32 results about "Spectral transformation" patented technology

The invention discloses a hyperspectral image classification method based on double-flow heterogeneous spatial feature collaboration. The method comprises the following steps: step 1, obtaining a hyperspectral image classification data set; step 2, designing an overall structure of a double-flow heterogeneous spatial feature collaborative modeling method (DHSFCM); step 3, setting experimental parameters; according to the hyperspectral image classification method based on double-flow heterogeneous spatial feature collaboration, noise reduction and feature enhancement are carried out through a spectrum transformation layer, then complementary features are extracted through a non-Euclidean feature extraction network (NEFEN) and an Euclidean feature extraction network (EFEN), and the hyperspectral image classification method based on double-flow heterogeneous spatial feature collaboration is obtained. And effective feature integration and interaction are realized by means of a cross attention fusion module (CAFM), and finally a classification result is output. According to the technology, accurate classification support can be provided for intelligent agriculture, resource exploration and ecological monitoring, and a new thought with engineering landing value is provided for intelligent analysis of hyperspectral images.

The invention discloses a pavement crack accurate extraction method and system based on multi-scale image segmentation, and belongs to the technical field of pavement detection, and the method comprises the steps: obtaining a to-be-detected pavement image, carrying out the brightness normalization and geometric correction of the pavement image, and constructing a multi-scale image set with different resolutions; generating a spatial texture channel and a spectral domain response channel for each scale image in the multi-scale image set, and fusing the spatial texture channel and the spectral domain response channel to obtain a spectral-space coupling input tensor, wherein the spectral domain response channel is obtained by extracting energy characteristics of a plurality of frequency bands after performing spectrum transformation on the scale image; a multi-scale image set is constructed for road surface images, and coupling expression of spatial texture information and spectral domain response information is introduced, so that feature consistency of fine cracks under different resolutions is enhanced, and scale drift and response distortion caused by shadows, light reflection and seam textures are inhibited at the same time; therefore, the separability and the stability of the crack in a complex scene are improved.

A system may generate embedded subsets by projecting each subset of the subsets into a vector space to generate a corresponding embedded subset. A system may encode the embedded subsets into an encoded image using a dataset encoder including a gated spectral state space model, the gated spectral state space model being a gated neural network that includes a spectral state space model, the spectral state space model being a state space model that represents features of the input dataset using at least a spectral transformation of each embedded subset of the embedded subsets. A system may predict a classification for the input dataset using the encoded image.

To provide a method and an audio processing unit for decoding an encoded audio bitstream.SOLUTION: A method for decoding an audio bitstream includes receiving an encoded audio bitstream, decoding audio data to generate a decoded low-band audio signal, extracting high frequency reconstruction metadata and filtering the decoded low-band audio signal with an analysis filterbank to generate a filtered low-band audio signal, and extracting a flag indicating whether spectral transformation or harmonic transposition is to be performed on the audio data and regenerating a high-band portion of the audio signal using the high frequency reconstruction metadata and the filtered low-band audio signal according to the flag.SELECTED DRAWING: Figure 1

The invention discloses an adversarial sample generation method and an evaluation method for robustness evaluation of an AIGI detector, and relates to the technical field of robustness evaluation. A pre-trained substitution model is selected and comprises a feature extractor and a classifier, K additional models are added behind the feature extractor in parallel, and a Bayesian model is constructed and used for simulating an attacked model; performing frequency domain attack on the Bayesian model by using the adversarial sample, during each attack, adding disturbance to the spatial domain of the original adversarial sample, converting the original adversarial sample from the spatial domain to the frequency domain, performing random spectrum transformation, and according to an attack optimization target, calculating a frequency domain gradient for updating the adversarial sample; furthermore, the frequency domain attack and the space domain attack are mixed, the space domain gradient is calculated during each attack, the frequency domain gradient and the space domain gradient are added and averaged to obtain a uniform gradient direction, the gradient direction is used to update the adversarial sample, and the final adversarial sample is obtained after the set iteration attack times. According to the method, an adversarial sample is generated by using a frequency-based post-training Bayesian attack (FPBA), so that high-quality attacks with certain generalization ability are performed on the AIGI detector, and the robustness of the AIGI detector is evaluated under white-box attacks and black-box attacks.

This invention discloses a method for classifying abnormal heart sounds, comprising: a data input module receiving collected raw heart sound data; a data preprocessing module preprocessing the raw heart sound data, including reading feature parameters of the raw data, pre-emphasizing the heart sound data, eliminating high-frequency signals, and controlling the degree of pre-emphasis by adjusting the pre-emphasis coefficient until high-frequency signals are eliminated; normalizing the data to a value range of 0 to 1; obtaining spectral data through spectral transformation and generating Mel-spectral cepstral coefficients; generating a training dataset and a validation dataset; a training module generating a gated recurrent unit network (GRN) model and training and optimizing the parameters of the GRN model using the training dataset; a validation module importing the parameters of the training model, inputting the validation dataset into the validation module, and validating the accuracy of the GRN model; and an output module outputting the classification results. This method enables automatic classification of abnormal heart sounds, reducing the workload of doctors and improving work efficiency; it also avoids the influence of human factors and enhances the reliability of diagnostic results.

The application relates to the technical field of soil monitoring, and discloses a method and system for monitoring soil pollutants in a power transmission and transformation project, wherein the method comprises the following steps: collecting hyperspectral image data in a power transmission and transformation project area; processing the hyperspectral image data by adopting a multistage spectral transformation mechanism to obtain characteristic wave bands representing pollutant contents; inputting the characteristic wave bands into a preset spectral inversion model to generate a concentration spatial distribution map of soil pollutants; constructing a soil and groundwater diffusion model according to the concentration spatial distribution map; and generating a migration path identification map of soil pollutants and an underground pollution risk area identification map based on the soil and groundwater diffusion model. The method has the beneficial effect of improving pollution identification precision and migration prediction accuracy.

The invention discloses a binocular stereo matching depth estimation method and system based on hyperspectral reconstruction feature enhancement. The method comprises the following steps: S1, receiving left and right binocular RGB images; s2, performing high-dimensional spectral feature reconstruction on the left binocular RGB image and the right binocular RGB image by using a multi-stage spectral transformation module, and outputting a multi-channel hyperspectral feature map covering a visible light wave band by learning spectral reflectivity priori of a material; s3, adaptive channel selection and feature recombination are carried out on the multi-channel hyperspectral feature map through a learnable spectrum dimension reduction layer, and a left pseudo-color feature map and a right pseudo-color feature map with enhanced material physical attribute differences are generated; and S4, performing geometric feature extraction and parallax iterative optimization based on the left and right pseudo-color feature maps by using a stereo matching module, and outputting a final compact parallax map. According to the method, optical physical priori is learned through model training, so that low-cost RGB hardware can reproduce high-fidelity spectral features, and the problem of depth perception under complex materials and extreme shadows is effectively solved.

To disclose a method for decoding an encoded audio bitstream.SOLUTION: A method includes receiving an encoded audio bitstream and decoding audio data to generate a decoded lowband audio signal. The method further includes extracting high frequency reconstruction metadata and filtering the decoded lowband audio signal with an analytical filterbank to generate a filtered lowband audio signal. The method also includes extracting a flag indicating whether either spectral translation or harmonic transposition is to be performed on the audio data, and regenerating a highband portion of the audio signal using the filtered lowband audio signal and the high frequency reconstruction metadata in accordance with the flag. The high frequency regeneration is performed as a post-processing operation with a delay of 3010 samples per audio channel.SELECTED DRAWING: Figure 5

The application discloses a method for detecting thiocyanate in a complex serum sample and application. The method comprises the following steps: pretreating a serum sample, performing SERS detection, performing on-chip spectral quality control, and performing ratio type quantification. The application realizes rapid, trace, accurate and reliable quantitative detection of SCN ‑ In the complex serum sample by combining an interface self-assembly gold nano array base, 4-MBA ratio type internal standard correction and an on-chip quality control strategy based on ultrafiltration induced spectral transformation, and can be applied to the fields of smoking exposure evaluation and smoking cessation process monitoring.

This invention relates to the field of crack detection technology, and discloses a method and apparatus for detecting warehouse wall cracks based on three-dimensional laser point cloud data. The method includes: acquiring cross-sectional data of the target warehouse wall; performing spectral transformation on the cross-sectional data to obtain the cross-sectional data spectrum; acquiring three-dimensional laser point cloud data of the target warehouse wall; separating the warehouse wall control contour from the three-dimensional laser point cloud data based on the cross-sectional data spectrum; determining the texture distribution features of each cross-section of the target warehouse wall based on the control contour elevation; calculating a segmentation threshold for the cross-section based on the texture distribution features; determining suspected crack points for each cross-section based on the segmentation threshold and the control contour elevation; combining each suspected crack point to obtain a binary image representing the crack; performing region recognition on the binary image representing the crack to obtain crack regions; and detecting cracks in the target warehouse wall based on the crack regions. This invention can improve the accuracy of warehouse wall crack detection.

An adaptive near-infrared spectral transformation method based on correlation and Gaussian curve fitting is proposed, belonging to the field of near-infrared spectral transformation technology. This invention addresses the significant errors in the quantitative analysis of analytes caused by peak overlap in near-infrared spectra. The method determines the number of Gaussian functions involved in Gaussian curve fitting using the number of discrete points in the near-infrared spectrum of the analyte, and determines the center position of the Gaussian function using the wavelength positions of the discrete points. Correlation analysis is used to determine the optimal bandwidth of the Gaussian function for extracting overlap information. Based on this, a set of equations for curve fitting is constructed. The height of the Gaussian function is determined by solving the equations, and the area integral of the Gaussian function is performed to obtain the transformed spectrum of the analyte, thereby constructing a content prediction model for the analyte. The proposed method has been applied to the prediction of COD content in wastewater and moisture content in corn, respectively. The prediction mean square error is reduced by at least 25% compared to before the transformation, indicating that the Gaussian function involved in curve fitting does not need to correspond to the true peak information to effectively decompose and recombine the overlap information in the original spectrum, thus reducing quantitative analysis errors.

This invention discloses a method for monitoring salinized arable land based on spectral analysis, comprising the following steps: Step 1: Data preparation: Collect and process data covering the target area; Step 2: Spectral index calculation and fusion: Fuse the salinity index after ground hyperspectral transformation with the salinity index calculated by Sentinel-2 multispectral analysis at the same time through univariate linear regression to generate a hyperspectral-multispectral fusion index dataset and construct a feature variable library; Step 3: Feature variable selection: Select key features from the feature variable library; Step 4: Support vector machine model construction and training: Train the support vector machine regression model; Step 5: Salinity inversion and accuracy verification: Output the predicted soil salinity value, generate a spatial distribution map of soil salinity in the Hetao Irrigation Area based on the predicted soil salinity value, and classify the salinization level of arable land based on the inversion results; Step 6: Dynamic monitoring: Repeat steps up to step 5 to generate a dynamic change map of salinization.

This invention discloses a hyperspectral image classification method, medium, device, and product, relating to the field of image processing. The method includes: performing standardized preprocessing and superpixel segmentation on a hyperspectral image, and constructing a superpixel graph structure; constructing a hyperspectral image classification model including three parallel branches; each branch includes a spectral transform sub-network, a superpixel-level graph sub-network, and a pixel-level convolutional sub-network, wherein the spectral transform sub-network is a convolutional network with different kernel sizes, used to extract features from the hyperspectral image; based on the hyperspectral image features and graph structure, the parallel superpixel-level graph and pixel-level convolutional sub-networks extract superpixel-level and pixel-level features and perform feature concatenation and fusion, and obtain a classification probability distribution based on the fused features; the three parallel branches of the classification model are trained independently, and the trained model is used to classify hyperspectral images; the classification results of the three branches are determined by a majority voting method to determine the final classification result.

The invention discloses a soil heavy metal hyperspectral remote sensing inversion method, device and system and a storage medium. The method comprises the following steps: acquiring indoor hyperspectral data, heavy metal copper content and hyperspectral remote sensing image spectral data; removing noise wave bands from the indoor hyperspectral data, and then carrying out SG spectrum smoothing; correcting spectral data of the hyperspectral image according to the indoor hyperspectral data subjected to SG spectrum smoothing processing; performing fractional order differential spectral transformation on the corrected spectral data of the hyperspectral image; taking the content of heavy metal copper as a dependent variable, taking spectral data of the hyperspectral image after fractional differential spectral transformation as an independent variable, inputting the dependent variable into a Transform feature selection framework, and establishing a soil copper content inversion model by using XGBoost to verify the effectiveness of Transform feature selection. By adopting the technical scheme of the invention, the characteristic waves of the heavy metals can be quickly and effectively identified and extracted.

The present application relates to a high-fidelity key cache compression method and system based on spectral quantization, comprising: step S1: spectral transformation and feature analysis process; receiving original key cache data from the model, mapping the original key cache data from the time domain to the frequency domain, and deeply analyzing the signal structure by using the energy concentration characteristics unique to the discrete cosine transform (DCT); step S2: frequency domain hybrid quantization encoding process; by using the main frequency extraction mechanism, the high frequency pre-emphasis technology and the hybrid bit width allocation strategy, the frequency domain coefficient is compressed into a sparse representation with extremely low bit; step S3: fusion decoding and sparse accumulation process; by using the hardware-aware fusion operator and the delay inverse transform technology, the attention score is calculated directly in the compression domain. The present application realizes the limit compression and efficient inference under the lossless precision.

The invention relates to the technical field of mineral resource exploration, in particular to a bauxite lithium-rich ore rapid optimization method and system based on an imaging hyperspectrum. The method comprises the following steps: collecting standard hyperspectral data according to a bauxite sample; using the standard hyperspectral data to construct a multi-dimensional fusion spectral data set; screening optimal characteristic wave band spectral data from the multi-dimensional fusion spectral data set; and constructing a lithium ore rapid optimization model based on the optimal characteristic wave band spectral data, and generating a lithium content spatial distribution thermodynamic diagram through the lithium ore rapid optimization model. According to the method, efficient and accurate recognition of the lithium-rich ore in the bauxite is achieved through multi-dimensional spectrum transformation, characteristic wave band screening and integrated estimation model construction, and the problems that a traditional method is high in cost, low in efficiency and insufficient in optimization precision are solved.

The application discloses a graph data backdoor defense method and device based on spectral domain transformation and edge weight learning, equipment, medium and program product, relates to the technical field of network security, and includes: obtaining and constructing a feature vector matrix and an eigenvalue matrix of an original graph data set; mapping each feature vector matrix from a spatial domain to a spectral space through spectral transformation to obtain a plurality of spectral domain node features, so as to construct a normal data distribution of the original graph data set, and determine whether each to-be-tested graph data is abnormal graph data; if it is determined that the graph data is abnormal, learning abnormal node features and abnormal adjacency relationships of the to-be-tested graph data through a multilayer perceptron to obtain abnormal edge weights; based on the abnormal edge weights, performing adaptive clipping on abnormal spectral components in the abnormal graph data through a Gaussian mixture model to obtain modified spectral domain node features; and performing inverse spectral transformation on each modified spectral domain node feature to reconstruct a target graph data set. The application can effectively resist multiple types of graph backdoor attacks.

The invention relates to the technical field of ecological environment monitoring, and discloses a mikania micrantha and eupatorium odoratum classification and identification method based on unmanned aerial vehicle hyperspectrum, and the method comprises the steps: obtaining unmanned aerial vehicle hyperspectral original image data of a target region, carrying out the preprocessing operation, carrying out the spectrum transformation and dimension reduction processing of an obtained standardized hyperspectral data set, and obtaining a target hyperspectral image; the method comprises the following steps: acquiring a plurality of groups of optimized feature data sets formed by combining different spectral transformations and dimension reduction processing, constructing a multi-type classification model based on the plurality of groups of optimized feature data sets, and screening out an optimal classification model through model training and performance evaluation so as to realize classification and identification of mikania micrantha and fragrant eupatorium herb in a target area and output a classification result. Therefore, the invention proposes and verifies a set of complete and optimized technical process from data acquisition, multiple pre-processing, PCA dimension reduction to construction of multiple classifiers, screens and determines an optimal technical combination for accurately identifying mikania micrantha and eupatorium odoratum, can effectively suppress noise, retains key distinguishing features, and realizes high-precision classification.

The invention discloses a deep soil nitrogen content monitoring method, a server and a storage medium, and belongs to the field of soil component monitoring, and the method comprises the following steps: obtaining remote sensing image data of surface soil of a soil sampling point, and collecting a soil sample of the corresponding soil sampling point; performing preprocessing, spectral transformation and spectral feature extraction on the remote sensing image data, and constructing a multi-dimensional feature space; analyzing the corresponding soil sample to obtain a change rule of surface nitrogen and deep nitrogen, and obtaining a surface-deep nitrogen relation model; screening a characteristic wave band combination of which the correlation with the surface nitrogen content exceeds a threshold value from the multi-dimensional characteristic space, and establishing a surface nitrogen content inversion model; in the actual use process, after remote sensing image data are collected, the nitrogen content of deep soil is monitored through the surface layer nitrogen content inversion model and the surface layer-deep layer nitrogen element relation model. The method avoids the problems of high cost and low timeliness of traditional sampling analysis.

The present application provides a kind of vegetation classification method, device, equipment and medium based on space-spectrum neural network, method includes: high spectral dataset is divided into training set, verification set and test set, and quantitative evaluation index is set;Based on the information separation of space-spectrum mixed search space, the differentiable architecture search strategy based on gradient optimization is used to search the neural network architecture, and the target network architecture is obtained, the information separation spectrum transformation operator for extracting spectral features is included in the space-spectrum mixed search space, and the information separation space attention depth convolution operator for extracting spatial features;Based on training set and verification set, the target network architecture is trained to obtain a classification model;Test set is input into the classification model to obtain vegetation classification result, and the classification result is evaluated according to quantitative evaluation index, solves the problem that classification accuracy and efficiency still have room for improvement in complex vegetation fine classification task, cannot meet the demand of large-scale, high-precision remote sensing monitoring.

The invention discloses a graph data backdoor defense method and device based on spectral domain transformation and edge weight learning, equipment, a medium and a program product, and relates to the technical field of network security, and the method comprises the steps: obtaining and constructing a feature vector matrix and a feature value matrix of an original graph data set; mapping each feature vector matrix from a spatial domain to a spectral space through spectral transformation to obtain a plurality of spectral domain node features so as to construct normal data distribution of the original graph data set, and judging whether each piece of to-be-detected graph data is abnormal graph data or not; if the to-be-detected graph data is judged to be abnormal graph data, abnormal node features and abnormal adjacency relations of the to-be-detected graph data are learned through a multi-layer perceptron, and abnormal edge weights are obtained; on the basis of each abnormal edge weight, performing adaptive cutting on abnormal spectrum components in the abnormal graph data through a Gaussian mixture model to obtain corrected spectral domain node features; and performing inverse spectral transformation on each corrected spectral domain node feature, and reconstructing to obtain a target graph data set. According to the method, multi-type graph backdoor attacks can be effectively resisted.

The invention belongs to the technical field of remote sensing data processing, and particularly discloses a hyperspectrum-based soil oxide content extraction and analysis method, which comprises the following steps of: performing at least two spectrum transformations on a hyperspectral original reflectivity spectrum of a target area to obtain a plurality of transformed spectrums, respectively extracting at least two spectrum characteristic parameters from the original reflectivity spectrum and the plurality of converted spectrums to generate a candidate characteristic set; for specific soil oxides, selecting at least one spectral characteristic parameter from the candidate characteristic set as an independent variable, constructing a plurality of candidate inversion models by adopting at least two wave band combination algorithms, and determining an optimal inversion model; and performing pixel-by-pixel calculation on the remote sensing image data of the target area by using the optimal inversion model to generate a spatial distribution content diagram of the soil oxides. The method can comprehensively, systematically and efficiently extract the contents of various soil components, and further analyze the response relationship between the contents of various soil components and environmental factors.

This application provides a music retrieval method, a music retrieval device, an electronic device, and a storage medium, belonging to the field of artificial intelligence technology. The method includes: acquiring target descriptive text and candidate music, wherein the target descriptive text includes the target object's description of the music; performing word recognition on the target descriptive text to obtain genre description words; performing spectral transformation on the candidate music to obtain candidate music spectrum sequences; based on the candidate music spectrum sequences, obtaining candidate music genre representation vectors corresponding to the candidate music; performing genre identification on the candidate music based on the candidate music genre representation vectors to obtain genre tag data for the candidate music; filtering the candidate music based on the genre description words and genre tag data to obtain target music; and feeding the target music back to the target object. This application embodiment can improve the accuracy of music retrieval.

Method for generating a multi-tone radio frequency signal and associated radar transceiver device. This method for generating a radio frequency signal comprising N distinct elementary frequency components includes at least one transmission-reception, in calibration mode, of a calibration radio frequency signal and obtaining a spectral transformation of said calibration radio frequency signal into a line spectrum, performed by a processing module of the radar transceiver device, and a calibration of attenuation signals, comprising: - a determination (38), by measurement in the line spectrum or by calculation, of at least one intermodulation frequency of an intermodulation line to be attenuated, - for each intermodulation frequency, a determination (40) of amplitude and phase parameters of a corresponding attenuation signal, as a function of the amplitude levels of the lines in the line spectrum.the attenuation signal having a frequency equal to said intermodulation frequency. The generated multi-tone radio frequency signal (50) comprises respectively said N components at elementary frequencies and said attenuation signals. Figure for the abbreviation: Figure 3,

The invention relates to the technical field of fire detection and visual identification, in particular to an intelligent fire detection and positioning method based on visual identification, which comprises the following steps: collecting scene panorama continuous frame images through a video sensor group to generate a video source signal, and separating the video source signal into a basic visual stream and an auxiliary visual stream; and performing background dynamic modeling on the basic visual flow to obtain a steady-state characteristic model, performing spectral transformation processing on the auxiliary visual flow to generate an enhanced spectral response flow, extracting fire related spectral characteristics, comparing the fire related spectral characteristics with the steady-state model point by point, and marking a disturbance area. And carrying out spatial domain and time domain two-dimensional analysis on the disturbance area, calculating a fire behavior score in combination with spatial features such as contours, textures and colors and time domain features such as area, shape and position drift, and outputting a fire judgment result and a geographic positioning coordinate according to spatial-temporal information of the high-score disturbance area. The method can reduce environmental interference and improve fire identification accuracy and positioning precision.

This invention belongs to the field of signal processing technology, specifically relating to a jitter analysis method and system for high-speed serial signals. By identifying and separating the data jitter sequence DDJ from the total jitter sequence TIE in a high-speed serial signal, a first jitter sequence containing random jitter data RJ is obtained. A first spectrum is obtained by spectral transformation of the first jitter sequence, and a first power spectrum is calculated from the first spectrum. The total power and root mean square (RMS) value of the random jitter are obtained by calculating a portion of the first power spectrum, thus achieving RMS estimation of random jitter in high-speed serial systems, improving accuracy, and being applicable to signals with various code types. By selectively accumulating power spectrum segments, stable random jitter estimation is achieved, reducing estimation complexity and saving system resources.

The invention discloses a CDOM absorption coefficient inversion method and system based on a liquid neural network, and relates to the technical field of water ecological environment remote sensing monitoring. The method comprises the following steps: carrying out spectrum transformation on in-situ hyperspectral observation data based on a matching relationship between remote sensing data and hyperspectral data, and carrying out feature reconstruction on equivalent multispectral reflectivity by utilizing a mapping relationship between a remote sensing reflectivity spectrum and CDOM optical characteristics; constructing an LNN network with continuous time dynamic characteristics, and performing multi-round iterative training on the LNN network based on a remote sensing spectrum change process by using the training sample; and performing inversion on to-be-predicted remote sensing observation data by using the trained LNN network to obtain a CDOM absorption coefficient inversion result of the corresponding time step. According to the method, the LNN with continuous time dynamic modeling capability is utilized to simulate a nonlinear relationship between remote sensing spectral information and a CDOM absorption coefficient, and a CDOM remote sensing inversion model with both accuracy and reliability is constructed.

The invention discloses a full-performance test real-time monitoring method and system based on edge computing. The method comprises the following steps: acquiring electric energy meter data through test equipment, performing smoothing, interpolation and adaptive adjustment on the acquired data in edge calculation to supplement signal loss, performing frequency spectrum transformation on signals for multiple times, and performing weighted fusion on frequency spectrum results of each stage to obtain a composite frequency spectrum; the adaptive signal filtering algorithm based on composite limit detection is adopted to filter interference and abnormal signals, the abnormal quantification method is adopted to carry out quantitative evaluation on the filtered signals and the composite frequency spectrum, and real-time early warning is carried out on the fused signal features based on the neural-decision network model to detect performance abnormity and make a response in time. And real-time monitoring of the full-performance test is realized. According to the scheme of the invention, real-time and accurate detection and evaluation of the electric energy meter signal are realized.