98 results about "Total variation model" patented technology

The invention discloses a self-adaptive regularized super-resolution image reconstruction method which can keep marginal definition, which mainly solves the problem that the prior method has edge fog in reconstruction of a degraded image. The method comprises the following steps: an imaging model is constructed; on the basis of an unconstrained objective function constructed by a Lagrangian multiplier method, gradient is increased to approach a bound term; the objective function is expanded; L1 norm is adopted to measure a data approximation term; a self-adaptive bilateral total variation model which can carry out local adaptive control on the smoothing effect is utilized to construct a self-adaptive regular term; a gradient approximation term is added to be as constraint of gradient consistency; edge information is kept; the self-adaptive regular term and a gradient consistency bound term are introduced as constraint conditions; an expanded Lagrangian objective function is constructed and optimized; and an optimized unconstrained objective function is utilized to reconstruct an image, thereby obtaining a high-resolution image of which the edge is kept. The method can keep image edge clear, can inhibit noise and is suitable for restoration treatment on the degraded image.

The invention relates to an intra-frame prediction video coding method based on image inpainting and vector prediction operators. The method is technically characterized by comprising the following steps of: (1) calculating the rate distortion cost value RD1 of a current block based on a traditional intra-frame prediction mode in high efficiency video coding (HEVC); (2) calculating the rate distortion cost value RD2 of the current block based on an intra-frame prediction mode of a Laplace equation image inpainting method, and if the RD2 is smaller than the RD1, calculating the rate distortion cost value RD2 of the current block based on an intra-frame prediction mode of a total variation model image inpainting method; (3) calculating the rate distortion cost value RD3 of the current block based on an intra-frame prediction mode of the vector prediction operators; and (4) calculating the prediction pixel value of the current block according to the comparison result of the RD1, the RD2 and the RD3 by an encoder end, and predicting, compressing and coding the current block. The method is reasonable in design; the prediction accuracy of the conventional intra-frame prediction mode based on image inpainting is improved; and the coding rate can be reduced under the condition that the quality of the coded and decoded video is almost not changed, so that the compression efficiency of video coding is improved.

The invention discloses a two-dimensional image-assisted depth image enhancement method. The method includes the steps of utilizing a ToF camera and a high-resolution color camera to obtain a low-resolution depth image and a high-resolution color image of a same scene, and mapping a low-resolution depth map to a high-resolution color space; constructing data items of an objective function according to reconstruction constraints of a depth image; utilizing edge information of the color image to perform weighting on a regularization model, introducing diffusion tensor to a second-order generalized total variation model, and forming a regularization model for characteristics of the depth image; and solving through iterative reweighting and a primal-dual algorithm, so as to obtain a reconstruction result. The two-dimensional image-assisted depth image enhancement method provided by the invention can effectively solve the problems of low resolution, unclear margin and existence of noise of the depth image, and effectively protects the edge structure of the image.

The invention provides a Split Bregman weight iteration image blind restoration method based on a non-convex higher-order total variation model, and belongs to the technical field of image processing. The method is characterized in that firstly, a non-convex higher-order total variation regularization blind restoration cost function is obtained by introducing image border sparse prior information meeting a hyper-Laplacian model and by combining a high-order filter bank capable of generating piecewise linear solutions; secondly, a weight iteration strategy is provided, a minimization problem of the non-convex higher-order total variation regularization blind restoration cost function is converted into a minimization problem of an approximate convexity cost function with the updated weight; thirdly, the minimization problem of the approximate convexity cost function with the updated weight is converted into a new constraint solving problem through an operator split technology, and the constraint solving problem is converted into a split cost function through the method of adding a penalty term; fourthly, the split cost function is solved through a Split Bregman iteration solving frame. According to the Split Bregman weight iteration image blind restoration method based on the non-convex higher-order total variation model, an image can be restored effectively and rapidly, the shortage that a staircase effect is generated in a traditional total variation regularization blind restoration method is overcome, and meanwhile a better restoration effect on manually degraded images and actually degraded images is achieved.

The invention discloses a non-local-restriction-based total variation image deblurring method, which mainly solves the problems that the edge of an image cannot be sharpened and the high-frequency details of the image cannot be recovered in the process of deblurring the image in the prior art. The method comprises the following steps of: (1) initiating a deblurring result image by a Wiener filtering method; (2) calculating the non-local weight coefficient matrix of the deblurring result image; (3) updating the deblurring result image by using a non-local-restriction-based threshold value iterative formula; (4) performing noise suppression on the deblurring result image by a total-variation-model-based denoising method; and (5) judging whether a stopping condition is met or not, if the stopping condition is met, acquiring a final deblurring result image, and if the stopping condition is not met, returning to the step (2) until the stopping condition is met. In the process of deblurring the image by the method, the edge of the image can be sharpened, and the high-frequency details of the image can be recovered; and the method can be used for deblurring various known blurred types of blurred images.

The invention relates to an image denoising method based on a Shearlet contraction and an improved TV model, wherein a TV-denoising model is improved and a novel mixed denoising method by combining the Shearlet contraction is proposed. The method organically combines the sparse representation capability of Shearlet for a high dimension function with the protection capability of the TV-denoising model for an edge, wherein the method obtains a first-denoising image through a hard threshold function contraction, and then improves fidelity terms of a total variation model, and then makes a second denoising of the false Gibbs effect of the first-denoising image by combing the improved total variation model. On the prerequisite of protecting important information such as edges, etc., the method effectively inhibits the false Gibbs oscillation caused by the Shearlet contraction, and realizes a better visual effect and a lower computation complexity.

A technical scheme of the invention includes a method of recognizing circular defects based on weld X-ray digital images; the method comprises: S1, acquiring a weld X-ray image with a circular defect; S2, using a display differential algorithm of a total variation model to denoise the weld X-ray image acquired in step A; S3, preliminarily determining the edge of the circular defect through Canny edge detection operator; S4, performing mathematical morphological processing on the image acquired in step C to further determine the edge contour, mass center and area of the circular defect. The method has the advantage that weld X-ray circular defects can be detected more quickly and accurately.

The invention relates to a turbulence-degraded image blind restoration method based on dark channel and Alternating Direction Method of Multipliers. The method includes the following steps: firstly on the basis of the multiple dimension theory, in each dimension, applying dark channel prior constraint on an image, applying sparse constraint and energy constrain on a point spread function, then using the coordinate descent method and conducting alternating iteration to estimate a fuzzy kernel and the image in current dimension, if the dimensions arrive at the maximum thereof, a final estimated fuzzy kernel is obtained, finally, in combination with a total variation model, using a derivative Alternating Direction Method of Multipliers to make details of the image restored quickly. According to the invention, the method, by using the dark channel prior information of a clear image as a constraint item, can help a cost function to converge to a clear solution in the iteration process, addresses the susceptibility of obtaining a fuzzy solution by using tapered prior information under the Maximum posterior probability in current blind restoration algorithm, such that the method herein can restore more image details, has less ring effect, and effectively increases restoring quality.

The invention discloses a textural feature and color feature fusion-based image saliency detection method, and mainly aims at solving the problem that the textural feature utilization is insufficient and the high-texture image saliency detection result is relatively bad in the prior art. The scheme is that the method comprises the following steps of: 1) inputting an image and removing textures of the image by utilizing a total variation model-based filtering method so as to obtain an image with color textures; 2) filtering the input image by utilizing a Garbor filter so as to obtain an image with image texture information; 3) calculating a preliminary contrast value according to the image with the color features; 4) calculating a background probability according to the image with the texture information; and 5) fusing the preliminary contrast value and the background probability to obtain a new contrast and then obtaining a saliency map on the basis of color and texture features. According to the method disclosed by the invention, the color and texture information of the image is sufficiently utilized, so that the complicated-texture image detection effect is improved; and the method can be applied to computer vision tasks.

The invention discloses a de-noising processing method for three-dimensional seismic images. By the de-noising processing method, advantages of a mixed norm iterative algorithm and accuracy of continuity factors of the seismic images are sufficiently utilized, so that a de-noising effect for the seismic images is obviously enhanced. The de-noising processing method has the advantages that the timing for carrying out image boundary protecting processing by 1-1 norms and the timing for carrying out smoothing for the images by 1-2 norms are adaptive in a seismic image de-noising process owing to introduction of the mixed norms; adaptability of the algorithm is obvious and accurate in the de-noising process owing to the introduction of the continuity factors, and a better effect in the aspect of accurately reserving useful pattern information in the original patterns is realized as compared with a traditional de-noising method; and as shown in test results, the convergence rate of the mixed norm iterative algorithm for solving a total variation model is high, computation time is short, and the mixed norm iterative algorithm is good in robustness.

The invention discloses an image denoising method and system based on total variation and wavelet transformation. The method comprises the steps of performing wavelet transformation of an obtained to-be-denoised original image, so as to obtain a low frequency wavelet coefficient, and high frequency coefficients in a horizontal direction, a vertical direction and an oblique direction of the original image; establishing a total variation model within a wavelet domain; and solving the established total variation model based on an iterative algorithm to obtain the optimal solution. According to the invention, the total variation model is directly established in the wavelet domain and solved for image denoising for the first time, so that good combination of two types of image processing methods is achieved; the staircase effect caused by the total variation method and the Gibbs phenomenon caused by the wavelet threshold shrinkage can be well overcome; and edge feature information of the image can be effectively maintained during denoising, and the good image quality is provided for subsequent processing.

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 provides an image enhancement method for lensless digital holography microscopy imaging. The image enhancement method comprises the following steps of representing a holographic image through diffraction; removing the zero-level spot of the holographic image; restoring a phase result from a TIE method which is restored based on a gaussian process, enabling the phase result to be used as an initial phase value of the iterative algorithm, and then carrying out phase retrieval by using a hybrid input / output iterative algorithm so as to remove the conjugate image; carrying out preprocessing on the noise image by adopting a neighborhood average filtering algorithm, then performing secondary denoising on the processed image by adopting a partial differential equation of a total variation model to remove the speckle noise and finally to obtain the de-noised holographic image. The mage enhancement method and apparatus provided by the invention have a good de-noising effect, high efficiency and capability of obtaining the clear holographic image.

The invention discloses a method for unifying face illuminations on pictures with different size characteristics. Firstly, a log-domain total variation model is adopted to decompose an original face picture into small size characteristic pictures and big size characteristic pictures; then an illumination processing is carried out to the big size characteristic pictures which are greatly impacted by illumination changes and a minimal value filtering wave processing with threshold value is carried out to the small size characteristic pictures; finally, the processed pictures with different size characteristics are composed to obtain a face picture with unified illumination. The invention mainly carries out the illumination unification on the big size characteristic pictures which are greatly impacted by illumination changes to avoid the impact on face identification rate caused by changing the small size characteristics with no illumination change. Moreover, the invention does not give up the big size characteristics which are greatly impacted by illumination so as to avoid the identification information lack caused by just adopting the small size characteristics for face identification. The method of the invention can be realized easily without strict alignment to the face pictures and without any training samples, meets various practical application requirements.

The invention discloses a human face light re-adding method based on a total variation model, comprising the following steps of: (1) establishing a multi-input logarithm total variation model; (2) inputting a human face image series of the same person into the multi-input logarithm total variation model; (3) decomposing the human face image series into a reflection component and a light illumination component by the multi-input logarithm total variation model; (4) inputting a human face light illumination image in a human face database into the multi-input logarithm total variation model; (5) decomposing the human face light illumination image in the human face database into a reflection component and a light illumination component by the multi-input logarithm total variation model; (6) combining the light illumination component in the step (3) and the light illumination component in the step (5) into a new light illumination component; and (7) combining the reflection component in the step (3) and the light illumination component generated by the step (6). According to the human face light re-adding method based disclosed by the invention, a human face image under extreme change light illumination can be formed; and single-image or multi-image input is supported, and the light illumination component and the reflection component can be accurately estimated.

In order to accurately and effectively realize compressed sensing reconstruction of a natural image, the invention provides a compressed sensing reconstruction algorithm based on image non-local low rank and total variation. According to the algorithm, non-local self-similarity and local smoothness characteristics of images are considered, a traditional total variation model is improved, weights are only set for high-frequency components of the images, and an edge detection operator with a differential curvature is used for constructing a weight coefficient. Furthermore, the algorithm takes animproved total variation model and a non-local low-rank model as constraints to construct an optimization model; and a smooth non-convex function log det (.) and a soft threshold function are respectively adopted to solve the low-rank and total variation optimization problems, so that the properties of the image are well utilized, the detail information of the image is protected, and the noise resistance and the adaptability of the algorithm are improved.

The invention discloses a tone mapping method based on an edge preservation total variation model. The method includes: inputting a high dynamic image; conducting light intensity reconstruction on the input high dynamic image; obtaining a logarithm value of the reconstructed light intensity value; using an edge preservation total variation filter to filter the obtained logarithm value of the light intensity value to obtain an image base layer; enabling the logarithm value of the light intensity value to deduct the image base layer to obtain an image detailed layer; calculating a compression scale factor and image output light intensity; and enabling the compressed light intensity to divide the input image light intensity to obtain a new scale factor M. The scale factor M is acted on a red channel, a green channel and a blue channel respectively to obtain a color-compressed image. The color-compressed image is corrected through gamma to obtain a low-dynamic image. By means of the tone mapping method based on the edge preservation total variation model, the high-quality low-dynamic image can be obtained, and a phenomenon of halo can not occur.

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 single image quick defogging method based on an improved atmospheric scattering model. The method, to solve the detects of an atmospheric scattering model, is characterized by, redefining atmospheric light and transmittance in an original scattering model respectively, and on the basis of the improved model, to begin with, carrying out scene segmentation on a fogged image according to fog concentration characteristics; then, by means of a corrosion method proposed in a He method, and through an averaging strategy, estimating scene incident light L; next, by maximizing contrast information of each scene, roughly estimating the scene transmissivity T, and providing a transmissivity adaptive adjustment mechanism based on environment characteristics to realize flexible defogging; and finally, providing a guided total variation model to carry out edge optimization on the initial estimation result, and importing the optimization result to the improved scattering model to restore a clear image. The method can effectively solve the problems of illumination non-uniformity, over enhancement and oversaturation and the like, and is superior to existing mainstream defogging methods in the aspects of defogging capacity, universality and defogging speed.

The invention discloses an image rendering method based on color shift and correction. The image rendering method comprises the steps of carrying out space conversion and obtaining component information of brightness and color at any pixel point of a target gray scale image; reading out any one pixel point t in the target gray scale image; selecting a pixel point s in a source color image by using a non-local approach, with a window determined by the pixel point s having similar variance features and discrete Fourier transform features to those of a window determined by the pixel point t; carrying out initial rendering for a color value of the pixel t according to a color value of the pixel s; establishing a coupling total variation model, solving the model to obtain an optimal color value, taking the solved optimal value as a corrected color value, and keeping a brightness value unchanged; and repeating the above steps, correcting each pixel point t of the target gray scale image, and outputting the corrected image. The invention can well overcome the problem of inconsistent color space after rendering, and achieves automatic, quick and accurate rendering.

The invention provides an atmospheric scattering model based infrared image enhancing method. The method includes steps of performing inversion operation on an infrared image so as to convert the infrared image into a virtual foggy image by utilizing the similarity between an inversion infrared image and a foggy image; utilizing a quadtree decomposition technology to decompose the inversion infrared image into a series of sub blocks and performing independent defogging treatment within each sub block; performing inversion on the inversion infrared image after defogging treatment again and obtaining an enhanced infrared image; based on an orientation total variation model, performing edge repairing on the infrared image subjected to enhancement; performing luminance adjustment on the infrared image subjected to enhancement based on a Retainex model, and thus the final enhanced infrared image is obtained. By adopting the method provided by the invention, a large amount of scene details hidden in the infrared image can be restored and no negative effect is caused.

The invention discloses a quick image restoration method of a total variation model coupled with a gradient fidelity term. The quick image restoration method aims at the TV (Total Variation) model coupled with the gradient fidelity term, and utilizes a splitting algorithm to solve the model in a minimum way to finally realize the quick image restoration. The quick image restoration method comprises the following steps: transforming the model, obtaining an unconstrained transformation model, discretizing the model, and decomposing the model. The quick image restoration method provided by the invention greatly shortens consumed time and realizes the quick image restoration, speed improvement is about five times of a traditional gradient descent, and cost performance is high.

The invention discloses a tone mapping method based on an edge-preserving total variation model, which comprises the following steps of: inputting a high dynamic image; performing light intensity reconstruction on the input high dynamic image; obtaining a logarithm value of a reconstructed light intensity value; using an edge-preserving total variation filter to filter the obtained light intensity logarithm value to obtain a basic layer of the image; subtracting the basic layer of the image by the light intensity logarithm value to obtain a detail layer of the image; calculating a compression scaling factor and the output light intensity of the image; dividing the input light intensity of the image by the compressed light intensity to obtain a new scaling factor M; acting the scaling factor M respectively on each channel of a red channel, a green channel and a blue channel to obtain the image with compressed colors; and performing gamma calibration on the obtained image with the compressed colors to obtain a low dynamic image. The tone mapping method based on the edge-preserving total variation model can obtain the low dynamic image with very high quality, and does not produce halos.

The invention relates to a multi-temporal POLSAR image registration method based on SAR-SIFT features. In the multi-temporal POLSAR image registration process, a multi-dimensional polarization scattering vector is constructed by combining polarization scattering components to calculate the gradient of a pixel point. According to the method, the problem that polarization information in the image cannot be fully utilized during POLSAR image registration in the prior art is solved. The feature point extraction precision in the registration process is effectively improved, a registration algorithmhas high performance, and an efficient registration task is achieved. In a process of constructing a scale space during multi-temporal POLSAR image registration, the total variation model is adoptedto replace Gaussian convolution, so that more detail information is reserved in the constructed scale space, speckle noise in the image can be effectively filtered out, stable feature points are extracted, mismatching point pairs are eliminated, and the accuracy and robustness of the multi-temporal POLSAR image registration algorithm are improved.

The invention discloses an image denoising method based on a separable total variation model. The method includes the steps that elements on an image are projected to (p, q) space, shrinkage projection is conducted on the (p, q) space, and the denoised image is obtained after iteration. The method particularly includes the steps of building the total variation model with separable elements, solving the separable total variation model, and obtaining the denoised image. An energy function of the image is determined based on the variation method thought, the image is in a smooth state by balancing the energy function, the Pseudo-Gibbs phenomenon is eliminated, the defects that in the prior art, an algorithm is complex and large in operand are overcome by building the discrete total variation model, the operation speed is increased, and the convergence speed and the signal to noise ratio are increased. The calculation speed and accuracy are adjusted by changing the number k of iteration times, the method is flexible to use and has the high peak value signal to noise ratio, random noise in the image can be well removed, details and textures are effectively reserved, and the method is especially suitable for processing random noise.

The invention discloses a different resolution infrared and visible light image fusion method and system based on total variation, which comprises establishing a total variation model of image fusion,respectively constraining the fusion image and the infrared image with a data fidelity term and a regularization term, and fusing the relationship between the image and the visible light image, according to the source image, solving the optimal solution of the total variational model and obtaining the fusion result. The invention builds a model according to the situation that the original infrared image and the visible image have different resolutions, and the obtained fusion image has high resolution and clear texture details, while retaining the characteristic that the pixel intensity distribution of the infrared image protrudes out of the thermal target, and the fusion result is convenient for human eyes to discriminate the thermal target and has good visual effect.

The invention provides a total variation-based mixed weighted Wiener filter image denoising method. The method comprises the following steps of: 1, acquiring a to-be-processed original image by utilizing a video camera which comprises a charge-coupled device; 2, carrying out grayscale image conversion on the original image acquired by the video camera; 3, carrying out noise addition on the converted grayscale image; and 4, denoising the image after noise addition by adoption of a hybrid model which is formed by a Wiener filter model and a total variation model. According to the method, the integrity of internal texture information of images can be ensured, and the loss of border angular point feature information of the images can be decreased.

The invention discloses a fabric fuzzy ball grade evaluation method based on a relative total variation model and MSER, and the method comprises the steps: firstly, collecting an original image of a fabric, carrying out the preprocessing, and enhancing the fuzzy ball edge information while eliminating the phenomenon of uneven illumination of the image; then, adopting a relative total variation model to inhibit the texture information of the fabric for the processed image, and effectively eliminating the texture structure of the fabric; then, segmenting a fuzzy ball area in the fabric by adopting an MSER algorithm, and eliminating isolated pixel points in the fuzzy ball image area; and finally, calculating a hair ball proportion, and establishing an evaluation grade of the hair ball to realize objective evaluation of the hair ball grade. The method solves the problem that the grade of the fuzzy ball cannot be accurately evaluated because the fuzzy ball cannot be accurately segmented dueto the influence of uneven illumination and texture information during the fuzzy ball segmentation in the existing method, can replace manpower to effectively complete the evaluation of the grade ofthe fuzzy ball of the fabric, and can meet the actual industrial requirements.

The embodiment of the invention provides a multi-target image restoration method, which comprises the following steps of: constructing a parameterized wavelet basis function; Introducing the wavelet basis function into a strip wave function to form an interpolation multi-scale strip wave function; Forming a parameterized interpolation strip wave learning dictionary for different texture structuresof the image; Replacing the BV (Omega) space in the total variation model with the (shown in the description) of the BV space, introducing a parameterized interpolation strip wave learning dictionaryto form an image strip wave sparse representation variation model; Based on the image stripe wave sparse representation variation model, performing homotopy analysis on the multi-target image, and accurately segmenting the target image; And based on a homotopy segmentation result of the image, establishing a target-controllable image segmentation and restoration coupling variation model, and restoring the image. According to the method, an image target controllable segmentation and restoration coupling model is constructed based on interpolation Bendels (strip waves) sparse representation, and accurate restoration of a damaged target object in an image is realized.

The invention provides a method for establishing a calligraphy and painting paper output quality prediction model. The method comprises the following steps: measuring calligraphy and painting paper surface performance parameters; measuring surface curtain parameters of the calligraphy and painting paper; establishing a total variation model in an MATLAB environment; determining values in the totalvariation model parameters; measuring a chromatic value of the calligraphy and painting paper subjected to curtain removal processing, and calculating a chromatic aberration value between the chromatic value and a standard set value; establishing a GRNN prediction model; constructing a GRNN prediction model to train a data sample; obtaining an optimal smoothing factor; verifying prediction accuracy. According to the method, the total variation model can be used for removing the paper curtain lines of calligraphy and painting, a color difference evaluation mode more conforming to visual characteristics is obtained, a printing output quality prediction model is constructed, and the prediction result is accurate and reliable.