is used to reconstruct the signal to remove noise and retain the spectral features.

[0054] Furthermore, in the present invention, the specific process of step 1 is as follows:

[0055] Wavelet transform decomposes the spectral signal x(t) into low-frequency approximate components A j and high frequency detail component D j ,in j Represents the number of layers of wavelet decomposition, and the decomposition formula is as follows:

[0056] Continuous wavelet transform,

[0057] in a is the scale factor, b is the time translation, and ψ is the mother wavelet function;

[0058] Discrete Wavelet Transform,

[0059] where h[n] is the low-pass filter coefficient used to approximate component A j , g[n] is the high-pass filter coefficient for the detail component D j , A j and D jare the low-frequency component and high-frequency component of the jth layer respectively;

[0060] The input spectral signal x(t) is decomposed layer by layer through wavelet transform to separate the detail components D at multiple scales j and approximate component A j , high frequency component D j Contains noise information, while the low-frequency component A j Contains the main signal information.

[0061] Furthermore, in the present invention, the specific process of step 2 is as follows:

[0062] In the decomposed high frequency component D j In the above example, the threshold denoising method is used to remove noise. The soft threshold method and the hard threshold method are used. The hard threshold formula directly sets the coefficients less than the threshold to zero:

[0063]

[0064] The soft threshold formula reduces the coefficients that exceed the threshold:

[0065]

[0066] Where T j is the threshold value, which can be calculated by the adaptive threshold method:

[0067] σ j is the high frequency component D j The standard deviation of is estimated as: N is the signal length;

[0068] Then process the high-frequency components D of each scale layer by layer j , remove noise by soft threshold or hard threshold method, and adaptively adjust the threshold T j , in order to ensure the removal effect of noise of different scales.

[0069] Furthermore, in the present invention, the specific process of step 3 is as follows:

[0070] For the high frequency components after denoising and the original low-frequency component A j Perform wavelet reconstruction to restore the spectral signal after denoising The wavelet reconstruction formula is as follows:

[0071] Where g[n] and h[n] are the high-pass and low-pass reconstruction filter coefficients, respectively;

[0072] Starting from the finest layer, the signal is reconstructed layer by layer, and the denoised detail components are With low frequency component Aj Synthesize layer by layer to finally get the denoised spectral signal

[0073] The work flow of the material spectrum signal acquisition and detection system of the present invention is as follows: Step 1, spectrum signal acquisition is performed, and the system acquires the spectrum signal of the material to be tested through a high-sensitivity spectrometer, a sensor and a light source. The high-resolution spectrometer is responsible for capturing the spectral information of the material, and the photoelectric sensor is used to detect the changes in the spectrum signal in real time, further supplementing the acquisition data of the spectrometer. After the material is irradiated with light of a specific wavelength emitted by a light source such as a laser, a xenon lamp or an LED, the spectrum signal will change, which will be received by the spectrum detector and converted into an analog signal.

[0074] Step 2: Signal conversion and preliminary processing: The collected spectral signal is converted into a digital signal through an analog signal converter and enters the signal processing unit. The processor in the signal processing unit executes a series of data filtering, denoising and feature extraction algorithms. First, the data filtering module performs low-pass, high-pass or band-pass filtering on the signal to remove unnecessary frequency band signals.

[0075] Step 3: Denoising and multi-scale signal processing. In the denoising module, the signal is decomposed by wavelet to separate the low-frequency approximate components (A) at multiple scales. j ) and high-frequency detail components (D j ). The high-frequency component contains noise components, and the low-frequency component contains the main signal information. The high-frequency component is denoised by soft threshold or hard threshold method to remove noise while retaining the effective signal. The specific steps are as follows: Wavelet decomposition: Decompose the spectral signal x(t) into low-frequency components (A) by continuous wavelet transform or discrete wavelet transform. j ) and high frequency components (D j ). The high-frequency detail component mainly contains noise information, while the low-frequency approximate component contains the main effective signal. Threshold denoising: For each layer of decomposed high-frequency components, the hard threshold method or soft threshold method is used to remove noise. The threshold is calculated by an adaptive algorithm to ensure that the denoising effect can be optimized at different scales. Signal reconstruction: Through wavelet reconstruction, the denoised high-frequency component is combined with the low-frequency component to restore the denoised spectral signal.

[0076] Step 4: Feature extraction and multi-channel fusion. In the denoised signal, effective spectral features such as peak position, intensity, shape, etc. are further extracted for material analysis and identification. For multi-channel spectral signals, the present invention uses a deep learning model based on the attention mechanism to perform weighted fusion of the signals of each channel. The system automatically adjusts the weight of each signal channel, strengthens important information and suppresses redundant or irrelevant information, thereby improving the accuracy and robustness of material analysis.

[0077] Step 5: Result display and user interaction The processed spectral data and analysis results are displayed through the display unit. Users can view the collected signals, denoised spectra, analysis results and other system status in real time through the graphical interface. The touch screen interface allows users to adjust system parameters and set acquisition conditions as needed, ensuring convenience and flexibility of operation.

[0078] Step 6: Real-time processing and feedback. The system supports real-time signal processing throughout the process, and the processor's high performance ensures rapid response during data acquisition and analysis. Under high load or multi-channel signal processing, the system can still maintain stable operation and provide real-time feedback on spectral data processing results.

[0079] The material spectrum signal acquisition and detection system of this embodiment can achieve high-precision and high-efficiency spectrum signal processing through innovative wavelet transform denoising algorithm, multi-channel feature fusion technology and efficient real-time processing architecture. The system's adaptive denoising, precise feature extraction and flexible multi-channel fusion scheme effectively improve the accuracy and processing efficiency of spectrum signals, solve multiple bottleneck problems in existing technologies, and have broad application prospects and market value.

[0080] Although the present invention has been disclosed as above with preferred embodiments, it is not intended to limit the present invention. A person with ordinary knowledge in the technical field to which the present invention belongs may make various changes and modifications without departing from the spirit and scope of the present invention. Therefore, the protection scope of the present invention shall be determined by the definition of the claims.

Claims

1. A material spectrum signal acquisition and detection system, characterized in that: include: A spectral signal acquisition unit, the spectral signal acquisition unit includes a high-sensitivity spectrometer, a sensor and a light source, the high-sensitivity spectrometer is used to capture the spectral signal of the substance, the sensor is a photoelectric sensor, the photoelectric sensor is used to detect the change of the spectral signal and supplement the data of the spectrometer, the light source emits light of a specific wavelength through the sample or the substance to be tested; A signal processing unit, the signal processing unit includes an analog signal converter, a processor, a memory module, a data filtering module and a signal denoising module. The analog signal converter converts the spectral signal from an analog signal into a digital signal. The processor is used to process the signal and execute data filtering, denoising and feature extraction algorithms. The memory module is used to temporarily store the collected spectral data and processing results, and supports caching and calculations during signal processing. The data filtering module has a built-in hardware module or a microprocessor to implement low-pass, high-pass or band-pass filtering of the collected signal. The signal denoising module uses a digital signal processing algorithm to remove noise and retain valid signals. The control unit includes a controller, a touch screen interface and a communication interface. The controller is used to adjust various system parameters. The touch screen interface is used for user input and system settings, allowing the user to set acquisition parameters and view system status. The communication interface is used to exchange data and transmit control signals with other units. The display unit includes a display screen and a graphics processing unit. The display screen is used to display the spectral signal processing results and provide a visual interface. Users can view real-time spectral data and analysis results through the display screen. The graphics processing unit is used to accelerate graphics rendering and improve system response speed.

2. The material spectrum signal acquisition and detection system according to claim 1, characterized in that: The high-sensitivity spectrometer adopts a high-resolution spectral detector, which is one of a photodiode array and a spectral fiber detector, to capture the spectral signal of a substance.

3. The material spectrum signal acquisition and detection system according to claim 1, characterized in that: The light source is one of laser, xenon lamp or LED.

4. The material spectrum signal acquisition and detection system according to claim 1, characterized in that: The processor uses a multi-scale wavelet transform denoising algorithm to perform denoising, and the multi-scale wavelet transform denoising algorithm includes the following steps: Step 1: Perform wavelet decomposition on the original spectral signal and use a multi-scale wavelet transform method to separate the high-frequency noise and low-frequency useful signal in the signal; Step 2: By selecting the best wavelet basis and adjusting the parameters of the wavelet basis according to the local characteristics of the signal, adaptive wavelet denoising is performed. Signal noise removal at each scale is performed using hard threshold or soft threshold methods; Step 3: In the wavelet reconstruction phase, an adaptive algorithm is used to reconstruct the signal to remove noise and retain the spectral features.

5. The material spectrum signal acquisition and detection system according to claim 4, characterized in that: The specific process of step 1 is as follows: Wavelet transform decomposes the spectral signal x(t) into low-frequency approximate components A j and high frequency detail component D j ,in j Represents the number of layers of wavelet decomposition, and the decomposition formula is as follows: Continuous wavelet transform, in a is the scale factor, b is the time translation, and ψ is the mother wavelet function; Discrete Wavelet Transform, where h[n] is the low-pass filter coefficient used to approximate component A j , g[n] is the high-pass filter coefficient for the detail component D j , A j and D j are the low-frequency component and high-frequency component of the jth layer respectively; The input spectral signal x(t) is decomposed layer by layer through wavelet transform to separate the detail components D at multiple scales j and approximate component A j , high frequency component D j Contains noise information, while the low-frequency component A j Contains the main signal information.

6. A material spectrum signal acquisition and detection system according to claim 5, characterized in that: The specific process of step 2 is as follows: In the decomposed high frequency component D j In the above example, the threshold denoising method is used to remove noise. The soft threshold method and the hard threshold method are used. The hard threshold formula directly sets the coefficients less than the threshold to zero: The soft threshold formula reduces the coefficients that exceed the threshold: Where T j is the threshold value, which can be calculated by the adaptive threshold method: σ j is the high frequency component D j The standard deviation of is estimated as: N is the signal length; Then process the high-frequency components D of each scale layer by layer j , remove noise by soft threshold or hard threshold method, and adaptively adjust the threshold T j , in order to ensure the removal effect of noise of different scales.

7. A material spectrum signal acquisition and detection system according to claim 6, characterized in that: The specific process of step 3 is as follows: For the high frequency components after denoising and the original low-frequency component A j Perform wavelet reconstruction to restore the spectral signal after denoising The wavelet reconstruction formula is as follows: Where g[n] and h[n] are the high-pass and low-pass reconstruction filter coefficients, respectively; Starting from the finest layer, the signal is reconstructed layer by layer, and the denoised detail components are With low frequency component A j Synthesize layer by layer to finally get the denoised spectral signal

Images