47 results about "Spectral structure" patented technology

The embodiment of the invention provides an account fraud user identification method in an operation transaction network, and belongs to the technical field of data processing, and the method specifically comprises the steps: 1, collecting basic data in a target network, and constructing graph structure data according to the basic data; 2, extracting spectral structure features and spatial structure features of the graph structure data to obtain multi-scale representation of each node in the graph structure data; 3, introducing a multi-scale attention mechanism to screen and combine multi-scale representation to obtain fusion representation; 4, introducing comparative learning as an auxiliary task, and constructing a training strategy; 5, training the graph neural network and the attention network based on the fusion representation and the training strategy to obtain a recognition model; and step 6, inputting node data to be identified in the target network into the identification model to obtain a fraud prediction result. Through the scheme disclosed by the invention, the identification efficiency and accuracy are improved.

The invention provides a bone conduction speech enhancement method based on a double-flow self-attention fusion network. The bone conduction speech enhancement method comprises the following steps: acquiring and preprocessing a bone conduction speech signal to be processed; a double-flow self-attention fusion network is constructed, the double-flow self-attention fusion network comprises a double-flow encoder, a cross-modal feature fusion module, a multi-scale decoder and a waveform reconstruction module which are connected in sequence, the double-flow self-attention fusion network is trained, a trained model is obtained, and the performance of the model under different conditions is evaluated. According to the method, a double-flow structure is used for processing time domain information and frequency domain information respectively, a self-attention mechanism is introduced, so that the model can capture a long-distance dependency relationship in each mode, and the feature quality is improved. According to the invention, the time-frequency information can be better integrated through the cross-modal feature fusion module. According to the invention, the multi-scale loss function supervises the output of different levels of the decoder, so that the network can learn effective spectrogram representation at different resolutions, and a finer spectrum structure can be recovered.

The invention relates to the technical field of traditional Chinese medicine complex system component analysis, in particular to a rapid identification and analysis method and system for traditional Chinese medicine compound components. Performing step-by-step expansion on the spectral signals to construct a multi-level feature structure; meanwhile, trajectory tracking is conducted on mass spectrum fragments, and a fracture path network is established. And establishing an interactive verification channel between the two to transmit a matching degree parameter, and screening candidate compounds according to the matching degree parameter. And calculating the characteristic contribution degree of each candidate in the spectrum structure and the path integrity of each candidate in the mass spectrum network, merging to generate a confidence index, and finally completing component identification. According to the method, through deep coupling of spectrum multi-scale decomposition and a mass spectrum fracture network, cross validation of multi-dimensional information is achieved, and the accuracy and reliability of component identification of a complex traditional Chinese medicine system are improved.

This invention discloses a portable multi-parameter water quality detection system based on an improved CT spectral structure, comprising an improved CT spectral structure, an S11639-01 CMOS image sensor, an RK3588 embedded platform, and a water quality parameter analysis module. The water sample to be tested is transmitted to form outgoing light carrying water quality characteristic information. The improved CT spectral structure collimates, disperses, converges, and corrects the image plane of the outgoing light, enabling the 200–1000 nm band line spectrum to be imaged within the effective pixel area of ​​the sensor. The sensor acquires raw spectral data and transmits it to the RK3588 for caching and processing. The system establishes a pixel-wavelength cubic polynomial mapping relationship based on the peak pixel position of the standard wavelength point, and completes intensity calibration through dark field subtraction, reference spectrum normalization, and response compensation to obtain corrected one-dimensional spectral data. This data is input into the water quality parameter analysis module, which outputs multiple detection results including COD, turbidity, SS, TN, and TP. This invention features a versatile structure and high integration, making it suitable for rapid on-site water quality detection and environmental monitoring.

This invention discloses a retrieval enhancement generation method for terahertz absorption spectroscopy diagnostics. The method includes: collecting terahertz absorption spectral data of biochemical substances; establishing a terahertz absorption spectral text structure database; establishing a terahertz absorption spectral image structure database; establishing a terahertz absorption spectral structure database; constructing a text vector library in the field of terahertz spectroscopy detection; determining the user question type and extracting structured information; if the user question is a terahertz absorption spectroscopy diagnostic question, then analyzing influencing factors and generating a query question, rewriting the user question into a query question; retrieving the text vector library in the field of terahertz spectroscopy detection based on the query question; generating an answer through a large language model and returning it. This method analyzes and calculates the influencing factors of terahertz absorption spectroscopy based on a structured database and rewrites the query question into a query question more suitable for RAG retrieval, improving the efficiency of researchers in diagnosing absorption spectra and retrieving knowledge in the field of terahertz spectroscopy detection.

The application provides a kind of multi-agent-based spectral data autonomous analysis method, it is related to the technical field of data processing, method includes: through multi-agent cooperation mechanism, the whole process of spectral data from acquisition, quality evaluation, pretreatment, spectral peak analysis to phase inference is unified modeling and scheduling, data quality information is converted into constraintable pretreatment strategy by central coordination agent, and consistency check and conflict arbitration are implemented on spectral peak parameters and phase inference results at the end of analysis, so that adaptive processing and result check are realized under the premise of maintaining spectral structure correlation relationship.The application can improve the accuracy of spectral data analysis.

The present invention relates to a polarimetric (PT) measurement device for anisotropic quantities of a medium (M), comprising a spectrally structured light source (SSS) delivering a series of pulses with reference spectral grooves, spectral polarization coding means (MPG) configured to spectrally polarize each of the reference pulses delivered by the light source with a measurement phase delay identical to the reference phase delay, a light detector (D) configured to capture the light intensity reflected by the medium (M) for each of the reference pulses delivered by the light source, and a measurement unit configured to determine the anisotropic quantities of the medium (M) from the images captured by the detector. Figure 3

The invention discloses a segmentation and optimization method and system based on a multispectral spectral image structure, and relates to the technical field of spectral image processing and image optimization. Multispectral spectral data of a slice to be detected are collected; the multi-spectral-band spectral data are preprocessed, and preprocessed multi-spectral-band spectral data are obtained; fusing the preprocessed multi-spectral-band spectral data to obtain tower type spectral data; dividing the tower-shaped spectral data into a plurality of spectral sub-segments according to wavelength intervals, calculating a spectral anomaly evaluation score for each spectral sub-segment, screening abnormal spectral sub-segments based on an adaptive threshold, and forming a primary spectral structure region based on the abnormal spectral sub-segments; based on the primary spectrum structure area, adaptive fine tuning is carried out on acquisition parameters, and multi-spectrum spectral data are acquired again; and repeating the steps on the new multi-spectral-band spectral data until the region boundary converges to obtain a final spectral region of interest, thereby realizing dynamic enhancement and segmentation optimization of the spectral image.

The present invention relates to a polarimetric (PT) measurement device for anisotropic quantities of a medium (M), comprising a spectrally structured light source (SSS) delivering a series of pulses with reference spectral grooves, spectral polarization coding means (MPG) configured to spectrally polarize each of the reference pulses delivered by the light source with a measurement phase delay identical to the reference phase delay, a light detector (D) configured to capture the light intensity reflected by the medium (M) for each of the reference pulses delivered by the light source, and a measurement unit configured to determine the anisotropic quantities of the medium (M) from the images captured by the detector. Figure 3

The invention discloses a hyperspectral image classification method based on a tensor width learning network, and relates to the technical field of image classification. According to the method, high-order tensor data can be directly processed in different scenes, the three-dimensional structure of hyperspectral imaging data can be well matched, space and spectral structure information is captured, and high stability is achieved; feature nodes are generated from a space dictionary and a spectrum dictionary, uncertainty caused by network random feature mapping can be well avoided, and the features of the feature nodes are more interpretable; the width network architecture adopted by the method has the characteristic of low time complexity, and meanwhile, the spatial-spectral characteristics of the hyperspectral image are considered, so that the training speed is higher, and the classification accuracy is higher.

The invention relates to the technical field of hyperspectral three-dimensional reconstruction, in particular to a skin tumor layered structure three-dimensional reconstruction system based on a hyperspectral image. The system is used for collecting multi-source hyperspectral skin monitoring data; performing spectrum registration and fusion processing on the multi-source hyperspectral skin monitoring data to generate fused spectrum skin monitoring data; analyzing skin monitoring spectrum structure feature data based on the fused spectrum skin monitoring data; performing skin deep tissue component characterization characteristic analysis processing according to the skin monitoring spectrum structure characteristic data to generate skin deep tissue component characterization characteristic data; according to the skin deep tissue component characterization characteristic data, carrying out skin tumor layered structure characterization and flowing three-dimensional modeling processing to generate a skin tumor layered structure model; and performing skin tumor trend evolution processing on the skin tumor layered structure model to generate a skin tumor trend layered structure model. According to the invention, the three-dimensional modeling of the skin tumor layered structure with precise representation is realized.

The application relates to the field of computer vision, and particularly discloses an image segmentation method and system, wherein a boundary distribution map is predicted by a multi-scale boundary distribution generation module as prior knowledge, a space-frequency domain joint supervision strategy is adopted, the boundary distribution map simultaneously approximates a real boundary in a spatial position and a spectral structure, false edge responses caused by local high-frequency interference are effectively inhibited, and the continuity and stability of the boundary are improved; an effective receptive field is expanded by a state space semantic enhancement module, and global semantic consistency is enhanced; a boundary-guided cross-scale decoder is combined with a boundary feature enhancement module, boundary-related responses are strengthened in multi-resolution fusion, and collaborative optimization of region positioning and boundary refinement is realized, so that accurate and complete polyp segmentation results are obtained under complex imaging conditions such as low contrast, various morphologies, and interference such as reflection and wrinkles, and reasonable calculation overhead is maintained.

The application discloses a fast area mapping heterospectral structured light measurement method, which can realize the area contour acquisition of a measured object only according to a square grid pattern, and then complete the area coordinate mapping of a projection device plane. Compared with the current method for acquiring the phase through horizontal and vertical stripes, a large number of stripe images are no longer needed for coordinate conversion, the efficiency can be significantly improved, and the fast measurement of a dynamic scene can be realized. Different regions are set to different spectrums for simultaneous projection, compared with the current method for sequentially projecting different regions, the measurement efficiency can be obviously improved. The application provides the feasibility for the dynamic measurement of a mutual reflection scene.

This invention discloses a method and system for calculating the ultra-fine spectrum of a sampled signal, belonging to the field of weak signal detection technology. Calculating the spectrum of signals with extremely long sampling times presents problems such as large data volume and high computer memory requirements. This method divides the sampled data sequence of an ultra-low bandwidth modulated signal into segments of equal duration. Each segment is then subjected to quadrature demodulation, filtering, and decimation concatenation. The processed segments are then sequentially concatenated, and finally, a time-frequency transformation is performed to obtain the signal spectrum with ultra-fine frequency resolution. This calculation method significantly reduces the amount of data required for time-frequency transformation, allowing the processing of massive amounts of sampled data to be completed on a conventionally configured computer, yielding signal spectra with frequency resolution at the millihertz or even microhertz level. This allows for the observation and understanding of the extremely low-frequency spectral structure or near-end phase noise of the signal. This calculation method has strong versatility.

The invention provides a device, method and equipment for generating a wireless baseband signal and a medium, and the device comprises a frequency domain data input module which is used for receiving frequency domain subcarrier data; the phase rotation module is used for applying complex phase rotation related to a subcarrier index to input frequency domain subcarrier data to generate at least one group of rotated frequency domain data; the inverse fast Fourier transform module is used for performing inverse fast Fourier transform of a smaller point number on the original frequency domain data and the rotated frequency domain data for at least two times, and outputting multiple time domain data segments; and the time domain splicing module is used for splicing the time domain segments in sequence to form a final time domain baseband signal of which the length is multiplied. By means of the structure, on the premise that the hardware scale is not remarkably increased, the signal with the frequency spectrum structure equivalent to large-point inverse fast Fourier transform output can be generated, the digital domain oversampling effect is achieved, and therefore the hardware implementation complexity and the design difficulty of a post-stage digital filter are effectively reduced.

The invention relates to the field of cell structure visualization, in particular to a macrophage structure feature visualization method and system based on image processing. The method comprises the following steps: acquiring a multi-modal macrophage image, carrying out multispectral structural feature analysis and fusion, and constructing a cross-modal feature fusion macrophage image; adaptive detail feature optimization processing is carried out on the cross-modal feature fusion macrophage image, and global brightness distribution enhancement is carried out, so that a global brightness optimization image is generated; performing microstructure morphological feature accurate quantization and three-dimensional point cloud modeling on the global brightness optimized image to construct a three-dimensional microstructure morphological model; and carrying out dynamic microstructure spatial-temporal change tracking and dynamic evolution trend analysis on the three-dimensional microstructure morphological model so as to generate a macrophage dynamic evolution trend chart. According to the method, the structure change of the macrophage is displayed through a clear image, so that the morphological characteristics and the evolution process of the cells can be conveniently and intuitively analyzed.

The application discloses a soil organic matter inversion method and system based on unmanned aerial vehicle multispectrum, relates to the technical field of soil spectrum detection, and comprises the following steps: S1, collecting multispectrum remote sensing data and soil characteristic data, and performing pretreatment on the collected data; S2, constructing a wave band reflectivity vector and a spectral form offset value, and generating a spectral structure matrix; S3, constructing a spectral feature matrix, performing correlation screening on the spectral feature matrix to obtain an effective feature matrix, performing resampling screening processing on the effective feature matrix, and constructing an inversion input feature matrix; and S4, outputting a soil organic matter inversion sequence based on a random forest regression model, performing spatial interpolation and grid construction on the soil organic matter inversion sequence, and generating a soil organic matter inversion image. The application solves the problems that in the existing unmanned aerial vehicle multispectrum soil organic matter inversion, feature redundancy and remote sensing reflectivity disturbance are superimposed, the feature screening result is unstable, and the model cross-plot generalization consistency is insufficient.

A computer-implemented method for generating a prediction of a molecular formula and molecular structure of an unknown chemical compound is provided. Nuclear magnetic resonance (NMR) spectral data of the unknown chemical compound comprising at least one of a 13C NMR spectrum and a 1H NMR spectrum corresponding to the unknown chemical compound is received as input. The NMR spectral data is provided to a machine learning framework pretrained on a plurality of spectra-structure pairs, wherein each pair comprises (i) a NMR spectral representation and (ii) a corresponding molecular structure or molecular formula. The NMR spectral data corresponding to the unknown chemical compound is processed by the machine learning model to generate a list of one or more candidate molecular formulas and corresponding candidate molecular structures. The list of candidate molecular formulas and corresponding candidate molecular structures for the unknown chemical compound is provided as output.

The application discloses a segmentation and optimization method and system based on multispectral spectral image structure, relates to the technical field of spectral image processing and image optimization, and collects multispectral spectral data of a to-be-measured slice; the multispectral spectral data is preprocessed to obtain preprocessed multispectral spectral data; the preprocessed multispectral spectral data is fused to obtain tower spectral data; the tower spectral data is divided into a plurality of spectral subsegments according to wavelength intervals, a spectral anomaly evaluation score is calculated for each spectral subsegment, an abnormal spectral subsegment is screened based on an adaptive threshold, and a primary spectral structure region is formed based on the abnormal spectral subsegment; based on the primary spectral structure region, adaptive fine tuning is performed on collection parameters, and multispectral spectral data is re-collected; the above steps are repeated on the new multispectral spectral data until the region boundary converges, and a final spectral region of interest is obtained, so that dynamic enhancement and segmentation optimization of the spectral image are realized.

The invention provides a hyperspectral data dimension reduction method and device capable of reserving spatial information, a terminal and a medium. The method comprises the following steps: acquiring an initial hyperspectral image; adding a space structure retention degree as an optimization item, and carrying out minimum noise separation transformation on the initial hyperspectral image to obtain a preliminary dimension reduction image; and based on the spectral structure similarity between the preliminary dimension-reduced image and the initial hyperspectral image, selecting a wave band in the preliminary dimension-reduced spectral image to obtain a final dimension-reduced image. According to the method, a better dimension reduction effect can be realized while space information is reserved.

The invention provides an antibacterial peptide de novo design method and system based on a spectrum structure-activity relationship. The method comprises the following steps: obtaining structures, sequences and simulated spectrum data of functional peptides including antibacterial peptides, anti-cancer peptides, anti-inflammatory peptides and the like; constructing a spectral diffusion generation model, and generating spectral data corresponding to the target function; mapping the spectrum into a CA atomic distance matrix through a neural network model, and generating an initial three-dimensional structure in combination with PyRosetta; optimizing the initial structure by using an RFdiffusion model to obtain a high-resolution full-atomic structure; inputting the optimized structure into a retrained ProteinMPNN model, and generating a candidate polypeptide sequence in combination with a function guidance strategy; and finally, performing function screening and structure verification on the sequence by using a deep learning model. The technical problems that an existing antibacterial peptide design method neglects peptide three-dimensional structure information and lacks an optimal design strategy for multiple functions such as antibiosis, anticancer and anti-inflammation, so that design flexibility and functional diversity are low are solved.

The application belongs to the field of underwater target recognition by sonar, and particularly relates to a target material classification method based on a time-frequency dual-path attention fusion network, which comprises a data acquisition module, an original echo data acquisition module based on an image sonar system; a preprocessing module, which pre-processes the original echo data; and a TF-CTAM-InceptionTime model, which outputs after processing the input pre-processed data. The application applies time-domain multi-scale convolution and frequency-domain FFT feature extraction to sonar target classification, fully utilizes the complementary information of time-domain waveform details and frequency-domain spectral structure, adopts a double attention mechanism to adaptively focus on the key frequency components and important moments of sonar echoes, enhances the distinguishing ability of target materials, introduces residual connection and bottleneck structure on the basis of InceptionTime to improve the stability of deep network training and accelerate convergence, avoids information loss caused by image-level processing, retains complete amplitude and phase information, and realizes efficient recognition and classification of target shape and material.

The application provides a hyperspectral data dimension reduction method and device reserving spatial information, a terminal and a medium. The method comprises the following steps: obtaining an initial hyperspectral image; adding a spatial structure reservation degree as an optimization item, performing minimum noise separation transformation on the initial hyperspectral image to obtain a preliminary dimension reduction image; and selecting bands in the preliminary dimension reduction image based on the spectral structure similarity between the preliminary dimension reduction image and the initial hyperspectral image to obtain a final dimension reduction image. The application can achieve better dimension reduction effect while reserving spatial information.

The invention discloses a different-spectrum structured light measurement method for rapid region mapping, which can realize acquisition of a region contour of a measured object only according to a square pattern, and further complete region coordinate mapping of a projection equipment surface. Compared with a current method for obtaining the phase through transverse and longitudinal stripes, a large number of stripe images are not needed for coordinate conversion any more, efficiency can be remarkably improved, and rapid measurement of a dynamic scene is achieved. Different spectrums are set in different areas for simultaneous projection, and compared with a current method of sequentially projecting different areas, the measurement efficiency can be obviously improved. And feasibility is provided for dynamic measurement of a mutual reflection scene.

The application discloses a soil organic matter inversion method and system based on unmanned aerial vehicle multispectrum, relates to the technical field of soil spectrum detection, and comprises the following steps: S1, collecting multispectrum remote sensing data and soil characteristic data, and performing pretreatment on the collected data; S2, constructing a wave band reflectivity vector and a spectral form offset value, and generating a spectral structure matrix; S3, constructing a spectral feature matrix, performing correlation screening on the spectral feature matrix to obtain an effective feature matrix, performing resampling screening processing on the effective feature matrix, and constructing an inversion input feature matrix; and S4, outputting a soil organic matter inversion sequence based on a random forest regression model, performing spatial interpolation and grid construction on the soil organic matter inversion sequence, and generating a soil organic matter inversion image. The application solves the problems that in the existing unmanned aerial vehicle multispectrum soil organic matter inversion, feature redundancy and remote sensing reflectivity disturbance are superimposed, the feature screening result is unstable, and the model cross-plot generalization consistency is insufficient.

This application relates to the field of lighting technology, and discloses a spectral control method, device, computer equipment, and storage medium, including receiving a brightness adjustment command, the brightness adjustment command carrying target brightness level information; searching for the driving parameters of each light-emitting channel from a preset driving parameter table according to the target brightness level information, or calculating the driving parameters of each light-emitting channel according to a preset algorithm; outputting the driving parameters to a multi-channel LED driving circuit, so that the driving circuit adjusts the driving current or driving duty cycle of each channel, controlling the multi-channel LED to output a composite spectrum that matches the target brightness level and has high color recognition; the composite spectrum has a preset target spectral structure at the target brightness level, so as to solve the technical problem that the color rendering effect and brightness perception of existing 2700k–3500k warm white LED light sources deteriorate in low-illuminance environments.

This invention discloses a method and system for detecting operational anomalies in electromechanical equipment based on multidimensional vibration spectrum feature analysis, belonging to the field of electromechanical equipment condition monitoring technology. This method performs layered modeling of the spectrum based on its spectral structure characteristics, extracting harmonic structure features, sideband cluster features, and impact residual features to construct multidimensional spectral structure characterization data. The spectral structure features are object-orientedly encapsulated into spectral elements, and combined with the spatial location of measurement points and the consistency of frequency band response, a cross-measurement point spectral element correlation diagram is constructed to extract anomaly propagation features. Finally, anomaly determination and location analysis are performed on the operating status of the electromechanical equipment. This invention can achieve unified expression and structured analysis of spectral features under complex operating conditions, improving the accuracy and stability of anomaly detection and enhancing the ability to characterize anomaly propagation paths, thus possessing significant engineering application value.

The invention relates to the field of computer vision, and particularly discloses an image segmentation method and system, a boundary distribution diagram is predicted through a multi-scale boundary distribution generation module to serve as prior, and a space-frequency domain joint supervision strategy is adopted, so that the boundary distribution diagram approaches a real boundary at the spatial position and the frequency spectrum structure at the same time; the pseudo edge response caused by local high-frequency interference is effectively suppressed, and the continuity and stability of the boundary are improved; an effective receptive field is expanded through a state space semantic enhancement module, and global semantic consistency is enhanced; according to the method, a boundary guide cross-scale decoder is combined with a boundary feature enhancement module, boundary related response is enhanced in multi-resolution fusion, collaborative optimization of area positioning and boundary refining is achieved, and therefore an accurate and complete polyp segmentation result is obtained under the complex imaging conditions of low contrast, various forms and interference of reflection, plica and the like. And reasonable calculation overhead is kept.

The invention discloses a self-adaptive Raman measurement method and system based on spectral structure decoupling. The method comprises the following steps: firstly, defining a parameter space containing laser power and integral time, and carrying out a small amount of initial measurement in the parameter space to obtain an initial spectrum data set; decoupling processing is carried out on each spectrum, and each spectrum is decomposed into a scale component representing the overall intensity level and a structure component representing the relative morphological characteristics; constructing a probability spectrum proxy model capable of predicting spectrum distribution under any parameter based on the measurement parameters and the decoupled scale and structure components; performing spectrum quality prediction on candidate measurement parameters by using the model and calculating uncertainty; selecting a target parameter of the next measurement by fusing a decision function of prediction quality and uncertainty; after measurement is executed, a data set is updated, a model is reconstructed, iteration is performed until a termination condition is met, optimal measurement parameters and spectra are output, and the automation degree, efficiency and result stability of Raman measurement are remarkably improved.

A method and a device for generating a multidimensional XCT measurement data set ((Kx, Ky, Kz)) of a sample (P) by means of spectrally multi-structured scattering imaging are disclosed. A plurality of scattering images (19) are generated by a measurement process comprising the steps of: a) Structuring a spectrum of an EUV / X-ray radiation (6) as a structured frequency range to provide an EUV / X-ray measuring beam (12) with a measuring beam spectrum comprising narrow-band structured frequency ranges, wherein at least one axial frequency space coordinate (Kz) in the frequency space (Kx, Ky, Kz) is assigned to each of the narrow-band structured frequency ranges, b) irradiating the EUV / X-ray measuring beam (12) onto a spatially laterally extended surface section of the sample (P), c) detecting scattered radiation which is scattered back by the sample (P) with a detector (17) which outputs intensity values ​​for the structured frequency range belonging to the measuring process, each of which is assigned at least one lateral frequency space coordinate (Kx, Ky) in the frequency space (Kx, Ky, Kz), the intensity values ​​forming the scattering pattern (19) for the associated measuring process. The intensity values ​​can be combined to form a multidimensional XCT measurement data set ((Kx, Ky, Kz)), wherein the positions of the intensity values ​​in the frequency space (Kx, Ky, Kz) are given by at least one lateral frequency space coordinate (Kx, Ky) and one axial frequency space coordinate (Kz). Furthermore, computer-implemented methods for reconstructing a multidimensional structure of a sample from a multidimensional XCT measurement data set are disclosed.