Seismic data denoising method, device and equipment based on optical computing, medium and product

Abstract

The application discloses a kind of based on optical computing seismic data denoising method, device, equipment, medium and product, the method includes: obtaining seismic data, and the seismic data is preprocessed;Based on spatial light modulator, the seismic data after preprocessing is encoded, corresponding light intensity image is generated;The light intensity image is projected randomly and is dimensionally reduced, and low-dimensional matrix is generated;The low-dimensional matrix is sequentially decomposed orthogonally and singular value decomposition;The singular value after decomposition is truncated, and low rank reconstruction is carried out based on the singular value after truncation, and the seismic data after denoising is obtained.The application is realized to the fast dimension reduction of original high-dimensional data by optical random projection, replace the time-consuming matrix multiplication operation in traditional electronic architecture, not only effectively reduce the calculation complexity, and make full use of the inherent high parallelism and ultra-wideband characteristics of optical computing, suitable for large-scale seismic data efficient denoising.

Description

Technical Field

[0001] This invention relates to the field of optical signal processing technology, and in particular to a method, apparatus, device, medium and product for denoising seismic data based on optical computing. Background Technology

[0002] Seismic data acquired in the field during oil and gas exploration is inevitably subject to various types of noise interference, which seriously affects the accuracy of subsequent data processing and geological interpretation. Therefore, seismic signal denoising technology has attracted much attention in the field of geophysical research. Currently, noise suppression methods are mainly divided into two categories: one is based on the sparsity of seismic data, which uses curvelet transform, Radon transform, or wavelet transform to sparsely represent noisy data, thereby achieving effective separation of effective signals from irregular noise; the other is to decompose seismic data into several singular value vectors or intrinsic mode functions through singular value decomposition (SVD), variational mode decomposition, etc., and achieve noise suppression by extracting the singular values ​​or intrinsic mode functions corresponding to the effective signals.

[0003] However, traditional Singular Value Decomposition (SVD) typically requires full or partial singular value decomposition of the constructed Hankel or Toeplitz matrices. When processing large-scale seismic data volumes, the matrix reconstruction and feature extraction processes consume significant computational resources and storage space, making it difficult to meet the high-efficiency processing requirements in oil and gas exploration. Compressed Singular Value Decomposition (CSVD) offers higher computational efficiency and can significantly save time costs. However, this method usually incorporates compressed sensing theory, assuming that the seismic signal has a sparse representation in a certain transform domain (such as wavelet domain, curvelet domain, or Radon domain) before sampling. Its effectiveness is highly dependent on data sparsity, leading to unstable denoising results for complex wavefields. Summary of the Invention

[0004] The purpose of this invention is to provide a method, apparatus, device, medium, and product for denoising seismic data based on optical computing. It achieves rapid dimensionality reduction of the original high-dimensional data through optical random projection, replacing the time-consuming matrix multiplication operation in traditional electronic architecture. This not only effectively reduces computational complexity but also fully utilizes the inherent high parallelism and ultra-wideband characteristics of optical computing, making it suitable for efficient denoising of large-scale seismic data.

[0005] To achieve the above objectives, embodiments of the present invention provide a seismic data denoising method based on optical computing, comprising: Acquire earthquake data and preprocess the earthquake data; The preprocessed seismic data is encoded using a spatial light modulator to generate a corresponding light intensity image. The light intensity image is subjected to random projection dimensionality reduction to generate a low-dimensional matrix; The low-dimensional matrix is ​​then subjected to orthogonal decomposition and singular value decomposition in sequence. The singular values ​​after decomposition are truncated, and low-rank reconstruction is performed based on the truncated singular values ​​to obtain the denoised seismic data.

[0006] As an improvement to the above scheme, the acquisition of seismic data and the preprocessing of the seismic data include: Two-dimensional seismic data is acquired, and the two-dimensional seismic data is divided into blocks using a sliding time window to obtain multiple partially overlapping data blocks; wherein, the window size of the sliding time window is related to the resolution of the spatial light modulator; The data blocks are normalized to obtain the normalization matrix of each data block.

[0007] As an improvement to the above scheme, the step of encoding the preprocessed seismic data based on a spatial light modulator to generate a corresponding light intensity image includes: The normalized matrix of each data block is input into the spatial light modulator, and each element in the matrix is ​​mapped to the brightness value of the macropixel of the spatial light modulator. Spatial combination modulation is used to generate the light intensity image corresponding to each data block; wherein, the size of the macropixel is related to the size of the data block.

[0008] As an improvement to the above scheme, the step of performing random projection dimensionality reduction on the light intensity image to generate a low-dimensional matrix includes: The light intensity image is projected onto a scattering medium for random projection dimensionality reduction to generate a dimensionality-reduced light signal. The reduced-dimensional optical signal is sampled, and pixels are randomly selected to generate the low-dimensional matrix.

[0009] As an improvement to the above scheme, the step of performing random projection dimensionality reduction on the light intensity image to generate a low-dimensional matrix includes: The light intensity image is subjected to random projection dimensionality reduction based on the random projection matrix to generate a low-dimensional matrix; wherein the random projection matrix is ​​obtained by training a programmable optical neural network.

[0010] As an improvement to the above scheme, the step of sequentially performing orthogonal decomposition and singular value decomposition on the low-dimensional matrix includes: The low-dimensional matrix is ​​orthogonally decomposed to obtain an orthogonal matrix; Construct a target matrix based on the orthogonal matrix, and then perform singular value decomposition on the target matrix.

[0011] As an improvement to the above scheme, the step of truncating the decomposed singular values ​​and performing low-rank reconstruction based on the truncated singular values ​​to obtain denoised seismic data includes: The singular values ​​after decomposition are truncated according to a preset energy retention ratio, and low-rank reconstruction is performed based on the truncated singular values. The reconstructed signal is subjected to inverse normalization and weighted fusion of overlapping regions to obtain denoised seismic data.

[0012] This invention also provides a seismic data denoising device based on optical computing, comprising: The data acquisition module is used to acquire earthquake data and preprocess the earthquake data. The modulation and coding module is used to encode the preprocessed seismic data based on the spatial light modulator to generate the corresponding light intensity image; The projection dimensionality reduction module is used to perform random projection dimensionality reduction on the light intensity image to generate a low-dimensional matrix. The matrix decomposition module is used to perform orthogonal decomposition and singular value decomposition on the low-dimensional matrix in sequence. The low-rank reconstruction module is used to truncate the singular values ​​after decomposition and perform low-rank reconstruction based on the truncated singular values ​​to obtain the denoised seismic data.

[0013] This invention also provides a terminal device, including a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor. When the processor executes the computer program, it implements the optical computing-based seismic data denoising method described above.

[0014] This invention also provides a computer-readable storage medium, which includes a stored computer program, wherein the computer program, when running, controls the device where the computer-readable storage medium is located to execute the optical computing-based seismic data denoising method described above.

[0015] This invention also provides a computer program product, which includes a computer program or computer instructions. When the computer program or computer instructions are executed by a processor, they implement the optical computing-based seismic data denoising method described above.

[0016] Compared to existing technologies, the beneficial effects of the seismic data denoising method, apparatus, device, medium, and product based on optical computing provided in this invention are as follows: Seismic data is acquired and preprocessed; the preprocessed seismic data is encoded using a spatial light modulator to generate a corresponding light intensity image; the light intensity image is subjected to random projection dimensionality reduction to generate a low-dimensional matrix; the low-dimensional matrix is ​​sequentially subjected to orthogonal decomposition and singular value decomposition; the singular values ​​after decomposition are truncated, and low-rank reconstruction is performed based on the truncated singular values ​​to obtain the denoised seismic data. This invention achieves rapid dimensionality reduction of the original high-dimensional data through optical random projection, replacing the time-consuming matrix multiplication operations in traditional electronic architectures. This not only effectively reduces computational complexity but also fully utilizes the inherent high parallelism and ultra-wideband characteristics of optical computing, making it suitable for efficient denoising of large-scale seismic data. Attached Figure Description

[0017] Figure 1 This is a flowchart illustrating a preferred embodiment of a seismic data denoising method based on optical computing provided by the present invention. Figure 2 This is a schematic diagram of a preferred embodiment of a seismic data denoising device based on optical computing provided by the present invention; Figure 3 This is a schematic diagram of a preferred embodiment of a terminal device provided by the present invention. Detailed Implementation

[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] Please see Figure 1 , Figure 1 This is a flowchart illustrating a preferred embodiment of a seismic data denoising method based on optical computing provided by the present invention. The seismic data denoising method based on optical computing includes: S1, acquire seismic data and preprocess the seismic data; S2, based on the spatial light modulator, encodes the preprocessed seismic data to generate the corresponding light intensity image; S3, perform random projection dimensionality reduction on the light intensity image to generate a low-dimensional matrix; S4, perform orthogonal decomposition and singular value decomposition on the low-dimensional matrix in sequence; S5 truncates the singular values ​​after decomposition and performs low-rank reconstruction based on the truncated singular values ​​to obtain the denoised seismic data.

[0020] It's important to note that the basic principle of Singular Value Decomposition (SVD) is the assumption that effective seismic information is primarily concentrated in eigenvectors corresponding to larger singular values. Seismic data is transformed into a Hankel matrix form. The rank of the Hankel matrix is ​​closely related to the number of phase axes in the seismic data, and the magnitude of the eigenvalues ​​on the matrix diagonal is directly related to the energy intensity of the phase axes; the stronger the energy, the larger the corresponding eigenvalue. Therefore, rank reduction can be used to discard smaller singular values ​​corresponding to noise, retaining larger singular values ​​corresponding to the effective signal, thus achieving noise suppression. However, traditional SVD methods still have some limitations in practical applications. SVD has high computational complexity, especially when dealing with large-scale 3D seismic data. Matrix reconstruction and singular value calculation are extremely time-consuming and memory-intensive, limiting its application in efficient seismic data processing. In complex geological environments, factors such as multiple waveform aliasing and nonlinear distribution of phase axes can cause the low-rank characteristics of the effective signal itself to be insignificant, thereby reducing the effectiveness of the SVD method. To improve computational efficiency, a seismic denoising method based on Compressed Singular Value Decomposition (CSVD) is proposed, which samples and compresses the data to reduce computational load. However, this method is dependent on the sparse structure of the signal and its denoising effect is unstable for complex wavefields.

[0021] Based on this, this invention proposes a seismic data denoising method based on optical computing. Optical computing is introduced into Randomized Singular Value Decomposition (RSVD), which is independent of signal sparsity and easier to implement in parallel, making it suitable for large-scale seismic data. Specifically, this invention first acquires seismic data and preprocesses it to fit the dynamic range of a spatial light modulator. Then, the preprocessed seismic data is encoded based on the spatial light modulator to generate a corresponding light intensity image. The spatial light modulator (SLM) is a dynamic optical device that can load information onto a one-dimensional or two-dimensional light field. Its core is an array composed of a large number of independent pixel units, each of which can be independently controlled by electrical or optical signals to change a certain parameter of the incident light. A Digital Micromirror Device (DMD) is a representative amplitude modulation type spatial light modulator (SLM). Preprocessed seismic data is input into the DMD, which encodes the data into an image based on pixel values ​​corresponding to light intensities and modulates the incident light to generate a corresponding light intensity image. Next, random projection dimensionality reduction is performed on the light intensity image to generate a low-dimensional matrix, thus reducing the dimensionality of the original high-dimensional signal. Finally, the generated low-dimensional matrix undergoes orthogonal decomposition (QR decomposition) and singular value decomposition. The singular values ​​after decomposition are truncated, and low-rank reconstruction is performed based on the truncated singular values ​​to obtain the denoised seismic data.

[0022] This invention integrates the random singular value decomposition (RSVD) algorithm with the optical random projection mechanism. While maintaining the denoising accuracy of traditional RSVD, it encodes seismic data through spatial light modulation, achieving rapid physical dimensionality reduction of high-dimensional seismic data blocks. At the same time, the inherent parallelism and ultra-high bandwidth of the optical computing architecture significantly reduce time costs and do not require the signal to have sparse characteristics in a specific transform domain. This effectively solves the technical problems of high computational overhead and strong dependence on sparsity in existing low-rank denoising methods, providing reliable technical support for achieving efficient, low-power, and parallelizable seismic data processing.

[0023] In a preferred embodiment, step S1, acquiring seismic data and preprocessing the seismic data, includes: S11, acquire two-dimensional seismic data, and use a sliding time window to divide the two-dimensional seismic data into blocks to obtain multiple partially overlapping data blocks; wherein, the window size of the sliding time window is related to the resolution of the spatial light modulator; S12, normalize each data block to obtain a normalization matrix for each data block.

[0024] Specifically, embodiments of the present invention acquire two-dimensional seismic data. In this model, the horizontal axis t represents the time sampling point, and the vertical axis x represents the seismic trace. Considering the local low-rank nature of complex data, a sliding time window is used in practical processing to divide the 2D seismic data into blocks, reducing edge effects while enhancing model robustness. The size of the sliding time window is related to the resolution of the spatial light modulator. Normalization is performed on each data block to adapt to the dynamic range of the spatial light modulator, resulting in a normalized matrix for each block. An example of the normalization expression is as follows:

[0025] In the formula, Represents data block The normalized matrix; , These represent the local maximum and minimum values, respectively. This parameter needs to be recorded for inverse normalization.

[0026] In another preferred embodiment, step S2, encoding the preprocessed seismic data based on a spatial light modulator to generate a corresponding light intensity image, includes: The normalized matrix of each data block is input into the spatial light modulator, and each element in the matrix is ​​mapped to the brightness value of the macropixel of the spatial light modulator. Spatial combination modulation is used to generate the light intensity image corresponding to each data block; wherein, the size of the macropixel is related to the size of the data block.

[0027] Specifically, in this embodiment of the invention, the normalization matrix of each data block is... The input is fed into a spatial light modulator, mapping each element in the matrix to the brightness value of a macropixel of the spatial light modulator, where the size of the macropixel is... (like The larger the size, the more accurate the calculation. When selecting the pixel values ​​for the spatial light modulator, it is necessary to balance the size of the data block and the size of the macropixel, and it should be as large as possible. The value of is chosen to reduce computational errors. Spatial combination modulation is used to generate a binary image (i.e., a light intensity image) for each data block to avoid temporal multiplexing errors. The expression for generating the light intensity image is as follows: ; In the formula, This represents the modulation function; redundant regions of the DMD are padded with zeros.

[0028] In yet another preferred embodiment, step S3, which involves performing random projection dimensionality reduction on the light intensity image to generate a low-dimensional matrix, includes: S31, the light intensity image is projected onto the scattering medium for random projection dimensionality reduction to generate a dimensionality-reduced light signal; S32, sample the dimensionality-reduced optical signal and randomly select pixels to generate the low-dimensional matrix.

[0029] Specifically, in this embodiment of the invention, the modulated optical signal (i.e., the light intensity image) is projected onto a scattering medium (such as frosted glass). The transmission matrix of the scattering medium naturally satisfies the Gaussian random distribution property, which is equivalent to multiplying the original matrix by a random projection matrix, i.e.: ; In the formula, This represents the optical signal after dimensionality reduction; The vectorized representation of the modulated optical signal; This represents the random projection matrix.

[0030] The light signal after dimensionality reduction through the scattering medium forms a two-dimensional speckle image. A high-resolution charge-coupled device (CCD) sensor is used for sampling, and pixels are randomly selected to form a low-dimensional matrix. The sampling rate should be greater than 0.2.

[0031] This invention encodes seismic data using spatial light modulation, constructs a random projection matrix using a scattering medium, and achieves rapid physical dimensionality reduction of high-dimensional seismic data blocks. The reduced-dimensional matrix is ​​then optically output via a CCD. Compared to traditional SVD methods, this invention performs random matrix multiplication directly in the optical domain, effectively reducing the matrix processing scale and possessing inherent advantages in high parallelism and bandwidth, making it more suitable for efficient processing of large-scale seismic data volumes. Compared to traditional RSVD methods, this invention eliminates the need to generate and store random Gaussian matrices. The process of light passing through the scattering medium is completed at the speed of light, avoiding the enormous computational and memory consumption of generating huge Gaussian random matrices on a computer, thus overcoming the application limitations of traditional RSVD, saving time and improving efficiency. Compared to the strong dependence of CSVD on sparse prior structures, this invention is suitable for seismic denoising tasks involving various complex geological structures.

[0032] In yet another preferred embodiment, step S3, which involves performing random projection dimensionality reduction on the light intensity image to generate a low-dimensional matrix, includes: The light intensity image is subjected to random projection dimensionality reduction based on the random projection matrix to generate a low-dimensional matrix; wherein the random projection matrix is ​​obtained by training a programmable optical neural network.

[0033] Specifically, embodiments of the present invention can also use programmable optical neural networks (such as diffraction neural networks based on MEMS or liquid crystal phase modulators) to replace the scattering medium. By training the network weights to simulate the characteristics of the random projection matrix, both the parallelism of optical computation and dynamic anti-interference can be achieved through electronic control adjustment. The light intensity image is multiplied by the trained random projection matrix to generate a low-dimensional matrix, realizing rapid physical dimensionality reduction of high-dimensional seismic data blocks.

[0034] In another preferred embodiment, step S4 involves sequentially performing orthogonal decomposition and singular value decomposition on the low-dimensional matrix, including: The low-dimensional matrix is ​​orthogonally decomposed to obtain an orthogonal matrix; Construct a target matrix based on the orthogonal matrix, and then perform singular value decomposition on the target matrix.

[0035] Specifically, in this embodiment of the invention, the optical projection results are post-processed using electrical computation. First, the low-dimensional matrix is ​​orthogonally decomposed to obtain an orthogonal matrix. For example, the expression for orthogonal decomposition is as follows: ; In the formula, This represents the low-dimensional matrix after dimensionality reduction, with dimension 1. Obtained by CCD sampling; Representing dimensions An orthogonal matrix that satisfies , representing an orthogonal basis; for The upper triangular matrix. The purpose of QR decomposition is to... Mapped to orthogonal space, preserving its main orientation information.

[0036] The target matrix is ​​constructed based on orthogonal matrices, and then singular value decomposition is performed on the target matrix. For example, the expression for constructing the target matrix is ​​as follows: ; In the formula, Represents the target matrix to be constructed. Represents an orthogonal matrix The transpose of the high-dimensional matrix. Projected onto an orthogonal matrix The orthogonal subspace represented by this is used to reduce computational complexity.

[0037] For example, the expression for singular value decomposition is as follows: ; In the formula, Represents a left singular vector matrix; This represents a diagonal matrix containing singular values ​​arranged in descending order. ; This represents the transpose of the right singular vector matrix.

[0038] In another preferred embodiment, step S5 involves truncating the decomposed singular values ​​and performing low-rank reconstruction based on the truncated singular values ​​to obtain denoised seismic data, including: S51, the singular values ​​after decomposition are truncated according to the preset energy retention ratio, and low-rank reconstruction is performed based on the truncated singular values. S52 performs inverse normalization and weighted fusion processing on the reconstructed signal to obtain denoised seismic data.

[0039] Specifically, in this embodiment of the invention, the singular values ​​after decomposition are truncated using an energy thresholding method to select an appropriate rank, thereby preserving the main signal corresponding to the low-rank structure and suppressing random noise corresponding to higher-order perturbations. For example, the expression for the preset energy retention ratio in the energy thresholding method is as follows: ; In the formula, The preset energy retention ratio is usually set to 0.9 or higher.

[0040] The signal is reconstructed based on the truncated singular values, i.e.: ; The reconstructed signal is then inversely normalized, i.e.: ; Data blocks that have completed inverse normalization During the data reconstruction process, overlapping blocks are weighted and averaged to eliminate boundary effects, resulting in denoised seismic data.

[0041] Furthermore, this embodiment of the invention also evaluates the denoising effect of the denoised seismic wavefield data based on the signal-to-noise ratio, and the calculation method is as follows: ; In the formula, Indicates the signal-to-noise ratio; This represents synthesized, noise-free seismic record data; This indicates the noisy signal obtained by adding random noise to the data. This refers to the denoised signal output through the technical solution in the embodiments of the present invention.

[0042] Meanwhile, the accuracy of optically computed RSVD is evaluated based on relative error, and the calculation method is as follows: ; In the formula, Based on accuracy; This represents the denoised signal calculated based on traditional RSVD.

[0043] This invention encodes normalized seismic data blocks using a spatial light modulator and utilizes a scattering medium to achieve high-speed random projection of the original high-dimensional data. This replaces the time-consuming matrix multiplication operations in traditional electronic architectures, effectively reducing computational complexity and fully leveraging the inherent parallelism and ultra-wideband characteristics of optical computing, making it suitable for efficient denoising of large-scale seismic data. By combining the advantages of optical dimensionality reduction and electronic SVD reconstruction, an efficient low-rank approximation of the seismic signal is achieved. After physical dimensionality reduction at the optical end, a CCD is used to acquire the reduced light intensity image, and then the electronic system performs QR decomposition, target matrix construction, and singular value truncation to finally complete signal reconstruction. Furthermore, this invention introduces a window overlap weighted average and a normalization-inverse normalization mechanism to further suppress edge effects, improve reconstruction accuracy, and adapt to the wavefield characteristics of nonlinear phase axes under complex geological conditions.

[0044] This invention integrates the high parallelism of optical computing with low-rank approximation techniques in signal processing to achieve efficient noise suppression during seismic wavefield denoising. In complex oil and gas reservoir exploration, where seismic data acquisition is large and noise interference is severe, this invention can achieve efficient data denoising at the front end, enhancing the ability to identify details such as fault structures and concealed reservoirs, thereby improving exploration efficiency and success rate. In the fields of seismic monitoring and disaster prevention and mitigation, it effectively suppresses unstructured noise, highlighting responses to real seismic events and improving the robustness of the monitoring system. With the gradual maturation of optical computing chips and optical devices, and the widespread deployment of optical computing simulation platforms, this invention will demonstrate significant advantages in high speed, low power consumption, and parallelism in large-scale data processing scenarios.

[0045] Accordingly, the present invention also provides a seismic data denoising device based on optical computing, which can implement all the processes of the seismic data denoising method based on optical computing in the above embodiments.

[0046] Please see Figure 2 , Figure 2 This is a schematic diagram of a preferred embodiment of a seismic data denoising device based on optical computing provided by the present invention. The seismic data denoising device based on optical computing includes: The data acquisition module 201 is used to acquire earthquake data and preprocess the earthquake data; The modulation and coding module 202 is used to encode the preprocessed seismic data based on the spatial light modulator to generate the corresponding light intensity image; The projection dimensionality reduction module 203 is used to perform random projection dimensionality reduction on the light intensity image to generate a low-dimensional matrix. The matrix decomposition module 204 is used to perform orthogonal decomposition and singular value decomposition on the low-dimensional matrix in sequence. The low-rank reconstruction module 205 is used to truncate the decomposed singular values ​​and perform low-rank reconstruction based on the truncated singular values ​​to obtain the denoised seismic data.

[0047] Preferably, the data acquisition module 201 is specifically used for: Two-dimensional seismic data is acquired, and the two-dimensional seismic data is divided into blocks using a sliding time window to obtain multiple partially overlapping data blocks; wherein, the window size of the sliding time window is related to the resolution of the spatial light modulator; The data blocks are normalized to obtain the normalization matrix of each data block.

[0048] Preferably, the modulation and coding module 202 is specifically used for: The normalized matrix of each data block is input into the spatial light modulator, and each element in the matrix is ​​mapped to the brightness value of the macropixel of the spatial light modulator. Spatial combination modulation is used to generate the light intensity image corresponding to each data block; wherein, the size of the macropixel is related to the size of the data block.

[0049] Preferably, the projection dimensionality reduction module 203 is specifically used for: The light intensity image is projected onto a scattering medium for random projection dimensionality reduction to generate a dimensionality-reduced light signal. The reduced-dimensional optical signal is sampled, and pixels are randomly selected to generate the low-dimensional matrix.

[0050] Preferably, the projection dimensionality reduction module 203 is specifically used for: The light intensity image is subjected to random projection dimensionality reduction based on the random projection matrix to generate a low-dimensional matrix; wherein the random projection matrix is ​​obtained by training a programmable optical neural network.

[0051] Preferably, the matrix decomposition module 204 is specifically used for: The low-dimensional matrix is ​​orthogonally decomposed to obtain an orthogonal matrix; Construct a target matrix based on the orthogonal matrix, and then perform singular value decomposition on the target matrix.

[0052] Preferably, the low-rank reconstruction module 205 is specifically used for: The singular values ​​after decomposition are truncated according to a preset energy retention ratio, and low-rank reconstruction is performed based on the truncated singular values. The reconstructed signal is subjected to inverse normalization and weighted fusion of overlapping regions to obtain denoised seismic data.

[0053] In specific implementation, the working principle, control process and technical effects of the seismic data denoising device based on optical computing provided in this embodiment of the invention are the same as those of the seismic data denoising method based on optical computing in the above embodiments, and will not be repeated here.

[0054] Please see Figure 3 , Figure 3 This is a schematic diagram of a preferred embodiment of a terminal device provided by the present invention. The terminal device includes a processor 301, a memory 302, and a computer program stored in the memory 302 and configured to be executed by the processor 301. When the processor 301 executes the computer program, it implements the optical computing-based seismic data denoising method described in any of the above embodiments.

[0055] Preferably, the computer program can be divided into one or more modules / units (such as computer program 1, computer program 2, ...), and the one or more modules / units are stored in the memory 302 and executed by the processor 301 to complete the present invention. The one or more modules / units can be a series of computer program instruction segments capable of performing specific functions, and the instruction segments are used to describe the execution process of the computer program in the terminal device.

[0056] The processor 301 can be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor, or the processor 301 can be any conventional processor. The processor 301 is the control center of the terminal device, connecting various parts of the terminal device through various interfaces and lines.

[0057] The memory 302 mainly includes a program storage area and a data storage area. The program storage area can store the operating system, applications required for at least one function, etc., and the data storage area can store related data, etc. In addition, the memory 302 can be a high-speed random access memory, or a non-volatile memory, such as a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, and a flash card, etc., or the memory 302 can also be other volatile solid-state storage devices.

[0058] It should be noted that the aforementioned terminal devices may include, but are not limited to, processors and memory, as will be understood by those skilled in the art. Figure 3 The structural diagram is merely an example of the terminal device described above and does not constitute a limitation on the terminal device described above. It may include more or fewer components than shown in the diagram, or combine certain components, or use different components.

[0059] This invention also provides a computer-readable storage medium, which includes a stored computer program, wherein the computer program, when running, controls the device where the computer-readable storage medium is located to execute the optical computing-based seismic data denoising method described in any of the above embodiments.

[0060] This invention also provides a computer program product, which includes a computer program or computer instructions. When the computer program or computer instructions are executed by a processor, they implement the optical computing-based seismic data denoising method described in any of the above embodiments.

[0061] This invention provides a method, apparatus, device, medium, and product for denoising seismic data based on optical computing. The method involves acquiring seismic data and preprocessing it; encoding the preprocessed seismic data using a spatial light modulator to generate a corresponding light intensity image; performing random projection dimensionality reduction on the light intensity image to generate a low-dimensional matrix; sequentially performing orthogonal decomposition and singular value decomposition on the low-dimensional matrix; truncating the singular values ​​after decomposition; and performing low-rank reconstruction based on the truncated singular values ​​to obtain the denoised seismic data. This invention achieves rapid dimensionality reduction of the original high-dimensional data through optical random projection, replacing the time-consuming matrix multiplication operations in traditional electronic architectures. This not only effectively reduces computational complexity but also fully utilizes the inherent high parallelism and ultra-wideband characteristics of optical computing, making it suitable for efficient denoising of large-scale seismic data.

[0062] It should be noted that the system embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Furthermore, in the accompanying drawings of the system embodiments provided by this invention, the connection relationships between modules indicate that they have communication connections, which can be specifically implemented as one or more communication buses or signal lines. Those skilled in the art can understand and implement this without any creative effort.

[0063] The above description represents the preferred embodiments of the present invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of the present invention, and these improvements and modifications are also considered to be within the scope of protection of the present invention.

Claims

1. A method for denoising seismic data based on optical computing, characterized in that, include: Acquire earthquake data and preprocess the earthquake data; The preprocessed seismic data is encoded using a spatial light modulator to generate a corresponding light intensity image. The light intensity image is subjected to random projection dimensionality reduction to generate a low-dimensional matrix; The low-dimensional matrix is ​​then subjected to orthogonal decomposition and singular value decomposition in sequence. The singular values ​​after decomposition are truncated, and low-rank reconstruction is performed based on the truncated singular values ​​to obtain the denoised seismic data.

2. The seismic data denoising method based on optical computing as described in claim 1, characterized in that, The acquisition of seismic data and the preprocessing of the seismic data include: Two-dimensional seismic data is acquired, and the two-dimensional seismic data is divided into blocks using a sliding time window to obtain multiple partially overlapping data blocks; wherein, the window size of the sliding time window is related to the resolution of the spatial light modulator; The data blocks are normalized to obtain the normalization matrix of each data block.

3. The seismic data denoising method based on optical computing as described in claim 2, characterized in that, The process of encoding the preprocessed seismic data using a spatial light modulator to generate a corresponding light intensity image includes: The normalized matrix of each data block is input into the spatial light modulator, and each element in the matrix is ​​mapped to the brightness value of the macropixel of the spatial light modulator. Spatial combination modulation is used to generate the light intensity image corresponding to each data block; wherein, the size of the macropixel is related to the size of the data block.

4. The seismic data denoising method based on optical computing as described in claim 1, characterized in that, The step of performing random projection dimensionality reduction on the light intensity image to generate a low-dimensional matrix includes: The light intensity image is projected onto a scattering medium for random projection dimensionality reduction to generate a dimensionality-reduced light signal. The reduced-dimensional optical signal is sampled, and pixels are randomly selected to generate the low-dimensional matrix.

5. The seismic data denoising method based on optical computing as described in claim 1, characterized in that, The step of performing random projection dimensionality reduction on the light intensity image to generate a low-dimensional matrix includes: The light intensity image is subjected to random projection dimensionality reduction based on the random projection matrix to generate a low-dimensional matrix; wherein the random projection matrix is ​​obtained by training a programmable optical neural network.

6. The seismic data denoising method based on optical computing as described in claim 1, characterized in that, The step of sequentially performing orthogonal decomposition and singular value decomposition on the low-dimensional matrix includes: The low-dimensional matrix is ​​orthogonally decomposed to obtain an orthogonal matrix; Construct a target matrix based on the orthogonal matrix, and then perform singular value decomposition on the target matrix.

7. The seismic data denoising method based on optical computing as described in claim 1, characterized in that, The process of truncating the decomposed singular values ​​and performing low-rank reconstruction based on the truncated singular values ​​to obtain denoised seismic data includes: The singular values ​​after decomposition are truncated according to a preset energy retention ratio, and low-rank reconstruction is performed based on the truncated singular values. The reconstructed signal is subjected to inverse normalization and weighted fusion of overlapping regions to obtain denoised seismic data.

8. A seismic data denoising device based on optical computing, characterized in that, include: The data acquisition module is used to acquire earthquake data and preprocess the earthquake data. The modulation and coding module is used to encode the preprocessed seismic data based on the spatial light modulator to generate the corresponding light intensity image; The projection dimensionality reduction module is used to perform random projection dimensionality reduction on the light intensity image to generate a low-dimensional matrix. The matrix decomposition module is used to perform orthogonal decomposition and singular value decomposition on the low-dimensional matrix in sequence. The low-rank reconstruction module is used to truncate the singular values ​​after decomposition and perform low-rank reconstruction based on the truncated singular values ​​to obtain the denoised seismic data.

9. A terminal device, characterized in that, The device includes a processor and a memory, wherein the memory stores a computer program and the computer program is configured to be executed by the processor, wherein the processor, when executing the computer program, implements the optical computing-based seismic data denoising method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program, wherein when the device containing the computer-readable storage medium executes the computer program, it implements the seismic data denoising method based on optical computing as described in any one of claims 1 to 7.

11. A computer program product, characterized in that, The computer program product includes a computer program or computer instructions, which, when executed by a processor, implement the optical computing-based seismic data denoising method as described in any one of claims 1 to 7.

Images