63 results about "Multispectral image fusion" patented technology

Typical inventive practice provides for electronic communication of a computer with a display, an active radar device (for transmitting target-location data and environmental data), a light measurement device (for transmitting visual light data), and passive imaging devices covering bands in the visual, infrared (MWIR and / or LWIR), and millimeter wave regions of the electromagnetic spectrum. Inventive software in the computer's memory establishes “operational modes.” Each operational mode is defined at least by a predominant environmental (obscuration and lighting) character, ascribes “modal indices” to individual imaging devices, and carries its own multispectral image fusion algorithm (which, pursuant to the ascribed modal indices, attributes weights to the imaging data from the respective imaging devices). The inventive software aims the imaging devices toward the target, selects an operational mode (based on the environmental data), and executes the image fusion algorithm associated with the selected operational mode so that a fused multispectral image is displayed.

The invention discloses a multi-spectral and panchromatic image super-resolution fusion method based on compressive sensing, which comprises the following steps: 1, respectively inputting a low-resolution multi-spectral image and a high-resolution panchromatic image; 2, obtaining a high-resolution joint sampling value matrix; 3, obtaining a high-low observation matrix; 4, obtaining a high-high observation matrix; 5, obtaining a high-resolution joint observation matrix; 6, studying a high-resolution over-complete dictionary; 7, obtaining a high-resolution rarefaction coefficient; and 8, reconstructing a high-resolution multi-spectral image. According to the method provided by the invention, the compression sensing technology is introduced into the fusion field of the multi-spectral images, the defects that the color is distorted and the spectrum is contorted in the prior art are solved, and the resolution of the fused multi-spectral image is about one time of that of the high-resolution panchromatic image.

Satellite image fusion method and system are provided. The satellite image fusion method includes matching sizes of a panchromatic image and a multispectral image captured from a satellite image; dividing the panchromatic image and the multispectral image into a plurality of blocks; calculating coefficients to acquire Intensity (I) component image data using pixel values of each block of the multispectral image; and generating fused multispectral image data by applying the acquired I component image data to a fusion algorithm. In the multispectral image fusion, the distortion of the color information can be minimized and the multispectral image data of the high resolution can be attained. In addition, the present invention is applicable to not only the IKONOS images but also other satellite images, and the present image fusion can be carried out fast.

The invention discloses a multispectral image and panchromatic image fusion method based on a generative adversarial network. The method comprises: firstly, collecting multispectral image and panchromatic image data sets, carrying out registration processing on images, and dividing the data sets into a training set and a test set; constructing a feature extraction network, and inputting the feature extraction network as a panchromatic image; secondly, constructing a generative adversarial network, optimizing a discriminator by utilizing a Wassertein distance, inputting a multispectral image, outputting a fused high-resolution multispectral image, finally training the generative adversarial network, and performing testing by utilizing a test set. Particularly, the method does not need an additional processing flow, and is an end-to-end method for realizing multispectral image fusion. Moreover, according to the method, a large amount of data is used as a fusion mapping relation for driving learning, and for most images with different data sets, the method can better maintain spectral information of an original multispectral image while spatial detail information of the fused image isinjected.

The invention discloses a positioning method of a working object and a target of a substation based on infrared and visible light fusion. The technical scheme is as follows: by using a visible light and infrared sensor system deployed in the substation site, real-time images are collected, and a distributed image processing platform is used to remove the noise, fuse and enhance the pretreatment, then the target and the background are segmented, target features are extracted to detect a site intrusion target, and then dynamic target identification, lactation and tracking are performed. It is divided into three stages: dynamic target detection and identification based on 3D scene and multi-spectral image fusion, target location based on a multi-camera and target tracking based on a 3D scene,and a corresponding algorithm prototype is given. The invention adopts a fusion method based on a three-dimensional scene model, utilizes the spatial corresponding relationship between visible lightand infrared images to transform, solves the fusion problem under the condition of large field of view and multiple angles, obtains a better effect, and obtains more complete and reliable image information.

The invention discloses a hyperspectral image and multispectral image fusion method based on a two-way dense residual network. A hyperspectral image with low spatial resolution and a hyperspectral image with high spatial resolution in the same scene are fused to reconstruct a super-resolution hyperspectral image. Frequency domain and spatial texture information in hyperspectral and multispectral images can be more fully utilized through the double-path dense residual network, the spatial resolution is improved, meanwhile, spectrum distortion is restrained, and the method can be applied to thefields of satellite remote sensing, agricultural geology general survey, medical imaging, environment monitoring and the like.

The invention discloses a hyperspectral and multispectral image fusion method and system and a medium. The hyperspectral and multispectral image fusion method comprises the following steps: inputtinga hyperspectral image Y and a multispectral image Z; extracting a spectral subspace S from the hyperspectral image Y by adopting singular value decomposition; estimating a subspace coefficient A according to the multispectral image Z and the obtained spectral subspace S; and multiplying the spectral subspace S and the subspace coefficient A to obtain a fused hyperspectral image X. According to themethod, the low-resolution hyperspectral image and the high-resolution multispectral image can be effectively fused to obtain the high-resolution hyperspectral image, and the method has the advantages of being high in reconstruction precision, high in calculation efficiency, high in universality and robustness and resistant to noise interference.

Satellite image fusion method and system are provided. The satellite image fusion method includes matching sizes of a panchromatic image and a multispectral image captured from a satellite image; dividing the panchromatic image and the multispectral image into a plurality of blocks; calculating coefficients to acquire Intensity (I) component image data using pixel values of each block of the multispectral image; and generating fused multispectral image data by applying the acquired I component image data to a fusion algorithm. In the multispectral image fusion, the distortion of the color information can be minimized and the multispectral image data of the high resolution can be attained. In addition, the present invention is applicable to not only the IKONOS images but also other satellite images, and the present image fusion can be carried out fast.

The invention relates to an infusion method of a full-color image and a multi-spectral image based on low-frequency correlation analysis. The method comprises the following steps: expressing a gradient field of the matched and updated full-color image as image space geometric information to form space geometric information constraint on a high-resolution fusion image; introducing spectral information constraint so that the low-frequency part of the high-resolution fusion image coincides with the original multi-spectral image; automatically determining the proportion relationship between two types of the constraints by low-frequency correlation analysis of the full-color image and the multi-spectral image so as to guarantee rational fusion of the space geometric information and spectral information in different image areas; and finally adding an adjustable parameter between two types of the constraints to adjust relative content of the geometric information and the spectral information in a fusion result. The fusion effect of the method is superior to the currently popular algorithms such as IHS, PCA, DWT and the like; and the relative content of the geometric information and the spectral information can be adjusted freely, thus improving practicability of the method.

The invention provides a satellite panchromatic and multispectral image fusion method of a multi-scale convolutional neural network. The method comprises the following steps: constructing a basic convolutional neural network, constructing a required training set and a required test set, sharpening the multispectral image, performing data enhancement processing on the training set and the test set,constructing a multi-scale convolutional neural network model, and averaging fusion results of the multi-scale convolutional neural network model to finally obtain a fusion result image; the three-layer convolutional neural network is constructed by selecting a construction method of the three-layer convolutional neural network with simple construction; Compared with the prior art, the method hasthe advantages that the implementation probability of the convolutional neural network is improved, the fusion efficiency is improved, when the network input is formed, the multispectral image is sharpened firstly, the spatial detail information in the multispectral image which is easily ignored in a traditional fusion method can be fully utilized, and the fusion effect of the spatial detail information is improved.

The invention provides a hyperspectral and multispectral image fusion method based on joint optimization, a computer readable storage medium and electronic equipment, and solves the problem that the spatial resolution of a hyperspectral image is low due to the fact that an existing hyperspectral and multispectral image fusion method depends on a spatial degradation matrix. The method comprises thefollowing steps: step 1, inputting a real value; 2, preprocessing the real value to obtain an observable low-spatial-resolution hyperspectral image and an observable high-spatial-resolution multispectral image; step 3, performing spectral demixing on the image Ih and the image Im by utilizing non-negative matrix decomposition according to the linear spectral aliasing model, and performing joint optimization on the required end member matrix and the required abundance matrix to obtain an end member matrix E and an abundance matrix Ah of the image Ih and an abundance matrix A of the image Im; and step 4, estimating a high-spatial-resolution hyperspectral image, and recording the image as an image Z.

The present invention discloses a method for realizing three-dimensional reconstruction by multi-spectral image fusion, comprising the following steps of: realizing biorthogonal wavelet decomposition; determining fusion operators; implementing HIS transform, and separation of space and spectral characteristics; combining bands; fusing a resulted image. After processing with the method for realizing three-dimensional reconstruction by multi-spectral image fusion, an internal structure of a three-dimensional structure is quite clear, and therefore digital effect of the multi-spectral image is optimal, and complementation of advantages of single spectrum information is realized.

The invention discloses a fusion method for a terahertz image and a visible light image, and belongs to the field of multi-spectral image fusion. The fusion method is characterized by comprising the steps of superposing the centers of targets formed by a terahertz image acquisition device and a visible light image acquisition device; shooting a detected object to obtain the terahertz image and thevisible light image of the detected object, and carrying out image preprocessing; performing image registration operation on the preprocessed terahertz image and visible light image, extracting imagefeatures, and performing geometric correction on the visible light image; and carrying out fusion according to a result of the registration operation, thereby obtaining a complete target area of thedetected object. The terahertz image acquisition device and the visible light image acquisition device shoot the detected object to carry out image information acquisition, so that the fused image isricher in information and more comprehensive in detail. The image fusion effect is good; the information of the visible light image and the terahertz image is integrated, so that a target scenery canbe displayed more comprehensively and clearly; and the method can be applied to numerous fields of security protection and the like.

The invention discloses a method for fusing full-color and multi-spectral images on the basis of fitting for substituted components. The method includes stretching the brightness of a full-color image at first, and interpolating a multi-spectral image to form an interpolated image with the resolution identical to that of the full-color image; transforming a matrix of the multi-spectral interpolated image to obtain a to-be-substituted component and a retaining component, and subtracting the to-be-substituted component from the full-color image to obtain a difference image; performing Gaussian filter for the difference image to obtain a difference trend image, and subtracting the difference trend image from the full-color image to obtain a fit full-color image; and substituting the fit full-color image for the to-be-substituted component, and inversely transforming a matrix to obtain a fused image. Compared with the prior art, the method has the advantages that the problem of spectral distortion caused by an existing method for fusing images by means of substituting components is solved, and the method is high in adaptability.

The invention proposes a full-color multispectral image fusion method and system based on adaptive Gaussian filtering, and the method comprises the steps: carrying out the downsampling of a full-colorimage till the size of the full-color image reaches the size of an original multispectral image; carrying out the statistical operation of the mean values and mean gradients of all bands of the downsampled full-color image and the original multispectral image, taking the mean value of the downsampled full-color image as the standard, and adjusting the mean value and mean gradient value of all bands of the multispectral image; carrying out the fitting calculation of the optimal parameters for the Gaussian filtering of the downsampled full-color image; carrying out the upsampling of the downsampled full-color image and the original multispectral image after filtering, carrying out the sampling till the size reaches the size of the original full-color image, obtaining an analog full-color image and an upsampled multispectral image, and carrying out the full-color multispectral fusion. The method enables the resolution to be high, enables the spectrum fidelity to be high, and is good in adaptability.

The invention discloses a multispectral image fusion imaging method and system based on fuzzy kernel estimation. The method comprises the steps that a multispectral image and a panchromatic image areacquired; inputting the multispectral image and the panchromatic image into a pre-trained panchromatic and multispectral image depth fusion network to obtain a fused multispectral image, and pre-training the panchromatic and multispectral image depth fusion network to establish a mapping relationship between the multispectral image, the panchromatic image and the fused multispectral image, and a training data set adopted when the panchromatic and multispectral image deep fusion network is trained being a fusion training data set constructed based on a spatial fuzzy kernel and a spectral fuzzykernel. According to the invention, a high-spatial-resolution multispectral image can be reconstructed efficiently; the method can be applied to the practical application fields of remote sensing satellite natural disaster monitoring, crop detection, mineral exploration, land resource investigation and the like.

The invention discloses a multispectral remote sensing image and panchromatic image fusion method based on adaptive spectrum-spatial gradient sparse regularization. The method comprises the followingsteps: step 1, obtaining the multispectral image and the panchromatic image which are registered in the same geographic area at the same time; step 2, based on the result of the step 1, performing down-sampling on the fused image to obtain l2 norms of the fused image and the multispectral image; 3, calculating the gradient and the weight matrix of the difference image of the fused image and the copied panchromatic image in the spatial and spectral directions so as to obtain the l1 norm of the fused image and the panchromatic image; and step 4, based on the step 2 and the step 3, obtaining energy functions of the fused image, the multispectral image and the panchromatic image, and carrying out iterative solution to obtain a fused image. The method mainly aims at the application requirementof fusion of the multispectral remote sensing image and the panchromatic image, and the spectral consistency characteristic and the spatial gradient sparsity characteristic of the multispectral imageare considered.

The invention discloses a discrete cosine transform based panchromatic and multispectral image fusion method and overcomes defects and shortcomings of existing panchromatic and multispectral image fusion method. The discrete cosine transform based panchromatic and multispectral image fusion method is characterized in that high-frequency information in panchromatic images is extracted by introduction of DCT (discrete cosine transform) transform extensively adopted in an image compression standard and successfully injected into low-resolution multispectral images through DCT inverse transform. Compared with the prior art, the method has advantages that fusion speed is evidently increased on the premise that low spatial detail distortion and low spectral distortion are guaranteed, and accordingly powerful technical support is provided for panchromatic and multispectral image fusion.

The utility model provides a surgery navigation system and method based on multispectral image fusion of reflecting mirror faces with adjustable angles and adjustable positions. The system comprises a C-shaped arm machine frame (11), an X-ray emission bulb tube (12), an image enhancement and acquisition device (13), an image processing workstation and a liquid crystal display monitor and further comprises two parallel reflecting mirrors (14), hinges (15), reflecting mirror guide rails (16), inserting grooves, a visible light CCD camera (17), a camera guide rail (18) and a calibration body with mark points. According to the surgery navigation method, a fluoroscopic image and an image acquired by the camera have the same visual angle and frame area, late image registration and other algorithm and the calculation process are simplified, the image with the optimal visual angle can be obtained through the adjustable reflecting mirror faces, and the method can be applied to surgeries of different positions flexibly. Registration is performed on the fluoroscopic image and a visible light image before a surgery, the X-ray emission bulb tube is turned off, a visible light real-time image is used for guiding the surgery, ionization radiation to doctors and patients is reduced, and positioning accuracy and treatment efficiency are improved.

The invention discloses a panchromatic and multispectral image fusion method and device and application thereof, and the method comprises the steps: carrying out the wave band superposition of a panchromatic image, carrying out the interpolation of a multispectral image into an interpolation image with the same resolution as the panchromatic image, carrying out the subtraction of the two images, and carrying out the Gaussian filtering, and obtaining an interpolation trend chart; obtaining a fusion model with high-fidelity fusion capability by constructing an unsupervised fusion model with a spectrum discriminator and a spatial spectrum discriminator; and automatically processing the panchromatic and multispectral images by using the fusion model to obtain a high-fidelity fusion image. The fusion model obtained through the fusion method is suitable for processing the remote sensing image with the original resolution, the common detail distortion problem of a current fusion method based on deep learning is effectively reduced, the operation speed is high, and the fusion effect is good.

The invention belongs to the technical field of garbage classification, and particularly relates to a garbage classification method and device based on depth image and multispectral image fusion, andthe method comprises the steps: obtaining a garbage image data set of to-be-classified garbage, wherein the garbage image data set comprises a to-be-classified garbage depth image and a to-be-classified garbage multispectral image; performing image registration processing on the to-be-classified garbage depth image and the to-be-classified garbage multispectral image; performing classification labeling processing on the registered garbage depth image and the registered garbage multispectral image; loading the labeled garbage depth image and the labeled garbage multispectral image into a YOLOv4neural network based on a Pytorch framework for model training, and generating an optimal neural network model; and generating a sorting result according to the optimal neural network model. According to the invention, automatic grading and automatic classification of the garbage are achieved, a category label with the maximum probability is output, high-precision and high-efficiency classification of the garbage is further achieved, and the production requirement of garbage classification is met.

The invention provides a real-time inspection method and equipment for a power transmission line, and belongs to the technical field of power transmission line detection. The method comprises the following steps: acquiring first inspection environment information from a multispectral acquisition module and second inspection environment information from a laser radar; wherein the first inspection environment information comprises a multispectral image of the power transmission line; the second inspection environment information comprises point cloud data of the power transmission line. And sequentially determining poses corresponding to the laser radar according to the acquisition timestamp of the second inspection environment information. And according to the pose of the laser radar and a preset point cloud map, determining the point cloud coordinate of each point cloud data point. And on the basis of a preset coordinate unified matrix and the point cloud coordinates of the point cloud data points, determining the point cloud coordinates of the point cloud data points in the world coordinate system, and fusing the point cloud data and the multispectral image through the point cloud coordinates of the point cloud data points in the world coordinate system to obtain a fused line inspection image.

The invention discloses a gradient filtering and PCA-based unmanned aerial vehicle image and multispectral image fusion method which comprises the following steps: images of the same pixel size are obtained via image registration operation and resampling operation, then two sets of independent multiband images that are same in spatial range are obtained via cutting operation, the two sets of independent multiband images are respectively a multispectral remote sensing image which is large in wave band quantity and low in spatial resolution and a three-wave-band unmanned aerial vehicle image which is high in spatial resolution, the two sets of independent multiband images are subjected to principal component transformation operation, specific filtering operators of all principal components of the unmanned aerial vehicle image are chosen to be subjected to gradient filtering operation, and therefore texture information of three principal components of the unmanned aerial vehicle image can be obtained; the texture information, as a certain weight, is superposed onto first three principal components of the multispectral remote sensing image which is large in wave band quantity and low in spatial resolution, then the principal components can be reinforced, the reinforced principal components are subjected to principal component inverse transformation operation, and therefore a multispectral fusion result which is high in wave band quantity and high in spatial resolution can be obtained. Via use of the method, limitation of a tradition fusion method in single wave band panchromatic data and multispectral image fusion, diverse data can take part in the image fusion, and the fusion result is enabled to have rich spatial detail information.

The invention discloses a complex scene recognition method based on multispectral image fusion, and the method comprises the steps: dividing an image of a scene data set into a training data set and atest data set, and enabling the image of the scene data set to be an image after the fusion of infrared light and visible light; constructing a DL-FME convolutional neural network for scene recognition according to the training data set, and segmenting the fused image by using the DL-FME convolutional neural network to obtain a segmented image; training the segmented image by using a DL-FME convolutional neural network to generate a scene recognition training model; inputting the images of the test data set into a training model to generate a scene recognition model; and identifying the to-be-identified scene by using the scene identification model. The invention further discloses a complex scene recognition system based on multispectral image fusion. The method is advantaged in that calculation cost is reduced, and scene identification accuracy is high.

The invention provides a panchromatic and multispectral image fusion method based on two-step sparse coding and wavelet transformation. The method comprises the following steps of data preprocessing,high-low resolution dictionary, one-step sparse coding, two-step sparse coding, two-step sparse coefficient merging, sparse coefficient reconstruction to obtain a fused image, wavelet transformation,high-frequency component fusion, low-frequency component fusion and inverse transformation. According to the invention, the algorithm efficiency can be improved through the two-step sparse coding algorithm, at the same time, the sparse coefficient is solved through two-step sparse coding, and the precision of sparse representation is improved; the problem of spectral distortion is solved, the fusion result can be improved through fusion according to the principle of maximum absolute value, the fusion result is clear in image and rich in detail information, the spectral information and the original image are the highest in consistency, the operation time is short, wavelet transformation is carried out through full-color and multispectral images, and the method is a rapid, efficient and effective fusion method.

The invention discloses a method for fusing a moon surface multispectral image based on angular point measurement, which is characterized by comprising the following steps of: (1) respectively carrying out N-level wavelet decomposition on two source images; (2) solving coefficient average values of corresponding positions of source image similar planes as coefficients of corresponding positions of fusion image similar planes; (3) calculating an angular point measurement response amplitude graph with the same size as that of an original image; (4) giving a sliding window with the size of K*L, wherein K and L are odd numbers, and defining a region counter; (5) determining values of central position coefficients of corresponding regions on a fusion image corresponding wavlet plane; and (6) carrying out redundancy wavelet inversion to solve the fusion image. The method has the advantages of describing significance degrees of the angular point and the edge point of the corresponding positions of the source image wavelet planes, using the angular point measurement response value as the selection reference of the fusion coefficient of the corresponding position of the wavelet plane so as to effectively maintain the detail characteristics such as edges, angular points and the like of the source images.

The invention discloses a hyper-spectral image super-resolution method based on hyper-parameter fidelity and depth prior joint learning. The method comprises the following steps: establishing a hyper-spectral and multi-spectral image fusion variational model based on depth prior regularization of a hyper-parameter fidelity model; optimizing a hyperspectral multispectral image fusion variation model; carrying out tensor representation on the model optimization iteration process; performing network expansion on the iterative process of variational model optimization, and executing the iterative process of optimization; and training the network by using the L1 norm as a loss function. The method has the capability of representing the hyperspectral image degradation model and the data prior in the network at the same time, and has excellent performance when being applied to hyperspectral multispectral image fusion.

The invention belongs to the field of image processing, and particularly relates to a time sequence remote sensing image fusion method. The method comprises the following steps: reconstructing a coarse resolution remote sensing image by using a time sequence model to obtain day-by-day coarse resolution time sequence data; performing panchromatic multispectral image fusion on the fine-resolution remote sensing image to obtain a high-resolution multispectral image; and carrying out space-time fusion on the existing high-resolution image and the reconstructed coarse-resolution image at the corresponding position to obtain a day-by-day high-resolution remote sensing image. According to the method, the space-time fusion technology, the panchromatic multispectral image fusion technology and the time sequence modeling interpolation technology are combined, so that the reconstruction of a high time sequence is realized, and the spatial resolution of the image is improved.

The invention provides a satellite-borne panchromatic and multispectral image fusion method. The satellite-borne panchromatic and multispectral image fusion method includes the steps: using Gaussian guided filtering to combine with a panchromatic image to complete resampling amplification of an original multispectral image; injecting the spatial detail information of the panchromatic image into the amplified multispectral image through the combination upsampling process, and then performing hyperspherical color transform (HCT) on the original multispectral image to extract the spectral information of the original multispectral image; and fusing the spectral information with the spatial information of the panchromatic image in the hyperspherical color space, and obtaining a multispectral image with high spatial resolution through HCT inverse transform. The satellite-borne panchromatic and multispectral image fusion method improves the spatial resolution of the resampled multispectral image, can be directly extended to fusion processing of the hyperspectral image, and is relatively higher in universality and practicality.