62results about How to "Achieve precise segmentation" patented technology

The invention discloses a principal vector analysis based broken bone section segmentation and broken bone model registration method. The method comprises the following steps of S1: performing axis extraction of a broken bone model by adopting a principal vector analysis algorithm; S2: extracting a grid vertex set of a broken bone section according to normal vector mutation of a triangular patch on a broken bone three-dimensional grid model and an included angle between the triangular patch and the broken bone axis; S3: performing spatial rough registration on the broken bone three-dimensional grid model in combination with methods for alignment of the broken bone axis and alignment of a principal vector of the grid vertex set of the broken bone section; S4: performing multi-time iterative computation for grid vertex sets of two broken bone sections by utilizing an iterative closest point algorithm to realize spatial fine registration of the broken bone three-dimensional grid model; and S5: according to a complete skeleton model obtained by broken bone three-dimensional grid model registration, performing fitting of a fracture steel plate model. According to the method, the grid vertex set of the broken bone section can be accurately segmented, rough consistency of the sections is realized in rough registration, the precision of fine registration is improved, and the success rate of fine registration is increased.

The invention discloses an auroral oval segmenting method based on a brightness self-adaptive level set, which mainly solves the defects of the existing auroral oval segmenting method that the segmentation precision is low, the robustness is poor and the application range is small. The auroral oval segmenting method comprises the following steps of (1) adopting a morphology component analysis method to preprocess an ultraviolet aurora image; (2) establishing a morphology saliency map to be used as shape characteristics of the auroral oval; (3) utilizing the marginal curve of the morphology saliency map to initialize a level set function; (4) calculating the brightness self-adaptive level set evolution speed and a stop function; (5) updating the level set function according to the brightness self-adaptive level set evolution equation; and (6) extracting a zero level set curve after being updated and utilizing the zero level set curve as the auroral oval margin to be outputted. Due to adopting the auroral oval segmenting method, the phenomenon of the traditional segmenting method such as result deviation and margin leakage can be avoided, advantages such as high segmentation precision and strong robustness can be achieved, and the method is applicable to the segmentation of different ultraviolet auroral images.

The invention relates to an eye fundus image vessel segmentation method based on a local enhancement active contour module. The method comprises: according to a feature vector of a Hessian matrix, vessel enhancement in an eye fundus image is carried out; curvature distribution statistics is carried out on the enhanced image to obtain an eyeball edge and get rid of the eyeball edge; and with a local enhancement active contour module, enhanced image segmentation is carried out by combining global energy information, thereby extracting an eye fundus vessel. According to the invention, on the basis of gray level distribution characteristics of the vessel in the medical eye fundus image, a local energy function is established and a global energy functional item is combined, so that the curve evolution process becomes stable; the speed is accelerated; and the vessel in the eye fundus image can be extracted precisely and effectively.

The invention discloses an image segmentation method of a local energy functional and a non-convex regular term combined with local entropy, comprising the following steps: (1) reading an image I (x,y); (2) initializing the parameters of the local energy functional and the non-convex regular term based on the local entropy in the image segmentation model; (3) calculating the local entropy hx of image I (x, y); (4) segmenting the image I (x, y) on the basis of the initialization parameters in the step (2), segmenting the image I (x, y) using a local energy functional based on the local entropyand an image segmentation model of a non-convex regular term, and updating the level set function phi in the segmentation process; 5) carry out evolution on that level set function phi according to the equation, jud whether the level set function phi converges or not, if so, stopping the evolution of the level set function phi, and outputting a segmented image; Otherwise, the process returns to step (4) to continue. The invention can efficiently and accurately segment gray-level non-uniform images.

The invention provides an auricle anaphora segmentation method and system, and the method comprises the steps: inputting a to-be-segmented human ear image and text description into an auricle anaphora segmentation model which comprises a text coding module, a text-guided visual coding module and visual decoding module, and an angle transformation module; the text coding module obtains a text feature # imgabs0 #; a visual coding module realizes fusion of text features and image features through a structure organized into four stages, a visual encoder in each stage generates visual features # imgabs1 #, a cross-modal sensing module aligns # imgabs2 # and # imgabs3 # to obtain multi-modal features # imgabs4 # and # imgabs5 #, each element in the multi-modal features # imgabs4 # and # imgabs5 # is weighted by an attention gating module to obtain weighted multi-modal features # imgabs6 #, and the weighted multi-modal features # imgabs6 # and # imgabs5 # are weighted by an attention gating module. Adding the elements and # imgabs7 # to generate an enhanced visual feature # imgabs8 #, inputting # imgabs9 # into a visual decoding module to gradually recover the spatial resolution of the image, and further fusing the text and the visual feature to output a multi-scale feature; the angle transformation module carries out angle transformation on the multi-scale features and outputs a segmentation mask of a text description related region. According to the invention, the auricle can be subjected to anaphora segmentation.

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 embodiment of the invention discloses a brain perfusion image segmentation method and device, a server and a storage medium. The method comprises the following steps: performing brain image segmentation on a pre-processed time sequence image to obtain a brain mask; determining a feature image according to the brain mask and the pre-processing time sequence image; Using the maximum gray scale projection image and the gray scale average image in the feature image to obtain a blood vessel mask; performing image standardization on the gray average image to obtain a standardized image; and segmenting the pre-treatment time sequence image superposed with the brain mask and the blood vessel mask according to the standardized image, the gray average image of the brain, the maximum gray projection image and the baseline mean value image before flowing into the contrast agent to obtain one or more of cerebrospinal fluid, gray matter and white matter. According to the embodiment of the invention, the problem of poor edge segmentation effect of different brain tissues segmented by the brain perfusion image in the prior art is solved, and the accurate segmentation of different brain tissuesin the brain perfusion image and the automation of image processing are realized.

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 discloses an occlusion tissue stripping method for a three-dimensional ultrasonic image, which comprises the steps of performing segmentation processing on a three-dimensional ultrasonicimage from a sagittal plane direction to obtain a plurality of slices; Carrying out contour key point identification on the plurality of standard slices by adopting a convolutional neural network toobtain key points of the standard slices; Fitting a cross section curve where each key point is located according to the same key point in each standard slice, and determining the key point at the corresponding position in the non-standard slice according to each cross section curve; Connecting the key points of the slices to obtain contour boundaries; And cutting the corresponding slices according to all the contour boundaries to obtain a plurality of cut slices, and synthesizing the cut slices to obtain a target three-dimensional ultrasonic image. The key points of the non-standard slices can be determined through the standard slices, so that the non-standard slices are cut. The invention further provides a shielded tissue stripping system, an ultrasonic detection device and a computer readable storage medium which have the above beneficial effects.

The invention discloses an image processing method and device and a storage medium. The method comprises the steps of extracting a target area image in an original pathological image; performing imagesegmentation on the target area image to obtain a focus area image, the focus area image comprising a focus area of the target area image; and marking a focus boundary of the target area image in thetarget area image or the original pathological image according to the focus area image. According to the image processing method, after the target area image is extracted from the original pathological image, accurate focus boundary detection can be conducted on the target area image, and therefore the image processing method used for achieving area-level accurate focus boundary detection is provided.

The invention discloses a device for splitting albedo and exocarp of tangerine peel, which belongs to the technical field of fruit processing apparatus. The device consists of an upper tangerine peel conveyor belt (1), a lower tangerine peel conveyor belt (2), a roll (3), a power unit (4) for the conveyor belts, a cutting tool (5), a cutting tool positioning device (6) and a power transmission unit (7) of the cutting tool. When in use, the tangerine peel with a radian structure is firstly pressed to be even by the device, then the exact position of the cutting tool is adjusted and fixed based on the thickness proportion between the albedo and the exocarp of the tangerine peel to obtain the best splitting effect, which avoids the problem of unevenly splitting the tangerine peel with different thicknesses by the common plane cutting tools. The device lays a foundation for splitting the tangerine peel effectively for respectively extracting and manufacturing pectin and tangerine essential oil with low cost and high quality. The device can be popularized and applied in fruit processing enterprises.

The invention discloses a semi-supervised industrial image defect segmentation method based on an adversarial generative network. A neural network is trained by using a small number of marked negativesamples with defects and a large number of positive samples without defects so as to obtain a segmentation network capable of automatically identifying the defects. In the construction process of theneural network, a segmentation network based on D-LinkNet and a reconstruction network based on U-net are respectively used, and feature spaces of a negative sample and a positive sample are separated in a cross training mode, so that the segmentation network can correctly segment defects in the negative sample. According to the method, dependence on industrial defect sample images can be greatlyreduced, and errors of a segmentation model during defect segmentation can be greatly reduced.

The invention discloses a Chinese ancient book character recognition method, a Chinese ancient book character segmentation, a layout reconstruction method, a medium and equipment, and the Chinese ancient book character recognition method comprises the steps: firstly obtaining a Chinese ancient book document image marked with a character bounding box and a character category, and taking the image as an original training sample; acquiring an annotation file of the original training sample; randomly selecting a plurality of original training samples, and processing the original training samples to obtain new training samples: processing the original training samples and the new training samples in an online random cutting mode to obtain a training sample set; training a character level detection classification model through training samples in the training sample set; and inputting a Chinese ancient book document image of which characters are to be recognized into the character level detection classification model to obtain a prediction bounding box and a prediction category of each character of the Chinese ancient book document image. According to the method, common characters can be recognized, some uncommon special characters in the ancient books can be recognized very accurately, and the problems of misjudgment, omission and the like existing in ancient book document recognition in the prior art are solved.

The invention relates to a human body part segmentation method and system based on SPECT imaging, based on a bone imaging data matrix, a human body region is extracted from a bone imaging image to obtain effective information of the bone imaging data matrix, the calculation amount is reduced, and the segmentation precision is improved. Then the human body area is divided into a head area, a chest area, a pelvic bone area and a leg area based on the data matrix corresponding to the human body area and human morphological characteristics, then the chest area is divided into a trunk area, a left upper limb and a right upper limb based on the data matrix corresponding to the chest area, and finally based on the data matrix corresponding to the leg area, the leg area is divided into the left lower limb and the right lower limb, so that the human body is divided into the head, the trunk, the four limbs, the pelvis and other parts, accurate segmentation of the human body parts is achieved, and accurate segmentation of the head, the trunk, the four limbs, the pelvis and other parts can be effectively conducted on any SPCET bone imaging image.

The invention discloses a multi-light-spot centroid extraction method and device based on an FPGA, a medium and a product, and relates to the field of digital image processing, and the method comprises the steps: firstly, caching image line data through a first-stage FIFO and a second-stage FIFO, forming three continuous groups of image line data streams, effectively solving the problems of data synchronization and processing rate matching, and improving the data processing efficiency; then, effective pixels of the light spots are accurately determined by means of a pixel threshold value, and different light spot areas are accurately divided by means of line continuity judgment and boundary marking, so that repeated recognition and error recognition of the light spots are avoided, and the accuracy of light spot recognition is greatly improved. And finally, according to the weighted average positions of the gray values of the effective pixels of all the light spots, the target centroid coordinates of the target light spots are calculated, the internal gray difference of the light spots is fully considered, and compared with a conventional method, the light spot energy distribution center is more accurately reflected, so that the calculation precision of the light spot centroid is remarkably improved.

The invention provides an early-stage esophageal squamous carcinoma detection method and system based on deep learning, and relates to the technical field of image processing, and the method comprises the steps: obtaining an esophageal endoscope image, carrying out the enhancement processing, carrying out the multi-stage decomposition, generating a first image, carrying out the background separation of the first image, generating a region of interest, and carrying out the morphological filtering and edge extraction; generating a binary mask image and performing normalization processing to obtain a standard image; extracting semantic features of the standard image, performing pixel-level classification on the semantic features to obtain a region segmentation result, calculating morphological features and textural features of an esophageal squamous cell carcinoma region, performing classification to obtain an esophageal squamous cell carcinoma classification result, adding the esophageal squamous cell carcinoma classification result as an input feature into a first classification model, and judging esophageal tissues according to a judgment standard to obtain an esophageal squamous cell carcinoma classification result; generating a first judgment result, performing nonlinear mapping and classification to obtain a second judgment result, and performing fine segmentation to obtain an accurate boundary of the early esophageal squamous carcinoma tissue.

A method for segmenting optic disc based on color retinal fundus image of lesion focus includes the following steps that the blood vessel detection model and optic disc detection model were established, the probabilistic map of blood vessel and the probabilistic map of optic disc were obtained from the detection model of blood vessel and the detection model of optic disc respectively; the probabilistic bubble map of the fitting straight line map of main blood vessel was obtained from the probabilistic map of blood vessel; and the optic disc region was selected from the optic disc communicationarea and the center and radius of optic disc were estimated. The method of the invention can effectively avoid interference of lesions, blood vessel occlusion, brightness change and the like in an image, thereby realizing accurate segmentation of an optic disc.

The invention discloses a microcirculation blood vessel automatic segmentation and quantification method based on enhanced feature fusion. The method comprises the following steps: step 1, data preparation and preprocessing; 2, designing an enhanced feature fusion network model; step 3, automatically segmenting microcirculation blood vessels; and 4, quantifying the diameter of the blood vessel based on the segmentation result. According to the enhanced feature fusion network model provided by the invention, accurate segmentation of microvessels is realized through fusion and enhancement of multi-scale features. The model shows good generalization ability on diversified clinical image data through effective data enhancement, and can adapt to image data with different illumination conditions, contrast ratios and resolutions. Therefore, the model has high practicability and reliability in practical application. According to the method, the diameter of the blood vessel is quantified through the segmentation result, the diameter of the microvessel can be accurately measured, and powerful support is provided for evaluation of the clinical microcirculation function.

The embodiment of the invention provides an optic nerve segmentation method and device in a magnetic resonance image, and the method comprises the steps: carrying out the image registration of a target magnetic resonance image, obtaining the spatial probability distribution information of the target magnetic resonance image, wherein the spatial probability distribution information comprises shapeinformation and position information of a visual pathway; predicting the target magnetic resonance image and the spatial probability distribution information based on a trained visual pathway segmentation model to obtain a visual pathway segmentation image of the target magnetic resonance image; wherein the trained visual pathway segmentation model is obtained by training a sample magnetic resonance image and sample space probability distribution information. According to the embodiment of the invention, the spatial probability distribution information of the visual pathway in the magnetic resonance image is acquired, and the visual pathway in the magnetic resonance image is segmented according to the shape information and the position information of the spatial probability distribution information, so that the problem that the boundary of the visual pathway is fuzzy is effectively overcome, and accurate segmentation of the visual pathway is realized.

The invention provides a fetal heart measurement method and device, and the method comprises the steps of obtaining a to-be-measured fetal heart ultrasonic cardiac four-cavity cardiac tangent plane image, and obtaining a to-be-segmented image; inputting a to-be-segmented image into the trained segmentation model to obtain a segmentation result image; and automatically measuring the size of each segmentation region in the segmentation result image. According to the invention, accurate segmentation of each structure of the fetal heart in the to-be-segmented image is realized. The size of each segmentation area in the segmentation result image is measured, automatic measurement is realized, the problem of low accuracy of the measurement result due to the fact that the measurement process is tedious and the operation individual only depends on other nonstandards in the prior art is avoided, the accuracy of the measurement result can be ensured, therefore, an expert can carry out rapid andaccurate diagnosis according to an automatic measurement result, conditions are provided for the ability of the expert to benefit more patients and regions, and the prenatal detection rate of the fetal congenital heart disease is improved.

The invention belongs to the technical field of image segmentation, and provides an image segmentation method and device, and a server, and the method comprises the steps: inputting a to-be-segmentedimage into an image segmentation model, carrying out the feature extraction of the to-be-segmented image to generate a feature map, and calculating the spatial position relation between pixel points in the feature map to obtain spatial position information; fusing the feature map and the spatial position information to obtain a feature map containing the spatial position information; and segmenting the to-be-segmented image according to the feature map containing the spatial position information, and outputting a target image. According to the embodiment of the invention, the problem that theboundary between different to-be-segmented targets cannot be segmented accurately is solved.

The invention discloses a defogging method and equipment for a foggy day traffic scene image, and the method comprises the steps: A, respectively calculating the atmospheric light values of corresponding regions in the foggy day traffic scene image according to different fog concentrations of a distant view region, a close view region and a transition region; then calculating a traffic scene imagesubjected to preliminary defogging according to an atmospheric scattering model by utilizing the transmission graph and the atmospheric light value of each channel in an HSI color space; B, on the basis of a preset I channel threshold value, performing global brightness improvement on the traffic scene image after preliminary defogging; wherein the preset I channel threshold value is obtained bysetting I channel pixels of a sky area in the traffic scene image subjected to preliminary defogging; wherein the sky area is obtained by segmenting the foggy day traffic scene image based on the darkchannel characteristics and the relative energy characteristics; and C, performing contrast-limited adaptive histogram equalization and guided filtering processing on the image obtained in the step Bto obtain a finally defogged traffic scene image. The method can be used for quickly and effectively defogging the traffic scene image.

The invention discloses an enterprise work task distribution system based on artificial intelligence, and relates to the field of enterprise task distribution, the technical scheme comprises a task system and a database, the task system internally comprises a central processing module, a task input module, a task identification system, a task matching system, a cloud storage module, a database and a supervision system, the external of the task system is in communication connection with a cloud storage module, the cloud storage module is in communication connection with a database, and a task input module inputs tasks inside an enterprise and transmits the input task information to a central processing module; task types can be identified through the task identification system, and accurate segmentation of the tasks can be realized through matching use of a task segmentation module, an employee ability evaluation module and an ability and task fitting module in the task matching system and comparison with previous tasks in a database.

The invention discloses a user grouping method. The method comprises the steps of determining one or more exclusive attributes of a category according to browsing behaviors of users purchasing commodities in the category in a first predetermined browsing time range; performing traversal combination on commodity attribute values of the exclusive attributes and commodity attribute values of other exclusive attributes to obtain user groups corresponding to all combinations; and allocating users with intentions of purchasing the commodities in the category in a second predetermined browsing time range to corresponding user groups. The exclusive attribute is a commodity attribute and has multiple commodity attribute values; the multiple commodity attribute values completely cover all the commodities in the category; and the commodity attribute values of the same commodity attributes selected by the same user group in a process of browsing the commodities in the category are most concentrated and have relatively low coincidence degree with other commodity attribute values of the commodity attributes. By adopting the method, the user groups can be distinguished according to real purchasedemands of the users.

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 discloses a method and a device for identifying molten steel at the tail end of a long nozzle, which are used for identifying the molten steel so as to judge the starting time of an automatic control program of the liquid level of an intermediate tank: in the production process, shooting and uploading an image at the position of the long nozzle above the intermediate tank in real time, detecting whether the long nozzle exists in the image by using a deep learning target detection algorithm, and after it is detected that the long nozzle exists in the image, a molten steel area is extracted from the tail end of the long nozzle in the image through the image processing technology, whether the molten steel flows out normally or not is recognized, and when it is recognized that the molten steel flows out normally, an intermediate tank liquid level automatic control program is started. The method not only can automatically judge and start the automatic control of the liquid level of the intermediate tank, but also can reduce the working intensity of operators.

The invention discloses a blood vessel segmentation method based on a geodesic distance map and a process function equation, and the method comprises the steps: enabling the front propagation based on the geodesic distance map to be expanded to the application of an asymmetric Finsler measurement condition, considering the anisotropy and asymmetry of a boundary, strengthening the description of a measurement function, and setting a freezing criterion of the front propagation; the situation of leakage in the anterior propagation process is effectively prevented, smooth anterior propagation is guaranteed, the target object can be found more accurately and quickly, and accurate segmentation of the blood vessel is achieved.

The embodiment of the invention relates to the field of image recognition, and discloses a super-capacitor electrode surface area detection method and system and electronic equipment, and the method comprises the steps: carrying out the multi-layer discrete wavelet decomposition of a collected image, and carrying out the nonlinear enhancement of a high-frequency coefficient of each layer according to different decomposition layers; performing reconstruction and noise reduction on the enhanced image by using a soft threshold method, wherein a threshold parameter is determined according to a decomposition level and a noise standard deviation; constructing a plurality of circular structure elements with different scales, extracting different aperture features correspondingly, and calculating a morphological gradient of each scale by adopting a Gaussian kernel; statistical features are calculated through a local window, a threshold value is determined, weighted fusion is conducted on the features of multiple scales, and the weight coefficient is decreased progressively along with increasing of the aperture. Through the scheme of the invention, the identifiability of micropore and mesoporous structures in the image can be improved, and accurate nondestructive detection of the surface area of the electrode is realized.

The invention provides a CT image tumor segmentation method and device based on anatomy guidance and a medium, and belongs to the technical field of medical image tumor segmentation. The method comprises the following steps: acquiring a human body CT original image, and performing pixel-level marking on the CT original image; performing data enhancement processing on the CT original image and the marked image; constructing a CT image tumor segmentation model, wherein the CT image tumor segmentation model sequentially comprises a feature sharing encoder, a feature guiding decoder and a pixel sharing output module; and training a CT image tumor segmentation model by using the data set subjected to data enhancement processing, collecting a to-be-segmented human body CT original image, preprocessing, and inputting into the CT image tumor segmentation model for tumor segmentation. According to the method, the complex mode of image data is fully utilized, the segmentation precision is improved, and the segmentation accuracy is remarkably improved.

According to the multi-modal zero-sample defect detection method based on the search engine and human guidance, the exogenous background image is automatically acquired through the search engine based on the keyword, and the acquisition process of the background image is greatly simplified; meanwhile, a textual feature extractor is enhanced by introducing a texture element network, so that the textual feature extractor can be combined with texture information in an exogenous image, accurate segmentation of the defect is realized when an unseen material or an unseen defect category is processed, and the limitation of the prior art in facing rare defects is overcome; that is to say, the problem that a traditional defect detection algorithm depends on a large-scale data set is solved by introducing a multi-modal zero sample learning method and combining feature alignment of the text and the exogenous texture image.