51results about How to "Preserve edge detail" patented technology

The invention discloses an image saliency detection method based on region description and priori knowledge. The method comprises the following steps that: (1) an image to be detected is subjected to pre-segmentation, superpixels are generated, and a pre-segmentation image is obtained; (2) a fusion feature covariance matrix of the superpixels is generated; (3) feature different region descriptors and space distribution region descriptors of each superpixel are calculated; (4) the initial saliency value of each pixel point of the image to be detected is calculated; (5) a priori saliency region and a background region of the image are obtained; (6) the saliency weight of each pixel point of the image to be detected is calculated; and (7) the final saliency value of each pixel point is calculated. The image saliency detection method has the advantages that the saliency region can be uniformly highlighted in an obtained final saliency map; the background noise interference is inhibited; a good saliency detection effect can be achieved in ordinary images, and the processing on the saliency detection of complicated images can also be realized; and the processing on subsequent image key region extraction and the like can also be favorably carried out.

A system and method descreen halftone images into a continuous tone image while preserving edge detail and reducing initial image blur. The descreening can be applied to monochrome or color images. Descreening is achieved by first using a low pass filter to form a blurred image of the original, which is used to guide future image filtering, but which further filtering is applied to the original image. The intelligent filtering is provided by a Sigma filter, which can be performed in a single iteration and sized and shaped according to values of the blurred image. The system and method can take a block-oriented approach that performs analysis for an entire block of pixels rather than on individual pixels. This further improves the implementation speed of the process.

The invention provides an object significance detecting method based on color contrast and color distribution. The steps of the object significance detecting method based on the color contrast and the color distribution include that S1: an input image is divided into small size super-pixels, average color and position in the super-pixels are calculated; S2: the center-periphery color contrast of each super-pixel is calculated, the color contrast value is multiplied by a priori distribution, and at last color contrast significance diagram is obtained by using a significance smooth operation; S3: color distribution variance of each super-pixel is calculated, and thereby a color distribution significance diagram is obtained; S4: the color distribution significance diagrams obtained by the S2 and the S3 are multiplied and refined by using MeanShift division, edges of an object are enabled to be more fine, and the final significance diagram is output. According to the object significance detecting method based on the color contrast and the color distribution, the significance diagram obtained can evenly highlight the significant object in the significance diagram, the edge details of the object are well retained, the background interference is restrained, and the following up processes such as the target object division are benefited.

A method for restoring hyperspectral remote sensing image based on non-convex and low-rank sparse constraint belongs to the field of hyperspectral remote sensing image processing in remote sensing image processing. In order to solve the problem that the existing hyperspectral remote sensing image restoration technology can not effectively remove noise and improve the image restoration quality, themethod comprises the following steps: inputting a hyperspectral remote sensing image; initializing a weight coefficient matrix, iterative times and a convergence threshold, initializing sub-image size and scanning step, partitioning sub-blocks; establishing an image restoration model; the auxiliary variable and the coefficient of the regular term being introduced, and the maximum-minimum algorithm being used to solve the problem iteratively; judging whether the restoration result satisfies the convergence condition; obtaining a hyperspectral restored image that meets the requirements by iterative times, otherwise returning to corresponding steps to continue the iterative operation; calculating a weight coefficient matrix and assigning appropriate weights to each sub-block; hyperspectral remote sensing images being restored to obtain the final restored hyperspectral remote sensing images. The effect of denoising is obvious and the image details are preserved.

The invention discloses a shearlet transform and fast bilateral filter image denoising method, which comprises the steps of: 1) acquiring an envelope signal of a noise image by using a noise imaging system, and establishing a medical ultrasonic image model; 2) carrying out multiscale and multidirectional decomposition on the medical ultrasonic image model after logarithmic transformation obtained in the step 1) by utilizing a pyramid filter bank; 3) performing threshold method contraction processing on a two-dimensional discrete shearlet transform coefficient of a high-frequency part in each subband image obtained in the step 2); 4) using a fast bilateral filter for filtering shearlet coefficients of low-frequency parts in the step 2); 5) and conducting shearlet inverse transform processing on all the coefficients processed in the step 3) and the step 4), so as to obtain a denoised medical ultrasound images. The introduction of the fast bilateral filter in the shearlet transform and fast bilateral filter image denoising method can effectively improve the denoising performance, and greatly increase the processing efficiency.

The invention relates to the technical field of image processing and provides an image fusion method and system based on bilateral filter and weight reconstruction. The method comprises the steps that an image is divided into a base layer, a middle layer and a detail layer through Gaussian filter and bilateral filter; bilateral filter and weight processing is conducted on the base layer and the detail layer to fuse the base layer and the detail layer, and the base layer and the detail layer which are fused are superposed to obtain a final fused image. By the adoption of the technical means, the fused image can retain the important information of an original image and can be more natural, edge details of the image are well retained, and distortions of the image are also decreased.

The invention discloses a method for dynamically compensating a blind pixel. The method comprises the following steps of: arranging pixel gray values of a 3*3 window matrix dot picture element from small to large according to the remarkable abnormity of the blind pixel in a local window; if the absolute value of the difference value between a medium value and other picture element gray value is greater than a preset threshold value, determining the picture element as the blind pixel; detecting all picture elements of an M*N array by using a medium gray value instead of a blind pixel gray value; and compensating all blind pixels. Therefore, a high-quality image is obtained in a display device.

The invention discloses a Bayesian wide-residual neural network-based injurious insect image recognition method. The method comprises the following steps of: 1, preprocessing an injurious insect imagerecognition data set and carrying out injurious insect edge detection on a greyscale map by using a Rich-Edge detection algorithm; 2, constructing a Bayesian wide-residual neural network (NWResNet);3, inputting an injurious insect edge image obtained in the step 1 into the NWResNet constructed in the step 2, and obtaining a total error function of the NWResNet by utilizing an injurious insect edge image training set obtained in the step 1; 4, training the network through a Block-cg algorithm by utilizing the error function obtained in the step 3; 5, updating a hyper-parameter according to the network optimized in the step 4; and 6, repeating the steps 4 and 5 to obtain a final network, and inputting an injurious insect image verification set into the network to ensure that the classification correctness is higher.

The invention relates to a mathematical morphology-based image edge detection method. The conventional method has defects in aspects of inhibition of noise and fake edges. The method has the following steps of: performing morphological processing of an image by using structuring elements of various scales to obtain an edge information image; performing weighted coalition of the edge information image to obtain a total edge information image; and finally performing thinning and thresholding segmentation of the total edge information image to obtain a final binary edge image. In the method, the structuring elements of various scales and the improved edge detection operators are adopted, which not only effectively filters noise and inhibits fake edges, but also keeps edge details to the maximum extent, so that the good edge effect is achieved.

The invention discloses a translation invariance shearler transformation medical image denoising method. The method comprises the following steps that 1, a noise imaging system is used for collecting envelope signals of a noise image, and a medical ultrasound image model is built; 2, the medical ultrasound image model obtained after logarithm transformation is subjected to a multi-scale multi-direction decomposition through a pyramid filter bank; 3, the two-dimensional discrete shearlet transformation coefficient of each subband image medium-high frequency part obtained in the step 2 is subjected to threshold value method self-adaptive shrinkage; 4, a guider filter is used for filtering the shearlet coefficient of the low-frequency part in the step 2; 5, all the coefficients processed in the steps 3 and 4 are subjected to shearlet inverse transformation, and a denoised medical ultrasound image is obtained. The guide filter is used for filtering the low-frequency part, and the operation time problem of a trilateral filter is solved.

The invention relates to a feature point location and edge detection-based image fusion test method. The method comprises the following steps of: (1) building a test device, wherein the test device comprises an infrared camera and an ultraviolet camera, and the infrared camera is parallel to the ultraviolet camera in the aspect of optic axis; (2) putting fluorescent light tubes at four corners of a field of view of the ultraviolet camera to carry out feature point location so as to obtain image fusion parameters; (3) shooting an infrared image and an ultraviolet image and respectively preprocessing the infrared image and the ultraviolet image; and (4) fusing the infrared image and the ultraviolet image by utilizing the fusion parameters obtained in the step (2). Compared with the prior art, the method provided by the invention is good in fusion effect.

The invention discloses a medical ultrasonic image de-noising method based on a normal inverse Gaussian model. The method comprises the following steps of 1), collecting an envelope signal of a noise image by means of a noise imaging system, and establishing a medical ultrasonic image model; 2), performing multidirectional decomposition on the medical ultrasonic image model after logarithmic transformation by means of a pyramid filter set; 3), performing threshold method adaptive contraction processing on a two-dimensional discrete shearlet transformation coefficient of a high-frequency part in each sub-band image; 4), performing filtering on the shearlet coefficient of the low-frequency part in the step 2) by means of a three-side filter; and 5), performing shearlet inverse transformation processing on all coefficients after processing of the step 3) and the step 4), thereby obtaining the medical ultrasonic image after de-noising. The medical ultrasonic image de-noising method can well restrain spot noise and furthermore can facilitate a doctor in performing disease condition analysis.

The invention discloses a super-resolution reconstruction method of multi-frame images. A steering kernel regression method is introduced separately to a clustering phase of an adaptive sparse learning model and a regularized reconstruction process; the two methods are enabled to complement each other's advantages and disadvantages, forming a reconstruction algorithm of better performance. Motionregistration is not required, and therefore, reconstruction errors due to registration errors are avoided and the quality of a reconstructed image can be significantly improved; the method is applicable to places including any motion mode. Compared with existing learning-based multi-frame super-resolution reconstruction methods, the simple efficient integrated reconstruction model that integratessteering kernel regression and sparse learning is provided which can make use of both global structural self-similar priori constraints and sparse constraints to carry out regressive estimation, so that edge details of an image can be better maintained, and image high-frequency distortion can be relieved; the method is not limited to motion amplitude or application scenes and accordingly is adaptive to complex environments.

The invention discloses a remote sensing image classification method and system based on self-adaptive spatial information. The method comprises the steps of obtaining a remote sensing image; Performing initial classification on the remote sensing image by adopting a fuzzy C-means algorithm based on a Markov random field to obtain an initial fuzzy membership matrix; Calculating spatial attractionbetween a current central pixel and each neighborhood pixel in the remote sensing image under the current iteration frequency b by utilizing a spatial gravitation model; Performing edge detection on the remote sensing image by adopting a Sobel operator to obtain spatial structure characteristics; Calculating an edge coefficient of the current central pixel by adopting a gradient reciprocal smoothing method according to the spatial structure characteristics; Constructing a Markov random field with adaptive weight according to the space attraction and the edge coefficient; And combining the Markov random field with adaptive weight with a fuzzy C-means algorithm to determine a classification result of the remote sensing image. According to the method, the problem of boundary pixel and spatialinformation weight coefficient estimation can be effectively solved, and the classification precision is improved.

The invention discloses a bidirectional iteration bilateral filtering method for asphalt images obtained through an UAV-borne infrared imaging device. The method includes a first step of taking original infrared asphalt images by utilizing an unmanned aerial vehicle (UAV), a second step of switching original infrared asphalt images in RGB color space to a Lab color space, a third step of expanding the image boundaries by using the boundary pixels of the infrared asphalt images in the Lab color space obtained in the second step, a forth step of conducting lateral filtering on the images obtained in the third step, a fifth step of conducting vertical filtering on the images obtained in the fourth step, and a sixth step of switching images in the Lab color space processed in the step 5 to the RGB color space and outputting filtered infrared asphalt images. Lateral and vertical bi-directional iteration bilateral filtering means is employed, infrared asphalt image noises can be rapidly eliminated, and the edge detail can be retained. The infrared asphalt image filtering effect is satisfactory, and edge information of images can be retained.

The invention discloses a thresholding image denoising method based on nonsubsampled Contourlet transformation, and the problem that the effect of an existing denoising method is bad is mainly solved. The realization method comprises the steps of inputting a noisy image, conducting the nonsubsampled Contourlet transformation on the noisy image, conducting a sub-band noise variance estimation and an estimation of a standard error of a coefficient in a transform domain, setting a reasonable threshold, conducting threshold processing on an NSCT coefficient through an improved semi-soft threshold function without processing a low frequent component, and obtaining a de-noised image through conducting nonsubsampled Contourlet inverse transformation on a processed NSCT coefficient. According to the thresholding image denoising method based on the nonsubsampled Contourlet transformation, noise in a natural image with white Gaussian noise can be effectively removed, and image information edges are kept as much as possible at the same time, so that the best recovery of an original image is obtained, and the thresholding image denoising method based on the nonsubsampled Contourlet transformation has good application prospect and great development potential.

Disclosed is a method for detecting a point target in an infrared image sequence. The method comprises the steps of (1) estimating a noise variance of a current frame image, adopting a bilateral filtering method to preprocess the image, (2) adopting the Robinson-Guard filtering method based on a template mid value to conduct filtering on the preprocessed image, (3) conducting binarization processing on the image, marking target points in the image, recording positional information of the target, (4) initializing parameters of a pipeline filter, (5) utilizing target positional information of first three frames for predicating the target position of the current frame, searching for the target, updating related information in a target position information sheet, conducting judgment to output the target information, and (6) repeating the steps until processing of all the images in the infrared image sequence is completed. According to the method for detecting the point target in the infrared image sequence, background suppression is effectively carried out, the false alarm probability is reduced, and the detection probability of the target is improved through the detection of the multi-frame sequence images.

The invention discloses an online diameter detection method of varnished wires. The online diameter detection method includes steps of acquiring diameter images of the varnished wires; filtering noise in the diameter images; subjecting the diameter images which are treated by noise filtering to edge detection to obtain edge images; calculating the diameter of the varnished wires according to the edge images. By adopting the edge detection method combining the iteration method to segment threshold of the images and the edge tracking method based on image scanning, background and target of the edge images obtained are well differentiated, edge continuity, no fake edges and burrs are generated, edge width is single pixel, and edge details are well reserved, so that the online diameter detection method has higher detection accuracy. Meanwhile, compared with a conventional edge detecting method, the edge detection method costs less time, so that the online diameter detection method has higher detection speed. In conclusion, the online diameter detection method has high theoretical value and engineering application population value.

The invention provides an improved shadow detection method based on a c3 space image, three aspects of aerial image enhancement, multilateral weighted smooth denoising and edge detection are improved, contrast enhancement is carried out on the c3 space image by adopting logarithmic transformation, and the pixel value range of a low-gray region is greatly expanded so as to improve the accuracy of extraction of a shadow pixel in a shadow region; de-noising processing is carried out on the image by adopting multilateral weighted filtering, and image edge details can be kept while image noise is removed; based on a Canny operator, a diagonal direction is increased to improve a gradient amplitude calculation method, a critical domain is determined by adopting adaptive double-critical domain selection, and finally edge detection is performed by adopting the double-critical domain; experiments prove that the shadow area can be successfully detected, interference of ground objects such as roofs and roads on detection results can be effectively weakened, and compared with an original algorithm before improvement, the shadow detection accuracy is obviously improved.

The invention belongs to the technical field of image scanning, and particularly relates to an image noise reduction method and application thereof. Under the condition of existing rapid imaging, reconstructed MRI images are too loud in noise, and diagnosis of a doctor is affected. The invention provides an image noise reduction method. The image noise reduction method comprises the steps of constructing a self-correction convolutional neural network; taking an L1 norm as a loss function; optimizing the self-correction convolutional neural network; taking an image with the noise as the input of the network, taking the image with the noise and a corresponding noiseless image as network labels, training the network, and obtaining a mapping relation from the image with the noise to the noiseless image; and performing noise reduction on the image needing noise reduction through the trained network to obtain a noise-reduced image. The problem of unclear images is solved.

The invention discloses an image composition method based on high-dynamic range principle, a device and a mobile terminal. The method comprises the following steps: acquiring multiple images with different exposure parameters; performing composition on multiple images with different exposure parameters based on the high-dynamic range principle to acquire a high-dynamic range image; performing tone mapping on the high-dynamic range image so as to enable the pixel of the high-dynamic range image to perform the recompression; the device comprises an image acquisition module, an image composition module and a tone mapping module; the mobile terminal comprises a processor and a storage for storing executable programs of the processor; when the processor executes the program stored in the storage, the operation of the above-mentioned method is realized. The high-dynamic range image is obtained by performing the composition on multiple images with different exposure parameters, and the tone mapping is performed, the image internal details are improved, and the image border details are reserved.

The invention provides an image filtering method and device, and the method comprises the steps: selecting a pixel point from a to-be-filtered image as a to-be-filtered point; selecting a matrix withthe size of (2N + 1) * (2N + 1) as a matrix to be filtered by taking the point to be filtered as a center, where N is a positive integer; selecting (2N + 1) pieces of data from the to-be-filtered matrix in the 0-degree direction, the 45-degree direction, the 90-degree direction and the 135-degree direction; performing directional matching on the 0-degree direction, the 45-degree direction, the 90-degree direction and the 135-degree direction by utilizing the selected (2N + 1) * 4 groups of data so as to obtain filtering output values in the four directions through calculation respectively; andaccording to the filtering output values in the four directions, calculating to obtain a filtering output value of the to-be-filtered point. According to the embodiment of the invention, the problemsof image detail loss and large calculation amount in the existing filtering method are solved, the edge details of the image are more truly reserved while the noise is effectively filtered out, and meanwhile, the calculation amount is reduced.

The invention discloses a wavelet domain interferometric synthetic aperture radar (InSAR) phase filtering method. The method comprises the following steps of carrying out N grade wavelet packet transformation on data in an InSAR phase diagram, wherein the N is greater than 1 and the N is a positive integer; identifying a Nth grade wavelet domain signal and marking a growth process of the identified signal; carrying out filtering on the InSAR phase diagram after the N grade wavelet transformation and carrying out inverse wavelet transformation on the InSAR phase diagram after the filtering. The invention also discloses an InSAR phase filtering apparatus.

The invention discloses a robust roundabout detection method for a lunar probe and an aircraft navigation method, solves the problems that a conventional roundabout detection method is influenced by illumination and is poor in robustness in the descending and landing process of the existing lunar probe, and belongs to the technical field of lunar detection. According to the method, firstly, an image is filtered, image edge details are reserved, and noise influence is removed; secondly, entropy distribution of the processed image is calculated, maximum entropy threshold segmentation is carried out, image information is subjected to ternary processing, the sensitivity of the image to a light source is removed, and meanwhile the image information is reserved to the maximum degree; according to the method, a normalized multi-index constraint ring-shaped mountain matching and fitting method is adopted to complete ring-shaped mountain extraction, and continuous, complete and clear edges of the lunar surface ring-shaped mountain can be extracted under different illumination conditions. The aircraft navigation method based on robust annular mountain extraction is better in real-time performance.

The invention discloses an infrared image hybrid noise reduction method based on noise recognition, which comprises the following steps: firstly, carrying out 3 * 3 median filtering on a near-infrared image by using a filter to obtain an image only containing Gaussian noise, and then carrying out wavelet transform denoising processing on the image; secondly, after wave decomposition, main information of the image being almost all distributed in a low-frequency sub-band, and processing of a low-frequency part being often omitted in order to prevent important information from being damaged during denoising; performing thresholding processing on the wavelet coefficient of the high-frequency region after image decomposition, and then performing median filtering again by adopting a 5 * 5 filter; and finally, performing wavelet reconstruction by using the wavelet coefficient of each sub-band to obtain a denoised image. Through the above mode, compared with a single denoising method, the hybrid noise reduction method can more effectively remove noise and improve the signal-to-noise ratio of the near-infrared face image in view of the denoising effect of the collected night infrared image, and edge details of the image are well reserved.

The invention discloses an image edge processing method based on guided filtering and application, and relates to the technical field of digital image processing. The method comprises the following steps of: acquiring an input image and a guide image; collecting a set first rectangular window wk1 used for weak denoising and a set second rectangular window wk2 used for strong denoising; using the window wk1 for guided filtering weak denoising, and using the window wk2 for guided filtering strong denoising; obtaining a denoising intensity absolute difference according to the filtering coefficients corresponding to the two windows, and obtaining a corresponding filtering weight according to the absolute difference; obtaining the value of the edge gradient of a strong denoising result image, and then judging pixel information needing edge smoothing processing; and based on a weak denoising result image and the filtering weight, carrying out edge smoothing processing on pixels to obtain a final result image. According to the method, edge smoothing can be achieved while image noise is suppressed and edge details are kept, the denoising intensity and the edge smoothing intensity can be independently controlled, the applicability is wide, and the flexibility is high.

The invention discloses a pest image recognition method based on a Bayesian width residual neural network. The method comprises the following steps: step 1, preprocessing the pest image recognition training data set, using a rich edge detection algorithm (Rich-Edge) Pest edge detection is performed on the grayscale image. Step 2. Construct a Bayesian Wide-Residual Network (BWResNet). Step 3: Input the pest edge image obtained in Step 1 into the BWResNet constructed in Step 2. Use the pest edge image training set obtained in step 1 to obtain the total error function of BWResNet. Step 4: Use the error function obtained in Step 3 to train the network. Here we propose the block-conjugate (Block‑cg) algorithm to train the network. Step 5. Update hyperparameters according to the optimized network in step 4. Step 6: Repeat steps 4 and 5 to obtain the final network, and the classification accuracy rate obtained after inputting the verification set of pest images into the network is higher.

The invention provides a method and a system for filtering image data. The distinguishing method comprises the following steps: sequentially selecting each pixel point in a plurality of pixel points in the image data to be processed as a target pixel point; constructing a filtering window by taking the target pixel point as a center, and performing vector addition on each phase in a plurality of phases in the filtering window to obtain a sum vector of the plurality of phases; calculating a phase angle difference between any two phases in the plurality of phases in the filtering window to obtain a plurality of phase angle differences and determine a median of the plurality of phase angle differences; determining a weight value for each phase angle difference, associating each normalized weight value with a corresponding pixel point in the filtering window, and calculating a weighted sum; and adding the weighted sum and the phase principal value of the sum vector to determine a filtering parameter of the target pixel point, and determining a filtering result of the target pixel point based on a comparison result of the filtering parameter of the target pixel point and a filtering threshold.

The invention discloses a thresholding image denoising method based on non-subsampling Contourlet transform, which mainly solves the problem of poor effect of the existing denoising methods. The implementation process is: 1) input a noisy image; 2) perform non-subsampling Contourlet transform on the noisy image; 3) estimate the noise variance of each subband and the coefficient standard deviation in the transform domain; 4) set reasonable 5) Threshold the NSCT coefficients using the improved semi-soft threshold function, and do not process the low-frequency components; 6) Perform non-subsampling Contourlet inverse transformation on the processed NSCT coefficients to obtain the denoised image. The invention can effectively remove the noise in the natural image containing Gaussian white noise, and while removing the noise, the edge of the image information is preserved as much as possible, so as to obtain the best restoration of the original image, and has a good application prospect and great development. potential.