52 results about "Variational regularization" patented technology

The invention discloses a seismic inversion method and system based on generalized total variation regularization, so as to solve the problem that step effects are generated inside a formation when the existing seismic inversion technology adopts total variation regularization. The method uses the characteristic that generalized total variation not only uses first-order partial derivative information of a to-be-inverted parameter but also uses second or higher-order information, the formation interface can be clearly expressed and the step effects inside the formation can also be weakened; besides, a phenomenon that a product of a wavelet matrix, a difference matrix and the natural logarithm of the to-be-inverted parameter can be used to be converted to the convolution among a convolutionkernel corresponding to the wavelet matrix, a convolution kernel corresponding to the difference matrix and the natural logarithm of the to-be-inverted parameter is found out, the convolution operation in a time domain is converted to point multiplication operation in a frequency domain, and the inversion speed is greatly improved; a reflection coefficient does not need to be inverted firstly, anddirect inversion is carried out in the frequency domain to obtain the to-be-inverted parameter; and the method and the system are applicable to the technical field of seismic exploration.

The invention relates to a scene flow estimation method based on 3D local rigidity and depth map guided anisotropic smoothing. The method comprises the following steps: S1, acquiring a texture image and a depth image which are aligned at the same time through an RGB-D sensor; S2, building a scene flow estimation energy functional, and calcualting a dense scene flow based on a 3D local rigidity surface hypothesis and a global constrained method, wherein the form of a scene flow energy function is shown in the description; S3, designing date items based on the texture image and the depth image as well as the 3D local rigidity surface hypothesis; S4, designing smoothing items based on a depth map driven anisotropic diffusion tensor and total variation regularization; S5, creating an image pyramid, and adopting a coarse-to-fine solution strategy; and S6, calculating a scene flow by use of a duality method, and introducing scene flow auxiliary variables. According to the invention, the weight of a space-domain filter is determined by both the chromatic aberration and location relationship between the pixels of a color image, and therefore, the edge distortion problem in the process of repair is solved. In order to reduce repair error, the weight of a value-domain filter is determined by the color information and the structural similarity coefficient.

The invention discloses a cartoon texture decomposition-based image super-resolution reconstruction method. The method comprises an offline training process and an online test process. The offline training process comprises training of a high and low-resolution texture dictionary pair, and offline texture anchor point mapping. In the online test process, a low-resolution input image is subjected to cartoon and texture decomposition; a high-resolution cartoon image corresponding to a low-resolution cartoon image is reconstructed based on an improved total variation regularization method; a high-resolution texture image corresponding to a low-resolution texture image is reconstructed through an improved external training sample texture anchor point mapping matrix-based super-resolution method; and the reconstructed cartoon image and texture image are subjected to addition so as to output a required high-resolution image. The method has the advantages of two algorithms, and not only has relatively good denoising capability and relatively good edge retention capability but also has high adaptability and robustness.

Provided is a complicated background SAR image naval ship tail track detection method. Firstly, morphological component analysis or relative total variation regularization is carried out on an SAR image, and the image is decomposed into a texture component representing a complication sea surface and a structure component containing naval ship tail tracks; then image enhancement is carried out on the structure component and thresholding is carried out on a gradient image, and accordingly a binary image of a naval ship tail track area is formed; finally, radon transformation or Hough transformation linear detection is carried out on the enhanced naval ship tail track area, inversion is carried out in a conversion domain, and continuous tail tracks and related parameters are obtained. Structure and texture decomposition is carried out on the SAR image, and influences of speckle noise and complicated textures on naval ship tail track detection are eliminated; image enhancement is carried out on the structure component, binaryzation is carried out on the gradient image, and the naval ship tail track area is further highlighted.

The invention discloses a total-variation regularization and variable splitting-based lens-less imaging rapid reconstruction method. Aiming at a thought that image reconstruction problems of lens-lessimaging systems adopt total-variation regularization and variable splitting, a to-be-solved target function is split to two sub-problems, and the sub-problems are alternately solved to obtain a finalresult. The method comprises the following steps of: firstly importing a total-variation regularization image reconstruction model according to a linear imaging mechanism in lens-less imaging; importing an auxiliary variable and splitting the to-be-solved target function into the two sub-problems by using a variable splitting method; solving the two sub-problems by using Tikhonov regularization and anisotropic total-variation (TV) regularization; and finally alternately solving the two sub-problems so as to obtain an optimum solution. According to the method, lens-less image reconstruction can be stably carried out under unsatisfactory factors, noises can be effectively removed, and detailed information such as edges and the like of reconstructed images can be kept at the same time.

The invention discloses a motion-blurred image restoration method based on rotary difference and weighted total variation, which is characterized by comprising the following steps: 1, estimating the motion direction of a motion-blurred image by using a motion-blurred direction space-domain estimation method based on a rotary difference operator; 2, estimating the motion length of the motion-blurred image by using a blurred image point spread function parameter estimation method based on error analysis; and 3, determining a blur kernel in accordance with the steps 1 and 2, and carrying out image restoration to obtain a sharp image by using a weighted total variation regularized image restoration method, i.e. using the weighted difference operator to detect the edges of the image in all directions, using the weighted difference operator as a total variation regular item to retain more image details, and solving the weighted total variation model by using a fixed point lagging and spread method so as to finish image restoration. By using the method to restore the motion-blurred image, the acquired image can well inhibit ringing, abundant image details can be retained, and the restored image can achieve high quality.

The invention discloses a non-local-total-variation image restoration method based on sparse overlapped group priori constraints. The method includes the following steps that 1, a degenerative-processmathematical model g=h*f+n is established; 2, a constraint item and a regularization term are established, and are the image representation constraint item phi<OGS>(f) and the image non-local-total-variation regularization term phi<NLTV>(f) under the sparse overlapped group priori constraints respectively; 3, an image restoration minimum energy functional model arg min lambada phi<fid>(g, f)+alpha phi<OGS>(f)+phi<NLTV>(f) is established; 4, a target function is optimized with the ADMM algorithm, and a middle restored image is solved and a updated. The end condition is met, iteration is end, and a final restored image is obtained. According to the method, image priori information can be fully used in the introduced sparse-overlapped-group-priori image representation constraint item, the identifying degree of the image similar structure is increased, the non-local-total-variation regularization defect is overcome, and more detail information is further reserved.

The invention discloses an L1 norm total variation regularization heterogeneity correction method based on anisotropy. The method comprises the following steps that: establishing anisotropy total variation regularization model based on a single image, converting a problem of removing infrared image heterogeneity stripe noise into a minimized total variation problem, then, adopting a separated Bragg iteration method to carry out optimization to obtain an optimal solution, and taking a last iteration result as a corrected infrared image. The method has an innovation point that a traditional total variation model aims at a structural feature improvement equation that the horizontal direction total variation of heterogeneity stripe noise is far greater than vertical direction total variation,a regularized constraint based on the L1 norm can be used to enable the method to be suitable for the heterogeneity correction of the infrared image, the separated Bragg iteration method replaces a steepest gradient descent method to carry out the optimal processing of the equation, processing speed is greatly improved, and the edge information of an object is kept to a maximum degree.

The invention discloses a rapid multi-resolution denoising method and device under a hybrid noise model. The method comprises the following steps: after acquiring an image with noise, building a multilayer gaussian pyramid and a multilayer Laplacian pyramid corresponding to theimage with the noise; performing a plurality of times of iteration calculation to obtain total variation regularization denoising target functions respectively by using a gradient descent method; performing the final iteration calculation to obtain a denoised image and the like. The device comprises a storage for storingat least one program, and a processor for loading at least one program for implementing the method. By adopting the rapid multi-resolution denoising method and device, the plurality of times of iteration calculations of the gaussian pyramid and the Laplacian pyramid are combined, the target functions can be solved corresponding to different image resolutions in each iteration calculation, the initial value of each iteration calculation is derived from the result of the previous iteration calculation and the Laplacian pyramid, denoising effect can be ensured, the calculation load of resolvingis lowered, the resolving speed is increased, and the calculation efficiency is increased.

The invention discloses an image conversion processing method in the technical field of image processing. The image processing method aims to solve the technical problems in the prior art that an image processing method based on the cyclic generative adversarial network often cannot focus on local features and has stripe-shaped noise in the image local feature migration process, and the overall color of a picture is easily changed. The method comprises the following steps: acquiring a real image containing processing characteristics and a real image not containing the processing characteristics, and constructing a training sample; inputting a training sample into the cyclic generative adversarial network, and training the cyclic generative adversarial network by taking a pre-constructed loss function minimization as a target, the loss function comprising a total variation regularization loss function; and processing the to-be-processed image by adopting the trained cyclic generative adversarial network.

The present invention relates to a solution for solving an ill-posed inverse problem in image analysis, e.g. in an electron tomography application in order to recover a structure of a sample. The solution is provided for instance as a method comprising steps of determining reliable prior knowledge about the solution, determining initial guess for the solution and determining the corresponding forward operator, deciding upon model of stochasticity, deciding on suitable regularization method, deciding on updating scheme, and producing a sequence using the set configuration.

The invention provides a robust CBCT reconstruction method based on low-rank tensor decomposition and total variation regularization, and the method comprises the steps: employing a Huber loss function as a data fidelity term in CBCT reconstruction, so that impulse noise re-constructed under low radiation dose has robustness; with the low-rank tensor characteristic as a priori term so that characteristic is beneficial to recovery of structural information loss caused by impulse noise; integrating a 3D TV priori item to reduce the influence of Gaussian noise, so that a CBCT reconstruction modelis proposed; and solving an optimization problem through an alternating minimization method, and obtaining and outputting a reconstructed image.Under the state that Gaussian noise and impulse noise are mixed, the influence of the noise can be effectively eliminated, the edge can be well reserved, and a high-quality three-dimensional image can be reconstructed.

The invention discloses a multi-energy-spectrum segmented sparse scanning iterative reconstruction method based on traditional single-energy CT. The method comprises the following steps of: 1, scanning a measured object in a plurality of corresponding limited angle ranges by applying X-rays with a plurality of energies by adopting a segmented sparse scanning mode on the traditional single-energy CT to obtain a plurality of segmented sparse projection data of the measured object under the plurality of energies; and step 2, applying an image iterative reconstruction method combining total variation regularization and robustness principal component analysis constraint to the plurality of segmented sparse projection data obtained in the step 1 to respectively reconstruct CT images of the measured object under a plurality of energies. According to the invention, a segmented sparse scanning mode is designed by utilizing the consistency of the structure of a measured object under different energies, so that the technical requirement on the voltage switching frequency of the bulb tube is greatly reduced. Meanwhile, in order to solve the problem of incompleteness of projection data, the invention provides an image iterative reconstruction method combining total variation regularization and robustness principal component analysis constraints, noise and artifacts are effectively removed,and a high-quality reconstruction result is obtained.

The invention discloses a bilateral total variation regularization correction method for the non-uniform response of an IRFPA detector. The method is characterized in that according to requirements of correction output of the non-uniform response of the IRFPA detector for artifact inhibition and detail maintaining, a bilateral total variation regularization term is led into an energy functional in cooperation with a correction parameter regression estimation method, diffusion properties of the bilateral total variation anisotropy are fully utilized, the artifact effect of response output after correction is effectively reduced, and detail information in an original scene is maintained better, so the temperature resolution of the IRFPA detector is effectively promoted.

The invention discloses a hyperspectral image denoising method based on non-convex low-rank matrix approximation and total variation regularization. The hyperspectral image denoising method comprisesthe following steps: 1) acquiring hyperspectral image data to be denoised, wherein M, N and K represent the number of width and height spectral bands of a hyperspectral image respectively; 2) constructing a denoising model based on non-convex low-rank matrix decomposition and TV regularization; (3) optimizing the model by adopting an Augmented Lagrange function (ALM) algorithm, and optimizing themodel by adopting an ALM algorithm; 4) optimizing and solving the model by adopting an augmented Lagrange function algorithm; and 5) outputting the hyperspectral image after the mixed noise is removed.

The invention relates to a partitioned bilateral total-variation regularization image noise elimination method. The method comprises the steps that (1) a pollution image X<0> is acquired and is used to initialize a denoised image subjected to the first iteration, and then the step (2) is entered; (2) a partitioned bilateral structure similar distance matrix DW<t> of a denoised image subjected to the t(th) iteration is calculated, and then the step (3) is entered; (3) partitioned bilateral total-variation regular terms of the denoised image subjected to the t(th) iteration are constructed, and then the step (4) is entered; (4) an energy functional E<t> composed of fidelity terms and the partitioned bilateral total-variation regular terms is constructed, and the step (5) is entered; (5) a steepest descent method is adopted to solve a minimization problem of the energy functional E<t>, a denoised image subjected to the (t+1)(th) iteration is obtained, and the step (6) is entered; and (6) whether the number of the iterations is smaller than the maximum number N of iterations is judged, if the number of the iterations is smaller than the maximum number N of iterations, t is made to be equal to t+1, and the step (2) is entered, and otherwise the denoised image subjected to the (t+1)(th) iteration is output to end the operation.

The invention proposes a cellular neural network edge detection method based on relative variation regularization. The method includes image preprocessing and edge detection after the preprocessing. The image preprocessing mainly adopts the relative variation to smooth noises and textures; and the edge detection after the preprocessing mainly adopts a cellular neural network method to detect images. The method is capable of extracting effective and reliable and accurate effective edge information from pictures which are rich in noises and complex in textures. Compared with a traditional Canny method, the method restrains effects from the noises and the textures and compared with a standard cellular neural network method, the method prevents advance design of complex CNN template parameters.

The invention discloses a sharp edge preserving resistance tomography image reconstruction method. The method comprises the steps of obtaining a relative boundary measurement value vector and a sensitivity matrix required by reconstruction according to a measured field domain; Setting initialization parameters; Determining an optimal regularization parameter by using an improved GCV method; Calculating a gradient and a Hessian matrix of the objective function; Updating the value of the solution by using a Gauss-Newton iteration method; Judging whether iteration is finished or not; And performing imaging according to the finally solved imaging gray value. According to the method, the defects that the edge of a reconstructed image is fuzzy and the image resolution is low in a traditional Tikhonov regularization algorithm and a total variation regularization algorithm are overcome, and the method has a good effect on improving the quality of an electrical tomography reconstructed image and keeping the keeping capability of a sharp edge.

The invention discloses a compression perceptual image reconstruction method based on multi-view images. Firstly, for each target reconstruction, a corresponding dynamic image set is selected by calculating a parallax compensation size between images, and then multi-view image reconstruction is realized based on the selected dynamic image set. In the reconstruction process, high-quality reconstruction results are obtained according to the adaptive total variational regularization constraint based on disparity compensation and non-local low-rank tensor constraint.

The invention discloses a tensor completion-based multi-energy CT (computed tomography) imaging method, a tensor completion-based multi-energy CT imaging device, storage equipment and a tensor completion-based multi-energy CT imaging system. The tensor completion-based multi-energy CT imaging method comprises the following steps of: firstly, respectively processing an obtained projection value ofeach section of narrow beam energy spectrum by using an FDK (frequency division duplex) algorithm to obtain a reconstructed image of each energy section; then, modeling the obtained reconstructed image of each energy segment into a third-order tensor, establishing a tensor nuclear norm and total variation regularization minimization model, and improving the precision of the reconstructed image ofeach energy segment; and finally, performing optimization weighting on each slice in the tensor obtained by modeling according to a weighting fusion algorithm to obtain a final imaging image. The beneficial effects of the invention are that the method achieves the data collection through the multifunctional CT simulation system based on GATE, combines the inherent multi-dimensional properties of aCT problem with the tensor, and achieves the more precise reconstruction of a CT scanning image.

The invention provides a low dose X-ray CT image reconstruction method based on completely generalized variational regularization. The method comprises the following steps that (1) the system parameters of CT equipment and projection data after logarithmic transformation under a low dose scanning protocol are acquired; (2) Anscombe transformation is performed on the projection data, and the projection data obeying Poission distribution are converted into Gaussian distribution data u of which approximate obeying variance is 1; (3) an ideal data restoration model based on completely generalized variational minimization is established, and the restored projection data f are obtained through solving by using a Chambolle-Pock algorithm; and (4) inverse Anscombe transformation is performed on the restored projection data f obtained in the step (3), and a CT reconstruction image is obtained through a filtering back projection algorithm. According to the method, noise and bar artifacts of the image can be removed under the premise that the projection data do not meet the piecewise constant hypothesis, and the spatial resolution of the image can be greatly maintained.

An atmospheric disturbance image recovering method based on variation regularization relates to image processing. The atmospheric disturbance image recovering method comprises: acquiring a video frame subjected to atmospheric interference of a fixed target or carrying out additional simulated disturbance on one picture by using simulation software to generate one group of video frame; carrying out low-rank decomposition on the video frame to obtain an initial reference image; optimizing the reference image by using an optimization model based on a non-partial total variation regular term and a controllable kernel regression regular term, and accelerating the optimization process by using a separated Bregman algorithm; carrying out B spline interpolation registering on the video frame by using an optimized reference image to obtain a registered video frame; using space weighted nuclear norm minimization to fuse the registered video frame to form a near diffraction limit picture; and carrying out deconvolution processing on the near diffraction limit picture to obtain the picture which is finally deblurred and has no noises. By virtue of the atmospheric disturbance image recovering method, the disturbance removing result is improved and a recovered image with clear vision and abundant details is obtained, so that the atmospheric disturbance image recovering method can be used for observing lands in the air, remotely monitoring and the like.

The invention discloses a low-dose CT projection domain denoising method based on composite regularization, and the method comprises the steps: employing a non-convex penalty function to enhance the sparsity of wavelet transformation based on a second-order fully generalized variational regularization method, and establishing a wavelet-TGV denoising model based on the non-convex penalty function for a projection domain; establishing a logarithmic penalty function, and performing wavelet-TGV denoising; adopting a split augmented Lagrange shrinkage algorithm, and finding an optimal solution of wavelet-TGV denoising; and inputting projection data, and reconstructing a CT image by using a filtered back projection algorithm. Most noise and strip artifacts are removed from the reconstructed image on the premise of keeping edge and texture detail information. The method has obvious noise reduction and artifact removal effects on low-dose CT images.

The invention provides a novel image restoration method based on self-adaptive anisotropy total variation regularization, which comprises the following steps: firstly, acquiring an original image by using image equipment, and calculating gradient mapping of the original image in an x direction and a y direction; secondly, improving the anisotropic total variation regularization model based on a weight adaptive theory to obtain a novel non-smooth and non-convex denoising model; and finally, calculating an approximate solution of the novel non-smooth and non-convex denoising model by using an iterative reweighting algorithm, and solving the approximate solution model by using an alternate minimization algorithm to obtain a final restored image. According to the method, the non-convex optimization problem is solved by adopting an iterative reweighting algorithm, so that the problem of imbalance of convergence rates in different regions is avoided, and an important geometric structure of the image is better reserved while the step effect is eliminated; and the restored image is used as an initial image of a new round of iteration, so that the denoising effect of the restored image is better and is closer to an original clear image.