61results about How to "Reliable segmentation" patented technology

In the invention, the only information depended on by the segmentation is the motion vector field based on 4*4 block uniform sampling and extracted from the H.264 video signals. In the first, the normalization is made for the motion vector field of continuous multi frames, and the iteration back-projection is made for the motion vector field in order to get the accumulated motion vector field used for enhancing the salient motion vector; then, making global motion compensation for the accumulated motion vector field, and meanwhile using fast region-growing algorithm to segment it into multi regions according to motion similarity. By the invention, a computer with 3.0 GHz CPU main frequency and 512M memory can smoothly process the CIF format video sequence and play video stream in 25 fps with good segmentation quality. The invention can also be used in motion object segmentation of MPEF compressed domain.

The invention provides a method for segmenting a white blood cell image, which is characterized by comprising the following steps: firstly carrying out binaryzation on the green component picture of a color white blood cell image to obtain an initial interested region based on a significance attention mechanism of human vision; carrying out cancellation and mergence on a region by labeling the initial interested region to obtain an adaptive significance window of each cell; and finally realizing segmentation of nucleuses and cytoplasm in each adaptive significance window by a boundary-extending method. The segmentation method is used to realize the fast segmentation of the white blood cells, particularly the accurate segmentation for high-capacity pictures which are the non-standard dyed and contain overlapped cells.

The invention discloses an aggregated white blood cell segmentation counting system. The system comprises an image acquisition module for dyeing white blood cells in a blood sample, dissolving red blood cells in the blood sample by using red blood cell lysate and acquiring a white blood cell image, an image preprocessing module used for performing image background removal on the white blood cellimage and obtaining an optimal segmentation threshold by using a maximum inter-class variance method and roughly segmenting a white blood cell region, an aggregated cell determination module used forobtaining a coarse segmentation image according to the rough segmentation of the white blood cell region, setting a discriminant function of a cell area and obtaining a multi-cell aggregation region,and an aggregated cell segmentation counting module used for extracting a cytoskeleton in each aggregation region and a gray curve at the cytoskeleton by using a morphological refinement method. According to the invention, by analyzing the gray scale characteristics of various white blood cell areas under a low power microscope, an adaptive threshold function is constructed, while a white blood cell count is obtained, the number of oxyphil cells is obtained, the cells in the aggregation region are quickly and accurately divided and counted, the method is quick and simple and is easy to implement.

The invention provides a multi-target interactive image segmentation method and device. The method comprises the following steps that (1) an image to be segmented is inputted, and a duplicate image of the image to be segmented is established; (2) an interactive seed point is selected in the image to be segmented, and a target area gray scale section is confirmed according to the interactive seed point and the initially set similarity rule; (3) a global threshold is obtained according to the gray scale histogram curve of the image to be segmented, and the duplicate image is segmented according to the relation between the target area gray scale section and the global threshold; (4) connection analysis is performed on a target area in which the interactive seed point is positioned according to the interactive seed point so that a target area template in which the interactive seed point is positioned is obtained; and (5) the steps (2)-(4) are repeated so that the required segmented multi-target template is obtained, and the image to be segmented is filled according to the multi-target template. Reliable segmentation of the target area can be guaranteed by the technical scheme, and interactive multi-target combined segmentation and management can also be realized.

The invention discloses a three-dimensional segmentation method for an intravascular ultrasound (IVUS) image sequence, which is used for improving the segmentation processing efficiency of an image sequence. The technical scheme comprises the following steps of: first, performing preprocessing of filtering noise and inhibiting a ring halo pseudomorphism on an original image; then, acquiring four longitudinal views of the IVUS image sequence and extracting intravascular cavity boundaries and intermediate-outer film boundaries from the longitudinal views; next, acquiring initial boundaries in transverse views by mapping boundary curves to IVUS images of each frame; and finally, acquiring the intravascular cavity boundaries and the intermediate-outer film boundaries in the IVUS images of the each frame finally by maximizing an energy function and discontinuously deforming the initial boundaries. Compared with a conventional method, the three-dimensional segmentation method has the following advantages of: first, capacity of utilizing the information of an overall image sequence; and second, capacity of finishing segmenting images of the each frame at the same time and realizing parallel processing of the overall image sequence, so as to greatly improve processing efficiency and shorten processing time.

The invention discloses a method for performing dentition segmentation on a cone beam CT image (CBCT). The method for performing dentition segmentation on a cone beam CT image comprises steps of defining an image structure according to an interest body image area in the cone beam CT image, obtaining a complete dentition through segmenting the CBCT image on the basis of a semi-supervised random walk algorithm and a soft constraint which is defined and registered by a three-dimensional deformable model, using the three-dimensional deformable model to introduce a soft constraint of a body image for processing noise in body image segmentation based on semi-supervised mark diffusion, adopting an iteration correction method to perform iteration solving problems on mark diffusion under the soft constrain and fitting on a surface voxel set by the three-dimensional model. The method for performing dentition segmentation on the cone beam CT image can effectively eliminate a segmentation error, improves dentition segmentation obtained through single time mark diffusion, improves segmentation accuracy and meets an accuracy requirement for clinic stomatology.

The invention relates to segmentation of cardiac acoustic signals, such as the heart sound signal, based on statistical algorithms. A duration-dependent Hidden Markov Model is disclosed which models the shifting states of the heart, based on the cardiac acoustic signal and the time spent in the given states relating to physiological events, e.g. the various states of the heart during the heart beat cycle.

The present invention discloses an automatic segmentation method for retinal nerve fiber layer in OCT image of ocular fundus, comprising the following steps: A. determining the inner layer outline u(s) of the retinal region as u(s)=[x(s), y (s)], and the outer layer outline v(s) as v(s)=[x(s), y(s)]; B. describing the retinal region in local structure data p (x, y) mode, and constructing the energy function (the formula is shown above) based on the local structure data p (x, y) after constructing the level set function phi; and C. minimizing the energy function, calculating the boundary curvem as m= {(x, y) / phi (x, y) = 0} while phi (x, y) = 0, and defining the region between the boundary curve m(s) and the outer layer outline v(s) as the retinal nerve fiber layer region. According to the invention, retinal nerve fiber layer in OCT image of ocular fundus can be segmented reliably and accurately.

The invention relates to segmentation of cardiac acoustic signals, such as the heart sound signal, based on statistical algorithms. A duration-dependent Hidden Markov Model is disclosed which models the shifting states of the heart, based on the cardiac acoustic signal and the time spent in the given states relating to physiological events, e.g. the various states of the heart during the heart beat cycle.

The present invention proposes an ultrasound system and method of identifying a vessel of a subject. The ultrasound system comprises: an ultrasound probe configured to simultaneously acquire a sequence of ultrasound blood flow data frames (such as a sequence of ultrasound Doppler data frames) and a sequence of ultrasound B-mode data frames of a region of interest including the vessel over a predetermined time period; a blood flow region selecting unit configured to select a blood flow region in the sequence of blood flow data frames; and a vessel segmenting unit configured to segment the vessel in at least one frame of the sequence of ultrasound B-mode data frames based on the selected blood flow region. Since there is no need to manually place any seed point for vessel segmentation any more, the user dependency is reduced and a fast measurement is made possible. Furthermore, since the segmentation of the vessel in a B-mode data frame makes use of information from multiple blood flow data frames, more robust, reliable segmentation can be achieved.

A method is disclosed for segmenting anatomical structures, in particular the coronary vessel tree, from 3D image data. In the method, a starting point is initially set in the 3D image data, and at least one known anatomically significant point and / or at least one known anatomically significant surface are / is identified in the 3D image data. Subsequently, proceeding from the starting point the structure is subsequently segmented pixel by pixel with the aid of a multiplicity of segmentation steps in such a way that an instantaneous distance is determined automatically relative to the anatomically significant point and / or to the anatomically significant surface in each segmentation step. Further, segmentation parameters and / or a selection of adjacent pixels for continuing the segmentation are / is established as a function of the distance, taking account of a model topology. The method enables an accurate and reliable segmentation of the anatomical structure.

Provided is a method for processing a substrate wherein occurrence of a defective article due to adhesion of an end material region to the side of terminal cut surface is eliminated. In a method for processing the substrate of a mother board, a mother board is scribed from the opposite sides of first and second substrates by means of a cutter wheel and divided for each unit display panel, and terminal processing for exposing the terminal region of each unit display panel is performed. When scribing of a terminal region (T) between a just cut surface (Ca) and a terminal cut surface (Cb) of a mother board is performed, (a) at first, a first cutter wheel is brought into pressure contact with a position on the terminal cut surface of a first substrate (G1) and a second cutter wheel is broughtinto pressure contact with a position on the just cut surface of a second substrate (G2) in order to perform scribing simultaneously on the opposite sides, (b) and then, the first cutter wheel is brought into pressure contact with the just cut surface of a first substrate (G1) and a backup roller having no edge is brought into pressure contact with the vicinity of just cut surface of a second substrate in order to perform scribing on one side.

The invention provides a method for processing a wafer, capable of dividing the wafer into a plurality of devices. The wafer has a device area and a peripheral residual area on the surface, the method for processing the wafer includes: a step of grinding the wafer for grinding the back side of the wafer corresponding to the device area to a predetermined thickness to form a ring-shaped enhancing part at the back side corresponding to the peripheral residual area, a step of supporting the wafer for bonding a cutting belt on the back side of the wafer and bonding the peripheral part of the cutting belt on a cutting frame so as to support the wafer by the cutting frame, a step of dividing the wafer for holding the wafer on a chuck workbench locating a cutting tool at the predetermined dividing line and cutting the predetermined dividing line so as to divide the wafer into a plurality of devices, a step of removing the ring-shaped enhancing part for rotating the chuck workbench when the cutting tool is located at the boundary of the device area and the peripheral residual area and cutting off the ring-shaped enhancing part, and a picking step for picking up the plurality of devices from the cutting belt.

A detection method for a spot image based thin line detection is disclosed. The method includes a step for constructing a band limited spot image from a transmitted and reflected optical image of the mask. The spot image is calibrated to reduce noise introduced by the one or more inspection systems. Based on the band limited spot image, a non-printable feature map is generated for the non-printable features and a printable feature map is generated for the printable features. One or more test images of the mask are analyzed to detect defects on such mask. A sensitivity level of defect detection is reduced in areas of the one or more test images defined by the non-printable feature map, as compared with areas of the one or more test images that are not defined by the non-printable features map

The invention discloses a video defogging method based on spectral clustering. The method specifically comprises the following steps: 1, acquiring, by a camera, a foggy video; 2, judging whether the current frame image Ik acquired in step 1 is a first frame image I1 of the video, if so, carrying out step 3, otherwise, carrying out step 4; 3, estimating global atmospheric light A for the first frame image I1, performing spectral clustering division, and calculating the transmittance of each cluster; 4, estimating the transmittance for video images from the second frame; and 5, recovering a frame of image according to the estimated global atmospheric light and transmittance. The video defogging method based on spectral clustering better ensures the spatial consistency of video frames, weakens the block effect of video images after defogging recovery, better ensures the continuity of video frames and avoids the scintillation effect among the video frames.

The invention provides an automatic image segmentation method and device. The method comprises the following steps of A, performing image pixel segmentation based on a graph theory method, so that theimage is segmented into different areas; B, performing under-segmentation detection on the segmented image areas, and correcting the segmented image areas; and C, analyzing the similarity between every two segmented image areas after the under-segmentation detection, and combining the segmented image areas if the similarity is sufficient. Through the automatic image segmentation method and device, rapid, accurate and reliable image area segmentation can be achieved, and a segmentation result is closer to a human visual effect.

The invention provides a warhead fragment perforation and pit detection method, which belongs to the technical field of image processing, and comprises the steps of performing geometric correction andactual physical size recovery on an equivalent target plate image acquired after static explosion of a warhead based on a homography matrix of photography transformation; linearly segmenting the equivalent target plate based on vertexes and boundaries; smoothing the equivalent target plate image based on guided filtering; generating a hyperboloid threshold value by adopting an image gradient in combination with changes of flash lamp illumination at different positions on the equivalent target plate; segmenting a fragment perforation region with bright gray scale and a fragment pit region withdark gray scale on the equivalent target plate by utilizing a hyperboloid threshold value; taking fragment perforation and pit areas as seed points of gray similarity to perform area growth; and removing the pseudo target region by using fragment perforation and pit area and shape feature criteria. According to the calculation method, fragment perforation and pit state information can be stored,fragment perforation and pit areas can be accurately segmented, and fragment perforation and pit detection can be conveniently and quickly realized.

According to the present invention, there is provided a cutting device capable of maintaining an appropriate indentation amount of a cutting bar without being affected by the thickness of a semiconductor substrate, a protective film, and a dicing tape. The cutting device presses the cutting bar (7) into a semiconductor substrate (W) having a scribe line (S) on the surface to divide the substrate, one side of the semiconductor substrate (W) is covered with a protective film (3), and a dicing tape (2) is attached to the opposite side. The structure of the cutting device includes a workbench (4) for placing the semiconductor substrate (W) together with the protective film (3) and the dicing tape (2); a displacement gauge (10) for measuring the surface height of the protective film (3) or the dicing tape (2) placed on the upper surface side of the semiconductor substrate (W) of the workbench (4), and a computer (C) control part (11) that subtracts the displacement amount to the specified indentation amount of the cutting bar (7) to enable the cutting bar (7) to be lifted and lowered under the condition that the measured value obtained by the displacement gauge (10) has displacement relative to the pressing starting position of the cutting bar (7).

An ultrasound system for identifying a vessel of a subject comprises: an ultrasound probe configured to simultaneously acquire a sequence of ultrasound blood flow data frames (such as a sequence of ultrasound Doppler data frames) and a sequence of ultrasound B-mode data frames of a region of interest including the vessel over a predetermined time period; a blood flow region selecting unit configured to select a blood flow region in the sequence of blood flow data frames; and a vessel segmenting unit configured to segment the vessel in at least one frame of the sequence of ultrasound B-mode data frames based on the selected blood flow region. Since there is no need to manually place any seed point for vessel segmentation any more, the user dependency is reduced and a fast measurement is made possible.

The invention provides a wafer processing method. A plurality of modified layers, which are not disposed in a wafer in a laminated manner, are capable of segmenting the wafer reliably. The wafer processing method comprises steps that the front surface of the wafer is provided with a plurality of segmentation predetermined lines in the shape of a grid, and the devices are formed on a plurality of areas divided by the above mentioned lines, and the wafer is segmented into various devices along the segmentation predetermined lines; according to an adhesive tape bonding process: the front surface of the wafer is provided with an adhesive tape in an adhesive manner; according to a modified layer forming process: the back surface side of the wafer is used to locate the focusing point of the pulse laser having the transmitting wavelength in the inner side, and is used to irradiate the pulse laser along the segmentation predetermined lines to form the modified layers; an adhesive tape heating process: the adhesive tape attached to the front surface of the wafer after the modified layer is heated, and then the cracks are extended from the modified layer to the front surface of the wafer; according to a segementation process: external force is applied to the wafer after the adhesive tape heating process, and then the wafer is segmented into various devices along the segmentation predetermined lines having the modified layers and the cracks.

The invention provides a laminated wafer processing method and an adhesive piece. In a plurality of chip laminated wafers stacked on wafers, in a state of adhering chip sides on the adhesive piece, wafers can also be reliably cut into a plurality of laminated chips. A paste layer (22) of an adhesive piece (2) correspondingly forms projections (3) on the peripheral residual area (1B) of a paste layer (22) and a laminated wafer (1) is adhered to a chip (15) of the laminated wafer (1); the projections (3) and the peripheral residual area (1B) are correspondingly adhered on the surface (10a) of a wafer (10); the peripheral residual area (1B) of the wafer (10) is supported through the projections (3). In the state, a cutting start point (modified layer 10c) of the wafer (10) is formed along a cutting predetermined line (11), and the wafer (10) is applied with outer force through the expansion of the adhesive piece (2), and accordingly, a laminated chip (1c) on the outmost side can be cut the same as other laminated chips (1c).

The invention relates to the technical field of medical image processing, and provides a brain nuclear magnetic resonance image segmentation method and system. The method mainly comprises the steps of brain nuclear magnetic resonance image data preprocessing, brain nuclear magnetic resonance image segmentation model construction and optimization, brain nuclear magnetic resonance image segmentation and result output. The system comprises a computer processor, a memory, a brain nuclear magnetic resonance image preprocessing unit, a brain nuclear magnetic resonance image segmentation model training unit and a brain nuclear magnetic resonance image segmentation and result output unit. According to the method, firstly, a space structure network of brain nuclear magnetic resonance images is constructed, pixel point information of brain function tissue activation images can be recorded, the space structure relation between the pixel point information and the space structure network can be effectively expressed, and then a brain nuclear magnetic resonance image segmentation model is established through an image variation auto-encoder structure; therefore, the model has a certain generation capability and can obtain a brain nuclear magnetic resonance image segmentation result with higher robustness and interpretability.

The invention provides upper and lower tooth occlusion simulation method and device and electronic equipment. The method comprises the following steps: acquiring a target three-dimensional image comprising a three-dimensional image of an oral cavity of a patient and a three-dimensional image of an imaging marker of an oral cavity positioning tool; according to the three-dimensional image of the imaging marker, determining an upper and lower tooth interface in the target three-dimensional image in the target three-dimensional image; dividing the target three-dimensional image through the upper and lower tooth interface to obtain an upper tooth area and a lower tooth area; and rotating the upper tooth area and / or the lower tooth area by taking a jaw joint axis in the target three-dimensional image as a rotation axis to generate an upper and lower tooth occlusion image. Therefore, when a doctor needs to observe the patient, the upper and lower tooth interface of the patient can be quickly determined by utilizing the three-dimensional image of the imaging marker in the target three-dimensional image, so that the upper tooth area and the lower tooth area can be quickly and accurately determined, quick upper and lower tooth occlusion simulation can be realized, the convenience of observing the upper and lower tooth occlusion condition of the patient in the diagnosis and treatment process is greatly improved.

There is provided a vertical crack forming method and a dividing method in a brittle material substrate, which can form a vertical crack on a brittle material substrate with high reliability. A method of forming a vertical crack on a brittle material substrate, comprising: a groove forming step of forming a groove line as a linear groove on a main surface; and an auxiliary line forming step of forming a groove And the outer circumference of the screed wheel is provided with a plurality of grooves arranged at equal intervals at intervals, and in the groove forming step, a groove line is formed so as to maintain a crack-free state in a straight line, In the forming step, a plurality of grooves are formed using a plurality of grooves formed on the outer circumference asymmetrically with respect to the outer circumference to form an auxiliary line, and the intersection of the two lines is used as a starting point for extending the vertical crack from the trough line.

A divider which divides a wafer having a division start points formed along the scheduled divisions into a plurality of device chips. The divider includes a placement table on which a wafer is placed, and division unit adapted to divide the wafer on the placement table into a plurality of device chips starting from the division start points. The placement table includes: a plurality of spherical bodies having the same diameter; a container that accommodates the plurality of spherical bodies in close contact with each other; and a placement surface formed by connecting vertices of spherical surfaces of the plurality of spherical bodies that are accommodated in close contact with each other.

An adaptive automobile meter detection method based on character segmentation cascade classifier which processes threshold segmentation, morphology and connected components analysis on the original image; extracts the pointer based on the contour analysis method, establishes a pointer information list; constructs a character segmentation cascade classifier by combining a HOG / SVM character segmentation classifier, a character filter and a CNN digit classifier. The character segmentation cascade classifiers is used to identify the digit character area of the automobile meter. Region analysis is performed to extract tick marks based on the center of the digit character area. The angular position corresponding to the tick mark is determined. The response value corresponding to the pointer is determined by establishing the Newton interpolation linear description relationship between the pointer angle and the response value and then the meter is classified as pass or fail. The invention is suitable for the field of visual inspection of automobile meter pointers.

The invention provides a laminated wafer processing method and an adhesive piece. In a plurality of chip laminated wafers stacked on wafers, in a state of adhering chip sides on the adhesive piece, wafers can also be reliably cut into a plurality of laminated chips. A paste layer (22) of an adhesive piece (2) correspondingly forms projections (3) on the peripheral residual area (1B) of a paste layer (22) and a laminated wafer (1) is adhered to a chip (15) of the laminated wafer (1); the projections (3) and the peripheral residual area (1B) are correspondingly adhered on the surface (10a) of a wafer (10); the peripheral residual area (1B) of the wafer (10) is supported through the projections (3). In the state, a cutting start point (modified layer 10c) of the wafer (10) is formed along a cutting predetermined line (11), and the wafer (10) is applied with outer force through the expansion of the adhesive piece (2), and accordingly, a laminated chip (1c) on the outmost side can be cut the same as other laminated chips (1c).

The invention discloses a multi-line laser radar ground segmentation method, a vehicle and a computer readable medium, the multi-line laser radar ground segmentation method is applied to automatic driving of the vehicle, and comprises the following steps: obtaining a preprocessing result of ground detection data of a multi-line laser radar; obtaining preliminary segmentation data of ground segmentation according to a preprocessing result; and obtaining final segmentation data of ground segmentation according to the preliminary segmentation data, and realizing ground segmentation according to the final segmentation data. According to the multi-line laser radar ground segmentation method disclosed by the invention, an automatic driving vehicle can adapt to a driving environment with violentpavement fluctuation and a complex road, and the adaptability is very strong; the ground object data of the next wire harness is segmented through the recorded actual elevation included angle betweenthe adjacent wire harnesses, so that the ground object segmentation is more accurate and reliable; in addition, through multiple times of confirmation of the ground object segmentation state and further ground object segmentation screening, the possibility of false identification of environmental perception of the vehicle can be greatly reduced.

The invention provides a method for manufacturing a chip, which can reliably divide a wafer by inhibiting abnormal extension of cracks and can reduce the number of times that the front surface and the back surface of the wafer are turned over. The chip manufacturing method comprises the following steps: a modified layer forming step of irradiating a first laser beam along a division predetermined line in a state that the back side of a substrate in a wafer having the substrate and a laminated body is exposed and the focal point of the first laser beam having a wavelength penetrating through the substrate is positioned in the substrate from the back side of the substrate, forming a modified layer and cracks; a grinding step of grinding the back surface side of the substrate exposed in the modified layer forming step after the modified layer forming step to thin the wafer to a predetermined thickness; and a laser processing groove forming step of irradiating a second laser beam having a wavelength absorbed by the substrate from the front surface side of the wafer along the division predetermined line after the grinding step, and forming a laser processing groove in the laminated body.