52results about How to "Good image segmentation" patented technology

The invention discloses a steel rail surface defect image adaptive segmentation method. The method comprises the following steps of S1, extracting a steel rail region by adopting a row grayscale mean successive summation method; S2, preprocessing a steel rail region image; S3, performing structure region and non-structure region division on the steel rail region image; S4, further distinguishing a defective region and a shadow region by utilizing a non-local feature of the image in the structure region; S5, adaptively building a background image model according to different features in the image; S6, performing image difference; and S7, performing dynamic threshold segmentation. The image is divided into the structure region and the non-structure region by utilizing image local information, the size of a pixel neighborhood window is adaptively adjusted by utilizing non-local information to calculate a mean, the accurate background image model is built, and the image difference and the dynamic threshold setting are performed, so that while a defective part of the image is highlighted, the influence of uneven illumination and steel rail surface reflection property on steel rail surface defect detection is effectively reduced, an ideal image segmentation effect is achieved, and the rail surface detection precision is ensured.

The invention discloses a remote sensing image segmentation method based on region clustering, belonging to the field of remote sensing image comprehensive utilization. The method comprises the following steps: carrying out region pre-segmentation by a MeanShift algorithm to remove noise and perform initial cluster on image elements; carrying out fuzzy clustering on images which are subject to the pre-segmentation by a fuzzy C-means algorithm (FCM), and initially inducing and identifying characteristics of each image object to obtain the probability that each object affiliates to some a category so as to constitute a land category probability space of the remote sensing images, thereby providing a basis of object combination for further region segmentation; and performing region segmentation in the probability space of clustering images, classifying image elements which are close in the probability space and similar in the category as the same objects by region labels. In the method of the invention, two defects in the existing segmentation method are overcome, the remote sensing images can be effectively and accurately segmented, segmentation tasks of the remote sensing images can be finished by batch by integration, and data support can be preferably provided for extraction of land information from the remote sensing images.

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

The invention discloses an image division method based on a watershed-quantum evolution clustering algorithm. The method has the following processes: (1) blocking and processing an input image to be divided, and seeking characteristics of a regional block as a clustering dataset; (2) setting population scale, number of distinct categories k and halt conditions, and randomly generating an initial quantum chromosome Q (t) as an initial clustering center; (3) observing the Q(t) to be a binary system chromosome p (t), calculating a fitness function value f<k> of each chromosome and reserving individuals in the current group; (4) carrying out mutation operation on the Q (t) to obtain Q<M>(t); (5) quantum crossover Q<M>(t) to obtain the Q<C>(t); (6) observing the Q<C>(t) to be a binary system chromosome p <c> (t) to obtain offspring chromosomes; (8) judging the halt condition of the offspring chromosomes, dividing an image kind which the chromosome with the highest affinity degree in the offspring chromosomes is corresponding to as output results if the halt condition is met, otherwise returning the process (3). The method has the advantages of good regional consistency, accurate edge preservation and can be used for target recognition in the image processing field.

The invention discloses a fuzzy clustering image segmenting method with transfer learning function. The method adopts the classic fuzzy C fuzzy-means algorithm as the study object and is special for overcoming the shortcoming of the C fuzzy-means algorithm that the low capacity is provided for resisting the noise while facing the image with noise. During processing the new image, the image segmenting method is mainly carried out for the image with noise pollution. With the adoption of the fuzzy clustering image segmenting method disclosed by the invention, the reliable clustering information obtained by summarizing lots of past similar images under the C fuzzy-means algorithm can be effectively learnt and utilized, such information is always considered as the clustering centre; by introducing the reliable information into the current new image segmenting task, the current clustering task can be effectively guided, and the noise resisting effect can be achieved, therefore, more precise clustering centre and more precise image segmenting result can be obtained.

The invention discloses an infrared camouflage material, which is characterized by the following: the infrared hidden paint is compound of launching rate less than 0.63 green visible dyes, 0.76-0.78 yellow earth visible dye and more than 0.92 deep green visible dyes. The paint possesses mechanic property, which is compatible with visible camouflage property.

The invention discloses an unsupervised image division method based on a multi-target immune cluster integration technology, which mainly solves the problems of poor global optimization capability, single evaluation index, single division scheme form and difficult selection of a plurality of division schemes in the traditional technology. The method comprises the following steps of: (1) extracting gray scale information and wavelet energy information of an image to be divided; (2) sampling the image by using an area-based sampling policy to generate a test sample set; (3) selecting different characteristic vectors to form a plurality of test sample sets; (3) generating a primary division scheme by using an evolution cluster based on a multi-target immune algorithm; (5) integrating and learning an optimal division scheme in the primary division scheme; (6) marking the class of the pixel points of the image according to the selected division scheme; and (7) outputting the image division result. The invention has advantages of high average accuracy in image division and strong robustness and is applicable to the obtaining of image information and the division of image texture.

The invention provides a GrabCut texture image segmentation method based on Gabor transform. The method comprises the following steps that: transforming a two-dimensional image into frequency domain space representation through Gabor wavelet transform, calculating the parametric transform of a frequency spectrum space two-dimensional Gabor filter group to obtain multi-scale multi-direction texture features, constructing a multi-scale texture energy generic function, calculating the GMM (Gaussian Mixture Model) probability distribution model of the texture features, and introducing the GMM of the texture features into GrabCut for image segmentation. The GrabCut texture image segmentation method based on the Gabor transform has the characteristics of being simple in artificial interaction, high in processing speed, good in image segmentation effect and the like.

The invention belongs to the technical field of image processing, and discloses a porous material hole automatic measurement method based on scanning electron microscope image segmentation, which comprises the following steps of 1, initializing parameters; 2, filtering processing; 3, obtaining a potential clustering number through the characteristic value information; 4, calculating histogram information of the filtered image; 5, completing image segmentation by using a fast fuzzy clustering algorithm; 6, extracting data distribution corresponding to the minimum clustering center as hole output to obtain a binary hole image; 7, performing morphological filling operation on the binary hole image; 8, automatically classifying the holes according to areas by utilizing multi-scale morphological closed reconstruction; and 9, counting the number, the area ratio and the actual average area of holes of different grades in the porous material. According to the method, automatic measurement andclassification of the holes of the porous material are realized, the porous material can be objectively evaluated, and an ideal image segmentation effect can be directly obtained.

The invention discloses an immunity chromatography test strip quantitation detection method based on a deep reliability network. The method comprises the following steps of collecting several immunity chromatography test strip images of different concentration sample liquids as training images and extracting a target area including a detection line and a quality control line after pretreatment; taking a pixel as a sample unit, selecting a proper network input characteristic quantity and calculating an input quantity of each sample so as to acquire the training sample; constructing the deep reliability network based on a restricted Boltzmann machine, inputting the training sample and completing training of the deep reliability network; preprocessing a sample liquid test strip image to be detected , calculating an input characteristic quantity and acquiring a test sample; inputting the test sample into the trained deep reliability network so as to acquire an image segmentation result of a sample liquid to be detected; and according to the image segmentation result, calculating a characteristic quantity and acquiring a quantitative detection concentration value. By using the method in the invention, a good image segmentation result can be acquired, concentration identification accuracy of the sample liquid to be detected is increased, and high applicability and robustness are possessed.

The invention relates to the technical field of medical image processing and particularly relates to a cell counting method based on skeleton extraction, which comprises the following steps: S1, obtaining a to-be-processed histopathology image, carrying out graying processing on each to-be-processed image, and then carrying out image segmentation on each grayed image to obtain a cell binary imageof each to-be-processed image; S2, performing morphological processing on each cell binary image to obtain a cell binary image with intracellular hole filling and cell edge noise and impurity noise point removal; And S3, performing cell skeletonization processing on each morphologically processed cell binary image to obtain the number of cell skeletonization nodes in each image, and obtaining a cell counting result of each to-be-processed image according to the relationship between the number of cell skeletonization nodes in each image and the number of cells. The obtained cell counting resultis good in accuracy, and the method is high in operation efficiency.

The invention relates to a preprocessing method and device for the image recognition of a DNA sequence, and the method comprises the steps: obtaining map information, and respectively obtaining a DNA map in a sampling time interval of each map; carrying out the gray scale linear stretching of the maps, and obtaining stretched DNA maps; obtaining a first pixel and a second pixel of each DNA map after stretching; calculating a global threshold T of gray scale mean values of the first and second pixels; calculating the variance sigma2 of the first and second pixels; and carrying out the segmentation of the maps through employing the global threshold T if the variance sigma2 is within a preset range. The method can achieve purposes of enabling a target and background to be easy to recognize and avoiding the wrong recognition of the target during the collection of an image of a reaction chip. Moreover, the method is short in operation time, is good in image segmentation effect, improves the accuracy of image recognition after the collection of the image of the reaction chip, and precisely determines the type of a basic group.

The invention provides an immunity cloning and multitarget optimizing texture image segmentation method. The method is mainly used for solving the problem in the prior art that the segmentation effect is poor caused by the fact that only spatial separation degree or category compactness is optimized. The method comprises the implementation steps: (1) reading a texture image and extracting a characteristic matrix G from the texture image; (2) generating an initial antibody group V(t) and carrying out initial setting; (3) calculating a clustering objective function f1 and a categorization objective function f2 according to the characteristic matrix G and the antibody group V(t); (4) carrying out immunity cloning operation on the antibody group V(t) so as to obtain a cloned antibody group Vc(t); (5) carrying out non-uniform mutation operation on the cloned antibody group Vc(t) so as to obtain an antibody group Vm(t) subjected to non-uniform mutation; (6) carrying out population updating operation on the antibody group Vm(t) subjected to non-uniform mutation so as to obtain an updated antibody group Vm(t+1); and (7) calculating the categories of all pixels in the texture image according to the updated antibody group Vm(t+1) and the characteristic matrix G. The method has the advantages of high segmentation efficiency and good image segmentation effect and can be used for extracting and obtaining detailed information on the texture image.

The invention discloses a texture image segmentation method based on Lamarck multi-target immune algorithm, mainly aiming at solving the problems of high operational data quantity, weak global optimization capability, one-sided evaluation index and poorer local searching capability in the prior art. The texture image segmentation method comprises the steps of: (1) extracting image grey-scale information and image small wave energy information; (2) based on watershed pre-segmentation, generating a test sample set for image sampling; (3) using the Lamarck multi-target immune algorithm for carrying out data clustering on the test sample set, and generating a data clustering scheme sets (4) according to the Minkowski index value, selecting the most satisfied data clustering scheme; (5) according to the selected data clustering scheme, marking image pixel point category attribution; and (6) outputting the image segmentation result. The texture image segmentation method has the advantages of low operational data quantity, lower calculation complexity, high image segmentation average accuracy rate and excellent segmentation result, and can be used for image information acquisition and image texture partition.

The invention discloses an image segmentation method and system based on dynamic multi-objective optimization. The method comprises the steps of: constructing a multi-objective function by using a K-Means algorithm and a Fuzzy C-means (FCM) algorithm; defining an environmental change rule by using a background difference method; constructing a self-adaptive inertial dynamic factor; optimizing a timely mutation factor; based on the self-adaptive inertial dynamic factor and the timely mutation factor, dynamically optimizing the multi-objective function by using a multi-objective optimization particle swarm method. The K-Means and the FCM are optimized by dynamically optimizing a particle swarm algorithm, thus a good aggregation result can be acquired; the problem of error segmentation of pixels or edge blur is avoided, thus a good image segmentation effect can be achieved; the image segmentation is high in accuracy rate; the image segmentation method and system based on dynamic multi-objective optimization can provide high-quality result data and technical reference for image recognition.

The invention discloses an image segmentation method based on a structure tensor and an image segmentation model, and belongs to the field of digital image processing. The method comprises the steps: taking an image as a hypersurface of a three-dimensional Euclidean space according the viewpoint of Riemannian geometry, and obtaining a classic ST (structure tensor); combining the obtained ST with the color information of the image, and obtaining an EST (extended structure tensor); carrying out the dimension reduction of the obtained EST through employing PCA, and obtaining a constrictive CST; carrying out the non-linear diffusion of the obtained CST through employing a vectorization mode of a PM equation; calculating the KL distance of two tensor spaces; substituting the obtained distance into a GrabCut model, and completing the segmentation of the image. The method is small in number of parameters, is simple in calculation, is high processing speed, is good in image segmentation effect, and is suitable for a condition that a to-be-segmented object is highly similar to the background.

The invention discloses a fast superpixel segmentation method which is a top-down iterative segmentation method. The divisibility of the region is judged by analyzing the color consistency of the regional pixel set according to the method. The method comprises the following steps that step 1: the corresponding grayscale image I<Gray> of the original color image is acquired according to the formula(1); I<Gray>(x,y)=[I<R>(x,y)+ I<G>(x,y)+ I(x,y)] / 3, wherein I<Gray>(x,y) refers to the pixel grayscale of the coordinates (x,y) on the image I<Gray>, I<R>(x,y) refers to the pixel grayscale of thecoordinates (x,y) on the image I<R>, I<G>(x,y) refers to the pixel grayscale of the coordinates (x,y) on the image I<G>, and I(x,y) refers to the pixel grayscale of the coordinates (x,y) on the image I; and I<R>, I<G> and I are the red, green and blue channel images of the original color image; and step 2: the superpixel region set S is calculated according to the grayscale image I<Gray>.

A bacteria microscopic image segmentation method based on a deep learning network comprises the following steps: 1) culturing bacteria, shooting a group of bacteria growth pictures under a microscope according to a fixed time interval, carrying out image preprocessing, constructing a training set, a verification set and a test set which are not intersected with one another, wherein the training set comprises original images and corresponding label images, and the verification set and the test set only comprise the original images respectively; (2) building a U-Net + + model, wherein the U-Net + + model is provided with an encoder module and a decoder module, the encoder module carries out feature extraction, the decoder module carries out feature reduction decoding to obtain the size of an original image; inputting the training set into the U-Net + + model for training, then inputting the verification set into the trained U-Net + + model for verification, and obtaining the trained U-Net + + model; and 3) inputting the test set into the trained U-Net + + model, and outputting a binary segmentation image. According to the method, the bacterial microscopic image can be automatically segmented quickly and accurately, too many complex image preprocessing links in the early stage are omitted, and time is saved.

The invention discloses a threshold segmentation method with minimal clustering distortion. Firstly, according to the set threshold, the image is segmented into a target part and a background part; then, the sum of the value clustering segmentation distortion in the target part and the background part is calculated when segmentation is carried out according to the set threshold; and the above process is repeated on all gray levels of the image, and the gray level corresponding to the minimal sum of the segmentation distortion is found out to be the evaluated threshold. When the method of the invention is adopted, segmentation is more accurate to carry out, and the segmentation distortion is minimal.

The invention relates to a segmentation method based on a reaction diffusion equation, characterized in that under a level set framework, based on a reaction diffusion mechanism, utilizing a diffusion term to smoothen a level set function; and through analyzing image properties and characteristics, constructing a reaction term having opposite signs inside and outside an image target. The equation can be realized based on a simple finite difference method; in addition, the level set function can be initialized to be any bounded function, thereby well solving the problem of contour initialization. The segmentation method can obtain sound image segmentation effects, and possesses the advantages of simple implementation and fast operation speed.

The invention discloses an artificial intelligence global optimization method based on a QUATRE architecture. The artificial intelligence global optimization method comprises the steps of: S1, establishing a quasi-affinity transformation evolution architecture which is shown in description; S2, setting an evaluation function, a global optimal value, an iteration threshold value, a maximum number of evaluation function calling times and a condition for carrying out iteration of the quasi-affinity transformation evolution architecture, and initializing the quasi-affinity transformation evolution architecture; S3, judging whether the condition for carrying out iteration is met, if so, continuing the execution of following steps, and if not, stopping the execution; S4, calculating a variation matrix B, and acquiring a next generation particle population X<G+1> of a particle population X<G> according to the quasi-affinity transformation evolution architecture; S5, updating the number of evaluation function calling times and the next generation particle population; S6, acquiring a Boolean evolution matrix M and a correlated evolution matrix M-bar, acquiring a current optical particle in the particle population in (G+1)th iteration according to an optimal value of the evaluation function, and returning to the step S3 for judgment after making G=G+1. The artificial intelligence global optimization method based on the QUATRE architecture has the advantages of few parameters and high efficiency.

The invention provides an image segmentation method and system, a medium and an electronic terminal. The method comprises the steps of obtaining to-be-segmented training images; inputting the paired to-be-segmented training images into a positioning network, and obtaining a positioning feature map, wherein the step of obtaining the positioning feature map comprises the substeps that the paired training images to be segmented are subjected to same point feature map extraction and overall attention feature map extraction, and the positioning feature map is obtained according to the overall attention feature map; inputting the positioning feature map into a segmentation network for training to obtain an image segmentation model; inputting the paired tumor images to be segmented into the image segmentation model, and performing tumor image segmentation. According to the image segmentation method, the paired training images to be segmented are input into the positioning network, the same semantic features are extracted, different-point semantic features can also be extracted, the obtained positioning feature map is input into the segmentation network for training, the image segmentation model is obtained, cross-image semantic extraction is achieved, and the segmentation accuracy is high.

The invention discloses a loess microstructure image processing method which comprises the following steps: step 1, preparing a loess sample, and acquiring an image of the loess sample; step 2, preprocessing the image; step 3, enhancing the image; 4, segmenting the image; and step 5, performing morphological processing on the image. According to the method, the enhancement effect is obvious through the comprehensive use of the average filter and the Laplacian filter. The gray level of the microstructure image can be expanded through a histeq function, and later measurement and analysis are facilitated; the image segmentation effect is obvious by using the maximum entropy algorithm of the mathematical statistics capability of the information amount.

The invention discloses an image segmentation network and an image segmentation method. In the image segmentation method, a welding pool image is input into an image segmentation network for image segmentation; wherein the image segmentation network is based on a Unet structure, the image segmentation network comprises a multi-layer feature extraction structure and a multi-layer feature fusion structure, the feature extraction structure is in jump connection with the feature fusion structure through an attention module, and the attention module focuses on target features of an image and filters redundant information generated in jump connection; the feature extraction structure comprises a plurality of branches and a feature splicing module; a plurality of branches extract features from different scales according to input information, and enrich semantic information; therefore, the image segmentation network has a better image segmentation effect.

The embodiment of the invention discloses an image segmentation method and device, electronic equipment, a processor and a computer readable storage medium. The method comprises the steps that U-shaped expansion is conducted on a preset YOLOv3 model, an improved YOLOv3 model containing a U-shaped sampling unit is obtained, the U-shaped sampling unit comprises a down-sampling network and an up-sampling network, the down-sampling network is composed of a Darknet53 network of the YOLOv3 model, and the up-sampling network is formed by connecting a plurality of up-sampling modules in the YOLOv3 model in series; training the improved YOLOv3 model by using a sample image of a preset training sample set to obtain an image segmentation model; and performing image segmentation processing on the to-be-processed image by using the image segmentation model. According to the embodiment, the Darknet53 is adopted as the down-sampling network, the depth of the sampling network is large, the model expression ability is high, the whole model is easy to train, and then the method can be suitable for complex image segmentation scenes.

The invention discloses an unsupervised image division method based on a multi-target immune cluster integration technology, which mainly solves the problems of poor global optimization capability, single evaluation index, single division scheme form and difficult selection of a plurality of division schemes in the traditional technology. The method comprises the following steps of: (1) extracting gray scale information and wavelet energy information of an image to be divided; (2) sampling the image by using an area-based sampling policy to generate a test sample set; (3) selecting different characteristic vectors to form a plurality of test sample sets; (3) generating a primary division scheme by using an evolution cluster based on a multi-target immune algorithm; (5) integrating and learning an optimal division scheme in the primary division scheme; (6) marking the class of the pixel points of the image according to the selected division scheme; and (7) outputting the image divisionresult. The invention has advantages of high average accuracy in image division and strong robustness and is applicable to the obtaining of image information and the division of image texture.

The invention proposes an image edge detection method based on threshold value segmentation, which is a new threshold value segmentation algorithm for taking threshold values ​​one by one for all points in the image and classifying them. Using the statistical information of the mean value of the gray value of the pixel neighborhood in the image as the threshold setting standard of the point, and introducing the variance of the gray value of the pixel in the neighborhood of the point as an additional judgment condition, so as to realize the global binarization that comprehensively considers the regional information, so that The extracted target points are the edges of the image. Moreover, the size of the structural elements for calculating the mean and variance can be adjusted, and structural elements of different sizes can be selected according to different needs. This method sets a threshold value for each pixel, so as to realize the global binarization comprehensively considering the area information, the image segmentation effect is good, and the edge location is accurate.

The invention discloses a dynamic video image segmentation method based on dual-channel convolution kernel and multi-frame feature fusion, and the method comprises the following steps: 1, converting an original image into a gray level image, violently extracting edge features through edge pixel transformation, and obtaining an edge feature image; 2, performing edge feature screening on the edge feature images through the dual-channel convolution kernels of different sizes, and performing multiplication operation on the two screened images to obtain edge images; 3, constructing two types of multi-frame feature target extraction; 4, obtaining a segmented image through filling and restoring operation. According to the method, the image can be effectively segmented through a double convolutionkernel and multi-frame feature fusion method, a complete target image is obtained without much impurity interference, the defect that a conventional edge segmentation boundary is not closed or continuous is overcome, multi-frame feature target extraction is dynamically updated, impurity accumulation is reduced, and a good image segmentation effect is obtained.

The invention discloses an image segmentation method for tail section pore feature extraction of a sintering machine, and belongs to the technical field of sintering production. A CCD image acquisition unit is used for performing image acquisition on the tail section of the sintering machine; and the acquired image is sent to a computer image processing unit, the image is processed by adopting image gray processing and a multi-threshold maximum between-class variance method of a three-time iteration RGB color channel, and the pore characteristics of the tail section are segmented from the image. According to the characteristics of the section of the tail of the sintering machine, the pore characteristics of the section of the tail of the sintering machine are extracted to the greatest extent by performing gray processing on the acquired image and performing multi-threshold maximum between-class variance method of the RGB color channel iteration for three times, so that the pore proportion in the image can be conveniently counted; and therefore, an operator can accurately judge the FeO content in sintered ore.