69 results about "Level set algorithm" patented technology

The invention discloses an unmanned aerial vehicle multi-overlapped-remote-sensing-image method for extracting a building contour line. The unmanned aerial vehicle multi-overlapped-remote-sensing-image method includes the steps that three-dimensional point cloud is generated with an aerial-triangulation and dense-matching combined method and filtered, and a building is detected from the point cloud; after the walls of the detected building are canceled, the general contour of the building is extracted from building top face information; the general contour of the building serves as a buffering area to be overlapped on spliced images and serves as shape prior information, evolution is carried out in the buffering area with a level set algorithm, and finally an accurate contour of the building is obtained. By means of the unmanned aerial vehicle multi-overlapped-remote-sensing-image method, as point cloud three-dimensional information generated by the multiple overlapped images is sufficiently used, and meanwhile the high-accuracy geometrical information of the high-resolution remote sensing images is used in a combined mode, the building contour extracting accuracy is remarkably improved, and the complexity of the method is lowered.

The invention discloses an image segmentation method for a dynamic contrast-enhanced mammary gland MRI sequence, pertaining to the field of magnetic resonance image processing techniques, which is characterized by comprising the following steps: a three-dimensional magnetic resonance image sequence of the section of the mammary gland is put into a computer; the image is divided into two parts including a mammary gland-air interface and a mammary gland-chest interface; a breast-air boundary is obtained by a splitting transaction in which a dynamic threshold controls the regional growth; an initial profile of the mammary gland and the chest is obtained in the same way, the complex profile of the breast and the chest is obtained with a method of controlling a level set; a three-dimensional magnetic resonance image sequence of a point-in-time is obtained by split jointing the segmentation results and taken as an initial position of the next group three-dimensional image segmentation. The image segmentation method of the invention increases the segmentation speed, solves the problem that a level set algorithm can not easily determine the initial profile and the velocity function and realizes an automatic segmentation of the complex dynamic contrast-enhanced mammary-gland magnetic resonance image with plenty of data.

The invention belongs to the field of treatment of medical images, and relates to a breast neoplasms ultrasonic image segmentation method based on improved level set algorithm. The method is characterized in that the original image is pre-treated and remains the effective area, the speckle noise is removed, so as to achieve the purpose of protecting the boundary; the image is adaptively segmented according to the threshold. The method comprises the steps of (1) inversing color for the image; (2) determining the threshold; (3) screening candidate areas; (4) arranging the rest candidate areas; (5) determining seed points. With the adoption of the method, the seed points can be quickly found, the seed points can be kept in a neoplasms area, and the accurate determination of the seed points ensures the accuracy of area growth and level set; in addition, the seed points grow in the area, and the initial contour can be found; the classic Chan-Vese (CV) algorithm is improved; the global statistics information of the curved line of the contour in the development process is considered during calculating the global statistics information; the accuracy of the segmentation result is ensured, and moreover, the automation level of the segmentation method can be further improved.

The invention provides a blood vessel extraction method comprising the steps that a medical image is acquired, and at least one blood vessel area is acquired through registration with a standard image; multiple seed points of the blood vessel areas are determined; the blood vessel enhancement image of the medical image is acquired by using a classifier method; the center line between the seed points is calculated based on a level set algorithm in the blood vessel enhancement image; and the blood vessels are grown based on the central line and the blood vessel extraction result is acquired. The blood vessels are extracted on the blood vessel enhancement image acquired by using the classifier so that the accuracy and the robustness of the result can be effectively improved.

The invention belongs to the technical field of medical CT (computed tomogram) image segmentation, and relates to a tooth CT image segmentation method based on deep learning. According to the technical scheme provided by the invention, a traditional Level Set algorithm and a U-net network model are combined, and a Level Set algorithm is used for solving the problem of a training set required by the neural network, so that the neural network can be trained by using unmarked tags, at the same time, the neural network model is used to complete the automatic segmentation of the image, the problemof non-convergence of curve evolution is avoided, and a sufficient and accurate segmentation effect can be obtained under the condition that a medical image training set is insufficient.

The present invention provides a time-optimal route planning method based on an improved level set algorithm. The method comprises: 1, constructing a three-dimensional route planning environmental space; 2, initializing a level set function as a symbol distance function, and setting the AUV navigation starting point on a zero level set; 3, constructing a level set evolution equation considering the influence of the ocean current on the AUV navigation, and constructing a narrow band at the starting point and setting a prohibited area; 4, evolving a level set function according to the level set equation established in the step 3, and storing the zero level set interface of each time step length; 5, determining whether a target point is in the current narrow-band range or not; 6, reconfiguring the narrow band, and employing an improved fast marching method to reinitialize the level set function to be a symbol distance function; and 7, employing a back iterative equation, obtaining the time-optimal route of the AUV, and outputting an optimal route. Through adoption of an ocean current reanalysis database, the time-optimal route planning method based on the improved level set algorithm generates an ocean current field and fully considers the influence of the ocean current in the route planning to allow the planning route to have high practicality.

The invention relates to a multidirectional moving population flow estimation method on the basis of a generalized regression neural network. A method on the basis of linear regression is difficult to respond to the complex conditions of serious shielding between pedestrians and poor population dividing quality. The multidirectional moving population flow estimation method comprises the following steps of: firstly, extracting dynamic textural features of a moving population by an optical flow field; then implementing division of the population according to the moving directions by the dynamic textural features and a level set algorithm to obtain ROIs (Region of Interest) representing different moving directions; and implementing regression analysis between ROI features and the population flow by utilizing the GRNN (Generalized Regression Neural Network) so as to acquire an accurate and real-time flow statistical result of the population with different moving directions in a scene. According to the multidirectional moving population flow estimation method provided by the invention, not only can the complex process of extracting and tracking personal features of the pedestrians be avoided and the capability of resisting the shielding between the pedestrians of the algorithm is greatly promoted, but also the integrity and difference of the pedestrian movement can be considered, so that the division of the population according to the moving directions is implemented.

The invention provides a medical image segmentation method based on the Markov. The method comprises the steps that 1, an original image is segmented initially through the Markov segmentation method; 2, the image is processed through the level set algorithm, and a level set function, a biased field and the gray average of areas are updated; 3, the original image divided by the currently updated biased field is a corrected image; 4, the corrected image is segmented through the Markov segmentation algorithm; 5, whether the sum of the number of times of iteration of the current level set algorithm and the Markov segmentation algorithm is greater than a preset value or not is judged, if yes, the corrected image which is finally obtained is segmented and then a final segmentation result is obtained, and if not, the step 2 is carried out again. The medical image segmentation method based on the Markov can process the medical image segmentation with the uneven gray level rapidly, steadily and accurately.

The invention discloses an eye fundus image optical cup automatic segmentation method based on improved PDE image repairing. The method is carried out with following steps: reading and normalizing an original image; selecting channels for the original image; enhancing the image based on a morphological principle; positioning blood vessels based on an image subtraction; filling and repairing blood vessel areas by the use of an improved BSCB model; performing median filtering and noise reduction for the image and highlighting object areas through gamma transformation; and preliminarily obtaining an optical cup contour through a level set algorithm under a condition that the mass center of the brightest pixel point set on the image serves as an initial seed point. According to the invention, the image in which the blood vessels are filled and repaired eliminate the influence on optical cup segmentation by the blood vessel areas to a certain extent; the accuracy of optical disk segmentation is improved; the seed point is automatically selected as the contour of an optical cup is obtained through the level set algorithm; and manual intervention in a conventional semi-automatic method is reduced.

The invention discloses a cardiac ultrasound imaging method, and aims at solving the problem of automatically segmenting of a left ventricle contour during a cardiac measurement process, and preventing the situation of inaccurate volume calculation which might be caused by manually specifying left ventricular segmentation points in the measurement process. Compared with a traditional left ventricular segmentation algorithm, a large number of images of the position of a left ventricle manually marked by a doctor are used to train a deep convolutional neural network, the position of the left ventricle can be accurately identified by the acquired detection network, and a level set algorithm can be performed within a detection position area to obtain a segmentation curve of a left ventricularendocardium; compared with a traditional algorithm, acquired results are more robust, the operation efficiency is higher. According to the cardiac ultrasound imaging method, a manual enhancement manner is adopted to enhance edge areas to improve a segmentation effect aiming at the situation of poor image quality of ultrasound mapping.

The invention relates to medical apparatuses and medical images and provides an improved hybrid multi-mode brain tumour image segmentation method. A brain tumour area is extracted through employing an FFCM algorithm, a hybrid level set algorithm is utilized to correct a boundary problem existing in the tumour area. The FFCM algorithm and the level set algorithm can be more effectively applied to MRI brain tumour images. The method is characterized by comprising steps that firstly, the MRI images in three modes including T1C, T2 and FLAIR are inputted, median filtering is employed to carry out filtering and initial segmentation of the images to acquire pre-processed images, then linear fusion is employed, lastly, FFCM clustering segmentation is carried out for the fused images, areas with relatively large gray values are automatically extracted, and hybrid level set segmentation is carried out for the acquired tumour under-segmentation areas. The method is mainly applied to medical image acquisition and processing.

The invention relates to a nondestructive detection device for the high energy X-ray of a nuclear fuel assembly. The nondestructive detection device comprises an accelerator system, a detection acquisition system, an image processing reconfiguration system and a remote control system, wherein the accelerator system comprises a ray machine head and a front collimator unit; the detection acquisition system comprises an acquisition output unit and an acquisition system shielding mechanism; the image processing reconfiguration system processes received image information through a watershed algorithm in combination with a level set algorithm; the remote control system controls the accelerator system, the detection acquisition system and image processing reconfiguration system, and sets pulse time sequences to different X-rays by the accelerator system through a Monte Carlo method, and sets the output time sequence of the detection acquisition system. The nondestructive detection device can perform X-ray detection on the nuclear fuel assembly in the environment of high radioactivity, so as to obtain high spatial resolution three-dimensional images.

The invention discloses a left-atrial-appendage CT image segmentation method, and relates to medical image processing. Firstly, left-atrial-appendage CT images are denoised, prior knowledge is utilized to construct missing boundary information of a left atrial appendage, and a region growing algorithm is used to obtain a single initial segmentation image to use the same as a single Atlas label graph; then information relationships of the single label graph and an adjacent left-atrial-appendage CT sequence are utilized for improved Atlas segmentation to obtain an initial segmentation sequence;contour information in the initial segmentation sequence is used as initial contours of a level set algorithm, and approximate Euclidean distance information of left-atrial-appendage CT sequence graphs is obtained to be used as feature input; and finally, a precise segmentation result sequence is obtained through a fast level set algorithm to achieve the purpose of left-atrial-appendage CT image segmentation.

The invention belongs to the technical field of medical imaging, and particularly relates to a cardiac magnetic resonance image segmentation method. A cardiac anatomic structure-based left and right ventricle level set segmentation method specifically comprises the following steps: performing geometry decomposition to a cardiac structure; resolving a convex hull of an initialized contour; producing a signaled distance function; and generating intima and extima through a dual-level set algorithm, and the like. By using the method, the simultaneous segmentation of the ventricular intima and extima and cardiac muscle can be realized, and the segmentation results can meet the physiological structure characteristics of a heart as far as possible.

The invention discloses a human liver feature modeling and vascular model space normalization method. The method comprises the steps of preprocessing multi-source medical image information data; extracting human liver edge features; performing three-dimensional modeling; and performing multi-source information and different-medium-based human liver and vascular feature coordinate normalization. According to the method, liver medical image feature extraction and three-dimensional model rebuilding are realized through gray level nonlinear transformation and multiphase piecewise smooth level set algorithms for gray level non-uniformity of human liver CT image and MRI image; and for the fact that accurate corresponding references do not exist in three-dimensional spaces of a multi-medium liver vascular model and a liver model, the problem in liver tissue organ coordinate normalization is solved through an external feature registration technology fusing multisource feature information.

The invention discloses a wavelet-decomposition-based SAR image change detecting algorithm of a multi-scale level set, and belongs to the field of remote sensing image processing. The wavelet-decomposition-based SAR image change detecting algorithm mainly solves the problem that a speckle noise effect is serious in an SAR image change detecting process. The wavelet-decomposition-based SAR image change detecting algorithm is implemented in the following steps that (1) a difference image of two registered SAR images in the same region and at different time phases is obtained by adopting logarithm ratio operators; (2) multilayer wavelet decomposition is conducted on the difference image through SWT so as to obtain images with different resolution ratios; (3) the images with the low resolution ratios are preliminarily segmented through a level set algorithm, and the outlines of obtained segmented images are used as initialization curves of the level set algorithm of the images with the higher resolution ratios; (4) the step (3) is repeated in a layer-by-layer mode until final segmented images are obtained by conducting level set segmentation on the images with the high resolution ratios. According to the wavelet-decomposition-based SAR image change detecting algorithm, the multi-scale application overcomes the defect that a closed curve is prone to getting into local optimum in a level set evolutionary process, and robustness on noise is enhanced; the wavelet-decomposition-based SAR image change detecting algorithm is applied to change detecting of the SAR images, the detecting effect and detecting accuracy are improved obviously, and the change detecting process is accelerated.

The invention relates to a method for automatically detecting cattle and sheep in a high resolution image. The method comprises a step of circling the cattle and the sheep, a step of marking the cattle and the sheep, a step of counting up quantities of the cattle and the sheep, a step of calculating area of the cattle and the sheep and other steps. In the step of circling the cattle and the sheep, the cattle and the sheep are circled in a dynamic manner through a level set algorithm thought via use of gradient information data of high mark images. In the step of marking the cattle and the sheep, the center of a cattle and sheep zone is found out, and the cattle and the sheep are marked. In the step of counting up the quantities of the cattle and the sheep, a cattle and sheep connection zone is found out via a breadth-first traversal, and the quantities of the cattle and the sheep can be determined based on area of the cattle and sheep connection zone. In the step of calculating the area of the cattle and the sheep, after a pixel size of the cattle and sheep connection zone is calculated, the area of the cattle and the sheep can be obtained. The method can be used for batch processing of image data, and cattle and sheep condition in a whole document folder can be calculated via one click.

The invention relates to a sea and land boundary detection method based on a level set algorithm. The sea and land boundary detection method based on the level set algorithm includes the following steps: (1) adopting a ground moving target (GMT) tool to generate a sea and land boundary module corresponding to a specific position satellite; (2) conducting pretreatment to a to-be-detected remote sensing image which is shot by the specific position satellite; (3) adopting the level set algorithm to conduct treatment to the to-be-detected remote sensing image so as to generate a sea and land boundary; (4) adopting a traditional edge detection method to conduct detection to the to-be-detected remote sensing image so as to generate a sea and land boundary; (5)conducting an intersection obtaining algorithm to the two different sea and land boundaries which are obtained from the step (3) and the step (4) to generate a final sea and land boundary. The sea and land boundary detection method based on the level set algorithm can be widely applied to remote sensing image process.

The invention provides a high density cell tracking method based on topological constraint and a Hungarian algorithm, which comprises: (1) segmenting a cell image sequence by using an image segmentation method which combines a level set algorithm and a local gray threshold process, and initially labeling segmented cells in each frame; (2) according to distance limitation, establishing a tracking search region for a cell to be matched in the kth frame in the k+1th frame, and listing the cells in the region as candidate cells; (3) establishing a coefficient matrix Q, and if a cell j in the k+1th frame is the candidate cell of a cell i in the kth frame, performing data association according to topological constraint to calculate the similarity Qij of the cell j, or assigning a larger value to the similarity of the cell j; (4) performing transformation on the coefficient matrix by using the Hungarian algorithm to find out independent zero elements, wherein the cells represented by the rows of the zero elements are matched; (5) finding out rows in which there are no zero elements after matrix transformation, and taking the cells corresponding to the rows into consideration respectively; and (6) adding 1 to the k, jumping to the step 2, and repeating the steps till the last frame of the image sequence. The method can realize high-efficiency cell tracking.

The invention discloses a slit-lamp anterior segment sectional image feature extraction method comprising the following steps: 1, using a slit-lamp to gather an anterior segment sectional image so asto obtain an anterior segment color image; 2, carrying out color space transformation for the anterior segment color image, carrying out multi-threshold binary, distinguishing cornea, iris and crystalline lens latent areas, and respectively using blob shape analysis to obtain the cornea, iris and crystalline lens areas; 3, using morphological operation to carry out hole filing and expansion process for the cornea, iris and crystalline lens areas; 4, using a level set algorithm to track the boundaries, precisely tracking surface boundaries of each part of the anterior segment, and finding the anterior segment. The method solves the problems that the level set algorithm is slow in speed when applied in the whole image, can realize real time extraction and analysis of the anterior segment sectional image features, thus providing reliable basis data for solving anterior segment clinic parameters.

The invention, which belongs to the field of digital surface model (DSM) data segmentation processing by applying a level set algorithm, relates to a level set-based method for constructing a level of detail 2 (LOD2) building model, so that a problem that the construction precision is not high due to a rough top surface structure in the existing two-dimensional image-based building model construction method can be solved. More specifically, the invention comprises the following steps: extracting a building outline mask omega m, selecting DSM data and distributing the building outline mask omega m and the DSM data into a unified coordinate system; obtaining building top surface data T; obtaining a characteristic space of the building top surface data T; carrying out multi-phase level set segmentation to obtain sub-areas; obtaining point sets of all the sub-areas, detecting a boundary point of each fragment and obtaining an image coordinate of an angular point of each primitive of the building; establishing a topological structure of the building top surface data T; and according to an aerial visible image, extracting texture data of the building surface and enabling the data to correspond to different primitives of the building, so that construction of the LOD2 building model is completed. The provided method is applied to a three-dimensional construction task of a large building with an LOD2 level.

The invention relates to an in-vivo bacterial colony counting method based on infrared and optical image dual-mode imaging information. The method includes following steps: obtaining an optical imageand a sequence infrared image of the same original target; processing the optical image by employing a level set algorithm; performing image registration on the optical image and the sequence infraredimage by employing affine transformation, and obtaining corresponding positions of connected domains in the sequence infrared image in sequence in the optical image according to obtained position information of the connected domains in the optical image; performing statistics on a connected domain local texture characteristic set of infrared image space lower sequence images corresponding to theconnected domains in the optical image one by one, wherein the texture characteristic set comprises five characteristics including a grayscale mean value, a grayscale variance, and the energy, the entropy and the correlation of a grayscale co-occurrence matrix; obtaining an in-vivo bacterial colony detection classification model; and counting the connected domains which are determined to have active characteristics in sequence, and obtaining the number of in-vivo bacterial colonies.

The invention discloses a shoeprint hole and insert feature detection and description method. The method comprises the steps of obtaining a shoeprint hole candidate set and extracting attribute information of the shoeprint hole candidate set; traversing the hole candidate set, and screening hole areas based on geometric and gray attributes to obtain a candidate area set; constructing a similarity matrix based on Euclidean distance by taking a long axis, a short axis, an eccentricity rate and an area as characteristics, and performing seed growth on the accurate candidate region set by adopting a seed growth algorithm after a growth rule is improved to obtain a growth region with hole characteristics; obtaining a candidate insert feature point set based on multi-scale difference Gaussian; traversing all candidate insert feature point sets, comparing a gray average value with a threshold value of the gray average value to obtain a screened insert feature initial region, performing texture consistency screening on the insert feature region, and determining a precise insert region based on a maximum extremum stable region; and carrying out boundary description on the precise embedded object region by adopting a level set algorithm.

The invention discloses a multi-category change dynamic threshold detection method based on multiple features of remote sensing images. First, the spectral change characteristics and the spectral gradient difference characteristics of images in two periods are extracted and superimposed by band, a principal component analysis is made of the characteristics after superimposition, the first two principal components are selected as final change characteristics, and the change intensity and the change direction are calculated. Then, the change intensity threshold is calculated under a polar coordinate frame, the pixels are divided into an unchanged part and a changed part, a threshold is set for the change direction of the changed pixels, and the change categories are identified. On the basis, the initial change detection result is analyzed, pixel sets containing only a single change are generated for all the change categories in turn, an optical change intensity threshold for each change category is acquired, and a change detection result based on a dynamic threshold is generated. Finally, the detection result is post-processed using a multi-phase level set algorithm. The detection accuracy of multi-category change is improved.

The invention relates to the technical field of image processing, in particular to an effective area extraction method of a fisheye image. The method comprises the steps of S1, obtaining the fisheye image, and performing image gray processing on the fisheye image to generate a gray fisheye image; S2, calculating a circle center coordinate (C'x, C'y) and a radius R' of an effective area of the grayscale fisheye image through a level set algorithm; and S3, calculating and extracting the effective area of the gray scale fisheye image according to the circle center coordinates (C'x, C'y) and theradius R ' of the effective area of the gray scale fisheye image. According to the effective area extraction method of the fisheye image, the effective area can be accurately extracted under the condition that the edge of the fisheye image is blurred or the light spots exist, and therefore the correction effect on the fisheye image can be improved in an auxiliary manner.

The invention discloses a multi-AUV three-dimensional collaborative route planning method. The method mainly comprises the steps of three-dimensional environment abstract modeling, level set functioninitialization, level set function evolution, optimal route point selection, collaborative route scheme design, route conflict judgment and re-planning and planning result output. According to the invention, the level set algorithm is improved; a plurality of airways can be simultaneously planned in a navigation space, thereby improving planning efficiency of algorithms, all AUVs are required to reach the end point at the same time during collaborative scheme design; each AUV needs to start one by one according to delay time, in addition, under the condition of considering rapidity and concealment, a collaborative planning scheme with the concealment and navigation time fused optimally is designed, the influence of ocean current and sound velocity factors is considered at the same time, conflict judgment and re-planning links are added, and a collaborative route is safer; compared with two-dimensional air route planning, the implementation of the three-dimensional air route planning ismore practical, and the actual navigation requirement can be better met.