49 results about "Lee filter" patented technology

The invention discloses a small-sample polarized SAR ground feature classification method based on a deep convolutional twin network, and mainly solves a problem that a conventional method is low in classification precision because the number of polarized SAR data mark samples is smaller. The method of the invention comprises the steps: 1), inputting a to-be-classified polarized SAR image and a real ground object mark of the to-be-classified polarized SAR image, and carrying out the Lee filtering; 2), extracting an input feature vector from the filtered to-be-classified polarized SAR data, andcarrying out the dividing of a training sample set and a test sample set; 3), carrying out the combination of each two samples in the training sample set, and obtaining a sample pair training set; 4), building the deep convolutional twin network, and carrying out the training of the deep convolutional twin network through the training sample set and the sample pair training set; 5), carrying outthe classification of the samples in the test set through the trained deep convolutional twin network, and obtaining the classes of ground features. According to the invention, the method expands thetraining set under the twin configuration, achieves the extraction of the difference features, enables the classification precision of a model to be higher, and can be used for the target classification, detection and recognition of a polarized SAR image.

The invention puts forward a polarized SAR (Synthetic Aperture Radar) image classification method based on an SLIC (Software Licensing Internal Code) and an improved CNN (Convolutional Neural Network), and is used for solving the technical problem of low classification speed and low classification accuracy in an existing supervised polarized SAR image classification method. The method comprises the following steps that: firstly, taking the Wishart distance and the polarization feature of a polarized SAR image as new data, carrying out Lee filtering on the new data, and inputting the new data into an improved CNN to be classified so as to obtain a preliminary classification result; then, carrying out SLIC superpixel segmentation on the pseudo-color image of the polarized SAR image to obtain a superpixel segmentation result; and finally, utilizing the superpixel segmentation result to carry out constraint post-processing on the preliminary classification result to obtain a final classification result. The method is high in classification speed and accuracy, and can be used for fields including polarized SAR terrain classification and the like.

The invention discloses a polarization SAR image classification method based on RBM and SVM, mainly solving the problem of the existing polarization SAR image classification method that the classification precision is low. The method comprises the steps as follows: (1) inputting the polarization SAR image to be classified, having delicate polarization Lee filtering operation; (2) resolving and extracting original feature of each pixel point and integrating based on the polarization coherence matrix, polarization covariance matrix and Cloude; (3) initializing and training RBM to obtain the related parameter; (4) classifying by using SVM according to the feature learned by the RBM to obtain the classification result. Compared with the existing method, the space dependency of the image is totally considered and the feature in favour of classifying polarization SAR image can be extracted, the polarization SAR image classification precision is obviously raised, the method can be used for terrain classification and object identification for polarization SAR image.

A semi-supervised polarization SAR terrain classification method based on full convolution GAN is implemented by the following steps: performing refined Lee filtering; inputting polarization SAR data;constructing a generation network G; constructing a discrimination network D; training and testing the networks and calculating classification accuracy rate. Compared with the semi-supervised Co-training classification method in the prior art, the invention improves the classification accuracy of polarization SAR data of the polarized synthetic aperture radar, solves the problem of insufficient utilization of polarization information, enhances the robustness of the model to the input data, and has the effect of end-to-end classification. The invention can be applied to the terrain classification of polarized synthetic aperture radar polarized SAR data.

The invention discloses a polarimetric SAR (synthetic aperture radar) image segmentation method based on a DBN (deep belief network). The advantages of learning features of the deep learning theory are applied to the polarimetric SAR image segmentation. The segmentation method includes subjecting polarimetric SAR data to fine Lee filtering; subjecting polarimetric SAR coherence matrix T to H / Alpha resolving to obtain the parameter feature; extracting gray-level co-occurrence matrixes from three channels on leading diagonals of the coherence matrix T, and calculating features including contrast, coherence, energy and inverse difference; combining the features and the elements of the coherence matrix to train a two-layer DBN; inputting the polarimetric SAR data in the DBN for classification and displaying a classification result image. According to the arrangement, the scattering feature and the gray-level co-occurrence matrix feature are integrated, information integrity is kept, and the layer-by-layer learning feature is applicable to polarimetric SAR image object recognition.

The invention discloses a method for detecting oil spill at sea based on a full-polarized synthetic aperture radar (SAR) image. The method includes the following steps: pre-treating polarized SAR data which require analyzing to obtain a full-polarized SAR covariance matrix; conducting refining polarized Lee filtering on the covariance matrix; based on the filtered covariance matrix or through a Stokes vector calculated by the covariance matrix, extracting 6 polarized features to constitute a feature combination; inputting training sample features with ground verification label information to a maximum likelihood (ML) classifier, training the classifier and optimizing parameters of the classifier; taking polarized features as input, utilizing the ML classifier to conduct detection and classification on an oil membrane; processing a classification result based on morphology, utilizing verification information of actual measurement to evaluate classification precision. According to the invention, the method can increase performances of oil spill at sea detection and classification methods and promote the application of the full-polarized SAR in actual engineering problems such as monitoring of oil spill at sea.

The invention provides a polarity SAR target detection method based on a FCN-CRF master-slave network. The method comprises steps of inputting a to-be-detected polarity SAR image, and carrying out delicate polarity Lee filtering on a polarity coherence matrix T of the polarity SAR image to filter coherent noise so as to obtain the filtered coherence matrix, wherein each element of the filtered coherence matrix is a 3*3 matrix, that is to say, each pixel point has nine-dimensional features. According to the invention, by expanding image block features into pixel-level features, the correlation degree of selected training samples through matching of pixel points of a region of interest is quite high and quite effective; the feature image blocks with the pixel points of the region of interest whose quantity is less than 50% of the whole image block will not participate in following calculation, so the operand is greatly reduced and the detection efficiency is improved; by using the Lee filtering to pre-process of the original polarity SAR image, coherence spot noise is effectively reduced and image quality and detection performance are improved; and by use of spiral scattering components corresponding to urban buildings obtained through the Yamaguchi decomposition, features of polarity SAR artificial targets are effectively extracted, and detection precision of the artificial targets is improved.

The present invention belongs to the field of image treating technology, and is especially improved image restoration method. The improved image restoration method includes the following steps: 1. Lee filtering the degenerate image to eliminate noise influence; 2. determining the image supporting area by means of threshold partitioning process; 3. inverse filtering to obtain estimated image; 4. projecting the estimated image to one real image space with one NL filter; 5. low pass filtering the NL filter output image; 6. finding out the difference values; 7. correcting the coefficients with the difference values; and 8. repeated correcting until the difference values are converged to minimum values to complete the image restoration. The present invention has raised support threshold precision, powerful anti jamming capacity and shortened running time.

The invention discloses a construction method of document image enhancement, which is characterized by comprising the following steps of: 1. combining global contrast enhancement and local contrast enhancement together, i.e. combining a CLAHE (Contrast-Limited Adaptive Histogram Equalization) method and an Lee filter method, and forming a detail-strengthened AHE method; 2. improving a DAHE method, and introducing a gradient amplitude to enable the DAHE method to have image space self-adaptive ability and obtain an IDAHE method; and 3. carrying out experiments by using sampled data to obtain the optimal experimental numerical values of parameters epsilon, k, lambda1 and lambda 2 of input images in different types. The invention has the advantages that the global contrast is improved, the local details are effectively amplified, and imaging is clear.

The invention discloses an ice cover radar image ice layer fine segmentation method based on an FCN-ASPP network, and relates to the field of computer vision and mode recognition. According to the invention, the radar amplitude image is used as a training sample of a network, corresponding data amplification is carried out for the problem of less ice layer image data, and the wide applicability ofthe method is expanded. Lee filtering is carried out on the ice cover image. In order to save edge information as much as possible, a threshold judgment process is added to a filtering process. FCN-is constructed, and FCN-is constructed; according to the ASPP ice layer segmentation deep network, the ASPP layer is improved, so that the extraction capability of the network on small-scale characteristics is enhanced. The preliminary classification result is further processed through CRF, and the segmentation result is further refined on the basis of achieving end-to-end pixel level segmentation.In addition, the network greatly realizes the autonomous learning process.

An image processor includes a Lee filter for processing image data to improve noise therefrom. The Lee filter includes a smoothing control signal. During filtering of the image data, the smoothing control signal is converted to a monochrome video signal that is provided to a display instead of, or in addition to, the filtered image data. The image projected by the display of the smoothing control signal highlights edge activity in the filtered image. This display of edge activity makes it easier to locate noisy areas in the image data and fine-tune the amount of filtering to further reduce noise present in the image data and yet leave the desired image details alone.

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 discloses a polarimetric SAR classification method on the basis of NSCT and discriminative dictionary learning and mainly solves the problems of low classification accuracy and low classification speed of an existing polarimetric SAR image classification method. The polarimetric SAR classification method comprises the following implementing steps: 1, acquiring a coherence matrix of a polarimetric SAR image to be classified and carrying out Lee filtering on the coherence matrix to obtain the de-noised coherence matrix; 2, carrying out Cloude decomposition on the de-noised coherence matrix and using three non-negative feature values of decomposition values and a scattering angle as classification features; 3, carrying out three-layer NSCT on the classification features and using a transformed low-frequency coefficient as a transform domain classification feature; 4, using the transform domain classification feature and combining a discriminative dictionary learning model to train a dictionary and a classifier; 5, using the dictionary and the classifier, which are obtained by training, to classify a test sample so as to obtain a classification result. The polarimetric SAR classification method improves classification accuracy and increases a classification speed and is suitable for image processing.

The invention discloses a selection method for counting an identical distribution space pixel based on a time sequence SAR (Synthetic Aperture Radar) image. The selection method comprises the following steps: firstly, carrying out pre-processing on an original SAR image data sequence to obtain a single-visual SAR strength image sequence; carrying out registration on the SAR strength image sequence; acquiring a rejection region of a likelihood ratio test under the assumption that a single-visual SAR strength image subjected to the registration complies with index statistical distribution; comparing the counting similarity of a time sample of each space pixel in a rectangular sliding window and a time sample of a central reference pixel; traversing each space pixel in a whole image and acquiring an identical statistical distribution sample of each space pixel; filtering the SAR image by adopting Lee filtering and the identical distribution sample of each space pixel. According to the selection method disclosed by the invention, samples with the same attribute of the reference pixel are selected through a likelihood ratio assumption test, so that a spatial distribution law and a backward scattering property of pixel points in a radar image are met better and a filtering image with the full resolution ratio which is closer to a real ground surface is obtained.

The invention discloses a polarized SAR image target detection method based on a deep stairway network, comprising the following steps: inputting a polarized SAR image to be detected and performing Lee filtering on the polarized coherent matrix T; solving the filtered T to obtain a polarized covariance matrix C; performing Yamahachi decomposition on the polarized covariance matrix C to form a pixel-based characteristics matrix F; normalizing the F and extracting blocks for each element of the normalized characteristics matrix F1 and listing them in a column; forming a characteristics matrix F2 based on the image blocks; obtaining a training set D according to F2; using the SLIC algorithm in the super pixels to obtain a test set T; constructing a target detection model based on a deep stairway network; using the training data set D to train the target detection model; and utilizing the trained target detection model to classify the test data set T. According to the invention, a deep stairway network is adopted, which only uses a small amount of labeled samples to obtain high target detection precision. The present invention can be used for the classification of ground targets.

The invention discloses a method for classifying polarimetric synthetic aperture radar (SAR) images on the basis of scattered power and intensity combined statistics. The method for classifying the polarimetric SAR images on the basis of the scattered power and the intensity combined statistics mainly solves the problem that in the prior art, zones with similar scattering properties are difficult to distinguish and classification numbers are fixed. The method is achieved by the following steps that: utilizing a Lee filter to conduct filtering on a coherence matrix T, utilizing Freeman decomposition to obtain a power matrix, utilizing eigenvalue decomposition to obtain an intensity matrix, conducting 8 neighborhood averaging on the power matrix and the intensity matrix respectively, selecting a k class homogeneous zone as a training sample, utilizing an EM algorithm to estimate parameters of probability density distribution functions of the power matrix and the intensity matrix of a k class sample, solving joint probability distribution of the power matrix and the intensity matrix of the k class sample, and conducting Bayesian classification on polarimetric SAR data to be classified to obtain classification results. The method for classifying the polarimetric SAR images on the basis of the scattered power and the intensity combined statistics has the advantages of being good in the classification effect of the polarimetric SAR images, and can be further used for target detection and target identification of the polarimetric SAR images.

The invention discloses a polarized SAR image classification method based on an extreme learning machine, mainly for solving the problem of low time efficiency of polarized SAR image classification. The method is realized by the following steps: inputting mark information of a polarized SAR image to be classified and a coherent matrix, and carrying out Lee filtering; extracting characteristics of the coherent matrix after the filtering, normalizing the characteristics to obtain a data set, and obtaining a mark set from the data set; dividing the mark set into a training set and a test set, and training the extreme learning machine by use of the training set; and predicating the type of the data set by use of the well trained extreme learning machine to obtain a polarized SAR image classification result. According to the invention, polarized SAR images are classified by use of the extreme learning machine, the classification precision is quite high, at the same time, the operation time is less, and the method can be applied to terrain classification and object identification.

An image processor includes a Lee filter for processing image data to improve noise therefrom. The Lee filter includes a smoothing control signal. During filtering of the image data, the smoothing control signal is converted to a monochrome video signal that is provided to a display instead of, or in addition to, the filtered image data. The image projected by the display of the smoothing control signal highlights edge activity in the filtered image. This display of edge activity makes it easier to locate noisy areas in the image data and fine-tune the amount of filtering to further reduce noise present in the image data and yet leave the desired image details alone.

The invention discloses a SAR image coastline extracting method based on a geometric active contour model. The method comprises steps of: reading a SAR image including a to-be-extracted coastline target by using Matlab software, filtering the SAR image including the to-be-extracted coastline target by using a Lee filter and adjusting the brightness of the SAR image; (2) performing convolution on the SAR image to generate a grid sampling point, drawing multiple small discs in the grid sampling point as the initial contour of the coastline; (3) using the initial contour of the coastline as an input of the geometric active contour model, vectorizing the coastline processed by the geometric active contour model by using area information combining images as a boundary stop condition of the geometric active contour model so as to finally obtain a continuous coastline. The method utilizes the global information of the images, has good expandability, and inhibits the influence of noise on coastline detection so as to accurately extract the coastline.

The invention discloses a polarized SAR (Specific Absorption Rate) image classifying method based on super-vector coding. The method comprises the following realizing steps: (1) inputting images; (2) refining Lee filter; (3) extracting a scattering characteristic vector; (4) coding super-sparsely; (5) obtaining super-vector characteristics; (6) normalizing and expanding the sample characteristic set; (7) selecting a training sample and a test sample; (8) training the classifier to classify the images; (9) calculating the classifying precision; (10) outputting a result. According to the method provided by the invention, the image characteristics extracted are unlikely to be affected by noise points, and are small in redundancy and good in representability, thereby being applicable to effectively improving the classification precision in the course of classifying, and detecting and identifying an SAR image objective of a polarized and synthetic aperture radar.

The invention discloses an SAR image speckle noise suppression method based on Gamma-Lee filtering. The method comprises the steps: carrying out the Gamma transformation of an original image so as toimprove the details of a dark part of the original image, and solving a problem that the retaining effect of an edge detail part of a filtered SAR image is not good; traversing each pixel point of themeasured image by adopting a translation-invariant local window, and realizing Lee filtering in a manner of counting the mean value and variance of the covered pixel points and a central pixel value;in combination with the thought of adjusting the Gamma coefficient and the Gamma index through negative feedback, inhibition of SAR image speckle noise is adjusted and enhanced step by step; finally,simulation experiment results show that the SAR image filtering method solves the problem that an existing SAR image filtering method is poor in image edge detail part retaining effect, and meanwhilethe speckle noise restraining effect is improved to a certain extent. The method is superior to an original Lee filtering method in noise suppression and edge detail protection effects, and can be applied to the field of SAR image processing.

The invention relates to the technical field of ultrasound, in particular to an ultrasonic image noise reduction method and electronic equipment, and is used for solving the problem of how to well suppress speckle noise and well maintain edge information during image filtering. According to the invention, only the image in the window is filtered by setting the filtering window. And if the image quality in the window is greater than the image quality threshold, it is proved that the image quality is relatively high, and at the moment, the Lee filtering algorithm is adopted for filtering, so that the edge information of the image can be ensured while the speckle noise is suppressed. If the image quality in the window is smaller than or equal to the image quality threshold value and the size of the filtering window is smaller than the window minimum value, it is proved that the image quality is low, at the moment, a gradient operator is adopted for filtering, the problem that the speckle noise suppression effect is not ideal is avoided, and edge information is prevented from being damaged.

The invention discloses a polarized SAR image classification method based on a long-short term memory recurrent neural network. The method comprises the following steps: carrying out delicate Lee filtering processing on polarized SAR data; carrying out super-pixel segmentation on the filtered polarization coherence matrix; obtaining a multi-dimensional characteristic polarization SAR image by using the spatial information of the polarization SAR image; obtaining sample data and test data of the polarimetric SAR image to be classified; carrying out deep learning on the long-short term memory network by using the sample data; classifying the test data by using the trained long-short time memory network; and obtaining a classification label, and obtaining a color classification result graph.According to the method, spatial information of the polarized SAR images is combined, the polarized SAR images with multi-dimensional features are obtained, the polarized SAR image data with the multi-dimensional features are used as input of the long-short time memory network, the classification accuracy of the polarized SAR images is effectively improved, and the method can be used for ground feature classification and recognition of the polarized SAR images.

The invention provides an image tracking method based on image frequency domain conversion. The method comprises the following steps: firstly, preprocessing input SAR image data; on the basis of a traditional delicate Lee filter, introducing a window size adaptive selection method based on equivalent opinions, and forming a filtering algorithm adaptive to the window size; secondly, establishing anSAR image dense sample application feature extraction algorithm based on a two-dimensional sinc function model; developing a multi-layer multi-window matching strategy, and determining the reliable matching window size in a self-adaptive mode according to the error measurement output by matching of different layers; developing a sub-pixel phase correlation algorithm based on multi-scale phase consistency characteristics, constructing multi-scale phase consistency structure images, and estimating the phase difference between the images in a frequency domain space. Through the verification of the specific embodiment, the tracking method provided by the invention can effectively track the target feature image, and improves the accuracy and robustness.

The invention discloses a full-polarization SAR image speckle suppression method. Polarization decomposition is carried out on a target by combining a polarization SAR scattering model on the basis ofnon-local Lee filtering, a more accurate similar set is obtained according to polarization similarity among pixels, and scene target polarization scattering characteristics and texture detail information are better kept while noise is filtered, so that a visualization result of coherent speckle suppression is obtained; in this way, the problems that an existing algorithm is high in complexity andtoo long in consumed time are solved through the improved weight acceleration algorithm, then, an image accumulation matrix is introduced to eliminate redundant calculation, the calculation complexity is greatly reduced on the basis that the filtering effect of the algorithm is kept, and the operation efficiency is remarkably improved.

The invention discloses novel tilapia feed produced from cassava lees through microbial fermentation and a preparation method of the tilapia feed and belongs to the technical field of tilapia feed processing. The feed is prepared from raw materials in parts by weight as follows: 75-93 parts of dry cassava lees filter cakes, 18-35 parts of pure insect pulp albumen powder, 12-18 parts of earthworm powder, 22-36 parts of soybean meal, 17-24 parts of peanut meal, 2.5-4.5 parts of a phagostimulant and 3.2-5.8 parts of a traditional Chinese medicine preparation through drying, crushing, sieving, mixing, granulation, drying, sterilization and other steps. The feed integrates efficacy of multiple traditional Chinese medicines and can have functions of resisting bacteria, diminishing inflammation, clearing heat, removing toxicity, enhancing the immunity of a body and the like. The feed is well absorbed by tilapia, the death rate of the tilapia suffering from streptococcicosis can be effectively decreased, and the culture risk is reduced. The feed is reasonable to prepare, contains rich nutrients and can increase the feed intake of the tilapia suffering from streptococcicosis, increase the growth speed and shorten the growth cycle.

The invention discloses an SAR (Synthetic Aperture Radar) image change detection method based on an FPGA (Field Programmable Gate Array), which mainly solves the problems that the prior art is easilyinfluenced by noise and the detection speed is low. The method comprises the following implementation steps of: 1) reading in two SAR images at different periods; 2) carrying out parallel Lee filtering on the two SAR images; 3) generating two difference graphs with different characteristics according to the filtered image, namely a direct difference graph and a logarithm ratio difference graph; 4)transforming the difference image into a frequency domain by adopting parallel wavelet transformation, and performing fusion and parallel wavelet inverse transformation in the frequency domain to obtain a fused difference image; 5) using a parallel difference K-means method to perform two clustering on a fusion difference map to obtain a change detection result. The invention adopts a method of FPGA parallel acceleration and difference graph fusion, has the advantages of fast detection speed and good anti-noise performance, and can be used for fast and accurate SAR image change detection in complex environments.

A SAR image speckle noise suppression method based on Gamma-Lee filtering, by performing Gamma transformation on the original image to improve its dark details, it solves the problem that the edge details of the filtered SAR image are not well preserved; the translation-invariant local The window traverses each pixel of the measured image, and implements Lee filtering by counting the mean and variance of the covered pixels and the central pixel value. Combined with the idea of ​​adjusting the Gamma coefficient and Gamma index with negative feedback, gradually adjust and strengthen the effect of speckle noise on SAR images. Finally, the simulation experiment results show that the present invention solves the problem of the existing SAR image filtering method that the image edge detail part is not well preserved, and at the same time, the suppression effect of speckle noise is improved to a certain extent; It is superior to the original Lee filtering method and can be used in the field of image processing of SAR images.

The invention belongs to the technical field of SAR image segmentation, and particularly relates to an SAR image Ostu segmentation method based on power transformation. The SAR image Ostu segmentation method based on the power transformation includes the following steps: obtaining an original SAR image, and carrying out speckle Lee filtering on the original SAR image to obtain a filtered image; respectively carrying out the power transformation with N different scales on the filtered image to obtain N images after the power transformation, wherein the N images after the power transformation are expressed as Z(1)-Z(N) respectively, and the Z(n) expresses the image obtained in the mode that the power deformation with the nth scale is carried out on the filtered image; obtaining the specific value of the interclass variance to the total variance of the Z(n) under different segmentation threshold values; obtaining the optimal segmentation threshold value and the optimal scale of the power transformation; obtaining a corresponding image after the optimal power transformation according to the optimal scale of the power transformation; segmenting the image after the optimal power transformation according to the optimal segmentation threshold value.

The present invention provides a SAR ship recognition method and system combining saliency and neural network, including data preprocessing, removing coherent speckle noise of SAR image through Lee filtering, while maintaining the edge information of the image, and then cropping the SAR image of each scene Obtain image blocks; construct a data set, including selecting SAR image blocks containing ships in different scenes, and mark the position information of ships with a rectangular minimum bounding box to obtain a SAR image ship data set containing label information; construct a fusion saliency perception Convolutional neural network, including extracting features through the Darknet53 network, calculating ship candidate frames and confidence based on the obtained multi-scale feature map, obtaining the salient feature map in the candidate frame based on the global contrast method, and taking the outer rectangle of the salient area as the detection Results; training network, for the SAR image to be recognized, the image block is obtained after preprocessing by row, and then the network model obtained by training is used for prediction, and the SAR image is re-stitched based on the predicted image block.