91 results about "Band selection" patented technology

The invention discloses a physical information neural network hyperspectral band selection method for target identification, which uses physical information SDI as prior information to be combined with a channel attention mechanism added in a deep neural network to guide band selection, and evaluates the advantages and disadvantages of band weights through a reconstruction module after the weights are generated. Wave band selection is carried out accordingly; according to the hyperspectral image target recognition waveband selection method based on the physical information neural network, compared with a traditional method, the process is simplified and the precision is improved through end-to-end learning, the utilization of spatial context information can be enhanced by adding an attention mechanism, and the model has interpretability by adding prior SDI information.

The invention discloses a point source methane emission wave band sensitivity enhancement signal identification method and system. The method aims at accurately identifying methane emission plume observed by a satellite. The identification method comprises the following key steps: data acquisition and preprocessing, refined construction of a methane unit absorption spectrum, sensitive wave band screening, and integration of a methane enhanced signal identification algorithm. The method solves the problems of suppressing false positive and enhancing methane signal detection, reduces the error of a methane concentration inversion result, improves the accuracy and reliability of methane identification by optimizing band selection and integrating an inversion algorithm especially in the aspect of point source methane emission plume identification, and has a good application prospect in the field of point source methane emission plume identification. And high-reliability data support is provided for refined methane emission identification and control.

The invention discloses a hyperspectral image waveband selection method and system based on a three-dimensional convolution auto-encoder. The method comprises the following steps: carrying out block processing on an original hyperspectral image; designing a three-dimensional space-spectrum reconstruction network based on a global spectrum sensing module; constructing a combined loss function of a reconstruction error term and a sparse regularization constraint term, and performing iterative updating on the network by using an Adam optimizer; constructing a wave band comprehensive evaluation criterion based on triple constraints and a dynamic fusion mechanism; and successively selecting an optimal wave band by adopting a greedy algorithm according to a comprehensive evaluation criterion, and outputting a wave band selection result of the hyperspectral image. Through fusion of a three-dimensional spatial spectrum reconstruction module and a multi-head self-attention mechanism, local and global association of hyperspectral data is comprehensively mined, and deep joint representation of spectrum and spatial information is realized. Meanwhile, sparse regularization constraint is introduced into a loss function, and the network is guided to focus the most valuable key wave band.

For user equipment (UE) that operates in a multi-link, single-radio (MLSR) mode, an access point (AP) of a (e.g., WiFi) local area network calculates link usage capacities (LUCs) for the different associated bands and performs comparisons based on those LUCs to the client traffic (CT) to select the band(s) to be used to transmit that CT during the next communication session between the UE and the AP. For example, if at least one LUC is greater than CT, then the AP selects the band with the greatest LUC. Otherwise, if two times the LUCs are greater than CT for at least two bands, then the AP distributes CT to the bands proportionately and sequentially. Otherwise, the AP uses conventional spraying to select the band. This band-selection technique provides improved throughput compared to using only conventional spraying for band selection.

This invention provides a feature band selection method based on neural network model pruning. The method includes the following steps: acquiring a hyperspectral dataset; dividing the dataset into a training set, a validation set, and a prediction set; constructing a classification network model based on the dataset; constructing a loss function for training the classification network model; inputting the training set and validation set into the classification network model for training, pruning the feature bands, and selecting discriminative hyperspectral data bands; and using the trained classification network model and the selected bands to classify the prediction set to obtain the classification result. This invention reduces the required bands while maintaining the accuracy of the analysis technique, retaining bands with stronger discriminative power for classification, thereby significantly improving classification accuracy, enhancing the generalization ability of the classification model, and reducing the cost of acquiring hyperspectral data in industrial settings.

The invention discloses a tree species identification method based on multi-modal spectrum and texture features of a wood cross section. The method comprises the following steps: acquiring hyperspectral image data of the wood cross section; constructing a comprehensive similarity matrix based on a plurality of spectrum similarity indexes, and selecting a representative wave band subset from the hyperspectral image data by adopting a multi-strategy wave band screening mechanism; performing multi-scale wavelet fusion on the representative wave band subset to generate a single-channel fusion image with consistent spatial resolution, and extracting points of interest and spectral features thereof from the single-channel fusion image; generating a gray-scale base map based on the hyperspectral image data, extracting various complementary texture feature maps from the gray-scale base map, and screening out significant points of interest from the various texture feature maps; and fusing the spectral features and the texture features, constructing feature vectors for representing wood tree species, and inputting the feature vectors to a classification model to complete tree species identification. According to the invention, rapid, accurate and intelligent identification of wood tree species can be realized.

The invention discloses a quantum behavior multi-target waveband selection method for hyperspectral target detection, and the method comprises the steps: obtaining a hyperspectral image, and initializing particle populations represented by different hyperspectral waveband subsets in the hyperspectral image; constructing a multi-objective optimization model for hyperspectral image waveband selection, wherein the model is used for measuring the information amount of each particle, the independence between wavebands and the data space retentivity; determining a non-dominated solution through an inter-particle game based on a multi-objective optimization model, generating global and local optimal particles, and updating binary candidate waveband vectors; removing low-separability wavebands in the binary candidate waveband vectors through a real-time constraint target waveband selection method to generate a new particle population; and obtaining a final non-dominated solution based on particle evolution, and selecting a final waveband subset from the non-dominated solution by calculating a target background separation perception robustness score.

An illustrative surgical system may access a plurality of images captured outside a structure within a patient; detect a difference between spectral reflectances of scenes captured in the plurality of images; and identify, based on the detected difference between the spectral reflectances of the scenes captured in the plurality of images, pixels in at least one of the plurality of images that correspond to structure tissue of the structure.

The application discloses a swarm intelligence optimization wave band selection method for hyperspectral image target detection, and comprises the following steps: S1, reading hyperspectral image data X, target signal d, and the number K of wave bands needed to be selected; S2, establishing a multi-objective optimization model based on information entropy, joint spectrum-space similarity and CEM error; S3, starting iteration, determining a global optimal solution gbest from a non-inferior solution set rep by adopting a roulette wheel operator strategy, calculating an individual optimal solution pbest, and updating the position and speed of each individual in the population; S4, performing crossover and mutation operations on each individual in the population; S5, calculating the fitness of each individual in the population P t+1 , updating the non-inferior solution set; and S6, performing evaluation by adopting an optimal / suboptimal ratio criterion, and selecting a best wave band subset. The application overcomes the one-sidedness of wave band evaluation of a single criterion method, and can realize effective dimension reduction of a hyperspectral image.

The application discloses an image sharpness evaluation method, comprising: obtaining an original image and an optical modulation transfer function; generating a multi-scale baseline sharpness response set, an information-physical sharpness kernel, an uncertainty measure and an evidence vector based on the original image and the optical modulation transfer function; estimating a failure posterior probability based on the evidence vector; generating a gating weight based on the failure posterior probability and the uncertainty measure; applying the failure posterior probability and the gating weight to perform a failure posterior coupling process on the multi-scale baseline sharpness response set and the information-physical sharpness kernel to generate a final sharpness evaluation result. The coupling process preferably includes applying an analytical bias correction and performing a cross-scale band selection. The application improves the evaluation robustness in a complex scene through a failure posterior coupling mechanism.

The application provides a hyperspectral image band selection method, device and electronic equipment. The method comprises: sampling each band image of a hyperspectral image in a three-dimensional space to obtain a sampling two-dimensional matrix; calculating the information entropy of each sampling band in the sampling two-dimensional matrix, grouping the information entropy, selecting a sampling band corresponding to the maximum information entropy in each group to form an initial clustering center matrix; performing bias processing on the initial clustering center matrix to obtain a target clustering center matrix; based on the sampling two-dimensional matrix and the target clustering center matrix, calculating the membership degree of each sampling band in the sampling two-dimensional matrix to each clustering center in the target clustering center matrix to obtain an initial membership matrix; calculating a kernel matrix based on the sampling two-dimensional matrix; and determining a target band selected in the hyperspectral image based on the kernel matrix and the initial membership matrix. The method is efficient in calculation and improves the local optimal solution problem.

The application provides a band selection method and device based on matrix calculation, equipment and medium, relates to hyperspectral image processing technical field, and includes obtaining a first ground object array and a second ground object array; calculating the channel value ratio of the first ground object array and the second ground object array under the same wavelength as a first adjustment factor matrix; after shifting the second ground object array to the right and left respectively, the obtained ratio is respectively taken as a second adjustment factor matrix and a third adjustment factor matrix; the first adjustment factor matrix, the second adjustment factor matrix and the third adjustment factor matrix are integrated into a target adjustment factor matrix; the wavelength with the highest similarity and the largest difference is determined from the evaluation value matrix.The application establishes the adjustment factor matrix by shifting the image of the ground object to the left and right, calculates the evaluation value matrix, and accurately determines the difference band and the similar band between the ground objects in a short time through the matrix resampling algorithm.

The application provides a fabric color difference grade determination method based on multi-dimensional spectral analysis, and belongs to the technical field of intelligent textile detection. The method comprises the following steps: acquiring a three-dimensional spectral data cube of a sock band under an integrating sphere diffuse light source environment by using a spectral imaging acquisition system, and converting an original image into spectral reflectance data through black and white board radiation correction; identifying a joint position by using the spectral response difference between the joint and the sock band, completing automatic segmentation of the sock band image, and acquiring effective detection areas of each segment; in combination with OCR-recognized character information, eliminating surface defects by using a deep learning target detection algorithm for a single sock band; separating texture shadows and background noise by using hyperspectral unmixing, and extracting pure spectra representing the true color of the sock band; and mapping the pure spectra to a CIELAB color space to calculate color differences and determine grades. The application realizes the intelligentization and objective evaluation of the sock band determination method by using spectral imaging and intelligent recognition technology.

The invention discloses a satellite edge hyperspectral image processing method and system, relates to the technical field of communication, and aims to solve the problems of resource limitation, communication bottleneck and unreasonable task scheduling faced by hyperspectral image processing in satellite edge calculation. The method comprises the following steps: taking a future data transmission rate of a satellite as a communication feature vector, and constructing a task unloading matrix in combination with a graph attention mechanism to realize accurate matching of tasks and satellite resources; the task unloading matrix is converted into task features, the task features and the hyperspectral image are fused to form a fusion tensor, a high-dimensional feature vector is generated based on the fusion tensor, key wavebands are screened, and a waveband selection matrix is obtained; and constructing a joint loss function containing a task unloading loss function and a band selection loss function, and iteratively optimizing the double matrixes until the loss is minimum to obtain an optimal matrix. Through collaborative optimization of task scheduling and image processing, system energy consumption, communication constraint and processing precision are balanced, the satellite resource utilization rate and image processing real-time performance are improved, and the method is suitable for low-orbit satellite edge calculation scenes.

The application belongs to the technical field of plant disease nondestructive detection, and discloses a nondestructive detection method and system for citrus Huanglongbing based on hyperspectral reflection and transmission fusion, which comprises the following steps: acquiring hyperspectral reflection spectrum data and hyperspectral transmission spectrum data of a to-be-detected citrus leaf, obtaining single reflection characteristic band data and single transmission characteristic band data after pretreatment and band selection; based on the single reflection characteristic band data, the single transmission characteristic band data and the spectrum data after feature-level fusion of the two, first, second and third prediction calculation results are obtained through respective optimal classification models; the corresponding weights are determined according to the classification accuracies of the optimal classification models, and the first, second and third prediction calculation results are subjected to decision-level fusion calculation to obtain the Huanglongbing grade of the citrus. The method introduces feature-level and decision-level fusion, has the discrimination ability of sample surface physical form and internal structure composition, and can quickly and nondestructively detect the Huanglongbing of the citrus.

The invention relates to the technical field of medical images and digital pathology, in particular to a COPD tissue pathology recognition method and system based on hyperspectral imaging. The method comprises the following steps: acquiring a hyperspectral image of COPD tissue through a spectrum self-calibration method; constructing a spectrum-morphological feature fusion model based on adaptive waveband selection, and inputting the hyperspectral image into the spectrum-morphological feature fusion model for principal component analysis and classification identification analysis; outputting pseudo-color spectral images of different wavebands according to an analysis result, fusing the pseudo-color spectral images of different wavebands to obtain a color image reflecting tissue heterogeneity, and converting the color image into a three-dimensional visual image; and updating the output result to the spectrum fingerprint database. Meanwhile, the invention discloses a system for realizing the method, and the COPD histopathology identification method and system based on hyperspectral imaging are adopted to realize accurate identification of COPD histopathology.

This invention discloses a method for detecting spoofed speech based on frequency band selection. The method includes: acquiring a target speech signal; transforming the target speech signal to obtain spectrogram features; performing frequency band segmentation on the spectrogram features to obtain low-frequency sub-band features and high-frequency sub-band features; training a speech synthesis spoofed speech detection model using the low-frequency sub-band features; training a recording playback spoofed speech detection model using the high-frequency sub-band features; inputting the low-frequency sub-band features into the speech synthesis spoofed speech detection model; and inputting the cross-matched high- and low-frequency sub-band features into the recording playback spoofed speech detection model to obtain the final speech detection result. In this invention, the robustness of the neural network spoofed speech detection system is improved under conditions such as dataset mismatch, and the feature size is reduced through sub-band selection, thereby reducing the number of parameters and computational load for spoofed speech detection.

The application relates to a hyperspectral image super-resolution method based on alternating diffusion band selection and degradation adaptation. The method comprises the following steps: constructing a blind image restoration target function and decomposing the function into a high-quality image and a degradation matrix solving subproblem. Then, a hyperspectral image super-resolution model containing three modules is constructed: a band selection and Gaussian dimension reduction module selects bands with large information quantity according to a variance distribution and a Gaussian attenuation model; a sampling priori and degradation kernel estimation module generates a random degradation matrix by Monte Carlo sampling, and combines a data fidelity loss function to optimize parameters to dynamically estimate the degradation matrix; a frequency band projection and diffusion guided image super-resolution module solves a high-quality image by means of a pre-trained diffusion model, band selection and degradation matrix solving, and optimizes an intermediate high-quality image through a data fidelity loss function and total variation regularization. Finally, the three modules are iterated to a preset number of times, and the obtained intermediate high-quality image is the result after restoration. The method can improve the image restoration performance and efficiency.

According to the multi-frequency-band selection method and device, the electronic equipment and the computer program product, a receiver ensures that the frequency band quality is reliably evaluated through long-resident asynchronous scanning, and then connection is initiated through active repeated invitation; the transmitter provides a predictable response opportunity through a periodic fixed monitoring window. Therefore, the two parties converge in the optimal frequency band through closed-loop handshake of'invitation-monitoring-confirmation '. According to the method, the strict time sequence synchronization requirement between receiving and transmitting equipment is thoroughly eliminated, so that the implementation complexity and cost of the system are remarkably reduced, meanwhile, the high reliability and robustness of the frequency band selection and link establishment process are ensured, and the method is particularly suitable for wireless communication scenes such as low-power-consumption and low-complexity Internet of Things and the like.

The invention discloses a spectral band selection method based on fractional domain transformation and differentiable information imbalance. The spectral band selection method is used for obtaining a band subset with diagnosis information. The method comprises the following steps: S1, collecting a medical hyperspectral image, and dividing the medical hyperspectral image into a training-verification set and a test set; s2, performing fractional Fourier transform on the hyperspectral data of the training-verification set to obtain transform domain spectral representations under different orders; s3, evaluating the retention degree of the candidate wave band subset on the transform domain spectral characteristics under different orders based on the differentiable information imbalance, and carrying out wave band selection; and S4, determining an optimal transformation order, performing fractional Fourier transform and waveband selection processes of the test set data, and generating an optimal waveband subset. According to the method, information most related to lesion judgment can be effectively reserved while the data dimension is reduced, the recognition accuracy and robustness of the lesion area in the medical hyperspectral image are effectively improved, and the method has important clinical application value.

This invention relates to a hyperspectral image band selection method applicable to complex terrain scenes, comprising: inputting a raw 3D hyperspectral image; extracting local spatial features and local spectral features through a heterogeneous dual-channel 3D convolutional network with a cross-attention mechanism; adding spatial location encoding and spectral location encoding to the local spatial features and local spectral features respectively; inputting the encoded features into a dual-channel Transformer module to generate band attention weights; constructing a band evaluation criterion based on the redundancy between bands attended by multi-head attention in the Transformer module and the band attention weights; and selecting the required bands after sorting the bands. The beneficial effects of this invention are: it fully considers the global-local and spatial-spectral features between bands, providing reliable support for subsequent applications. Therefore, the method proposed in this invention has significant practical application value.

The invention relates to a resource allocation method for optimizing C + L wave bands in an elastic optical network through a noise Q network, and belongs to the field of optical fiber communication. In order to improve the C + L wave band resource allocation performance of the elastic optical network, the resource allocation method based on the spectrum fragments of the noise Q network and the service transmission optical signal-to-noise ratio quality guarantee is designed. The frequency spectrum fragmentation degree is measured by designing a network frequency spectrum fragmentation index reflecting the C + L wave band characteristics of the elastic optical network; an optical signal-to-noise ratio and modulation format combined safe optical path selection method is used to construct an action space of a noise Q network meeting the consistency, continuity and non-overlapping performance of optical path frequency spectrums, and a reward function reflecting optical path frequency spectrum fragmentation and the optical signal-to-noise ratio is designed. And the action selection adopts a maximum Q value function method of cumulative rewards to optimize an optical path, a modulation format, waveband selection and spectrum allocation. According to the method, the transmission reliability of the service in the elastic optical network and the spectrum resource utilization rate can be improved, and spectrum fragments are reduced.

The invention provides a state offset identification method and system based on low-dimensional discrete perception, and the method comprises the steps: employing an MEMS wave band selection unit to switch between preset discrete wave bands, obtaining the near-infrared response features of a target object through multiband time multiplexing sampling, and constructing a low-dimensional discrete feature vector independent of an absolute value. Through individualized baseline modeling, the system learns feature distribution of daily equipment of a user, compares current state features with an individualized baseline model constructed in a normal state, and outputs a hierarchical or stateful health suitability conclusion. The system has a self-adaptive compensation mechanism, and can absorb the influence of device drift and environmental change. According to the invention, the problems of high cost, complex calculation and lack of individualized evaluation ability in the prior art are solved; state deviation recognition detection of daily equipment is achieved, and the method can be widely applied to liquid detection, equipment monitoring, environment evaluation and other scenes.

This application relates to the field of shortwave infrared imaging technology, and provides a method and apparatus for optimizing color bands in shortwave infrared images driven by both physical data and technical data. Addressing the problem of exponentially increasing combinations of hyperspectral / multispectral bands, this method abandons the computationally expensive exhaustive training method and utilizes a hierarchical screening framework. First, a physical model is used to determine candidate bands. Then, improved statistical indices are used to quickly perform unsupervised dimensionality reduction to obtain a small number of high-potential combinations. Finally, generative validation of these high-potential combinations yields the optimal band combination. This strategy ensures the physical interpretability of band selection while significantly reducing the time cost and computational consumption in finding the optimal solution.

The application discloses a kind of electric power equipment fault diagnosis methods based on artificial intelligence in the technical field of mechanical equipment fault diagnosis, comprising the following steps: using sensor to collect the original vibration signal of electric power equipment operation, based on fault category division label data set;The local spectral kurtosis of the original vibration signal is calculated, the frequency band of the fault feature is identified by constructing adaptive band selection function, and the target frequency band is enhanced and reconstructed to obtain enhanced vibration signal;The empirical mode decomposition is carried out to the enhanced vibration signal, and the effective intrinsic mode function is obtained, the differential entropy of the intrinsic mode function is calculated, based on the fusion of time domain and frequency domain statistical characteristics, obtain time-frequency statistical fusion feature vector.The application introduces adaptive spectral kurtosis calculation and feature band selection method, effectively solves the feature aliasing problem caused by fixed frequency band division in traditional frequency domain analysis method, so as to improve the separation and diagnosis precision of fault feature.

The invention relates to quantitative detection of a semiconductor wet etching solution, in particular to a near-infrared detection method for the content of acid and ammonium acid in an organic system, which comprises the following steps: carrying out uniform formula design on a modeling sample, and calculating the content of acid and ammonium acid in the modeling sample by applying a basic principle of a stoichiometric method through a partial least square method. The method for quantifying the mass fractions of the acid and the acid ammonium in the organic system is formed by associating near infrared spectrum data of a sample with the mass fractions of the acid and the acid ammonium through spectrum pretreatment, analysis waveband selection, determination of optimal modeling parameters such as the optimal number of main factors and a root-mean-square error (RESECV) of cross validation and the like and establishing a model. Compared with a traditional titration method and a gas-phase liquid-phase detection method, the method has the advantages that the time consumption is long (more than 15 minutes) and the precision is poor (more than 0.5%), the interference of similar structures of acid and acid ammonium can be effectively eliminated, and the detection precision is within 0.05%; moreover, the quantity of modeling samples required by the method is only 25-35, so that the modeling workload (the quantity of general near-infrared modeling samples is 100-300) is reduced, and the requirements on accuracy and high efficiency of product content inspection can be met.

The present application belongs to the field of spectral analysis and intelligent detection technology, and discloses a soil available phosphorus content hyperspectral imaging rapid detection method and system. The method sequentially performs adaptive baseline correction, target-oriented orthogonal signal correction combined with fractional order differential transformation phosphorus response enhanced multi-order feature extraction, sparse Bayesian sensitive band selection based on mutual information graph Laplace regularization, multi-scale hollow causal convolution regression network prediction of phosphorus adsorption state perception, and temperature and humidity environment compensation correction fused with Kubelka-Munk diffuse reflection theory physical constraints. The physical reasonableness of shared coding features is constrained by bidirectional cross-attention fusion of spectral and spatial features and phosphorus adsorption state classification auxiliary task, and the available phosphorus content, confidence interval and adsorption state type judgment result are output.

The invention relates to the technical field of hyperspectral remote sensing image intelligent processing and hydrocarbon substance identification, in particular to an absorption characteristic-based hyperspectral satellite image hydrocarbon substance intelligent identification algorithm, which comprises the following steps of: acquiring original data of a hyperspectral satellite image, and preprocessing to generate a surface reflectance data cube; a wave band near 1100 nm is selected as a feature recognition wave band, and spectrum confusion is avoided; dynamically searching a wave trough position in a range of 1095-1105 nm for each pixel spectrum curve, and establishing an absorption baseline; the absorption depth and the absorption area are calculated, and absorption characteristics are quantified; fusing the absorption depth, the absorption area and the trough screening result to generate a pixel-level absorption index map; and performing clustering analysis based on the absorption index spectrum, and outputting a hydrocarbon substance distribution confidence classification result. The algorithm adopts a dynamic wave band selection and multi-feature fusion mechanism, improves the accuracy and anti-interference performance of weak signal identification, and is suitable for oil-gas exploration and environment monitoring application.

Embodiments of the present disclosure relate to devices, methods, apparatuses and computer readable storage media of interference weighting based sub-band selection procedure. The method comprises receiving, from a second device in a radio access network, a configuration for determining a weighted inter-subnetwork interference between a first subnetwork of the radio access network and at least one second subnetwork of the radio access network, wherein the configuration is indicative of at least one of a first weighting or a second weighting for the weighted inter-subnetwork interference; determining the weighted inter-subnetwork interference at least based on the configuration; and transmitting the weighted inter-subnetwork interference to the second device for selecting at least one sub-band for the first subnetwork. In this way, weighting operations may be proposed for the inter-subnetwork interferences and the associated low-complexity sub-band selection procedure to minimize the sum weighted interference over all the subnetworks, which may flexibly optimize and balance various system performance.

The application discloses a kind of unsupervised hyperspectral image band selection methods, comprising: step one, using ResNet50 network to the feature extraction of each band graph in hyperspectral image, obtain each band feature map;Step two, construct Gaussian noise chart, and the feature extraction of Gaussian noise chart is carried out through ResNet50 network, obtain each negative sample feature map;Step three, obtain the average band distance between each band feature map and all negative sample feature map, obtain the band graph set of preliminary screening;Step four, based on the band graph set of preliminary screening, using contrast peak selection strategy, obtain the preferred band graph.The method of the application is reasonable, the feature extraction of band graph and Gaussian noise chart is carried out using ResNet50 network, and the average band distance between the two is used to carry out denoising and redundancy reduction, realize band graph selection.