531 results about "Image segmentation algorithm" patented technology

A computer based system and method of visualizing a region using multiple image data sets is provided. The method includes acquiring first volumetric image data of a region and acquiring at least second volumetric image data of the region. The first image data is generally selected such that the structural features of the region are readily visualized. At least one control point is determined in the region using an identifiable structural characteristic discernable in the first volumetric image data. The at least one control point is also located in the at least second image data of the region such that the first image data and the at least second image data can be registered to one another using the at least one control point. Once the image data sets are registered, the registered first image data and at least second image data can be fused into a common display data set. The multiple image data sets have different and complimentary information to differentiate the structures and the functions in the region such that image segmentation algorithms and user interactive editing tools can be applied to obtain 3d spatial relations of the components in the region. Methods to correct spatial inhomogeneity in MR image data is also provided.

The invention discloses a target identification and capture positioning method based on deep learning, and belongs to the field of machine vision. The method comprises the following steps that: firstly, utilizing a Kinect camera to collect the depth and the color image of a scene; then, using a Faster R-CNN (Regions with Convolutional Neural Network features) deep learning algorithm to identify ascene target; according to an identified category, selecting a captured target area as the input of a GrabCut image segmentation algorithm; through image segmentation, obtaining the outline of the target so as to obtain the specific position of the target as the input of a cascade neural network for carrying out optimal capture position detection; and finally, obtaining the capture position and the capture gesture of a mechanical arm. Through the method, the instantaneity, the accuracy and the intelligence of target identification and positioning can be improved.

The invention discloses a segmentation method for an image with deep image information, which is high in segmentation accuracy, and capable of still achieving a good segmentation effect under the condition of a very similar front background. The segmentation method comprises the steps of: (1) obtaining the image with deep image information via Kinect; (2) performing probability modelling on the colour information of the front background and the deep image information; (3) performing parameter estimation on the model by EM (Expectation-Maximization) algorithm; and (4) performing segmentation after the first image segmentation on the image by adopting an image segmentation algorithm, wherein an energy function is formula shown in the abstract, and according to the energy function, the smallest segmentation is evaluated by a maximum flow algorithm, so as to obtain the final segmentation object.

The invention discloses a design method of a target detection network based on fused image segmentation features, which is effective for large-scale targets. Based on the general target detection framework Mask RCNN and image segmentation feature fusion, the target segmentation feature and a ResNet-101 convolutional network are integrated into the rpn module, an RoI Pooling module and an RoI Alignmodule, the experiments show that the method is effective for large targets, and the image segmentation algorithm for small targets can be improved completely if the image segmentation effect is ideal.

An image processing apparatus, method and computer program that controls so that an image of image data is displayed on a display unit, controls so that a region of interest is indicated on the displayed image to acquire image data of the region of interest, generates an extraction region extracted from the image data by using each of the image segmentation algorithms to acquire the image data of the extraction region, calculates similarity by comparing the image data of the extraction region with the image data of the region of interest to select the image segmentation algorithm having highest similarity, and outputs image data extracted using the selected image segmentation algorithm to the display unit.

The invention provides a multi-scale image segmentation method for a remote sensing image, in particular for the remote sensing image with high spatial resolution, and constructs a relation among hierarchical structures of the segmentation results of different scales with consistent segmentation boundary. The method comprises the following steps: obtaining an initial segmentation result by a basic image segmentation method, scanning the segmented regions to establish an adjacency relation among segmented blocks, generating an initial bottom small-scale region structure, and subsequently adding features such as gray scale, texture, shape and the like on the basis to merge, adjust, and form a second-layer larger-scale segmented region structure and a third-layer large-scale segmented region structure. The course can be performed by iteration until the required-scale segmentation hierarchy and structure are formed. The generated hierarchical structures of the multi-scale segmentation regions can realize quick switching and access among the segmentation regions of different scales, and the structures are not only suitable for a watershed image segmentation algorithm, but also suitable for other segmentation methods to construct the hierarchical structures of the multi-scale segmented regions.

The invention is applicable to the technical field of image processing and provides a device and a method for high dynamic range image display. The method includes: segmenting a high dynamic range image according to preset image segmentation algorithm to acquire a plurality of image subblocks; adjusting the brightness value of any pixel point in each image subblock according to attribute information of each image subblock, the brightness value of the pixel point at the corresponding position of each image subblock and the preset image linear mapping relation; and generating and displaying a low dynamic range image according to the adjusted brightness value of each pixel point in each image subblock. Displaying effects of the generated low dynamic range image are greatly improved by adjusting the brightness value of any pixel point in each image subblock according to attribute information of each image subblock, the brightness value of the pixel point at the corresponding position of the high dynamic range image and the preset image linear mapping relation.

The invention discloses a method for automatically identifying a breast tumor area based on an ultrasound image. The method comprises the following steps of acquiring the ultrasound image of the breast, and preprocessing the ultrasound image; segmenting the ultrasound image subjected to preprocessing through an image segmentation method to obtain a plurality of segmented subareas; extracting a grey level histogram, texture features, gradient features and morphological features of the ultrasound image, and combining the grey level histogram, the texture features, the gradient features and the morphological features of the ultrasound image with two-dimensional position information to obtain high-dimensionality feature vectors; selecting the most effective feature subset of the high-dimensionality feature vectors through feature ordering based on biclustering and a selection method; performing learning classification on the selected most effective feature subset through a classifier, and then automatically identifying the breast tumor area. By means of the method, the breast tumor area can be identified automatically from segment results of the breast tumor ultrasound image, therefore, automation performance of computer-aided diagnosis is improved, manual operation of clinical doctors is reduced, and subjective influence of clinical doctors is reduced.

The invention discloses a method for selecting an autonomous landing area of an unmanned aerial vehicle (UAV) under a complex environment based on visual SLAM, which is used for solving the technical problem of poor practicability of the existing UAV landing area control method. According to the technical solution, the method comprises the steps of obtaining an image sequence via an overlooking monocular camera carried by a UAV mobile platform, calculating the pose of the UAV in real time via an SLAM algorithm and establishing a sparse point cloud map, and meshing the point cloud map to construct a two-dimensional grid height map; then dividing the grid map according to the height in combination with a Means shift image segmentation algorithm, and finally screening an area that is farthest from a potential obstacle and is suitable for landing of the UVA according to the landing height requirement. According to the method, the pose of the UAV is calculated by adopting the monocular visual SLAM and estimated in real time, the two-dimensional grid height map is constructed, and the area suitable for landing of the UVA is screened. The method, which does not depend on a landmark, has good practicability.

The invention provides an intelligent water-treatment microorganism machine vision identification system and a method. By using artificial intelligent technology, the system and the method can real-timely shoot microscopic images of microorganism in water and carry out the steps of automatic image pre-treatment, image segmentation, microorganism characteristic parameter extraction and selection, and microorganism classification and identification. The system and the method have the advantages that optimal segmentation threshold value can be searched automatically in HIS color space by using self-adaptive image segmentation algorithm; and the classifier is designed in a voting manner to obviate the low classification accuracy by using single classifier so as to effectively improve the entire classification accuracy and accurately identify microorganisms in drinking water and urban domestic sewage. The implementation of the method can further shorten the microorganism detection period in the water treatment process and accurately predict the condition of the water-treatment microorganisms to allow the operators to take measures in time. Accordingly, the method and the system can powerfully ensure the safety of drinking water and the normal operation of urban domestic sewage treatment facility so as to achieve considerable economic and social benefits.

The invention discloses a blood vessel segmentation method for a liver CTA sequence image. Firstly contrast enhancement and noise smoothing preprocessing are performed on an inputted three-dimensional liver sequence image; then liver blood vessels and the boundary thereof are enhanced and blood vessel centers are thinned by adopting OOF and OFA algorithms; seed points of the blood vessel center lines are automatically searched according of the geometrical structure of the blood vessels, and the center lines of the liver blood vessels are extracted so as to construct a liver blood vessel tree; and finally the liver blood vessels are preliminarily segmented through combination of a fast marching method and corresponding blood vessel and background gray scale histograms are calculated, and accurate segmentation of the liver blood vessels is realized by adopting an image segmentation algorithm. The liver blood vessels can be effectively and accurately segmented by fully utilizing the geometrical shape and gray scale information of the blood vessels for aiming at the CTA sequence image which is low in contrast, high in noise and fuzzy in boundary. The blood vessel segmentation method for the liver CTA sequence image can be popularized to other three-dimensional blood vessel segmentation.

An improved image segmentation algorithm is provided that identifies each object in an image. Pixels of the image are sorted based on a range of attribute values of the image. These pixels are then added to a labeling image one by one starting with an extreme point in the range of attribute values. Features are calculated for each object located and these features are matched with pre-defined acceptance criteria. If there is a match, the object is output to an output image. The steps of adding pixels to the image, evaluating the features of the resulting objects and outputting the objects are repeated until a stopping point is reached.

The invention discloses an image segmentation method and device and belongs to the field of image processing. The image segmentation method includes the following steps: establishing a significance model of an image; according to the significance model, obtaining foreground sample points and background sample points in the image; according to the significance model and the foreground sample points and the background sample points, establishing a foreground and background classification model; and according to a preset image segmentation algorithm, segmenting the image, wherein the preset image segmentation algorithm uses the foreground and background classification model and edge information between pixel points to segment the image. Through automatic determination of the foreground and background sample points in combination with the significance model, the foreground and background classification model is established and the foreground and background classification model is used to realize image segmentation. Therefore, problems, which exist in related technologies, that the foreground sample points and the background sample points must be selected roughly manually and the segmentation efficiency is comparatively low when a large quantity of images are segmented are solved so that effects of realizing automation selection of samples and improving the classification precision and segmentation efficiency are achieved.

The invention discloses a multi-visual-field convolutional neural network-based image feature identification method. The method comprises the steps of collecting CT images with positive and negative tags in a historical database, and establishing a data set; judging a position region of a specified feature in each CT image of the data set by utilizing an image segmentation algorithm, and extracting sensitive regions of different pixel sizes; constructing a multi-visual-field convolutional neural network; inputting the extracted sensitive regions of different pixel sizes as samples to the multi-visual-field convolutional neural network, and training the multi-visual-field convolutional neural network to obtain a trained multi-visual-field convolutional neural network; and processing the to-be-identified CT images, inputting the extracted sensitive regions of different pixel sizes to the trained multi-visual-field convolutional neural network for performing feature identification, and determining the positive and negative tags of the to-be-identified images according to an identification result. According to the scheme, the end-to-end image identification is realized and the identification accuracy is ensured.

The invention discloses a computer three-dimensional model establishing method based on Kinect. The method comprises the following steps that a candidate three-dimensional model set is established, consistency dividing is carried out, marking is carried out, a three-dimensional model set S with parts marked is obtained; an object to be modeled is scanned with Microsoft Kinect equipment, point cloud data and image data containing the object to be modeled are obtained; an image dividing algorithm is used to form corresponding foreground objects in the point cloud data and the image data of the object to be modeled in a dividing mode; a representative model is selected, the point cloud data and the image data of the foreground objects are driven to carry out marking dividing, component parts of the foreground objects are obtained; three-dimensional model corresponding parts which have highest similarity with the component parts of the foreground objects are searched in the three-dimensional model set after the parts are marked with part-level shaped descriptors; the contour driving deforming technology and a point set data registering algorithm are used for carrying out registration on the parts and the corresponding point cloud data again so that accurate position combinations can be obtained.

The present invention relates to a tree segmentation method based on a laser radar point cloud and a single tree extraction method based on the laser radar point cloud. A tree segmentation process is characterized by 1) separating the point clouds of a tree from an original laser radar point cloud; 2) projecting the three dimensional original laser radar point cloud to a two dimensional plane, and utilizing an image segmentation algorithm to segment the point clouds of the tree initially; 3) transferring a two dimensional initial segmentation result to a three dimensional grid environment to determine the root grids and the branch grids of each segmentation block, and utilizing a 3D grid fusion algorithm to fuse the segmented point clouds. According to the present invention, the image segmentation algorithm is utilized to obtain the initial segmentation result, then the over-segmentation is fused by the 3D grid fusion algorithm, the single tree is extracted by utilizing a three-dimensional model of the single tree, and the height, the size and the crown area of each tree are obtained, so that the over-segmentation problem in the conventional image segmentation algorithm is corrected, and the accurate single tree extraction results are provided. Moreover, the tree segmentation method based on the laser radar point cloud and the single tree extraction method based on the laser radar point cloud are not limited by the point cloud density, and the manual intervention is reduced greatly.

The invention discloses a heart left ventricle segmentation method based on a deep full convolutional neural network. According to the method, a deep learning idea is introduced into heart magnetic resonance short-axis image left ventricle segmentation; The process is mainly divided into a training stage and a prediction stage, in the training stage, a preprocessed 128 * 128 heart magnetic resonance image serves as input, a manually processed label serves as a label of a network to be used for calculating errors, and along with increase of training iteration times, the error of a training setand the error of a verification set are gradually reduced; And in the test stage, inputting data in the test set into the trained model, and finally outputting prediction of each pixel by the networkto generate a segmentation result. According to the method, segmentation of the heart magnetic resonance short-axis image is achieved from the perspective of data driving, the problem that manual outline drawing is time-consuming and labor-consuming is effectively solved, the defects of a traditional image segmentation algorithm can be overcome, and high-precision and high-robustness left ventricle segmentation is achieved.

The invention provides an image marking method based on a characteristic scene, which comprises the steps of: segmenting a marked image used for learning by adopting an image segmenting algorithm, and obtaining vision description of an image region; 2, carrying out automatic semantic scene clustering on the marked image used for learning; 3, generating a characteristic scene space for each scene semantic category; 4, establishing corresponding semantic tree for each characteristic scene space; and 5, projecting an image to be marked in each characteristic scene space, determining the semantic category of the image to be marked for the fitting degree of a mixing model by adopting the projected vision characteristic and obtaining the final marking of the image through the semantic tree for the image for determining the semantic content. The image marking method fully utilizes the marked character information of a learning image, is used for carrying out automatic semantic scene clustering on the marked image used for learning, and ensures that more complete vision description is obtained under the special scene, thereby improving the reliability of automatic semantic marking through the image vision characteristic.

The invention discloses a stereo matching method by utilizing a graph theory-based image segmentation algorithm. The method has the advantages of greatly improving real-time property and ensuring higher robustness on illumination noise compared with a mean-shift segmentation-based image segmentation algorithm. The method comprises the following steps of: 1) acquiring left and right images of an object to be matched and calibrating the images; 2) respectively calculating an initial parallax error of each pixel point in the calibrated left and right images by using a window method; 3) comparingcorresponding parallax error values in the two obtained parallax error images, and selecting a preferable parallax error of each pixel as an initial parallax error in a step 5); 4) segmenting the calibrated left and right images in the step 1) by utilizing the graph theory-based image segmentation algorithm; 5) performing median filtering on initial parallax error image information acquired in the step 3) by utilizing the segmented images; and 6) obtaining a parallax error image of the left and right images to be matched.

This paper provides a new image segmentation algorithm for object-based image retrieval. The system partitions multi-dimensional images into disjoint regions of coherent color and texture. In order to distinguish the object contour lines from texture edge, the description length of the line is used as discriminating criteria. Visual attribute values are assigned to each region.

The invention discloses an infrared thermal imaging temperature measuring method by correcting surface emissivity through image segmentation. The infrared thermal imaging temperature measuring method includes: aiming at infrared images of a target object surface, wherein the infrared images are shot by a thermal infrared imager, using an image segmentation algorithm, distinguishing areas with different emissivities in the infrared images; based on the emissivities of different areas measured by an emissivity measuring instrument, establishing emissivity distribution array corresponding to image pixels of the infrared images one to one, using the emissivity distribution array to correct gray level array of the infrared images of the target object surface, thus obtaining infrared images which fit emissivity distribution characteristics of the target object surface, thus calculating and obtaining temperature field distribution of the target object surface. Due to the fact the existing infrared thermal imaging temperature measuring method is not capable of accurately setting the emissivity distribution of the target object surface, errors of temperature measuring results are big. The infrared thermal imaging temperature measuring method by the correcting surface emissivity through the image segmentation is capable of effectively reducing the errors, the temperature measuring results are enabled to fit actual temperature distribution conditions of the target object surface. The infrared thermal imaging temperature measuring method by the correcting surface emissivity through the image segmentation is simple in requirement for equipment and easy to achieve.

A multiple pornographic image classification method based on an image segmentation algorithm and deep learning relates to the technical field of information, especially to the technical field of image identification. The method is characterized by comprising the following four major steps: skin color identification, principal component analysis of the skin color region, deep learning, and pornographic image identification based on a convolutional neural network. First, non-pornographic images are screened out through a skin color pixel detection and skin region partitioning algorithm based on the YCbCr theory. Undetermined images are input to a LeNet5-based convolutional neural network model for identification after feature extraction. Compared with the traditional identification based on skin color and features, the method can eliminate the noise influence of non-body-part images, is not constrained by light and human postures, and can greatly improve the accuracy of traditional pornographic image classification. Compared with general deep learning based on a convolutional neural network, the method does not need massive labeled images, and the characteristics of a deep residual network determine that the model can better analyze the characteristics of pornographic images. Only through about ten hours of training, an identification effect above 90% can be achieved.

The invention belongs to the technical field of ore granularity detection and analysis in mine development, and particularly relates to an ore image granularity detection algorithm based on an image processing technology. The algorithm is characterized by comprising the following steps: (1) completing acquisition of an ore image by using an industrial camera; (2) preprocessing the collected ore image, and sequentially performing image graying, median filtering denoising and binarization processing; (3) performing distance transformation and morphological reconstruction on the obtained binarized image by applying a chain code technology; (4) adopting an ore image segmentation algorithm combining Canny operator edge detection and a watershed algorithm based on region growth; and (5) analyzing and calculating the ore granularity by utilizing the obtained ore parameters and the granularity model. According to the invention, high-precision and high-accuracy granularity detection is obtainedthrough accurate ore image segmentation, the failure rate of the crusher is reduced, and the ore yield is increased.

The invention discloses an optical coherent tomographic image retinopathy intelligent testing system and a testing method. A current acquired retina image is mainly determined by an ophthalmology doctor by means of naked eye observation, and large-scale popularization is not facilitated. According to the system, a deep learning concept is used as a technical core; a migration learning strategy isutilized; a convolutional neural network algorithm in a deep learning model is used for establishing a classifier, thereby realizing classification of retinopathy; furthermore an image segmenting algorithm is used for realizing focus extraction and retina layering, thereby obtaining specific information of a pathology position in a picture and quantification information of shape parameters, and generating a related diagnosis report for further diagnosis by the doctor. The optical coherent tomographic image retinopathy intelligent testing system and the testing method have advantages of fillingin gaps in pathology intelligent identification and accurate positioning in a current optical coherent chromatography imaging system, effectively reducing working intensity of the doctor, and furtherpromoting clinical application and technical development of the optical coherent chromatography imaging system in ophthalmology disease diagnosis.

The invention discloses a projection full-convolution network three-dimensional model segmentation method based on fusion of multi-view-angle features, and the method comprises the steps: 1, collecting data of a data set inputted into a three-dimensional network model; 2, carrying out the sematic segmentation of a model projection rendering map through a FCN (full-convolution network) integrated with multi-view-angle features, and obtaining the probability that the pixels of the projection rendering map of a model in all view-point directions are predicted as labels; 3, carrying out the back projection of a projection rendering map sematic segmentation probability graph of the model in each view-point direction, employing the maximum view-angle pooling, and obtaining the probability that amodel patch is predicted as each label; 4, carrying out the optimization through employing a Graph Cut image segmentation algorithm, and obtaining a final prediction label of the model patch.

The invention relates to a binocular image and object contour-based virtual and actual sheltering treatment method, which comprises the following steps of: accurately calculating the regional contour of a real object by an interactive image segmentation algorithm, and taking the regional contour as the geometrical information of the real object; and according to the negatively-correlated relation between parallax and depth, estimating the parallax of a virtual object under a current visual angle, determining the relative depth information between the virtual object and the real object in a current scene, and hierarchically dividing the scene. The front-and-back sheltering relation between the virtual object and the real object is estimated and determined by using the relative depth information, the depth information is not required to be calculated in a pixel-by-pixel manner; and the virtual and actual sheltering treatment is achieved in a two-dimensional image space, whether the sheltering between the virtual object and the real object exists or not is judged, and corresponding treatment is carried out, so that the binocular image and object contour-based virtual and actual sheltering treatment method can be applicable to most sheltering circumstances and a better virtual and actual sheltering treatment effect is achieved. The binocular image and object contour-based virtual and actual sheltering treatment method can be widely applied to the spatial sheltering treatment of virtual reality systems, such as interactive digital entertainment, sports research and training simulation, distance education and training and the like.

The invention discloses an automatic image annotation and translation method based on decision tree learning. A new image is automatically annotated, and a text word list with a visualized content is translated by a machine so as to realize the machine retrieval of image data, comprising a training annotation image set and image automatic annotations, wherein the training annotation image set utilizes an image segmentation algorithm to segment a training image set into sub areas and extract low-level visual features of each sub area; the feature data is discretized, and then the training annotation image set is classified by a clustering algorithm based on a low-level feature discrete value to construct a semantic dictionary; the low-level feature discrete value is used as an input attribute of the decision tree learning; and self training learning is carried out on the constructed dictionary by a decision tree machine learning corresponding to preset semantic concepts so as to generate a decision tree and obtain a corresponding decision rule. The training annotation image set has expandability and robustness and can improve the recall ratio and the precision ratio of the retrieval when the training annotation image set is applied to semantic image retrievals.

The invention relates to a detecting and positioning method and a detecting and positioning system for aiming at underwater obstacles. The detecting and positioning system comprises a multi-beam image sonar and a computer. Firstly, an underwater sonar image in the forward direction is obtained through the multi-beam image sonar arranged on an autonomous underwater vehicle (AUV); then the image noise is reduced through a median filter so as to improve the image quality; after the obstacles and a background are segmented by utilizing an image segmenting algorithm, and according to a segmenting threshold value, each pixel point in a scene image is subjected to the binarization treatment; and finally an obstacle outline is obtained from a binary image, and according to the obtained outline, the size and the location of the obstacle are calculated. The detecting and positioning method can rapidly detect the obstacles of an unmanned surveying vessel or the underwater vehicle in the forward direction, can calculate the size and the location of the obstacle, and can be directly used for underwater navigation.

The invention discloses a shadow detection and removal algorithm based on image segmentation, and relates to the technical field of image processing. The algorithm mainly solves the problems about how to judge whether shadows exist in a region or not or whether an edge is a shadow or not and how to remove corresponding shadows. The algorithm includes the steps that firstly, through texture and brightness characteristics, the probability that each pixel point is a shadow edge is estimated through the combination of local information and overall information; an image is segmented through a watershed algorithm and contour information shown in the specification; a shadow region and a non-shadow region in the image are segmented through a region fusion algorithm based on edges, and meanwhile the shadow region and the non-shadow region are segmented into multiple sub-regions respectively; then classifiers SVM are trained respectively for recognizing shadows; then, a shadow detection energy equation is solved through an image segmentation algorithm, and then a final shadow detection result is acquired; finally, according to the shadow detection result, shadow labels are calculated through an image matting algorithm, the shadow region is lightened through the marks, and illumination of the shadow region is restored.

The invention discloses a three-dimensional facial reconstruction method based on multi-view cellphone selfie pictures. The three-dimensional facial reconstruction method includes automatically positioning mark points on selfie face images of a same person from n different angles; setting up a first energy function according to the mark points and solving a camera parameter Pk (k=1,....,N), wherein in the first energy function, I is pixel serial number, k is input image vision number, mi is the ith mark point, Xi is three-dimensional point coordinate corresponding to the ith mark point, and Pk is a projection transformation matrix from a three-dimensional point to the corresponding image; reconstructing a target function E=Edata+Ecolor+Esmooth according to the camera parameter, optimizing the target function and using a multi-tag image partition algorithm to solve the target function to obtain a three-dimensional facial model. By the three-dimensional facial reconstruction method, the accurate three-dimensional facial model can be reconstructed quickly by the cellphone selfie images.