57results about How to "Guaranteed edge" patented technology

The invention discloses a region-level and pixel-level fusion saliency detection method based on convolutional neural networks (CNN). The research object of the method is a static image of which the content can be arbitrary, and the research goal of the method is to find a target striking the eyes of a person from images and assign different saliency values to the target. An adaptive region generation technology is mainly proposed, and two CNN structures are designed and used for pixel-level saliency prediction and saliency fusion respectively. The two CNN models are used for training a network model with the image as input and the real result of the image as a supervisory signal and finally outputting a saliency map of the same size as the input image. By means of the method, region-level saliency estimation and pixel-level saliency prediction can be effectively carried out to obtain two saliency maps, and finally the two saliency maps and the original image are fused through the CNNs for saliency fusion to obtain the final saliency map.

The method comprises: using an extreme-value method to detect the image pixels polluted by the salt-and-pepper noise; using a self-adaptive filter method to correct the grey value of noise pixels to get an image removing the salt-and-pepper noise; making stationary wavelet de-composition to get relevant low-frequency component and high-frequency component; keeping an unchanged state for the low-frequency component, and using the coefficient correlation method for the high-frequency component to mark its pixels as noise or edge; if a pixels is marked as edge, then its value is unchanged; otherwise, using self-adaptive adjacent domain method to contract the wavelet coefficient; making stationary wavelet inverse transformation for the processed wavelet coefficient to get a noise-removed image.

The invention discloses a moving object extraction method based on superpixel division and an optical flow method, and mainly solves the problems of more noises, high-frequency information loss, inaccurate boundary and the like of the existing moving object extraction method. The implementation steps of the method are as follows: (1), inputting an image, and pre-dividing the image into a superpixel set S to obtain a mark sheet I 2; (2), taking images of two adjacent frames in a video sequence and determining a rough position of a moving object by a Horn-Schunck optical flow method; (3), using the optical flow method to obtain the speed u in the horizontal direction and the speed v in the vertical direction, wherein V is speed amplitude of the optical flow method; (4) performing median filtering, Gauss filtering, binarization operation and morphology opening and closing operation on the optical flow result V to obtain V4; (5) using a superpixel division result to further correct the optical flow result, and extracting to obtain the accurate moving object. Superpixels belonging to a moving area are extracted accurately. Simulation experiments show that compared with the prior art, the moving object extraction method has the advantages of simple operation, small noise, clear boundary and the like, and can be used for extracting the moving object in the video sequence.

The invention discloses a semi-automatic image cutting method based on nuclear transfer, which mainly solves the problems that the existing interactive image cutting method can not keep data consistency, the cutting result is not stable, and edges are not clear. The method is realized through the following steps: inputting an image to be cut, obtaining a superpixel set of the input image through a mean shift method, calculating a superpixel color column diagram feature set, building a similarity matrix W through a Bhattacharyya coefficient formula, performing interactive operation on the precut image by a user to obtain a seed superpixel set, building a must-link constrain set M and a cannot-link constrain set C, transferring constrain information M and C to the whole nuclear space through a nuclear transfer method, so as to obtain a nuclear matrix R, clustering the nuclear matrix R through a k mean clustering method, so as to obtain a clustering mark vector, and outputting and displaying a cutting result. The method has the advantages of good data consistency, stable cutting result, simplicity of operation and clear edges, and can be applied to image retrieval, the technology of converting 2D to 3D, target detection and tracking, and medical image analysis.

The invention discloses a remote sensing image change detection method based on controllable kernel regression and superpixel segmentation. The problems that only grey information of an image is considered when a difference chart is structured, other feature information is underused, k-means clustering is directly carried out on the difference chart, and therefore the situation that a weak declension area cannot be detected is easily caused are mainly solved. The method comprises the steps of adopting the controllable kernel regression on two input time phase images to respectively extract structural feature matrixes, combining feature matrixes of neighbourhoods with the structural feature matrixes respectively, obtaining a local structural feature matrix, decomposing the local structural feature matrix by using a non-negative matrix factorization algorithm, carrying out a difference chart structure on an obtained coefficient matrix, finally segmenting the difference chart to obtain an over-segmentation image by using a superpixel segmentation method, carrying out the K-means clustering on the over-segmentation image, and obtaining a change detection result. The remote sensing image change detection method based on the controllable kernel regression and the superpixel segmentation can keep marginal information of images, is good in noise proof performance, improves change detection precision, and can be applied to fields of disaster situation monitoring, land utilization, agricultural investigation and the like.

The invention discloses a speckle noise suppression method for polarized SAR (Synthetic Aperture Radar) data based on a non-local mean value fused with PCA (Polar Cap Absorption), which mainly solves the problems that speckle noises in a homogeneous region can not be well filtered and the edge detailed information can not be effectively maintained in the traditional polarized SAR filtering method. The implementation process of the method comprises the following steps of: (1) inputting a coherence matrix T of the polarized SAR data; (2) maintaining a bright target for the coherence matrix T; (3) solving a characteristic vector of span data by utilizing a PCA method; (4) filtering the non-local mean value for elements of the coherence matrix T, wherein a filtering weight value is obtained by calculating the characteristic vector of the span data; and (5) generating a pcolor by utilizing the filtered coherence matrix T through a Pauli vector method. Compared with the prior art, the speckle noise suppression method remarkably improves the capability of speckle noise suppression of the polarized SAR data, can effectively smoothen the homogeneous region and maintain the edge detailed information, and can be used for the pretreatment process of the polarized SAR data.

The invention discloses an SAR image speckle suppression method based on second generation curvilinear wave transformation, which mainly overcomes the defect of scratch effect and point target loss brought by a curvilinear wave to the SAR image speckle suppression. The SAR image speckle suppression method comprises the following steps: performing the second generation curvilinear wave transformation to a selected test image and partitioning the selected test image into 5 layers of subbands; keeping coefficients of the first layer unchangeable and zero-setting coefficients of the fifth layer; respectively evaluating parameter vectors of hybrid Gaussian models from the second layer to the fourth layer by an EM method; marking the coefficients from the second layer to the fourth layer; reconstructing the image, detecting the edge of the reconstructed image, and performing the average filtering to the uniform area of the reconstructed image to obtain the filtered image; and performing the nonlinear anisotropy dispersion iteration to a difference image obtained by the original image and the filtered image to obtain a speckle suppressed image. The invention has the advantages of keeping clean edge of the image, removing the scratch effect and remaining the point target characteristic information of the image, and can be used for preprocessing scene analysis and image understanding in the SAR image.

The present invention discloss a positron tomographic scanning image reconstruction method of multiphase level set. Said method mainly includes the following steps: obtaining projection data, defining number of level set functions, using K mean cluster method to obtain initial concentration value of image, calculating to obtain object function for reconstructing image, solving partial derivatives, obtaining current level set evolve correction, updating reconstructed image, obtaining estimated image and concentration estimated value, and converging reconstructed image. Said invention has the advantages of eliminating noise, retaining edge and eliminating edge pseudo-image, etc.

The invention discloses an intelligent positioning method for small ceramic files based on multi-feature fusion. The intelligent positioning method comprises the steps of S1, image acquisition; S2, image preprocessing; S3, feature extraction, S4, image target determination; and S5, ceramic file positioning. In the step of image acquisition, an image is acquired through a high-speed linear array CCD camera. In the step of image preprocessing, gray processing is firstly performed on the image, then filtering and denoising processing is carried out, the filtered image is segmented by adopting an edge detection algorithm, the profile of the small ceramic tile is extracted, and then a profile diagram of the ceramic tile is acquired. In the step of feature extraction, corresponding primitive features are generated based on geometric features of the ceramic tile profile in the profile diagram in the step S2; and a logical AND operation is performed by using a filtered image and the profile mask diagram in the step S2, transferring an operation result diagram into an HSV color space, and color features are generated; and combination features are generated according to the primitive features and the color features. The intelligent positioning method realizes high-precision positioning for the small ceramic tiles at a large FOV (Field of View), thereby providing accurate positioning data for visual inspection and robot grasping. In addition, multiple segmentation units can operate concurrently in a multi-threaded system, and the real-time performance of the system is improved.

The invention discloses an NSST (NonsubsampledShearlet Transform) domain MRF (Markov Random Field) and adaptive threshold fused remote sensing image change detection method, which solves the problem that edge information of a change region cannot be kept while a miscellaneous point is removed in the conventional change detection method. An implementation process for the method comprises the following steps of: inputting two remote sensing images of different time phases and constructing a difference image by using a difference value method; performing nonsubsampledShearlet decomposition on the difference image; combining a directional sub-band of each layer into a high-frequency sub-band; performing adaptive threshold classification on a high-frequency sub-band and a low-frequency sub-band of each layer to obtain a high-frequency adaptive threshold classification chart and a low-frequency adaptive threshold classification chart at each layer; performing MRF classification on the low-frequency sub-band of each layer respectively to obtain an MRF classification chart for each layer; and fusing the classification results to obtain a change detection result. The method has the advantages of high anti-noise property, high edge information retention capacity, less false drop of a detection result and high accuracy. The method is used for the fields such as urban area change monitoring, forestry and vegetation change monitoring and military target monitoring.

A detection device of an appearance of a mobile phone lens is used to accurately detect and distinguish unrepairable defects caused by material flakes and repairable punctuate defects caused by others such as dust, smudginess and the like in a transparent window area of a mobile phone lens. The detection device comprises a line-scan camera, a highlight strip light source, a strip high-transmittance astigmatic lens, a frosted astigmatic lens, and a strip convex lens, wherein the line-scan camera is disposed at the top side of the mobile phone lens, and the light source is disposed at the bottom side so as to allow the backlight and the line-scan camera to be on the same axis. Although the camera and the light source are on the same axis, the camera is guaranteed to aim at the edge of the light instead of the center of the light. Therefore, it is guaranteed that not only the gray scale contrast between a defect and the edge is sufficient, but also the light intensity of a week defect is sufficient to be identified by the line-scan camera.

The invention discloses a backed fabric image division method based on texture suppressing smoothing filtering and watershed algorithm. The backed fabric image division method based on the texture suppressing smoothing filtering and watershed algorithm includes that using a hybrid median filtering algorithm to filter and scan noise of features of a fabric image with backed weave based on color mode conversion; filtering through a texture suppressing smoothing filtering algorithm to remove the backed weave shadows and yarn textures with the same color from the fabric image and keep yarn color features; extracting the chromatic gradient of the fabric image, and performing image division through the watershed algorithm to obtain a regional mark image; combining the segmented regions with similar colors to obtain a color separation index image of the fabric image. By means of the backed fabric image division method based on the texture suppressing smoothing filtering and watershed algorithm, edges of different colors of yarns are effectively kept based on smoothing the yarn textures with the same color and backed weave edge shadows, and the problems that the edge details between regions are weakened after performing Gaussian filter, and the yarn textures are reserved as edges after performing bilateral filter are avoided.

The invention relates to a hybrid image filtering method based on target scale, which is mainly used for self-adaptive median filtering and anisotropic diffusion. The method is characterized in that: first anisotropic diffusion is carried out to a processed image, thereby inhibiting most noise other than pulse noise; then the target scale of all pixels of the diffused image are calculated; and finally the calculated target scale carries out self-adaptive median filtering to the diffused image so as to remove the rest pulse noise after diffusion and obtain the final result. The method can remove the noise of images with various gray levels, and simultaneously keeps the details of the images while removing the noise.

The invention discloses an X-ray image uneven display processing method based on distribution and reconstitution of measure spacing, which comprises the following steps: doing normalizing and logarithmic transformation to the inputted image, choosing linear or nonlinear wave filter method to do image measure spacing distribution, weighting the image information on different measures and reconstituting the weighted signals, doing normalizing and antilogarithm processing to the constructed signals to obtain the uneven image. It can adopt blank field smooth filter wave operator during the decomposable process.

The invention discloses a method for SAR image speckle suppression based on area division and non-local total variation, which mainly resolves the problem of prior art that inhomogeneous area is involved in the homogeneous area when area is being divided, and introduces the thought of non-local total variation therein. The achieving steps are (1) acquiring the area mark from the sketch drawing of the Primal Sketch sparse representation model by utilizing geometric structure block, and dividing the SAR image into homogeneous area and inhomogeneous area, (2)processing the inhomogeneous area by utilizing method of non-local total variation speckle suppression in combination with singularities, (3) processing the homogeneous area by utilizing self-adaptive window Lee filtering method, (4) combining the speckle suppressed inhomogeneous area with the speckle suppressed homogeneous area to obtain the result of SAR image speckle suppression. The invention accomplishes the balance between singularity keeping and speckle suppression in SAR image speckle noise suppression and enhances the effect of SAR image speckle suppression.

The invention provides an edge protection denoising method, which comprises the following steps: step 1, according to the gray-scale domain correlation between the image center pixel and the neighboring pixel, pixel classification is carried out on the input image, and pixels are divided into three types of pixels of the image homogeneous region or edge region, single-point noise pixel and noise block region pixel; Step 2, for three types of pixels with different attributes, the spatial domain and the gray domain under the multi-structure template are combined with adaptive filtering.

The invention discloses an overlapped protein point separation method and device based on concave point matching, and the method comprises the steps: solving all angular points on an overlapped protein point outline boundary in a gel image through employing a Harris angular point detection method; constructing an approximate polygon of the overlapped protein point outline boundary in the gel image based on a polygonal approximation algorithm of angular points through all solved angular points, thereby judging the concave and convex points of the top points of the approximate polygon and obtaining a concave angular point; selecting a matching rule through a preset concave point, selecting concave points from the concave angular points, constructing a separation line, and completing the segmentation of overlapped protein points in the gel image. The method can improve the segmentation effect of the overlapped protein points in a bidirectional gel image.

The invention discloses a singular value decomposition non-local mean-based polarized synthetic aperture radar (SAR) data speckle suppression method for mainly overcoming the defects that the conventional polarized SAR filter technology cannot well filter speckle noise of homogeneous areas and cannot effectively keep marginal detail information. The method comprises the following processes of: (1) inputting a covariance matrix C of polarized SAR data; (2) performing bright target retention on the covariance matrix C; (3) acquiring a logarithmic characteristic matrix from a span matrix, and performing singular value decomposition; (4) performing singular value decomposition non-local mean filtration on elements of the covariance matrix C one by one; and (5) generating a pseudo color graph through the filtered covariance matrix C by a Sinclair vector method to display and observe the filtration effect. Compared with the prior art, the method has the advantages of remarkably improving the speckle noise suppression capacity of the polarized SAR data and effectively smoothing the homogeneous areas and keeping the marginal detail information, and can be used for the pre-processing process of the polarized SAR data.

The invention discloses a color map guide-based depth map restoration and view synthesis optimization method, which comprises the following steps: firstly, detecting inconsistent regions, detecting the edge of an input depth map, performing swelling processing on the edge, and marking the swelled edge as a potential inconsistent region; secondly, constructing a weight based on an iterative reweighting-based least squares algorithm, and after weight construction is completed, performing integral solving, and updating a depth map; judging whether the iteration is performed for set times or not according to a result; and if YES, outputting the depth map, and ending the calculation, or redetecting the inconsistent regions. The color map guide-based depth map restoration and view synthesis optimization method disclosed by the invention can suppress strong noise and restore inconsistent regions between the depth map and a color map to improve the consistency between the depth map and the color map and restore a correct boundary of the depth map, and has important guiding significance for improving the quality of a synthesized view. Meanwhile, the denoising and edge retaining capacity forconsistent regions is high; and an adopted matured iterative weighted least squares model is high in adaptability to parameters, and the robustness of the model is improved.

The invention relates to a bilateral regression filter method for grayscale and colored images. As an effective information carrier, images can be used as an important manner for obtaining and exchanging effective information by users; but the images systematically import certain noises due to various interference factors in the forming, transmission and reception processes, some detail features are hidden in the noises, and great difficulties are brought to the image processing processes such as image observation as well as feature information extraction and analysis, so that adopting a proper method to remove pollution noises in the images is a very important pre-processing step. The bilateral regression filter method provided by the invention can be used for the filtering of grayscale and colored images.

The invention belongs to the technical field of optical detection and image processing, and aims to realize contrast enhancement of an electronic speckle interference image with low illumination, andstripes after image binaryzation can be kept complete as much as possible, so that extraction is more accurate when skeleton lines are extracted subsequently. The technical scheme adopted by the invention is as follows: the low-illumination electronic speckle interference fringe image enhancement method comprises the following steps of selecting a multi-scale Retinex algorithm (multi-scale retinex, MSR) based on center encircling by utilizing a Retinex theory. The low-quality low-illumination electronic speckle interference fringe image enhancement method is mainly applied to optical detection and image processing occasions, and can be used for processing low-quality low-illumination electronic speckle interference fringe images obtained through experiments, adjusting scale parameters andobtaining an optimal result.

The invention discloses a fuzzy clustering color image segmentation method based on multichannel weighted guided filtering, and the method comprises the steps: firstly effectively removing different types of noises with different intensities through multivariate morphological reconstruction for a color image with noises, thereby reducing the impact on image clustering from the noises; secondly, when multi-channel guided filtering is carried out on the membership degree of the color image with the noise, guiding and filtering the membership degree by using each channel of the original color image; the filtering result of each channel is weighted to obtain the final filtering output image, so that the mutual influence between different channels can be avoided, the edge of the final filteringresult can be kept more effectively, and the accuracy of color noise image segmentation is improved. In addition, due to the fact that filtering result weighting calculation of the three channels islinear operation, the fuzzy clustering color image segmentation method based on multi-channel weighting guided filtering has low time complexity.

The invention discloses an industrial character detection process based on imprecise segmentation, and belongs to the technical field of general machine vision. The process comprises an offline stageand an online detection stage. In the offline stage, a character reference model is obtained and is used in the online detection stage, in the online detection stage, an improved maximum stable extremum region algorithm is adopted in image segmentation so that the robustness of the algorithm is improved; in the character segmentation process, non-adhesion characters and adhesion equal-width characters are subjected to character segmentation in a projection segmentation mode; and if the character is the adhered non-equal-width character, character segmentation is performed by adopting a projection extreme value mode. The positions of the characters can be well determined in a mode of recognizing and segmenting at the same time so that the character segmentation difficulty in the online stage is reduced.

The invention discloses an image-forming method of weighting curvature flow filtering during the least square iterating for constructing image of positive electron emitting tomography, firstly getting observing data, determining the scale of original image, comparing each component in the observing data with 1, dividing each component in the error variance and filtering the intermediate image with curvature flow and getting the filtered image as original image, repeating this course until the image reconstructed image contracted and getting the finished image; the invention adding curvature flow filtering in the least square iterating depending on the current feature of PET image-forming, through which to inhibit pattern noise and improve the image quality; the invention is characterized by the simple process, fast treating speed and good effect for inhibiting the background disturb.

The invention discloses a GaN-based completely-vertical Schottky varactor based on a metal eave structure and a preparation method thereof. The varactor structurally comprises a silicon substrate layer; a substrate protection layer, a bonding layer, an ohmic metal layer, an n<+>-GaN layer, an n<->-GaN layer and a Schottky contact metal layer are sequentially arranged on the silicon substrate layerfrom bottom to top; an anode pad is arranged on the Schottky contact metal layer; and a cathode pad in contact with the bonding layer is further arranged on the bonding layer. The reverse bias characteristic is effectively improved, which is reflected by the increase of reverse breakdown voltage and the reduction of leakage current during reverse bias, so that the frequency multiplication efficiency is effectively improved.

The invention discloses a heavy fabric image segmentation method based on texture suppression smoothing filtering and watershed algorithm. This method is based on the characteristics of fabric images with heavy tissue and uses a hybrid median filtering algorithm to filter out scans based on color mode conversion. Noise; then filter through the texture suppression smoothing filter algorithm to remove heavy tissue shadows and same-color yarn textures in the fabric image, retaining the yarn color characteristics; then extract the color difference gradient of the fabric image, perform image segmentation through the watershed algorithm, and obtain the region Mark the image; finally, merge the segmented areas with similar colors to obtain the color separation index image of the fabric image; while smoothing the yarn texture of the same color and reorganizing the edge shadows, it effectively maintains the edges of yarns of different colors, avoiding Problems include the weakening of edge details between regions after Gaussian filtering and the retention of yarn texture as edges by bilateral filtering.

The invention belongs to the technical field of image analysis, and discloses an image saliency detection method based on FLIC superpixel segmentation. The input image uses a filtering method based ona total variation model to remove texture and obtain an image containing color features. Gabor filter matrix extraction is used to filter the image, and the image containing image texture informationis obtained. The image containing color features is segmented by FLIC and the contrast value is calculated. FLIC is used to segment the image which contains texture features to calculate the contrastvalue. By using linear fusion technique, the two contrast ratios are fused linearly through weights to obtain new contrast ratios, and the salient images based on color features and texture featuresare finally obtained. The detection effect of the invention for the high-texture image is obviously improved compared with the current representative method; the method can be used in computer visiontasks.

The invention discloses a ROAD-detection-based index similarity sorting filtering method which comprises the following steps: calculating statistical volume of the sum of first k pixels having a minimal difference value with a central pixel; comparing a ROAD value corresponding to each pixel with a threshold value T; if the ROAD value of the pixel is greater than the T, confirming the pixel as pulse noise; performing pulse noise removing treatment on the basis of index similarity sorting; and defining a similarity function and confirming the pixel with the maximum similarity with surrounding pixels as an output of a window. Compared with the prior art, the method has the advantages that the useful edge information and pulse noise can be efficiently identified, and meanwhile, the pulse noise having a small difference with the surrounding pixels can be detected; when the similarity of a certain pixel is evaluated, a quantity value difference and a direction difference between the pixels are simultaneously used as an evidence for evaluating the similarity of the pixels; the effect of noise removing capacity is excellent; the color information of an image before and after being filtered is maintained; and the distortion and the loss of details and colors are avoided.

The invention discloses an X-ray image uneven display processing method based on distribution and reconstitution of measure spacing, which comprises the following steps: doing normalizing and logarithmic transformation to the inputted image, choosing linear or nonlinear wave filter method to do image measure spacing distribution, weighting the image information on different measures and reconstituting the weighted signals, doing normalizing and antilogarithm processing to the constructed signals to obtain the uneven image. It can adopt blank field smooth filter wave operator during the decomposable process.

The invention discloses a polarimetric SAR (synthetic aperture radar) data speckle suppression method based on adaptive shape non-local means, mainly solving the problems that speckle noises in a homogeneous region can not be well filtered, margin detail information can not be effectively maintained and polarimetric information can not be maintained completely. The realization process of the method comprises the following steps: (1) inputting a covariance matrix C of polarimetric SAR data; (2) maintaining bright targets of the covariance matrix C; (3) carrying out shape adaptive non-local mean filtering on non-bright target elements of the covariance matrix C; (4) carrying out weighted mean on the estimated results of different shape blocks; and (5) generating the despeckled covariance matrix C into a pcolor by adopting a Sinclair vector method so as to display the filtering observation effect. Compared with the prior art, the polarimetric SAR data speckle suppression method disclosedby the invention has the advantages that polarimetric SAR data speckle suppression capability is obviously improved, the homogenous region can be effectively smoothed, the margin detail information can be effectively maintained, and the method can be applicable to preprocessing of the polarimetric SAR data.