51results about How to "Good regional consistency" patented technology

The invention discloses a method for segmenting a multi-dimensional texture image on the basis of fuzzy C-means FCM clustering and spatial information and mainly solves the problem of poor quality of image segmentation. The realizing process comprises the following steps of: inputting the texture image to be segmented, carrying out two-dimensional discrete wavelet transformation to the image, and calculating the characteristic vector corresponding to each wavelet coefficient; segmenting the coarsest scale of wavelet transformation; calculating spatial coordinate factors corresponding to the coefficients of the coarsest scale, adding the spatial coordinate factors into an objective function of a traditional FCM clustering algorithm and obtaining the segmenting result marker mapping and the marking field of the scale; obtaining the segmenting result marker mapping of the next scale by adopting the multiple dimensional segmenting method determined by an adaptive scale until the obtained segmenting result marker mapping is at the finest scale; and outputting the segmenting result of the finest scale as the final segmenting result. The method has the advantages of accurate segmenting edge and good consistency of segmenting regions and can be used for segmenting texture images, SAR images including texture information, remote sensing images and medical images.

The invention discloses a method for classifying polarimetric SAR (synthetic aperture radar) images on the basis of Cloude decomposition and K-wishart distribution and mainly solves the problem that the prior art is poor in classification effect and high in computation complexity. The method includes the steps of firstly, reading a polarimetric SAR image to be classified, subjecting each pixel of the image to Cloude decomposition to obtain an entropy H and a scattering angle alpha; secondly, initially partitioning the polarimetric SAR image according to the values of the entropy H and the scattering angle into eight classes; thirdly, subjecting results of eight classes of the whole polarimetric SAR image to iteration to obtain more accurate classification results. Compared with classic classification methods, the method is stricter in partitioning the polarimetric SAR image, better in classification effect, lower in computation complexity and applicable to terrain classification and target recognition of the polarimetric SAR images.

The invention discloses a polarization synthetic aperture radar (SAR) image classification method based on spectral clustering. The polarization SAR image classification method mainly solves the problem that an existing non-supervision polarization SAR classification method is low in accuracy. The polarization SAR image classification method comprises the steps of extracting scattering entropy H of representation polarization SAR target characteristics to serve as an input characteristic space of a Mean Shift algorithm combining with space coordination information; diving in the characteristic space with the Mean Shift algorithm to obtain M areas; choosing representation points of all areas on the M areas to serve as spectral clustering input to spectrally divide all areas, and further finishing spectral clustering on all pixel points to obtain pre-classification results; and finally classifying the whole image obtained from the pre classification with a Wishart classifier capable of reflecting polarization SAR distribution characteristics in an iteration mode to obtain classification results. Tests show that the polarization SAR image classification method is good in image classification effect and can be applied to non-supervision classification on various polarization SAR images.

The invention discloses a polarized synthetic aperture radar (SAR) image classification method based on Freeman decomposition and data distribution characteristics and mainly solves the problems of high computation complexity and poor classification effect in the prior art. The polarized SAR image classification method includes step: (1) performing Freeman decomposition for polarized SAR images to be classified to obtain plane scattering power, dihedral angle scattering power and volume scattering power; (2) initially dividing the polarized SAR images into three classes according to the three scattering powers; (3) calculating the distribution characteristic parameter xL of each pixel point in each class; (4) subdividing each of the three initially divided classes into three classes according to the distribution characteristic parameters xL to divide the whole polarized SAR images into nine classes; and (5) performing complex Wishart iteration for the obtained nine-class dividing results to obtain the final classification result. Compared with the typical classification method, the polarized SAR image classification method is rigorous in polarized SAR image dividing, good in classification effect and small in computation complexity and can be applied to terrain classification and object identification of the polarized SAR images.

The invention discloses an image dividing method based on immune multi-object clustering, relating to the technical field of image processing, and mainly solving the problems that the conventional method has single evaluation index, and easily has bad region consistency and disorder boundary. The method comprises the following realization steps of: (1) extracting the characteristic of an image to be divided, and primarily dividing the image by controlling the watershed of a mark; (2) setting a running parameter and initializing the population of an antibody; (3) combining the locally-searched immune multi-objective optimizing method with the population of the antibody to obtain an approximate Pareto solution set; (4) selecting an optimal solution in the approximate Pareto solution set obtained in the step (3) according to a PBM index; and (5) marking an image pixel point according to a primary dividing result obtained in the step (1) and a clustering result obtained in the step (4) to obtain a final classifying result. The image dividing method has the advantages of good dividing result region consistency, being capable of keeping complete information, and having fast computation speed, and can be used for identifying an image object.

The invention discloses a hyperspectral image classification and wave band selection method based on multi-target immune cloning. The hyperspectral image classification and wave band selection method based on multi-target immune cloning comprises the following steps that a sparse representation classifier is used for classifying hyperspectral remote sensing images so as to obtain classified images based on pixels; the hyperspectral remote sensing images are preprocessed by means of mean shift, the processed images are segmented into a plurality of clustering results through a multi-target immune cloning clustering and wave band selection algorithm, and the optimal clustering results are selected from the clustering results so as to constitute a segmentation image; the obtained classified images and the obtained segmentation image are fused by means of the maximum voting rule so as to obtain a final result image. The hyperspectral image classification and wave band selection method based on multi-target immune cloning has the advantages that the very high accuracy rate and Kappa coefficient values can be obtained under the circumstance that few training samples exist, region consistency is well kept, the hyperspectral image classification and wave band selection method is suitable for multiple hyperspectral data, and parameters are adjusted easily and conveniently.

The invention discloses a method for segmenting a gray scale image based on the multi-objective particle swarm optimization algorithm, relates to the technical field of image processing, and mainly solves a problem that a traditional method is single in evaluation index, has high possibility of leading poor consistency of regions, and is disordered in edge. The method comprises the following steps: (1) extracting characteristics of a gray scale image to be segmented, namely selecting a gray scale of the image as data for a cluster to be clustered; (2) setting an operating parameter and initializing a particle swarm; (3) performing multi-objective clustering on the data combining the multi-objective particle swarm optimization algorithm to obtain a solution set approximate to Pareto; (4) selecting a solution with the largest PBM index from the solution set obtained in the step (3) as the optimal solution, namely a clustering result; (5) allocating a clustering mark to each pixel of the image to be segmented according to the clustering result obtained in the step (4) to obtain a segmenting result. The method has the advantages that the consistency of regions is good according to the segmenting result, complete information can be reserved, and the computing speed is quick, and can be used for image target identification.

The invention discloses an SAR (synthetic aperture radar) image segmentation method based on total-variation spectral clustering, and mainly solves the problems of low accuracy and poor regional consistency of the existing spectral clustering method when applied to SAR image segmentation. The method comprises the following steps of: (1) performing total-variation image decomposition on the input SAR image; (2) extracting the gray characteristic G of the structure part of the SAR image; (3) extracting the wavelet characteristic T of the texture part of the SAR image; (4) calculating the similarity Wij between any two points i and j in a total sample set S by use of the gray characteristic G and the wavelet characteristic T; (5) approaching to the characteristic vector space after the spectral sampling of the total sample set S through the sampling sample set Sm according to an Nystrom approaching method by use of the sampling sample set Sm selected from the total sample set S, and taking the characteristic vectors corresponding to the first k maximum characteristic values as a dimension-reducing characteristic matrix Y; and (6) performing k-means clustering on the dimension-reducing characteristic vector matrix Y, and outputting the final segmentation result of the SAR image. The method disclosed by the invention has the advantages of high segmentation result accuracy and good regional consistency, and can be applied to the SAR image target detection and target segmentation and recognition.

The invention discloses a stochastic gradient Bayesian SAR image segmentation method based on a sketch structure, mainly used for solving the problem that SAR image segmentation in the prior art is inaccurate. The stochastic gradient Bayesian SAR image segmentation method comprises the following implementation steps of: (1), sketching an SAR image to obtain a sketch image of the SAR image; (2), according to an area chart of the SAR image, and dividing a pixel subspace of the SAR image; (3), performing hybrid aggregation structured surface feature pixel subspace segmentation through a method based on a stochastic gradient variational Bayesian network model; (4), performing independent target segmentation based on the sketch line aggregation feature; (5), performing line target segmentation based on a visual semantic rule; (6), performing segmentation of a pixel subspace in a homogeneous area by adopting a polynomial-based logistic regression prior model; and (7), combining segmentation results to obtain a segmentation result of the SAR image. By means of the stochastic gradient Bayesian SAR image segmentation method based on the sketch structure disclosed by the invention, the good segmentation effect of the SAR image is obtained; and the stochastic gradient Bayesian SAR image segmentation method can be used for semantic segmentation of the SAR image.

The invention discloses a two-stage clustering-based SAR image semantic segmentation method, and mainly solves the problem of inaccurate SAR image segmentation in the prior art. The method comprises the following steps of 1) extracting a sketch chart according to a sketch model of an SAR image; 2) regionalizing the sketch chart to obtain a regional chart, and dividing the SAR image into a mixed aggregation structure ground object pixel subspace, a structure pixel subspace and a homogeneous pixel subspace according to the regional chart; 3) designing a directional statistics vector of an extremely non-homogeneous region in the mixed aggregation structure ground object pixel subspace; 4) performing segmentation on the mixed aggregation structure ground object pixel subspace according to the directional statistics vector; 5) performing corresponding segmentation on the structure pixel subspace and the homogeneous pixel subspace in sequence; and 6) combining segmentation results of the three subspaces to obtain a final segmentation result of the SAR image. According to the method, a good segmentation effect of the SAR image can be obtained; and the method can be used for target classification and image interpretation.

The invention discloses a K distribution and texture feature-based SAR (Synthetic Aperture Radar) image segmentation method, which belongs to the technical field of image processing. The segmentation method comprises the following steps: intercepting category C training samples on an SAR image to be segmented; intercepting 100 training samples of 9*9 at a feature region; extracting feature parameters of the training samples by using a K distribution statistical model and a gray scale co-occurrence matrix; arranging each category of feature coefficients into a matrix, i.e., a dictionary of cth category targets; calculating the dictionary of each category of targets and arranging the dictionaries to form a global big dictionary in sequence according to the method; inputting the SAR image to be segmented, substituting each pixel point by 9*9 pixel points of a neighbor of the pixel point, and solving a feature coefficient inverse solution matrix operation of the pixel point to obtain weight a; making delta i (a), i=1, ellipsis, wherein C is a vector which only remains the coefficient corresponding to the cth category in a and enable zero setting of the rest coefficients; and calculating a residual error function and repeating the steps to obtain a segmentation result of each pixel point of the SAR image to be processed according to the fact that the minimum error function is a category label of the feature coefficients.

The invention discloses a polarimetric SAR image classification method based on residual learning and a conditional GAN, and the method comprises the steps: (1), constructing a generator of the conditional GAN; (2), constructing a discriminator of the conditional GAN; (3), filtering a to-be-classified polarization SAR image; (4), performing pauli decomposition of a filtered scattering matrix; (5),normalizing a feature matrix; (6), generating a training data set and a test data set; (7), performing residual learning of deep and shallow features in the generator; (8), classifying features afterresidual learning; (9), obtaining a classification correctness rate; (10), training the generator of the conditional GAN; (11), classifying test data set. The method achieves the residual learning ofthe deep and shallow features of a polarimetric SAR image in the generator, achieves the extraction of the comprehensive feature information, achieves the good regional consistence of a classification result image, and is high in classification precision.

The invention relates to an image segmentation method based on geometric block spacing symbiotic characteristics and semantic information. The method comprises the steps that an initial sketch model is used for obtaining a sketch image, geometric blocks are constructed by taking line segments forming a sketch line as a unit, then the geometric blocks are mapped to the corresponding position of an original image, spacing symbiotic matrixes based on the geometric blocks are extracted, the spacing symbiotic matrix of each geometric block is taken as the characteristics of the corresponding line segments, the characteristics are utilized for dividing the sketch line into the speckle semantic category and the general boundary category, semantic information classified by the sketch line is respectively utilized for guiding the image super pixels obtained by the over-segmentation method to be combined, statistic is carried out on color mean values of the super pixels which are combined by being guided by the sketch line with the speckle semantic category, according to the symbiotic statistic relation between each super pixel and the super pixel in the neighbourhood in color, further combining is carried out, and the final segmenting result is obtained.

The present invention discloses a quick density peak value clustering based polarimetric SAR image classification method, mainly to solve the problem of low classification precision of an existing unsupervised polarimetric SAR classification method. The method comprises: 1. filtering an to-be-classified polarimetric SAR image; 2. performing Yamaguchi decomposition on the filtered polarimetric SAR image, and calculating four kinds of scattering power of each pixel point; 3. extracting main scattering power of each pixel point, and initially dividing the whole image into four classifications; 4. according to the main scattering power of all pixel points in an initial classification, dividing the whole polarimetric SAR image into M classifications; 5. using center points of the M classifications as new pixel points for clustering, and converting a clustering result into a pre-classification result of the whole image; and 6. performing iterative classification on the pre-classification result to obtain a final classification result. The experiment shows that the method provided by the prevent invention has a better classification effect, and can be used to perform unsupervised classification on various polarimetric SAR images.

The present invention discloses a sketch structure-based mean field variational Bayes synthetic aperture radar (SAR) image segmentation method. The method mainly solves the problem in the prior art that the SAR image is not segmented accurately. The method comprises a first step of sketching the SAR image, so as to obtain a sketch of the SAR image; a second step of dividing pixel sub-spaces of the SAR image according to an area chart of the SAR image; a third step of segmenting based on the pixel sub-spaces of a hybrid gathering structure of a mean field variational Bayes deduction network model; a fourth step of segmenting based on an independent target of a sketch line gathering feature; a fifth step of segmenting based on a line target of a visual semantic rule; a sixth step of segmenting based on a homogeneous area pixel sub-space of a polynomial logic regression prior model; and a seventh step of combining segmentation results, so as to obtain a segmentation result of the SAR image. Through adoption of the method, a good segmentation effect of the SAR image is obtained and can be used for semantic segmentation of the SAR image.

The invention discloses a polarimetric SAR image classification method based on strip waves and a convolution neural network, and mainly solves a problem that a polarimetric SAR is low in image classification precision in the prior art. The method comprises the following specific steps: (1), inputting a polarimetric SAR image; (2), extracting Pauli decomposition features; (3), constructing a characteristic matrix; (4), selecting a training sample and a test sample; (5), building an initial matrix; (6), initializing the convolution neural network; (7), training the convolution neural network; (8), testing the convolution neural network; (9), carrying out the coloring; (10), outputting a classification result image. Compared with the prior art, the method effectively improves the classification precision of the polarimetric SAR image.

The invention discloses a (synthetic aperture radar) SAR image segmentation technical method based on GMM (gaussian mixture model) parameter transferring and clustering. According to the method, the problems of unstable segmentation results, and unsatisfactory precision and area consistency in the prior art are mainly solved. The SAR image segmentation technical method comprises the following implementation steps of: 1) inputting an image, and determining segmentation class numbers; 2) extracting features; 3) setting initial parameters; 4) clustering a sample for seven times to acquire seven groups of clustering results; 5) solving clustering consistency values; 6) dividing the sample into a source domain and a target domain according to the clustering consistency values; 7) estimating the parameters of the source domain by using an EM (expectation-maximization) algorithm; 8) searching K neighbor points of the target domain sample in the source domain sample, and solving the clustering consistency values of the points; 9) solving new parameters of the target domain according to the consistency values of the K neighbor points and the parameters of the source domain; and 10) solvinga probability value of each sample of the target domain according to the new parameters, and acquiring a final segmentation result of the image. The SAR image segmentation technical method has the advantages of stable and good segmentation effect and high area consistency, and can be used for detecting and recognizing radar targets.

The invention discloses a texture image segmenting method based on a reinforced airspace-transform domain statistical model, belonging to the technical field of image processing and mainly solving the problems of poor segmenting accuracy and high calculating complexity in the prior texture image segmenting method. The method comprises the following segmenting steps: (1) inputting a texture image and the texture category number; (2) dividing the texture image into thick image blocks with the length and the width of 16 and carrying out two-dimensional discrete wavelet transform on each image block; (3) training an EHMM-HMT parameter and calculating a likelihood value and a segmenting result in a thick scale of the thick image block; (4) dividing the texture image into thin image blocks with the lengths and the widths of respectively 8, 4 and 2; (5) calculating a likelihood value and a segmenting result in each thin scale of the thick image block; and (6) combining the multiple dimensioned MAP fusion of boundary information. The invention has the advantages of high segmenting accuracy and low calculating complexity of the texture image and can be used for segmenting a microtexture image and a macrotexture image.

The invention discloses a polarization SAR image classification method based on object orienting and spectral clustering. The method mainly solves the problem that the accuracy rate of image classification of a polarization synthetic aperture radar SAR in the prior art is low. The method comprises the implementation steps that (1), filtering is carried out on coherence matrixes of polarization SAR data, and the coherence matrixes are used for synthesizing a color image of the polarization synthetic aperture radar SAR; (2), related parameters of the polarization synthetic aperture radar SAR are set; (3), by means of combination with the related parameters of the polarization synthetic aperture radar SAR, all pixels of the color image of the polarization synthetic aperture radar SAR are combined to form super-pixel blocks; (4), all the super-pixel blocks of the color image of the polarization synthetic aperture radar SAR are combined; (5), class centers of the combined super-pixel blocks are calculated; (6), spectral clustering is carried out on the class centers of the super-pixel blocks to finish final classification. According to the method, the influence of noise is overcome, the accuracy rate of classification is increased, and the method can be applied to terrain classification and target recognition.

The invention relates to the field of image processing and discloses a multi-scale color texture image segmentation method combined with an MRF (Markov Random Field) and a neural network. The method is mainly used for solving the problem that a simple Gauss model can not be used for accurately describing a feature field distribution characteristic in the traditional MRF method. The method comprises the following steps: when modeling a multi-scale feature field, estimating a probability distribution of a texture feature by using the output of the neural network in a supervision environment on each scale; when modeling a multi-scale mark field based on a classic Potts model, considering the interaction of a mark set of a same-scale second-order neighborhood position and a corresponding mark on a lower resolution scale for each pixel position; by using the maximum likelihood criterion, gradually acting a segmentation result on the lower resolution scale on the next scale from top to bottom to finally realize the multi-scale image segmentation. The segmentation result obtained by the invention has good region homogeneity and Boundary authenticity, and the method can be used for segmenting the color texture image.

The invention discloses an SAR image statistical distribution and DBN-based SAR image classification method, and mainly solving the problem that traditional deep confidence networks DBN are easy to cause bad area consistency and incomplete edge information when being used for SAR image classification. The method comprises the following steps of: preprocessing a to-be-classified SAR image so as toobtain a DBN input matrix; designing a DBN formed by three limited Boltzmann machines; pre-training the designed network by using the input matrix so as to obtain a trained DBN; randomly selecting a set of partial pixels with category labels from labels of SAR image category mark sheet, and carrying out micro-adjustment on the trained DBN by using a counter-propagation algorithm; and carrying outpixel-by-pixel classification on the to-be-classified image by using the micro-adjusted DBN so as to obtain a classification result, and coloring and outputting the classification result. The method has the advantages of being excellent in classification result, good in area consistency and complete in edge information, and can be applied to terrain classification and target recognition of SAR images.

The invention discloses a remote sensing image change detection method and device based on DT-CWT (dual-tree complex wavelet transform) and MRF (Markov random field). According to the method and device of the invention, the multi-directional expression and multi-scale analysis of image information can be facilitated; correlations between pixels are fully utilized; the retention of high-frequency information and the removal of noises can be effectively balanced; edge detection is smoother; change detection results have good regional consistency; a false detection rate can be greatly reduced, and the influence of registration errors can be removed; and the accuracy of remote sensing image change detection can be greatly improved.

The invention discloses a Freeman decomposition and homo-polarization rate-based polarized synthetic aperture radar (SAR) image classification method for mainly solving the problems of higher calculation complexity and poor classification effect in the prior art. The method comprises the following steps of: (1) inputting a covariance matrix of polarized SAR data; (2) performing Freeman decomposition on the input matrix to acquire three types of scattering power matrixes of plane scattering, dihedral angle scattering and volume scattering; (3) performing initial division on the polarized SAR data according to the three types of scattering power matrixes; (4) calculating the homo-polarization rate of all pixel points of the polarized SAR data of each class; (5) selecting a threshold value, and dividing the polarized SAR data of each class in the step (3) into 3 classes according to the homo-polarization rate, so that the whole polarized SAR data are divided into 9 classes; and (6) performing repeated Wishart iteration and coloring on the division result of the whole polarized SAR data to obtain a final color classification result graph. Compared with the classical classification method, the method has the advantages that the division of the polarized SAR data is stricter, the classification result is obvious and the calculation complexity is relatively low.

The invention discloses a polarization SAR (synthetic aperture radar) image classification method based on Freeman decomposition and PSO (particle swarm optimization), and mainly aims to solve the problems of higher computational complexity and poor classification effect in the prior art. The implementation steps are as follows: (1) inputting a covariance matrix of polarization SAR data; (2) performing Freeman decomposition on the input matrix to obtain three scattering power matrixes of plane scattering, dihedral angle scattering and volume scattering; (3) initially dividing the SAR data according to the three scattering power matrixes; (4) obtaining two threshold values of each category by virtue of the two-dimensional double threshold value Otsu method on the basis of QPSO (quantum-behaved particle swarm optimization); (5) dividing each initially divided category of the polarization SAR data into three categories, thereby dividing the whole polarization SAR data into 9 categories; (6) carrying out Wishart iteration and coloring on the division result of the whole SAR data to obtain a final color classification result image. Compared with the classical classification method, the polarization SAR image classification method based on Freeman decomposition and PSO is more rigorous in dividing the polarization SAR data, the classification result is obvious, and the computational complexity is relatively small.

The invention discloses an SAR image segmentation method combining background information and a maximum posterior marginal probability standard, belonging to the technical field of image processing, and mainly solving the problem that the image margins with small gray level difference can not be acquired completely and segmented accurately by the existing watershed segmentation method based on a regional hierarchical mode. The segmentation process comprises the following steps: carrying out mean sampling the original image to obtain a quadtree hierarchical structure; intercepting a training sample on the original image to carry out feature extraction; carrying out gradient corrected watershed segmentation on each other hierarchies except the coarsest hierarchy; combining the regions obtained by the watershed segmentation; regarding a parent region and an adjacent region as the background information; training parameters by using the maximum expected value; and calculating to obtain the posterior marginal probability P (xA|y) of the regions at each hierarchy, so as to obtain a final segmentation result. The method of the invention has the advantages of good regional consistency of the segmentation result and accurate margin, and can be used in SAR image target identification.

A polarimetric SAR image classification method based on residual learning and conditional GAN, the steps of which are: (1) constructing a conditional GAN ​​generator; (2) constructing a conditional GAN ​​discriminator; (3) to classify the polarimetric SAR image (4) Perform Pauli decomposition on the filter scattering matrix; (5) Normalize the feature matrix; (6) Generate training data set and test data set; (7) Residue the deep and shallow features in the generator Difference learning; (8) Classify the features after residual learning; (9) Obtain the classification accuracy rate; (10) Train the generator of the conditional GAN; (11) Classify the test data set. The invention performs residual learning on the deep and shallow features of the polarimetric SAR image obtained in the generator, and extracts comprehensive feature information, so that the area consistency of the classification result map is good and the classification accuracy is high.

The invention relates to the field of image processing and discloses a multi-scale color texture image segmentation method combined with an MRF (Markov Random Field) and a neural network. The method is mainly used for solving the problem that a simple Gauss model can not be used for accurately describing a feature field distribution characteristic in the traditional MRF method. The method comprises the following steps: when modeling a multi-scale feature field, estimating a probability distribution of a texture feature by using the output of the neural network in a supervision environment on each scale; when modeling a multi-scale mark field based on a classic Potts model, considering the interaction of a mark set of a same-scale second-order neighborhood position and a corresponding mark on a lower resolution scale for each pixel position; by using the maximum likelihood criterion, gradually acting a segmentation result on the lower resolution scale on the next scale from top to bottom to finally realize the multi-scale image segmentation. The segmentation result obtained by the invention has good region homogeneity and Boundary authenticity, and the method can be used for segmenting the color texture image.

The invention discloses a K distribution and texture feature-based SAR (Synthetic Aperture Radar) image segmentation method, which belongs to the technical field of image processing. The segmentation method comprises the following steps: intercepting category C training samples on an SAR image to be segmented; intercepting 100 training samples of 9*9 at a feature region; extracting feature parameters of the training samples by using a K distribution statistical model and a gray scale co-occurrence matrix; arranging each category of feature coefficients into a matrix, i.e., a dictionary of cth category targets; calculating the dictionary of each category of targets and arranging the dictionaries to form a global big dictionary in sequence according to the method; inputting the SAR image to be segmented, substituting each pixel point by 9*9 pixel points of a neighbor of the pixel point, and solving a feature coefficient inverse solution matrix operation of the pixel point to obtain weight a; making delta i (a), i=1, ellipsis, wherein C is a vector which only remains the coefficient corresponding to the cth category in a and enable zero setting of the rest coefficients; and calculating a residual error function and repeating the steps to obtain a segmentation result of each pixel point of the SAR image to be processed according to the fact that the minimum error function is a category label of the feature coefficients.

The invention relates to an image segmentation method based on geometric block spacing symbiotic characteristics and semantic information. The method comprises the steps that an initial sketch model is used for obtaining a sketch image, geometric blocks are constructed by taking line segments forming a sketch line as a unit, then the geometric blocks are mapped to the corresponding position of an original image, spacing symbiotic matrixes based on the geometric blocks are extracted, the spacing symbiotic matrix of each geometric block is taken as the characteristics of the corresponding line segments, the characteristics are utilized for dividing the sketch line into the speckle semantic category and the general boundary category, semantic information classified by the sketch line is respectively utilized for guiding the image super pixels obtained by the over-segmentation method to be combined, statistic is carried out on color mean values of the super pixels which are combined by being guided by the sketch line with the speckle semantic category, according to the symbiotic statistic relation between each super pixel and the super pixel in the neighbourhood in color, further combining is carried out, and the final segmenting result is obtained.