354results about How to "The segmentation result is accurate" patented technology

The invention discloses a three-dimensional image segmentation method and system based on a full convolutional neural network, and the method comprises the following steps: step1, collecting and obtaining a sequence image, and carrying out the marking of the sequence image, and obtaining training sample data; step 2, performing normalization preprocessing on the training sample data obtained in the step 1; step 3, applying the sample data processed in the step 2 to carry out supervised training on pre-constructed 3-D full convolution residual U-net network model, training to a preset convergence condition, and obtaining a trained three-dimensional image segmentation model; and step 4, after normalization processing is carried out on the sequence image data to be segmented, inputting the sequence image data into the three-dimensional image segmentation model trained in the step 3, and obtaining a sequence image segmentation result. The continuity information of the sequence can be fullyutilized, and a relatively good result can be obtained in three-dimensional image segmentation.

The invention relates to medical image processing and aims to provide a multiple organ segmentation method based on a deep convolutional neural network and a regional competition model. The multiple organ segmentation method based on the deep convolutional neural network and the regional competition model comprises processes of training a three-dimensional convolutional neural network; using the trained three-dimensional convolutional neural network to learn prior probability images of liver, spleen, kidney and background in CTA volume data; determining an initial segmentation region of each tissue according to the prior probability image of each tissue; determining the probability of each pixel point belonging to each of the four tissues in the image; establishing a multiple region segmentation model based on regional competition; solving the model using the convex optimization method; and performing post-processing to obtain the contour of each organ. The invention uses the convolutional neural network to automatically and rapidly detect positions of liver, spleen and kidney at the belly, thereby obtaining the prior probability image of each organ is obtained. Then the invention uses the regional competition model, so that the contours of liver, spleen and kidney can be accurately segmented.

The invention relates to a new method for extracting moving objects and partitioning multiple objects, in particular to a method for extracting moving passengers and partitioning images applied to bus passenger statistics. The method comprises the following steps of: acquiring images; preprocessing the acquired images; detecting and partitioning moving objects; tracking and calculating the moving objects and the like. A method for extracting the moving objects based on inter-frame difference of edge information can keep the profile of the moving objects, so subsequent processes, such as opening filling, evaluation of the central position of the objects and the like are more correct and more convenient. Multiple moving objects are partitioned by a projection method, so a partition result is correct, and the method is simple and convenient and has high instantaneity.

The invention relates to the field of medical image processing and aims to provide a fully-automatic three-dimensional liver segmentation method based on a convolution nerve network. The fully-automatic three-dimensional liver segmentation method based on the convolution nerve network comprises the following processes: preparing a training set, and training the convolution nerve network, processing CTA volume data of an abdominal liver by utilizing the trained convolution nerve network to obtain a segmentation result of the liver. The liver is segmented by means of the three-dimensional convolution nerve network; the three-dimensional liver segmentation method is fully automatic and can also prevent under-segmentation and over-segmentation phenomena well; and the accurate liver segmentation result is obtained.

The invention discloses a method for partitioning a genetic fuzzy clustering image and provides a method for partitioning a fuzzy clustering image on the basis of a genetic algorithm, which aims to solve the problem that a fuzzy C mean value algorithm is sensitive to noise and is easy to generate an overclosed clustering center due to noise influence. The partitioning method comprises the following steps of: firstly, carrying out noise resistant pretreatment on an original image by a gray level and neighborhood information; then obtaining an initially optimal clustering center by utilizing a genetic fuzzy clustering algorithm; and finally calculating the membership degree of each pixel in an image according to the obtained initially optimal clustering center by a histogram amendment clustering center of the image after noise resistance to obtain a partition result. The method adopts an improved gray level similarity function in the noise resistant pretreatment and ensures the noise resistant effect in noise with larger strength; and a clustering center distance punitive measure is added into the genetic fuzzy clustering algorithm, thereby the image with serious noise interference and a smaller target to be partitioned can be effectively partitioned, and the correct clustering center and an accurate partition result can be obtained.

The invention provides an automatic division method for pulmonary parenchyma of a CT image. According to the automatic division method, the CT is divided through carrying out a random migration algorithm for two times to obtain the accurate pulmonary parenchyma; in the first time, the random migration algorithm is used for dividing to obtain a similar pulmonary parenchyma mask; and in the second time, the random migration algorithm is used for repairing defects of the periphery of a lung and dividing to obtain an accurate pulmonary parenchyma result. Seed points, which are set by adopting the random migration algorithm to divide, are rapidly and automatically obtained through methods including an Otsu threshold value, mathematical morphology and the like; and manual calibration is not needed so that the working amount and operation time of doctors are greatly reduced. According to the automatic division method provided by the invention, a process of 'selecting the seed points for two times and dividing for two times' is provided and is an automatic dividing process from a coarse size to a fine size; and finally, the dependence on the selection of the initial seed points by a dividing result is reduced, so that the accuracy, integrity, instantaneity and robustness of the dividing result are ensured. The automatic division method provided by the invention is funded by Natural Science Foundation of China (NO: 61375075).

The invention discloses an image semantic segmentation model generation method and device, equipment and a storage medium. The method comprises the steps of obtaining an image sample set; using the image sample set to train an image semantic segmentation model; wherein the image semantic segmentation model comprises a feature map extraction part and a feature map analysis part, the feature map extraction part comprises a plurality of cascaded hole convolution processing residual error modules, and the feature map analysis part is constructed based on an attention mechanism, a pixel correlationmechanism and multi-scale information. According to the technical scheme provided by the embodiment of the invention, the attention mechanism is effectively utilized to learn the dependency relationship in spatial position and channel dimension, the feature expression capability is enhanced, the pixel correlation mechanism is utilized to make the segmentation result more accurate, and meanwhile,the multi-scale feature information is also utilized to learn a global scene to improve the accuracy of pixel point classification.

The invention relates to a method for automatically partitioning tree point cloud data. The method comprises the following steps: acquiring and preprocessing point cloud, estimating the direction by a point cloud process, constructing a local coordinate system, fitting a conicoid by using a closest point process, calculating the principal curvature by using the conicoid, defining and calculating the axial distribution density, distinguishing the branch point cloud and the leaf point cloud by using the axial distribution density, carrying out region growing on the branch point cloud, and carrying out region merging on the branch point cloud. By using the tree scanning data of a laser scanner and the estimated principal curvature, the invention partitions the tree scanning point cloud according with the actual organ distribution conditions. The method automatically partitions the tree point cloud scanning data among different organs through the local direction of principal curvature, and has the advantages of simple algorithm and accurate calculation result. The calculation result has important application value in the fields of tree point cloud 3D reconstruction, forest measurement, tree point cloud registration and the like.

The invention relates to an organ sketching method and device, a computer device and a storage medium. The method comprises the following steps: inputting a to-be-processed medical image into a firstdeep convolutional neural network to obtain a target area image, and inputting the target area image into a second deep convolutional neural network to obtain an initial segmented image; and obtainingan organ sketching result based on the initial segmentation image. In the organ sketching method and device, a computer device and a storage medium, a deep learning model is adopted, a universal framework is established, different types of organs can be effectively and automatically sketched, consumed time is short, adaptability is high, efficiency is high, meanwhile, a target area is located firstly, then the target organ is segmented, and a segmentation result obtained within a small range is more accurate and higher in accuracy.

The invention is suitable for the technical field of remote sensing image processing, and provides a large-scale remote sensing image sea-land segmentation method. The method comprises the steps: A, carrying out the sea-land filling of a remote sensing image, and obtaining a sea-land rough segmentation result image; B, carrying out the segmented extraction of the remote sensing image through combining with the sea-land rough segmentation result image, and obtaining a sea-land precise segmentation result image; C, carrying out the fusion of the sea-land rough segmentation result image and the sea-land precise segmentation result image, and obtaining a sea-land segmentation result image. The method achieves the automatic sea-land segmentation of the large-scale remote sensing image, and better solves problems in the prior art of sea-land segmentation that the sea-land segmentation is poor in precision and very bad in effect in offshore and offshore multi-island regions and a sea-land region with complex landform features.

The invention provides a polarized SAR (synthetic aperture radar) image segmentation method with space adaptivity, mainly solving the problem that a segmentation result can not reflect image detail information caused by insufficient space complexity adaptivity by using the conventional segmentation technique. The polarized SAR image segmentation method comprises the following steps of: firstly, combining H/alpha-ML Wishart cluster and quadtree disassemble to obtain initial segmentation areas with unequal sizes and capability of adapting scene complexity, and regulating the sizes and the shapes of the initial segmentation areas by utilizing complex Wishart distribution and an MRF (markov random field) to obtain a final segmentation result. The method fully and effectively utilizes polarization information of a polarization SAR image with the excellent space adaptivity, the segmentation result can be used for well reserving detail information of a polarization SAR image, the segmentation speed is relatively quick and the segmentation result is relatively exact.

The invention provides an automatic segmentation method for the pre-background of a lepidopteron image based on a full convolutional neural network (CNN). A full convolutional network applied to pixel-level classified prediction is built via a fine tuned and pre-trained CNN model. Before the network is training, data enhancement is performed on an insect image data set to meet the requirement of deep neural network training on a sample quantity. Outputs of the different convolutional layers are fused, and a network model applicable to segmenting the pre-background of the lepidopteron image isacquired by exploration. Edge details are refined by further using a condition random field (CRF) based on an initial segmentation result of the CNN, and the maximum outline of the foreground is extracted and filled to remove noise interference existing in an output result of the network model and cavities in the foreground. According to the method, the pretreatment task of the insect image is completely automatic, and the lepidopteron species automatic recognition efficiency can be significantly improved.

The invention discloses a captive test (CT) image partitioning method based on adaptive learning. The CT image partitioning method comprises the following steps of: 1) acquiring a CT image; 2) extracting characteristics of the CT image; 3) inputting strokes for representing a lesion region and a non-lesion region on the CT image by a user; 4) constructing a region model of the image by taking the strokes input by the user as a basis according to the extracted characteristics of the CT image, and constructing an edge model of the image by adopting an edge detection method; and 5) combining the region model and the edge model to construct a new model, calculating the new model to obtain a partitioning result. By adopting the method, a difference between the lesion region and the non-lesion region on the CT image can be effectively described; the method adapts to the complexity of the CT image; the problems caused by low signal-to-noise ratio (high noise) of the CT image are solved; the user can quickly and precisely partition the lesion region on the CT image in an interactive manner with high efficiency; and therefore, the production efficiency of a medical department can be greatly improved.

The invention relates to a multi-target picture segmentation based on a level set, and the multi-target picture segmentation based on the level set is applied in the technical field of the image analysis. The method includes the following steps: firstly, drawing one closed curve or more than one closed curves on to-be-segmented images as an initial outline; secondly, utilizing an initiative outline model based on a region, carrying out an iteration revolution on the initiative outline, and obtaining a profile curve of a target. The initiative outline model based on the region fully takes partial gray level information of the image into account. Therefore, the multi-target picture segmentation based on the level set is capable of segmenting images with uneven gray levels, extending the model into multiple orders, and segmenting over the multi-target images. Compared with the prior art, the multi-target picture segmentation based on the level set has the advantages of being insensitive in initialization, quick in calculation speed, strong in anti-noise ability and accurate in segmentation results. Meanwhile, the multi-target picture segmentation based on the level set is capable of segmenting medical images provided with a plurality of different targets.

The invention relates to an image segmentation method based on improved intuitionistic fuzzy C-means clustering, belonging to the image segmentation field. Firstly, an improved non-membership functionis proposed to generate intuitionistic fuzzy sets, and a method based on gray feature is proposed to determine the initial clustering center, which highlights the role of uncertainty in intuitionistic fuzzy sets and improves the robustness to noise. Secondly, the improved nonlinear function is used to map the data to the kernel space in order to measure the distance between the data points and the clustering centers more precisely. Then we introduce the local space-Gray level information, considering membership degree, gray level features and spatial position information. Finally, the intuitionistic fuzzy entropy in the objective function is improved, and the fuzziness and intuitionism of the intuitionistic fuzzy set are considered. The invention can effectively overcome the influence ofnoise and blur in the image on the algorithm, improves the segmentation performance of the algorithm, the pixel clustering performance and the robustness, is suitable for various different types of gray-scale images, and can obtain more accurate segmentation results.

The invention relates to a fundus image blood vessel segmentation method based on Frangi enhancement and an attention mechanism UNet, and the method comprises the steps: firstly extracting a green component from an input image, and carrying out the contrast adjustment on the basis of the extracted green component through a contrast-limited histogram equalization method; calculating a Hessian matrix of each pixel point in the image after the contrast ratio is adjusted; constructing a Frangi vascular similarity function by utilizing the characteristic value of the Hessian matrix under the condition of a scale factor, and obtaining the maximum response; respectively subtracting the product of the maximum response value and the enhancement factor factor factor from the pixel values of the RGBthree same channels of each pixel point of the input image; then, carrying out gray scale transformation on the image after frangi enhancement, and carrying out zero mean normalization operation on each pixel value to be between [0, 1]; and finally, inputting the obtained training image blocks and label image blocks into an attention mechanism UNet network for training; and obtaining a segmentation result through testing. According to the invention, the generalization ability of the model is improved.

The invention provides a definition circle HSV color space based medical image segmentation method and a cancer cell identification method. The specific process comprises: step 1, finding out RGB values and position information of a slice image target color pixel P and a background color pixel Q in an RGB color space; step 2, converting an RGB color space based slice image to the HSV color space to obtain an HSV color space based image; step 3, according to position information of the stored pixel P, taking (H,S) corresponding to the pixel P as circle center coordinates of a definition circle, and setting a radius of the definition circle; according to the position information of the pixel Q, extracting H, S and V values corresponding to the pixel Q, assigning the values to all pixels in the definition circle, and removing a target color; and step 4. converting the HSV color space based slice image subjected to removal of the target color back to the RGB color space, and then segmenting the slice image subjected to removal of the target color. By utilizing the definition circle HSV color space based medical image segmentation method and the cancer cell identification method, an extremely accurate segmentation result can be obtained.

The invention belongs to the field of biomedical image processing and specifically relates to a PET-CT lung tumor segmentation method combining a three dimensional graph cut algorithm with a random walk algorithm. The method comprises the following steps of: performing linear up-sampling on an original PET image and performing affine registration on PET and CT images; calibrating tumor seed points and non-tumor seed point; performing random walk algorithm segmentation on the PET image in combination with the tumor seed points; acquiring a foreground target area Ro completely including a target lung tumor area, using the areas, except the Ro, as a background area Rb of a non-lung tumor area; establishing gauss mixture models for the foreground area Ro and the background area Rb separately; computing energy items according to the gauss mixture models of the foreground area Ro and the background area Rb and obtaining a final segmentation result by using an graph cut algorithm. The method fully utilizes the function information and the PET image and the structure information of the CT image, enables complements between the random walk algorithm and the graph cut algorithm, and achieves an accurate lung tumor segmentation result.

The invention relates to a video object tracking cutting method using a Snake profile model, comprising the following steps: based on a space-time fusion method, roughly locating a Snake profile at a time field passes through a sectional frame core vector prediction manner, and then evolving from the initial profile by using a modified Snake greedy method in an air field to obtain a precise profile of the video object. The method comprises the following specific steps: dividing a video sequence into cut units with every four frames as a unit in the time field; and selecting the two front frames in one unit as key frames, wherein the initial profile is an external rectangle of the movement area obtained through the detection of movement change, and the initial profiles of the third and fourth frames are obtained by mapping the previous frame of the premise profile and the two previous frames of movement vector reflections. In the air field, during profile point iteration updating, large errors are considered, the impossible profile point is eliminated in real time. Compared with the prior art, the method has the advantages that the disadvantages of manually drawing the initial profile are overcome and high precision and rapid speed are achieved.

The invention provides a method for automatically segmenting consistent damage areas for buddha figure Thangkas. The method includes the steps of conducting vertical projection for a head light area of a Thangka image, obtaining an image symmetric axis by using a one-dimensional function symmetry detection method, obtaining an initial segmentation result by employing a partitioning segmentation method based on the symmetric axis, obtaining the image covered by the damage areas, extracting texture characteristics by using Gabor transformation, constructing a multi-scale and multi-characteristic set in combination with Lab space color characteristics, employing KNN classification to obtain a secondary segmentation result, further refining the damage areas by using morphological operation, and removing small damage areas to finally obtain a template of consistent damage areas. Through the method, various large-scale linear and block falling areas in the buddha figure Thangkas can be automatically segmented, the segmentation speed is fast, the efficiency and accuracy are high, and rapid automatic segmentation of the damage areas of buddha figure Thangkas can be realized.

The invention discloses a level set SAR (Synthetic Aperture Radar) image segmentation method based on local and global area information, which mainly solves the problem that the existing level set method is influenced by speckle and cannot segment the SAR images with uneven gray. The method comprises the following implementation steps of: firstly, initializing a level set function phi, and segmenting the SAR image into an internal area omega 1 and an external area omega 2; secondly, convolving intensity information of the internal area and the external area of the image through a Gaussian Kernel Function, taking the convolved information as local area information, and forming an energy term based on the local area; then, solving intensity mean values c1 and c2 and probability densities p1and p2 of the internal area and the external area, and forming the energy term of the global area; and finally, adding a bound term L (phi) of level set length and a penalty P (phi) which avoids renewed initialization, forming a total energy function ESAR, solving a gradient sinking equation through a variation method, and updating the level set phi. The obtained new segmented area and the experimental result show that the segmentation method provided by the invention can get more ideal segmentation effects, and the method can be used for SAR image segmentation and target identification.