180 results about "Region growing algorithm" patented technology

The invention relates to a high-resolution remote sensing image variation detection method based on self-adaptive threshold division. The method comprises the following steps of: firstly, geometrically registering high-resolution remote sensing images during two periods; obtaining difference images of the high-resolution remote sensing images during two periods by a difference method; obtaining a gray value threshold extracted by a variation target by a maximum infra-class variance algorithm; extracting a variation target pixel by the gray value threshold; during the detection after-treatment period, eliminating most detected noise pixels through morphology conversion with respect to more disordered variation spots; extracting the sized variation region by a region growth algorithm in combination with the threshold setting of the pixels contained in the variation region to be extracted; and finally obtaining the border outline of the variation region by a border detection method, overlapping the border outline to an original image and drawing a complete variation region. The invention does not need too much manpower, has rapider speed and can achieve favorable detection precision and visual effect.

The invention discloses an image-brightness-characteristic-based pan / tilt / zoom (PTZ) video visibility detection method, which comprises the following steps of: acquiring a road condition video image by using a PTZ video camera, extracting a region of interest (ROI) of the road surface, and acquiring high consistency of the selected pixels; obtaining an accurate road surface region by utilizing a Nagao-filtering-based regional growth algorithm, and ensuring consistent illumination of the selected pixels in a world coordinate; in the road surface region, extracting a contrast curve reflecting the road surface brightness change, searching brightness curve characteristic points, and calculating the farthest pixel which can be identified by human eyes in the image through an extinction coefficient; and combining the camera to calibrate and convert the maximum visual range, and determining a visibility value. In the method, artificial markers are not needed to be set, the existing PTZ camera is fully utilized to shoot road condition and acquire images, can monitor the road condition in real time, and is low in monitoring cost; moreover, the requirement on large-area road condition monitoring is met, the monitoring is stable and is not interfered by external environment, and the visibility monitoring method is easy to implement, high in accuracy and good in effect.

The invention discloses a liver multi-phase CT image fusion method which comprises the steps of firstly performing coarse registering on a source image sequence according to a multi-resolution CT image registering method based on a combined histogram, then realizing automatic liver image dividing according to a region growing algorithm in combination with confidence connection and liver image dividing based on a gradient vector flow snake model, and effectively extracting the edge information of the liver; performing blood vessel extraction based on a directional region growing algorithm on the liver image, performing free deformation transformation based on a B spline and liver non-rigid registering based on space weighting mutual information on a liver essential image, thereby accurately finding an image pair at a same position of a space; and finally performing image fusion based on wavelet transformation. The liver multi-phase CT image fusion method aims at the characteristics of a liver CT image. An image separation process and an image registering process are combined with the image fusion process, thereby greatly improving fusion precision.

A method to estimate segmented motion uses phase correlation to identify local motion candidates and a region-growing algorithm to group small picture units into few distinct regions, each of which has its own motion according to optimal matching and grouping criteria. Phase correlation and region growing are combined which allows sharing of information. Using phase correlation to identify a small number of motion candidates allows the space of possible motions to be narrowed. The region growing uses efficient management of lists of matching criteria to avoid repetitively evaluating matching criteria.

A method of reducing noise in a volume-rendered image includes generating a volume-rendered image from data, identifying a pixel location of suspected noise in the volume-rendered image, and calculating a voxel location that corresponds to the pixel location and intersects a rendered surface in voxel space. The method includes implementing a region-growing algorithm using the voxel location as a seed point to identify a plurality of voxels in a suspected noisy region. The method includes modifying the data to generate modified data by assigning lower opacity values to the plurality of voxels. The method includes generating a modified volume-rendered image from the modified data and displaying the modified volume-rendered image.

The invention is a character extracting method in complex background color image based on runlength adjacency graph, which belongs to character extracting field in preprocess of color image character identification. All color connection sections are acquired with CRAG (color run-length adjacency graph) region growing algorithm after the digital color image is acquired, then the color average is carried on with color classification, acquires several color centers, forms different color layers with the color centers, then the color connection regions accordant to connection region differentiation rule are distributed onto several color layers. Finally the character is selected out through character analysis and size consistency differentiation, and acquires the character image. The method has a high speed and accuracy.

The invention relates to a ship monitoring method. The monitoring method comprises a target identification method and a target tracking method, wherein in the target identification method, after an intercepted real-time video image and a background image are subjected to differential operation and binarization, a ship target is obtained by using a region growing algorithm, and the identified ship target is tracked and locked. By matching the target identification method, the target tracking method, a target locking method and a pairing comparison and database management method, a monitoring system can capture a ship and acquire continuous ship video under the condition of fog or night so as to realize all-weather monitoring of a shipping lane, ensure smoothness of the shipping lane and safety of the ship, reduce manual operation and realize automatic management; and the data information and the like acquired by monitoring are stored into a database, so that the data information and the like are convenient for digital management of the ship, can be timely called as required and can assist managers in learning more information about the ship.

The invention relates to a water sky line detection method based on gradient saliency. The method comprises steps as follows: a frame image is acquired by an optical imaging instrument, and if the image is a color image acquired by an ordinary camera, the image is subjected to standardization processing, and a 24-bit RGB (red green blue) color image is obtained; if the image is a gray image acquired by an infrared imaging instrument, the image is subjected to standardization processing, and an 8-bit gray image is obtained; the obtained standardized image is subjected to Gaussian downsampling and the like. A gradient magnitude matrix and a gradient direction matrix of the image are calculated respectively according to the type of the image acquired by the optical imaging instrument, all gradient information of an original image is reflected in the result, and accuracy of the water sky line detection result is guaranteed. Line segment detection based on a region growing algorithm is performed sequentially from high to low according to the gradient saliency, and the problem of severe noise interference due to the fact that gradient information is directly used for detection is solved.

The invention discloses a segmentation method and system for an abdomen soft tissue nuclear magnetism image. The segmentation method comprises the steps that pre-segmentation is conducted on an area to be segmented through an area growing algorithm, then a morphological operator is adopted to conduct expansion and corrosion operations to carry out further processing on the pre-segmentation result, so that the pre-segmentation result forms an original segmentation outline. After rectification is conducted between a shape template set and the original segmentation outline, kernel principal component analysis is conducted, and prior shape information is obtained through a statistics model. The prior shape information is combined with data items of an energy function of a nuclear magnetism image segmentation model, and an energy function is built; a kernel graph cuts algorithm is used for carrying out segmentation on the original segmentation outline and an objective outline is obtained. The segmentation method and system can achieve semi-automatic segmentation, the system is simple, the robustness of the nuclear magnetism image segmentation algorithm is effectively improved so as to enable the segmentation result to be more accurate, and the segmentation method and system for the abdomen soft tissue nuclear magnetism image can be applied to nuclear magnetism image segmentation.

The invention relates to a method for partitioning brainstem areas automatically from MR (magnetic resonance) sequence images. The method is used for partitioning the brainstem areas automatically and continuously from the human body encephalic MR sequence images, so as to offer technological bases for reconstruction of 3D model of the encephalic tissues to doctors to judge the state of an illness of the patient efficiently and intuitively. The method comprises steps of (1) preprocessing: carrying out gray level clustering analysis on all MR nerve sequence images, and clustering the gray levels of all the MR images to be processed to be five-level gray level images; (2) selecting seed point pixel, obtaining the outline border of the brainstem area of the first MR image through a neighbor region growing algorithm based on the image area growing rate; (3) roughly partitioning the current MR image through a block region growing method; (4) repartitioning the roughly partitioned MR images through logical judgment treatment and the region growing method; and (5) repeating the step (3) and the step (4) till finishing partitioning all MR images.

The invention relates to an intelligent extraction method of a build-up area on the basis of nighttime light data. The intelligent extraction method comprises the following steps: adopting an adaptive particle swarm optimization algorithm to realize the optimal selection of the selection parameters of a VIIRS (Visible Infrared Imaging Radiometer Suite) nighttime light image sample and a MODIS (Moderate Resolution Imaging Spectroradiometer) normalized difference vegetation index image sample; adopting a region growing algorithm based on SVM (Support Vector Machine) classification to finish SVM model training, and adopting a cross validation method to carry out precision validation on the model; and according to an optimized parameter, determining a city sample and a non-city sample, and adopting the region growing algorithm based on the SVM to extract a city built-area range. By use of the intelligent extraction method, from a sample selection source, the adaptive optimization of a sample selection parameter is carried out, and the SVM and the region growing algorithm are adopted to improve processing efficiency and precision for the nighttime light data to extract the nighttime light data.

The present invention discloses a pulmonary nodule segmentation method based on the Hession matrix and three-dimensional shape indexes. According to the method, medical CT images are fully utilized; sequential pulmonary parenchymas are segmented through using an optimal threshold according to the gray values of sequential CT images, and the volume data of the three-dimensional pulmonary parenchymas are constructed; the Hession matrix feature values of each voxel point in the volume data of the three-dimensional pulmonary parenchymas are calculated; the three-dimensional shape indexes are constructed according to the shape features of a three-dimensional nodule model and on the basis of the Hession matrix feature values and two-dimensional shape indexes; and a three-dimensional sphere-like filter, namely, a 3D shape index nodule detection function is constructed finally and is adopted to perform nodule detection on the three-dimensional volume data of the pulmonary parenchymas, and with detected nodule regions adopted as a plurality of seed points of region growth, three-dimensional segmentation is performed on nodules on the basis of a confidence-based region growing algorithm. The method of the invention is simple in operation, can automatically detect and segment different types of suspected pulmonary nodules and has high stability and high accuracy.

The invention discloses a similar image colorization algorithm based on classification learning. The similar image colorization algorithm comprises the following steps: sample images are collected, an image gradation co-occurrence matrix attribute is extracted, the sample images are classified into five categories through the AP algorithm, superpixels of a target image and superpixels of a reference image are calculated respectively, then, colors are transferred from the reference image to the target image, colors of the superpixels are corrected afterwards according to continuity of image space, and finally the algorithm is used for conducting color diffusion to complete colorization. According to the similar image colorization algorithm, the influence on an image by a global attribute of the image is considered, the image gradation co-occurrence matrix attribute is extracted to conduct classification learning on parameters of a superpixel matching function, as a result, different parametric functions can be provided for superpixel matching on images with different compositions, and the universality of the similar image colorization algorithm on the images is improved; besides, after the matching process, region growing algorithm partition can be conducted at a superpixel level, and color correction can be conducted in a region.

The present invention discloses a Chinese character recognition method based on an optical character recognition technology. For an inputted gray scale text image, a cutting result is obtained by a hierarchical structure cutting method based on a connected region, then the binaryzation and denoising operations are carried out on the image by a region-growing algorithm based on the pixel distribution characteristics to obtain a to-be-recognized character, and then the to-be-recognized character is sent to a general single character recognizer to recognize to thereby obtain a candidate character set, and finally, a final recognition result is obtained by a similar character classification and recognition method. Compared with the prior art, the Chinese character recognition method based on the optical character recognition technology of the present invention enables the character recognition accuracy and efficiency to be improved effectively, enables the text information to be inputted to a computer at a high speed, solves the conflict between the low-speed information input and the high speed information processing, and enables the onerous keyboard recording work to be simplified.

The invention relates to an object-neural-network-oriented high-resolution remote-sensing image classifying method, aiming at solving the problems that the conventional remote-sensing image classifying method is low in classification precision and cannot effectively utilize information of all wave bands of a remote sensor. The method comprises the following steps that: an image of the ground is shot by a high-spatial-resolution sensor and is transmitted to a computer; the computer carries out primary image element division on the input image by a region growing algorithm; the primarily-divided image is subjected to multi-size division according to continuously-set neterogeny degree thresholds and shape features and spectral signatures of the image, thus forming divided images with different sizes; and the obtained divided images with different sizes are used for establishing a BP (Back Propagation) neural network, setting training parameters and establishing training samples to classify the image which is subjected to the multi-size division, thus obtaining a high-resolution image. The method is applicable to the field of obtaining of images with high spatial resolutions.

The present invention provides a system, method and computer program product for improving quality of images. The pixels of the input image are used to generate two lists of pixels based on two different scale space values. Thereafter, the two pixels lists are processed to obtain an output image with improved quality.

The invention discloses a method for measuring the size of a three-dimensional reconstruction liver model on the basis of an improved region growing algorithm. The method includes the steps that seed point selection and growth criteria in a traditional region growing algorithm are improved by means of a quasi-Monte Carlo method, and an abdomen CT image is segmented by means of an improved region growing segmentation method to extract a liver area; a segmented binary image is used for three-dimensional reconstruction to obtain the three-dimensional reconstruction liver model only provided with a surface mesh, and the model is sealed; a regular square bounding box is arranged, the bottommost face of the bounding box is provided with a projection plane, the sizes of pentahedrons defined by a triangle patch on the reconstruction model and a projection projected by the triangle patch on the projection plane in an enclosed mode are calculated, wherein the triangle patch is provided with a normal vector in the positive direction and a normal vector in the negative direction; finally, the algebraic sum of the sizes of the pentahedrons is the size of the liver model. The method can well represent the shape of the liver, and therefore size measurement is effectively carried out on the three-dimensional reconstruction liver model. Besides, the size of a part of the liver can be measured, and measurement precision is high.

The invention discloses a dynamic image moving object detection method. The method includes firstly, reading two images of time-consecutive frame and respectively subjecting the images to Gaussian filtering; secondly, calculating consecutive frame differential images to obtain a binary image with moving information; thirdly, utilizing the improved Chan-Vese algorithm to detect the moving object of the image; and finally, utilizing the improved region growing algorithm to detect the accuracy of the moving object. The two consecutive frame differential images are used as a seed pixel point set of the moving object, the Sobel edge detection operator is used for detecting the pixels around the seed pixel, the pixels around the seed pixel are marked as moving object pixels if the pixels around the seed pixel are detected as edge pixels, the detected edge pixels and the moving object detection result are mixed, and accordingly the moving object can be detected accurately.

A technique for automatically generating a virtual model of a branched structure using as an input a plurality of images taken of the branched structure. The technique employs an algorithm that avoids inaccuracies associated with sub-optimal threshold settings by “patching” holes or leaks created due to the inherent inconsistencies with imaging technology. By “patching” the holes, the algorithm may continue to run using a more sensitive threshold value than was previously possible.

The invention discloses a cerebrovascular image segmentation method based on a multi-angle serialized image space feature point set, comprising the following steps: S1, registering a live image and a mask image of cerebral vessels; S2, extracting a geometric feature point set of a serialized subtraction image; S3, locally adjusting the positions of the feature points at the edges of the blood vessels in the serialized subtraction image; S4, adopting a spatial rotating coordinate system to remove the feature points in non-vascular positions in the subtraction image; S5, determining the adaptive segmentation threshold of the serialized subtraction image; and S6, based on region growing and vascular image segmentation of adaptive threshold and by taking the feature points obtained in S4 as seed points and the adaptive segmentation threshold obtained in S5 as the standard of the law of growth, adopting a region growing algorithm to segment the serialized subtraction image to obtain a pure cerebrovascular image.

The invention discloses a method used for detecting polysilicon solar energy battery piece EL image surface dark patch defects. According to the method, binaryzation extraction of a target battery pieces is carried out based on a battery piece EL image collected using a near infrared camera, image segmentation is carried out through a region growth mode based on complex and various background interferences caused by polysilicon so as to obtain possible defect connected domains; two methods are adopted for elimination of false detection, wherein according to one method, connected domain analysis is carried out so as to extract connected domain area and opening area characteristics, and according to the other method, the images corresponding to the connected domains are subjected to curve detection so as to eliminate false detection through image texture analysis. The method is capable of determining solar energy battery piece dark patch defects accurately, and realizing locating of darkpatch defects.

The invention provides an object positioning method and device based on RGB-D information and a machine vision system, and relates to the technical field of digital image processing. The object positioning method based on RGB-D information comprises the following steps: obtaining environment images under a current viewpoint; based on an RANSAC algorithm, in geometrical similarity measurement criterion, introducing a color similarity metric to measure the similarity between points and a plane, carrying out partitioning on a point cloud graph, and selecting a seed point from each block in sequence to obtain plane parameters; carrying out local point judgment according to color information and geometry information of the environment images, carrying out re-estimation on the plane parameters and extracting a plane characteristic equation; and obtaining a point cloud cluster of each object under the current viewpoint through a region growing algorithm based on the RGB-D information to realize object positioning. The method realizes an efficient and stable object localization algorithm through combination of the RGB-D information and the region growing algorithm, can locate a plurality of objects having a certain mutual shielding, and has high real-time performance and robustness.

The invention provides a blood vessel extracting method comprising the steps of providing head and neck CT angiography data, dividing the head and neck CT angiography data into a plurality of parts, and adopting corresponding vessel extraction algorithms to perform blood vessel extraction on the plurality of parts. Considering that vessels of different parts of the head and neck region differ greatly in morphology, the head and neck region is divided into four blocks in the method, and extraction of the blood vessels of the head and neck region is realized by combining a region growing algorithm, a level set algorithm and a dynamic tracking algorithm. The method has good robustness, is high in blood vessel extraction speed, and is applicable to data of different manufacturers.

The invention provides a breast cancer image identification method and device, and a user terminal. The method comprises the following steps of using a threshold segmentation algorithm to segment a preprocessing pathology image and acquiring segmented binary images; taking lymphocytes as the seed of a region growing algorithm, taking the union set of the segmented binary images as a target picture, and using the region growing algorithm to segment and acquiring area binary images; and acquiring an area corresponding to the area binary images according to the area binary images, and calculatingand acquiring the proportion of infiltrating carcinoma interstitial lymphocytes through the area. By using the method of the invention, the identification of cancer cells in the pathology image of abreast cancer through a computer is realized, the accuracy of breast cancer pathology image diagnosis is increased, the errors of artificial naked eye determination are avoided, breast cancer pathology image diagnosis time is shortened, diagnosis efficiency is increased and convenience is brought for the diagnosis work of a doctor.

The invention discloses a Laser point cloud-oriented single tree segmentation method based on Faster R-CNN. The method comprises the steps of obtaining forest stand point cloud data; calculating pointcloud features of the scanned forest segments to realize branch and leaf separation of forest segment point cloud data; performing adaptive voxelization operation on the main point cloud data of theforest stand, and performing multi-angle projection on the main point cloud data to generate a corresponding depth image; detecting a trunk in the generated depth image by adopting a deep learning method; and obtaining the spatial three-dimensional point cloud of the corresponding trunk through back projection by utilizing the position information of the trunk in the detected depth image; and taking the obtained point cloud of the trunk part as a seed point, and combining a region growing algorithm to realize individual tree separation. According to the method, a deep learning method is adopted, learning is carried out by means of big data samples, the accuracy of individual tree segmentation is higher, and possibility is provided for accurately solving the problem of individual rubber tree segmentation based on LiDAR data of the ground by using deep learning.

The invention relates to a method for segmenting an inhomogeneous medical image. The method for segmenting the inhomogeneous medical image comprises the following steps that firstly, foreground seed points and background seed points on the image to be segmented are selected; secondly, the probability that each grey level belongs to the foreground or the background of the image to be segmented is evaluated according to grey level information of a selected seed point set, the grey levels are mapped on all pixel points of the image, and therefore a corresponding probability density distribution graph is obtained; thirdly, the selected foreground seed points and the selected background seed points are used as growing seed points respectively, one probability threshold on the corresponding probability density distribution graph is used as a growing condition, a region growing algorithm is executed, and therefore a foreground seed point group and a background seed point group which have grown automatically are obtained; finally, the obtained seed point groups which have grown automatically are used as seed points of a random walk algorithm, the random walk algorithm is executed, and a final segmentation result is obtained. By the adoption of the method for segmenting the inhomogeneous medical image, the sensitivity to the number and the position of initial seed points can be reduced, and the segmentation precision of the inhomogeneous medical image is obviously improved.

The invention provides a method for manufacturing femtosecond laser complex components in a large batch. The surface processing of the complex component comprises: setting, according to the area and the focal depth of a laser processing region, a constraint condition and a region growing algorithm by using a complex component model reconstruction and surface processing pattern splicing algorithm;dividing the processing region, simplifying the region boundary to obtain the processing region, extracting the key features of the processing pattern types and layout features; designing a slicing algorithm according to the process of periodic and non-periodic layout in order that the processing pattern is automatically and quickly laid out on the surface of the complex component; by using the processing region division and pattern segmentation algorithm and setting the element type and combination in the processing pattern, obtaining the intersection of the element and the processing regionboundary and classifying the sub-pattern to the corresponding processing region; by using processing pattern real-time detection and online splicing manufacturing technology and by setting and installing a high-precision CCD camera, obtaining the video image of a processed part, recognizing the boundary of the processed pattern in real time based on an image processing algorithm, and then according to the boundary information of the pattern to be processed, predetermining the registration error of the two processing patterns, obtaining a required correction transformation parameter, and achieving online splicing manufacture in a laser processing process.

The invention discloses a lead bonding welding spot defect positioning and classifying method. The method comprises the following steps: 1) obtaining a bonded welding spot image by using an industrialcamera; 2) carrying out initial positioning on an area where a welding spot is located by utilizing an algorithm based on pixel neighborhood variance; 3) removing redundant non-welding-spot areas byusing a gray projection algorithm; 4) performing primary extraction on a region where the welding spot is located by using a region growing algorithm, and performing defect segmentation by using a level set method on the basis; 5) extracting linearly separable main characteristics of the welding spots by utilizing kernel principal component analysis; and 6) sending the extracted main features to arandom forest classifier for defect type classification, and giving welding parameter adjustment suggestions according to multi-classification results. Compared with other welding spot detection technologies, the lead bonding welding spot defect positioning and classifying method based on image processing and machine learning has the advantages of being high in precision, high in speed, high in intelligent level and the like, and has great application prospects in actual electronic industrial production.

The invention provides a product surface defect detection method and device, and equipment. The method comprises: obtaining an image and determining whether the image is a to-be-detected image needingto detect a specified type of defect or not; if yes, carrying out sliding sampling on the to-be-detected image by utilizing a sliding window through a pre-trained defect detection model, carrying outdefect detection on the sampled window area, and outputting a position identifier of the window area with defects in the to-be-detected image; carrying out image segmentation on the window area withthe defects by utilizing an image segmentation algorithm to obtain contour areas of the defects in the window area with the defects; and connecting the adjacent contour regions through a region growing algorithm to obtain the form and number of defects existing in the to-be-detected image. According to the invention, a sliding sampling detection mode can be carried out on an image, and the problems of time and labor waste, incapability of detecting surface fine defects and poor positioning performance in the prior art are solved.

The invention discloses a method for estimating the defect degree of carbon fiber surface based on a region growing algorithm. Firstly, image preprocessing is performed on a carbon fiber profile imageto obtain a preprocessed surface defect enhancement image; then, an ROI region of the carbon fiber profile image is extracted based on the preprocessed image so that defect regions can be found out better; after that, the regional growth algorithm is used for segmenting the defect regions of the researched carbon fiber surface image from the image; finally, the proportion of defects in the imageto the entire image is calculated by using a matlab correlation program, and the number of the defects in the image can be counted. According to the method, not only can the tiny defects existing on the surface of a carbon fiber material be effectively segmented, but also the proportion of the tiny defects to the whole image can be calculated, a reliable measurement index is provided for evaluating the advantages and disadvantages of the carbon fiber material by professionals, and the effect of assisting in evaluation of the quality of the evaluated carbon fiber material is achieved.