24 results about "Image cube" patented technology

The invention discloses a large yellow croaker face recognition method and system based on target detection. The method comprises the steps that visible light image sequences, hyperspectral image cubes and environmental parameters of large yellow croakers in different growth stages are collected; and based on the visible light image sequence, positioning and segmenting a fish face region of interest by using a target detection model, and based on a reference frame selected in the visible light image sequence and a spatial corresponding relation extracted from common feature points of the hyperspectral image cube, performing spatial registration on the fish face region of interest. Obtaining a registered target fish face area and a hyperspectral fish face data block; and extracting a fish face texture feature vector from the target fish face region, extracting a spectral feature vector from the hyperspectral fish face data block, constructing a feature fusion image, inputting the feature fusion image into an image neural network model to generate a fusion feature vector, and inputting the fusion feature vector into an age recognition model to obtain an age recognition result. According to the invention, the accuracy of large yellow croaker age identification can be improved.

The invention relates to a circuit board defect identification method and system based on multi-dimensional image data, and belongs to the technical field of data identification processing, and the method comprises the following steps: obtaining synchronous image data of a circuit board to be detected in a plurality of imaging modes; performing space-spectrum joint registration on each modal image to generate a multi-dimensional image cube with a unified coordinate system and pixel alignment; inputting the multi-dimensional image cube into a pre-trained multi-branch heterogeneous fusion neural network; generating a pixel-level defect probability graph by utilizing a defect sensing context decoder, and performing geometric constraint optimization on the probability graph by combining prior information of a circuit board design layout; outputting defect types, positions and confidence coefficients, and establishing an interpretable defect fingerprint database according to the multi-dimensional response characteristics of the defects; the method has the beneficial effects that false defect signals generated by image noise and circuit board surface texture interference can be effectively inhibited, the omission ratio and the false detection ratio are greatly reduced, and pixel-level accurate defect positioning is realized.

The invention provides a hyperspectral spectrum data fusion method and a hyperspectral spectrum data fusion system capable of widening a spectral band, and the method comprises the steps: carrying out the wavelength domain matching of Vis-NIR hyperspectral data obtained by a ground object spectrometer based on the visible light band of a hyperspectral image, and storing the resampled Vis-NIR hyperspectral data in a spectrum library. Performing similarity measurement on the spectral vector of each pixel in the hyperspectral image and all standard spectrums in the spectrum library to generate a spectral feature fingerprint map and a spectral response coefficient map; and performing spectrum reconstruction on each pixel in the hyperspectral image data in combination with the spectral feature fingerprint map and the spectral response coefficient map. And finally, obtaining a hyperspectral image cube after broadened spectrum fusion. According to the scheme, on the basis of wavelength domain matching, the spectral feature fingerprint map, the spectral response coefficient map and the like, the hyperspectral image cube obtained through final fusion covers a wider spectral range, and the original high spatial resolution is reserved.

The invention relates to the technical field of spectral analysis and detection, and discloses a method and a system for rapidly detecting pesticide residues in food based on multispectral imaging. The method comprises the following steps: acquiring a multispectral image and correcting to obtain a reflectivity image cube; inputting into a variational auto-encoder for encoding to obtain a potential feature vector; after normalization, calculating a mahalanobis distance, screening residual candidate areas, generating space and spectrum attention weights, and performing outer product to construct a collaborative attention tensor; multiplying with the reflectivity image element by element to obtain an enhanced image, extracting features, and inputting the features into a regression network to predict the pesticide residue concentration. According to the invention, the accuracy and generalization ability of pesticide residue detection under different food varieties are improved.

The application discloses a landslide identification method and device based on a generative adversarial network data expansion strategy, and comprises the following steps: acquiring a multi-source data set and performing pretreatment, extracting landslide disaster factors, and constructing a landslide sample library after pretreatment and landslide label production are performed on the landslide disaster factors; after the pretreated landslide disaster factors and landslide labels are superimposed in layers, one-dimensional array data are produced through an image cube, and the one-dimensional array data are divided into a training set and a test set; landslide data in the training set are input into a network in batches for training, minority class sample expansion is completed, and the expanded samples are added to the original training set to form a balanced data set; the balanced data set is input into a back-end machine learning classification model for training, an optimal model generated by training is used to test the test set, and a landslide prediction result is obtained. The application solves the problem of imbalance between classes in the data set, and can make the landslide identification boundary more clear and complete, and improve the precision of landslide target identification.

A method for selecting a light source for illuminating defects, an electronic device, and a non-transitory storage medium are provided. The method includes acquire grayscale images of an object with a known defect and generates a pseudo-hyperspectral image cube based on the grayscale images, so that algorithms related to hyperspectral images can analyze the grayscale images collected under different light sources. A most effective or target light source can be automatically and quickly determined from the plurality of light sources, improving an efficiency of light source selection.

The application provides a kind of hidden frame glass curtain wall structural adhesive aging nondestructive testing method based on visual detection, comprising: obtaining hyperspectral image cube of the surface of the structural adhesive to be detected and end-member spectral feature information containing aging product components.According to the feature information, the hyperspectral image is spectrally decomposed, the abundance value of each target chemical component at each pixel point is calculated to generate a material component abundance map.Through clustering the abundance map, an aging area cluster composed of chemical component abundance abnormal and spatially adjacent pixel points is obtained and its position is determined.According to the component abundance vector of all pixel points in the aging area cluster, its area component vector is determined and input as an input vector into the aging degree evaluation model to obtain the aging degree index.Based on the position information and aging degree index of all aging area clusters, an aging state map of the structural adhesive is generated.The above method can effectively identify the early aging of the structural adhesive and improve the accuracy of aging assessment.

The application provides a kind of spatiotemporal modulation Fourier transform stokes polarization imaging spectrometer, including in turn along the direction of propagation of light beam: telescope system, polarization modulation system, converging system, polarization interference system and imaging system;Target scene emitted target light field is collimated into parallel light field in telescope system and is incident;Linearly polarized light field is obtained after collimated parallel light field is incident to polarization modulation system and is incident to converging system;Converging system is used to image linearly polarized light field to polarization interference system and produce superposition interference, form polarization interference imaging light field with specific optical path difference distribution characteristics and are incident to imaging system to obtain polarization interference image;By step scanning to target scene, obtain polarization interference image cube, by extracting each field of view corresponding interference order interference map, carry out stokes channel filtering and fourier transform, and then demodulate the spectral information of each stokes parameter of target light field.

The application relates to the technical field of spectral analysis detection, and discloses a food pesticide residue rapid detection method and system based on multispectral imaging. The method comprises the following steps: collecting a multispectral image and correcting to obtain a reflectivity image cube; inputting a variational autoencoder to encode to obtain a latent feature vector; calculating Mahalanobis distance after normalization to screen a residue candidate area, generating spatial and spectral attention weights and performing outer product to construct a collaborative attention tensor; multiplying the reflectivity image element by element to obtain an enhanced image, and extracting features to input a regression network to predict a pesticide residue concentration. The application improves the accuracy and generalization ability of pesticide residue detection under different food varieties.

The present disclosure relates to a hyperspectral band selection method based on space-spectrum combined graph learning, which comprises the following steps: firstly, pre-processing a hyperspectral image cube to obtain a hyperspectral image data matrix, and constructing a spatial affinity matrix by using a spatial pixel point set in the hyperspectral image data matrix, and constructing a spectral affinity matrix by using a spectral data point set in the hyperspectral image data matrix; then, embedding the spatial affinity matrix and the spectral affinity matrix into a self-representation model to obtain a graph self-representation model containing a coefficient matrix; finally, processing a neighboring band of the coefficient matrix to obtain a reconstruction target function, and obtaining an optimal band subset. The present disclosure can improve the robustness and effectiveness of the spectral band selection algorithm in the spatial and spectral dimensions, while ensuring the local spatial consistency between the hyperspectral bands.

The invention discloses a batch image encryption method based on a discrete hyperchaotic mapping and plane segmentation scrambling method. Plaintext images are obtained, and the plaintext images with different sizes are fused into image cube blocks. Generating a pseudo-random sequence by using double two-dimensional discrete hyper-chaotic mapping to perform point-to-point scrambling on the image cube blocks; performing inter-block scrambling on the scrambled cubic blocks by an improved Huaendao scrambling method; and finally, diffusion operation is performed on the cube blocks to obtain a ciphertext image which is visually similar to noise. According to the encryption method, the bidimensional hyper-chaotic mapping is adopted, the mapping has the advantages of being high in iteration speed, large in key space, high in complexity, high in sensitivity and the like, and the encryption method is suitable for secret communication of image information to guarantee the safety of important information. The multi-image cube block is adopted as the minimum unit in the encryption process, compared with traditional single-unit image encryption and decryption, the efficiency of batch encryption and decryption is greatly improved, the image cube is not limited by the image size, and the compatibility of the encryption method is improved.

The present application relates to a circuit board defect recognition method and system based on multi-dimensional image data, belonging to the technical field of data recognition processing, comprising the following steps: obtaining the synchronous image data of the circuit board to be detected under multiple imaging modalities; performing spatial-spectral joint registration on each modality image to generate a multi-dimensional image cube with a unified coordinate system and pixel alignment; inputting the multi-dimensional image cube into a pre-trained multi-branch heterogeneous fusion neural network; generating a pixel-level defect probability map using a defect perception context decoder, and performing geometric constraint optimization on the probability map in combination with prior information of the circuit board design layout; outputting the defect type, location and confidence, and establishing an interpretable defect fingerprint library according to the multi-dimensional response characteristics of the defects; the beneficial effects of the present application are: it can effectively suppress false defect signals caused by image noise and circuit board surface texture interference, greatly reduce the missed detection rate and false detection rate, and realize accurate pixel-level defect positioning.

The invention relates to an automatic extraction method, device and equipment for an ancient ceramic glaze splitting form and a medium. The method comprises the following steps: acquiring a hyperspectral image cube and three-dimensional point cloud data of an ancient ceramic glaze, and performing spatial alignment processing on the hyperspectral image cube and the three-dimensional point cloud data to obtain a hyperspectral data block and a three-dimensional point cloud data block; inputting the hyperspectral data block and the three-dimensional point cloud data block into a pre-trained double-flow coding cross-modal fusion model to obtain an initial three-dimensional point cloud slicing segmentation prediction result and a two-dimensional chemical anomaly confidence prediction map; on the basis of the two-dimensional chemical anomaly confidence prediction map, arbitration optimization is carried out on prediction points with fuzzy or isolated boundaries in the initial three-dimensional point cloud slicing segmentation prediction result, and three-dimensional point cloud optimization segmentation labels are obtained; and performing three-dimensional space smoothing processing on the three-dimensional point cloud optimization segmentation label to obtain an ancient ceramic glaze three-dimensional cracking network form. By adopting the method, the end-to-end automatic extraction of the three-dimensional slicing network form can be refined.

The invention provides a color block smooth boundary generation method and device, electronic equipment and a storage medium, and relates to the technical field of computers. The method comprises the following steps: acquiring a cubic texture image corresponding to the surface of a target object, wherein the cubic texture image comprises a plurality of color blocks; aiming at each current pixel in the cubic texture image, determining all target adjacent pixels which do not belong to the same color block with the current pixel in a preset neighborhood range of the current pixel; calculating the distance between each target adjacent pixel and the current pixel; determining a directed distance field value corresponding to the current pixel according to the distance between each target adjacent pixel and the current pixel; the directed distance field value is used for representing the approaching degree between the current pixel and the boundary of the color block to which the current pixel belongs; and rendering smooth boundaries among the plurality of color blocks according to the directed distance field values corresponding to the pixels in the cubic texture image. Smooth boundaries between the color blocks can be realized, and the rendering effect is further improved.

A method of the present disclosure, among others, involves illuminating a tissue sample with light across both a visible wavelength range and a near-infrared wavelength range, wherein the tissue sample has been administered with at least one fluorescent agent; acquiring a snapshot image of the tissue sample, wherein pixels of the snapshot image capture different 2D views of the tissue sample at a same time along with multiple wavelengths of light incident on a respective sensor pixel; creating a hyperspectral image cube of the tissue sample from the snapshot image; performing spectrally-resolved quantitative fluorescence measurements on the hyperspectral image cube; analyzing the spectrally-resolved quantitative fluorescence measurements to detect a presence of at least one fluorescent agent in the tissue sample; and / or identifying an optical fingerprint of an abnormal region of the tissue sample based on the detected at least one fluorescent agent in the tissue sample.

This application relates to image processing technology, and provides a method, apparatus, electronic device, and storage medium for determining a light source. The method includes: acquiring multiple grayscale images of an object carrying a target defect, wherein different grayscale images correspond to different light sources; generating a pseudo-hyperspectral image cube based on the multiple grayscale images; determining defect information and spectral wavelength corresponding to the target defect from the pseudo-hyperspectral image cube; generating a spectral feature curve based on the defect information and the spectral wavelength; and determining the target light source from the spectral feature curve using a hyperspectral image analysis algorithm. This application can improve the efficiency of light source selection.

The invention mainly relates to the technical field of remote sensing, and provides a chipping-coverage-free glacier remote sensing mapping method based on a multi-index probability threshold value in order to improve the image accuracy and precision of a glacier distribution diagram. Establishing a multi-temporal image cube data set, constructing a glacier region recognition model based on the normalized snow index, the normalized water body index and the surface temperature, training the glacier region recognition model based on the multi-temporal image cube data set to generate a glacier region recognition result, and generating a glacier contour according to the glacier region recognition result. The limitation of a traditional single-index fixed threshold value method in a complex environment is overcome through a multi-index collaborative judgment strategy, and the image accuracy and precision of the glacier distribution diagram are improved.

A hyperspectral image band selection method based on latent feature fusion comprises: S11, inputting a hyperspectral image cube and segmenting the inputted hyperspectral image cube into several regions by superpixel segmentation; S12, learning low-dimensional latent features corresponding to the several regions from the several regions respectively to obtain a latent feature matrix of all the regions; S13, calculating an average Laplacian matrix and an average latent feature matrix of the hyperspectral image cube; S14, fusing the latent feature matrix, the average Laplacian matrix, and the average latent feature matrix of all the regions to obtain a low-dimensional self-representation matrix of the hyperspectral image cube; and S15, clustering the low-dimensional self-representation matrix by a k-means algorithm to obtain an optimal band subset of the hyperspectral image cube.

The invention discloses a face image protection method based on semi-tensor product compressed sensing and three-color transformation, and the method comprises the steps: obtaining a plaintext image, carrying out the extraction and compression of face information of different sizes, and carrying out the fusion of the compressed face information images, and forming an image cube. And then a pseudo-random sequence generated by mapping iteration of the hyper-chaos neuron network is combined with a three-color shuffling algorithm, and scrambling and diffusion operation is performed on the image cube, so that a ciphertext image can be obtained. According to the method, firstly, the face information in the face image is extracted and compressed, so that the data volume required by encryption is reduced; hyper-chaos neuron network mapping has the characteristics of high iteration speed, large key space, high complexity, high sensitivity and the like, and is suitable for secure transmission of image information under the condition of ensuring the security of confidential information. The face information is wholly encrypted, so that the encryption efficiency is improved, and the method is not limited by the specification of the image.

The invention relates to the technical field of agricultural product quality detection and nondestructive testing, in particular to a mango hardness batch prediction method based on a hyperspectral imaging technology, which comprises the following steps: placing batch mangoes on a low-reflectivity sampling platform at equal intervals, and obtaining a hyperspectral image cube containing all mangoes in a darkroom by using a hyperspectral camera; extracting information from the hyperspectral image cube, calculating to obtain a full-wave band spectral vector of the mango in each label area, obtaining an actually measured hardness value of each mango, and establishing a'spectral vector-hardness' original data set; preprocessing the original data to obtain a training set and a verification set, and constructing a hardness prediction model by taking a ridge regression function as a target function; and inputting the spectrum vector of the to-be-detected mango into the model, and outputting a corresponding hardness prediction value to realize batch mango hardness nondestructive detection. The method constructs the mango high-luminosity hardness prediction model, has adaptability to mango samples, realizes nondestructive testing, and can accurately predict the mango hardness.

The invention provides a nondestructive testing method for aging of a structural adhesive of a hidden framing glass curtain wall based on visual inspection. The nondestructive testing method comprises the following steps: acquiring a hyperspectral image cube of the surface of the structural adhesive to be detected and end member spectral feature information containing aging product components; and performing spectral decomposition on the hyperspectral image according to the feature information, and calculating the abundance value of each target chemical component at each pixel point to generate a material component abundance map. And clustering the abundance map to obtain an aging region cluster formed by pixel points with abnormal chemical component abundance and adjacent space, and determining the position of the aging region cluster. And according to the component abundance vectors of all the pixel points in the aging region cluster, determining the region component vector, and inputting the region component vector as an input vector into the aging degree evaluation model to obtain an aging degree index. And generating an aging state diagram of the structural adhesive based on the position information and the aging degree indexes of all the aging region clusters. According to the method, the early aging of the structural adhesive can be effectively identified, and the aging evaluation accuracy is improved.

A method of measuring blood oxygenation including acquiring one or more images of a portion of a body with an RGB camera, converting RGB colors in the one or more images into a multispectral data imaging cube, wherein the multispectral date imaging cube comprises a red channel, a blue channel, and a green channel, decoupling an oxygenated blood information and a deoxygenated blood information from the multispectral data imaging cube based on a first reflectance of the green channel and a second reflectance of the red channel, and determining a blood measurement based on the oxygenated blood information and the deoxygenated blood information.

The invention relates to the technical field of spectral analysis and microscopic imaging, and discloses a dynamic surface scanning Raman imaging system and a spectral data acquisition method, and the system comprises a light beam output module which is used for outputting a collimated laser beam; the device comprises a spatial light modulator, an excitation light path module, a sample stage, a collection light path module, an imaging spectrometer and a control module, the control module generates a corresponding driving signal according to a preset image sequence, and the preset image sequence comprises a plurality of preset excitation images arranged according to a set sequence; controlling a spatial light modulator to generate a corresponding dynamic structure light field through a driving signal, obtaining Raman spectrum data of a plurality of excitation light field distribution diagrams through an imaging spectrometer, and reconstructing and decoupling the Raman spectrum data to obtain Raman imaging cube data. Complete Raman imaging cube data can be reconstructed through a small amount of pattern switching and exposure, and the spectrum acquisition efficiency is greatly improved.

The application discloses a big yellow croaker fish face recognition method and system based on target detection, comprising the following steps: collecting visible light image sequences, hyperspectral image cubes and environmental parameters of big yellow croakers in different growth stages; based on the visible light image sequences, a target detection model is used to locate and segment a fish face region of interest; based on the spatial correspondence of the common feature points extracted from the selected reference frame in the visible light image sequences and the hyperspectral image cubes, the fish face region of interest is spatially registered to obtain a registered target fish face region and a hyperspectral fish face data block; a fish face texture feature vector is extracted from the target fish face region, a spectral feature vector is extracted from the hyperspectral fish face data block, and a feature fusion graph is constructed; the feature fusion graph is input into a graph neural network model to generate a fusion feature vector, and the fusion feature vector is input into an age recognition model to obtain an age recognition result, so that the accuracy of big yellow croaker age recognition can be improved.