101 results about "Image deformation" patented technology

The invention discloses an identification photo generation method and electronic equipment, and relates to the field of image processing, after an original user image and an original template image (identification photo template) aiming at the original user image are acquired, image deformation is carried out on the original template image according to the original user image and an image deformation algorithm, and the identification photo is generated. A target template image which aims at an original user image and conforms to one or more of facial key point constraints, light and shadow consistency constraints and facial structure constraints is obtained, and the original user image and the target template image are input into a trained image generation model (obtained by optimizing and training a diffusion algorithm); the corresponding areas are processed according to the different processing weights corresponding to the face area and the non-face area, it is ensured that the obtained target identification photo is more natural, real and reliable in vision, the specification requirement of an actual scene for the identification photo is met, the calculation time and cost are reduced, and the requirement for large-scale identification photo batch generation is met.

The invention, which belongs to the technical field of remote sensing image processing and geological disaster monitoring, discloses a high-risk slope multi-temporal remote sensing image deformation monitoring and early warning system comprising a multi-platform remote sensing data acquisition module, a deep neural network registration module, a sub-pixel-level deformation detection module, a meteorological data fusion module and a dynamic early warning module. The centimeter-level registration precision is realized by introducing an adaptive feature registration mechanism based on a graph attention network, the deformation estimation precision of a texture sparse region is improved by adopting a time-space domain combined sub-pixel-level deformation field estimation algorithm, and the early warning timeliness is improved by establishing a dynamic early warning threshold self-adjustment mechanism fused with meteorological data. The system supports multi-scale multi-platform cooperative monitoring, the early warning advance period is prolonged by 7-15 days, and the system is successfully applied to high-risk slope monitoring projects along the expressway.

The invention belongs to the technical field of data processing, and provides a scale detection and scale conversion method and system based on an ultrasonic image, and the method comprises the steps: constructing a discrimination function fusing multi-modal features according to the spectral basis functions of different scales, and determining the scale contour of the scales in the ultrasonic image; the design can dynamically adapt to the spectral characteristics and structural characteristics of the scale, can effectively cope with the interference of scale diversity, ultrasonic speckle noise and the like of equipment of different manufacturers, and ensures the stability of scale positioning and subsequent conversion; the method comprises the following steps of: establishing a biomechanical-image deformation coupling model by taking a scale contour of a scale in an ultrasonic image as a geometric reference and combining biomechanical characteristics of an ultrasonic probe and tissues to obtain a corrected pixel distance between adjacent scale points of the scale; and curvature compensation for eliminating refraction errors of ultrasonic waves in curved tissues is introduced to correct a scale pixel-millimeter conversion coefficient, so that the pixel-millimeter conversion error rate is reduced to 1.2%, which is far lower than the level of a traditional method.

The invention relates to the technical field of automatic visual detection, and discloses a bottle body label quality detection-based model automatic generation method, which comprises the following steps of: acquiring a multi-angle image of a bottle body, carrying out denoising, enhancement and brightness equalization processing, extracting a curved surface label through target detection and a perspective transformation algorithm, and carrying out image processing on the curved surface label; further generating a distortionless complete label image through an image registration and fusion technology, and taking the distortionless complete label image as a training sample set of a label defect detection model constructed based on a double-branch deep learning network; and performing hot updating on the label defect detection model through a closed-loop feedback and continuous learning mechanism. According to the method, the technical problems of imaging deformation of the curved surface label, diverse defects, difficulty in detection and performance degradation after model deployment are effectively solved, and high-precision, high-robustness and sustainable-evolution automatic label quality detection is realized.

The invention provides an endoscope image stabilization control method, which comprises the following steps of: acquiring an original dithering image sequence and inertial data of an endoscope, fusing the original dithering image sequence and the inertial data, acquiring an initial mixed motion field, performing operation semantic segmentation on the original dithering image sequence, and acquiring a collaborative attention map in combination with a physical motion mechanism. Obtaining a space-time adaptive filtering kernel through the collaborative attention map, filtering the initial mixed motion field according to the space-time adaptive filtering kernel, obtaining a stable motion field, carrying out image deformation and interpolation synthesis processing on the stable motion field, and obtaining a stable image sequence; through the integrated technical scheme of operation semantic segmentation, dynamic image stabilization optimization and accurate jitter suppression, the problem of interference of endoscope jitter on the neurosurgery operation visual field is solved.

The invention provides a defect detection method and device, electronic equipment and a storage medium. The method comprises the following steps: acquiring a template image and a product image of a to-be-detected product; performing feature matching on the template image and the product image, and performing non-rigid alignment on the product image based on feature point pairs obtained by feature matching to obtain an alignment image of which the geometric view angle is consistent with that of the template image; and defect detection is carried out based on the aligned image and the template image to obtain a detection result, so that the problem of image deformation caused by the fact that the surface of the to-be-detected product is a curved surface and the shooting angle is not correct in a traditional scheme is effectively solved, and local and global deformation of the product image can be corrected, thereby greatly improving the alignment precision of defect detection, and improving the detection accuracy of the to-be-detected product. Therefore, the real difference of the product surface can be more focused during defect detection, non-defect geometric deformation interference is effectively shielded, and the accuracy and robustness of defect detection of complex surfaces of household electrical appliances and the like are greatly improved.

The invention relates to the technical field of medical treatment, in particular to a deformation registration method and system, a self-adaptive treatment planning method and a treatment planning system which are used for improving the accuracy of image deformation registration, and the method comprises the steps: obtaining a plan image, a fractional image and dose distribution data corresponding to an initial treatment plan of a patient; inputting the plan image, the graded image and the dose distribution data corresponding to the initial treatment plan of the patient into a trained deformation registration model, performing deformation registration on the graded image and the plan image, and outputting a deformation field; wherein the trained deformation registration model is obtained based on dose distribution consistency loss training between dose distribution data corresponding to the initial treatment plan and deformation dose distribution data corresponding to the deformation field.

The invention relates to an image registration method and device, electronic equipment and a storage medium, and the method comprises the steps: obtaining a first image and a second image obtained through image collection of a target physiological region, the first image data being a three-dimensional image, and the second image being a two-dimensional image; and determining a deformation field describing a mapping relation between the first image and the second image, and performing nonlinear deformation on the first image according to the deformation field to obtain a deformed image. And projecting the deformed image to obtain a two-dimensional third image capable of measuring a registration effect with the second image. According to the embodiment of the invention, through the architecture for image deformation and projection and image registration based on the deep learning automatic learning deformation field, the elastic registration between the images can be simply, quickly and accurately realized. Meanwhile, the image registration mode is realized by depending on a software algorithm without depending on the improvement of a hardware structure, so that the cost is low.

An apparatus includes a memory to store first volumetric image data representing a first volumetric image of an anatomical region having a volume of interest (VOI). The first volumetric image includes several first voxels at first locations in the first volumetric image. The first volumetric image data includes first uncertainty values corresponding to the several first voxels. The first uncertainty values represent probabilities that the corresponding plurality of first voxels are part of the VOI. The memory is to store second volumetric image data representing a second volumetric image of the anatomical region. The apparatus includes a processing device operatively coupled to the memory. The processing device is to determine a deformation field to map the several first voxels to second locations in the second volumetric image. The deformation field is based in part on the first uncertainty values.

The invention relates to the technical field of computers, in particular to a block chain right-confirming forestry remote sensing data ontology feature recognition method and system, and the method comprises the steps: extracting the multi-scale features of forestry remote sensing image data through a pre-trained deep convolutional neural network, and generating a feature pyramid with an unchanged rotation and scale; a time-space fusion feature vector is generated by combining seasonal attention weight adjustment, and forestry remote sensing image fingerprints are formed after principal component analysis and dimension reduction; a fuzzy commitment is adopted to bind the random key after error correction coding with a forestry remote sensing image fingerprint to generate a block chain evidence storage commitment, and a storage architecture with on-chain evidence storage and under-chain metadata collaboration is constructed; and efficient right confirmation is realized. The problem that a traditional Hash method is sensitive to image deformation is solved, the problem of feature drift caused by seasonal changes is solved, verifiable right confirmation is achieved, the method has the advantages of being high in storage efficiency, high in response speed and the like, a credible ownership authentication basis is provided for forestry remote sensing data sharing, and the data circulation efficiency is remarkably improved.

An information processing device includes: an own position / posture estimation unit that estimates a position and a posture of a device based on sensing information acquired by a sensor unit, and outputs own position / posture information; and an image deformation unit that performs deformation processing on an image based on the own position / posture information and distortion of an optical system included in the device.

The invention relates to the technical field of optical lenses, and discloses a small-head optical imaging system, and the optical imaging lens comprises a first lens, a second lens, a third lens, a fourth lens and a fifth lens which are sequentially arranged from the object side to the image side along an optical axis, and the object side surface of the first lens to the image side surface of the fifth lens are aspheric surfaces; the diaphragm is arranged in front of the first lens; the small-head high-definition imaging lens is of a five-lens type, the surface shape structures of all the lenses are combined with the optimal range of optical parameters, the head size of the imaging lens can be effectively shortened and the visual angle of the lens can be improved under the condition that imaging deformation is reduced while imaging is maintained, and the small-head high-definition imaging lens has high resolution brought by high pixels and is high in resolution ratio. The optical imaging lens is small in size and integrated, can adapt to the increasingly pursued light and thin design of a mobile phone, and enables a rear cover of the mobile phone to be flat as far as possible and not to protrude or only to slightly protrude.

The invention provides a chromosome straightening method and device, electronic equipment and a storage medium. The chromosome straightening method comprises the steps that a chromosome image forms a binary image; s2, acquiring a central axis of each single chromosome in the binary image; s3, segmenting the chromosome image into a plurality of first-level triangles; s4, mapping the ordinates of the first-level central axis sampling points corresponding to the first-level triangles to be the same to form second-level central axis sampling points; s5, according to transformation of each first-level central axis sampling point and the corresponding second-level central axis sampling point, obtaining a corresponding second-level triangle coordinate through the first-level triangle through a rigidity-keeping image deformation algorithm; calculating an affine transformation matrix corresponding to each first-level triangle according to transformation of the first-level triangles and the corresponding second-level triangles, and performing affine transformation on the chromosome image region corresponding to each first-level triangle to obtain a straightened chromosome image; the chromosome straightening method provided by the invention can effectively correct the chromosome image.

The embodiment of the invention provides a method and device for image processing, equipment, a storage medium and a system for augmented reality. In the method, grid information corresponding to the pose of the augmented reality device is acquired. The grid information is used for image deformation processing of the augmented reality equipment. A first set of pixels of a first image acquired by the augmented reality device in the first memory is caused to be stored to the second memory based on the grid information. The read speed of the second memory is higher than that of the first memory. Further, positional information of a first pixel in the first image corresponding to a second pixel in the image-deformed second image is determined, and a value of the second pixel is acquired by reading a value of the first pixel from the first memory or the second memory based on the positional information. The priority of reading the value of the first pixel from the second memory is higher than that of the first memory. In this way, the delay of video perspective of the augmented reality device can be effectively reduced.

In a learning data generation device, an acquisition means acquires a first image corresponding to an image obtained by photographing one or more objects arranged in a predetermined arrangement state, and first information corresponding to information indicating the predetermined arrangement state in the first image. An image deformation means deforms a second image corresponding to an image of a desired object. An image generation means generates a fourth image having an arrangement state that matches or is similar to the predetermined arrangement state indicated by the first image, by pasting a deformed second image to the first image or a third image corresponding to the image of a position where the desired object is arranged. A learning data generation means generates data including the fourth image and second information corresponding to the information indicating the arrangement state of the object in the fourth image, as learning data.

The invention discloses a full-automatic left atrial myocardial strain analysis method based on deep learning, and belongs to the field of medical image processing. The combined group registration and segmentation model comprises a group registration encoder, a segmentation encoder, an inter-task attention module, a group registration decoder, a segmentation decoder, an image deformation module and a mask deformation module; defining a loss function, wherein the loss function comprises registration loss, smoothing loss, segmentation loss and inter-task loss; constructing a heart movie image data set, and training and testing the joint group registration and segmentation model based on the heart movie image data set; and for a newly collected heart film image sequence, predicting a motion field sequence, an inverse motion field sequence and a mask sequence by using the trained model, and calculating a left atrial myocardial strain graph, a strain curve and a strain rate curve. According to the invention, the accuracy, efficiency and repeatability of strain analysis are improved.

Embodiments of the present disclosure provide a facial image deformation method and apparatuses, an electronic device, and a computer-readable media. A specific embodiment of the method comprises: determining an anchor point corresponding to a dragging operation in response to detecting the dragging operation on a target facial image, wherein the anchor point corresponds to a part displayed in the target facial image; determining an offset of the anchor point based on a movement amount corresponding to the dragging operation; and deforming the target facial image based on the offset of the anchor point.

The invention relates to the field of medical image processing, and discloses a medical image data processing method and system, and the method comprises the following steps: S1, obtaining a multi-time-point medical image sequence; s2, constructing a finite element mesh model and endowing the finite element mesh model with biophysical attributes; s3, solving a physical optimization problem to calculate a displacement field; s4, calculating a stress field, a strain field and a physical residual field; s5, outputting a basic analysis result; the system comprises an image acquisition module, a model construction module, a physical solving module, a multi-scale coupling module, a risk assessment module, a treatment planning module and an output module. According to the method, physical constraints based on a biomechanical model are constructed, and a physical residual field is calculated in combination with a reverse image deformation technology, so that the transformation from traditional geometric morphology measurement to deep information mining based on a biophysical model is realized, and the information dimension and accuracy of an image processing result are improved.

The invention discloses a three-dimensional non-rigid image splicing method, system and equipment applied to the medical field and a storage medium, and the method comprises the steps: enabling a collected three-dimensional image to form a two-dimensional redundant data maximum value intensity projection drawing based on a reserved redundant depth, and carrying out the registration of the projection drawing through employing an FFD transformation model, performing pixel resampling on the two-dimensional redundant data maximum value intensity projection drawing according to the coordinate offset to obtain a deformation field; performing nonlinear fitting on all the obtained deformation fields by using a sigmoid function to obtain a fitted deformation field; applying the fitted deformation field to an originally acquired three-dimensional image to perform XY direction deformation, and completing deformation correction processing of the three-dimensional image in the XY direction; according to the method, a three-dimensional non-rigid image with data dislocation is obtained, Z-direction deformation is converted into a maximum similarity search problem, then the number of similar layers of a three-dimensional data block under each grid is calculated through grid division, and correction of the three-dimensional non-rigid image with data dislocation is achieved through fusion in a distance weighting mode. Compared with a current image splicing technology, the method has the advantages that the splicing precision and the splicing speed are considered at the same time, and TB-level data can be efficiently processed while micron-level three-dimensional non-rigid image deformation correction can be achieved.

This invention discloses a method, apparatus, device, and storage medium for medical image deformation registration. The method includes: acquiring a first modal image and a second modal image, and generating a mask image based on the noise region features of the second modal image; constructing a hybrid similarity measure function based on the first and second modal images, the hybrid similarity measure function including a similarity measure term based on image context information and a similarity measure term based on local gradient features; iteratively optimizing the registration of the first and second modal images based on the hybrid similarity measure function and the mask image to determine the spatial transformation parameters for aligning the second modal image with the first modal image; and transforming the first modal image according to the spatial transformation parameters to obtain the registered first modal image. Compared with the prior art, this invention improves the accuracy and robustness of multimodal medical image deformation registration.

The invention belongs to the technical field of PET (positron emission tomography), and relates to a PET probe extensive field image multi-model adaptive deformation and FPGA (field programmable gate array) curing method, which comprises the following steps of: acquiring single-lift event data of a PET probe; generating an original extensive field image based on the single-lift event data; performing coordinate traversal transformation on the original extensive field images by adopting an image deformation model in a preset multi-model transformation library; performing off-line analysis on each model and the parameters thereof in the multi-model transformation library by utilizing a preset evaluation index, and screening out an optimal model and an optimal parameter; and transforming the original pan-field image by using the optimal model and the optimal parameters to generate an enhanced image and a crystal position lookup table, and solidifying the enhanced image and the crystal position lookup table to the FPGA. The method has the beneficial effects that the local adhesion area can be stretched in a targeted manner through the nonlinear transformation model, the distinction degree of crystal event clusters is remarkably improved, and the LUT calibration precision is fundamentally improved.

The present application relates to image processing, and provides an image processing method. The method comprises: acquiring an original image to be displayed; performing grid division on the original image to generate a target grid; stretching edge points of the target grid as control points to form an outer contour of a target image, and obtaining pixel coordinates corresponding to stretched edge points; traversing internal filling points in the target grid, and on the basis of the pixel coordinates corresponding to the stretched edge points and a grid difference algorithm, obtaining pixel coordinates of interpolation points corresponding to the target image and the internal filling points; and on the basis of the pixel coordinates corresponding to the stretched edge points and the pixel coordinates of the interpolation points, deforming the texture of the original image to obtain the texture of the target image. The image processing method of the present application can ensure the internal texture filling after image deformation to obtain a correct texture pattern. The present application further provides an image processing system, a vehicle, and a readable storage medium.

The invention relates to the technical field of image processing, in particular to a target picture and template picture specific area association matching method and system.The method comprises the steps that a template picture, a specific area mask corresponding to the template picture and a target picture are received as input, and multi-scale features are jointly extracted through a neural network model; and in combination with area prior information provided by the mask, predicting a central point offset proportion of a corresponding area in the target image. And calculating an accurate position coordinate in the target picture by combining the predicted offset based on the center coordinate and the width and height parameters of the specific region in the template, and further positioning the matched region and generating an association result. According to the method, the mask is fully utilized to guide the network to pay attention to the key area, and the matching robustness and the positioning precision under the condition that the image has deformation, illumination change or noise interference are improved. Compared with a traditional method, error accumulation caused by dependence on dense feature matching or key point detection is avoided, and end-to-end rapid and accurate area positioning is achieved.

This invention relates to a dynamic target detection and tracking method for unmanned surface vessel (USV) remote sensing images. It includes the following steps: a) employing adaptive sensor fusion technology to integrate radar, optical, and infrared sensors for all-weather, multimodal target detection; b) using deep learning spatiotemporal fusion technology, employing 3D convolutional neural networks or LSTM to capture the spatiotemporal features of dynamic targets; c) using unsupervised anomaly detection technology to train models for common maritime scenes to identify and track anomalous targets; d) implementing water surface reflection reduction technology to reduce the impact of water surface reflection and solar flare; e) applying an adaptive occlusion handling strategy, using a predictive model to maintain target tracking even when the target is occluded by waves or other objects; f) utilizing a multi-task deep network for target detection, classification, velocity estimation, and orientation prediction; g) implementing wave adaptive correction, using machine learning to correct image distortions caused by water surface waves in real time.

The invention relates to the technical field of oral cavity imaging, in particular to an intelligent electronic tongue depressor and a high-precision three-dimensional oral cavity imaging method. According to the structural features of different frames of optical slice images, the local deformation possibility of each matching feature point in each frame of optical slice image is obtained, and matching error feature points forming a deformation target area are screened out; for any frame of optical slice image, according to the local deformation possibility of different matching feature points in the deformation target area, the positions of the different matching feature points in the three-dimensional point cloud model and the change trend of a corresponding moving path when the tongue depressor intersects with the deformation target area, the matching error consistency of the deformation target area on each frame of optical slice image is obtained; and in combination with the point cloud coordinates of the matching error feature points in the deformation target area and the position distribution of the deformation gravity center, obtaining point cloud correction coordinates of the matching error feature points. According to the method, the point cloud correction coordinates of each image deformation area are accurately obtained, and the accuracy of oral cavity three-dimensional model construction is improved.

A projected image deformation correction result evaluation method and apparatus, a device, and a storage medium. The method comprises: obtaining corner point coordinates of a projected image, wherein the projected image is a quadrilateral, and the corner point coordinates are coordinates corresponding to four corners of the quadrilateral in a rectangular coordinate system; obtaining four included angles between four sides of the projected image and the corresponding sides of an absolute rectangle, wherein the absolute rectangle refers to that one group of opposite sides in the four sides of the rectangle is parallel to the X-axis in the rectangular coordinate system, the other group of opposite sides is parallel to the Y-axis in the rectangular coordinate system, the X-axis is the horizontal direction, and the Y-axis is the vertical direction; determining values of the four included angles on the basis of the corner point coordinates; and evaluating a correction result of the projected image on the basis of the values of the four included angles.

The invention belongs to the technical field of infrared-visible light real-time registration, and particularly relates to a real-time registration method for an infrared-visible light dual-spectrum camera, and the method comprises the following steps: collecting an infrared-visible light image to be registered; random affine transformation and random elastic deformation are applied to original infrared images in the strictly registered RoadScene and TNO data sets to obtain deformed infrared images, and the deformed infrared images and the deformed infrared images form an unregistered infrared-visible light image data set; the method comprises the following steps: establishing an improved Dense Matcher network; performing improved Dense Matcher network training by using the unregistered infrared-visible light image data set to obtain an optimal infrared-visible light image deformation field prediction model; and predicting a deformation field of the infrared-visible light image to be registered by using the optimal infrared-visible light image deformation field prediction model, and performing spatial transformation on the infrared image according to the deformation field to obtain a registered infrared-visible light image.

The embodiment of the invention relates to an image processing method and device based on face key points, equipment and a storage medium. The method comprises the steps of determining a reference face key point matrix and a reference effect point matrix based on a selected target special effect; the target special effect is a rendering effect generated by applying the target special effect file package to the face based on the position of the face key point; determining a target face key point matrix in the target face image; based on the reference face key point matrix, the reference effect point matrix and the target face key point matrix, image deformation processing is carried out in a vectorization mode, and a target effect point matrix corresponding to the target face image is determined; the target effect point matrix is a matrix formed by rendering key points when the target special effect file package is applied to the target face image. Therefore, the deformation calculation of the face key points is carried out in a vectorization mode, the efficiency of the deformation calculation of the face key points is greatly improved, and the application efficiency of the target special effect file package of the face is improved.

The invention discloses a longitudinal submerged arc welding pipe steel plate compression molding deformation detection image distortion repairing method. An image detection device, a PC terminal, an application server and a background database of the steel plate pressing deformation image online detection system are sequentially connected, application software is installed in the application server, and the PC terminal is in communication connection with the image detection device; the longitudinal submerged arc welding pipe steel plate compression molding deformation detection image distortion repairing method comprises the following steps: establishing a longitudinal submerged arc welding pipe steel plate compression molding model database; forming parameters are monitored, collected and analyzed in real time; calculating steel plate pressing deformation characteristic data; and predicting and guiding the rolling reduction of the forming die. The method has the beneficial effects that the steel plate pressing deformation detection image distortion repairing system is established, the welded pipe forming process is corrected in real time according to image deformation distortion monitored in the longitudinal submerged arc welded pipe forming steel plate pressing process, the oil and gas pipeline manufacturing process is optimized, the management and control capacity is improved, and the product quality is guaranteed.

The invention belongs to the technical field of image processing, and discloses a 4D-CBCT dynamic reconstruction method which comprises the following steps: constructing a priori motion model based on principal component analysis in an off-line manner for describing a main mode of organ respiratory motion; designing and training a deep learning network fusing a convolutional long short-term memory network and a double attention mechanism, wherein the network can learn nonlinear mapping from a two-dimensional projection image sequence to a PCA coefficient representing three-dimensional motion; in the online reconstruction stage, real-time projection is input into a network to predict a PCA coefficient, and a high-quality 4D-CBCT image sequence is quickly reconstructed through PCA inverse transformation and image deformation. According to the invention, the CBAM module enhances local features and inhibits noise in an encoder, and the SPA module carries out global context compensation after ConvLSTM. According to the method, the precision and robustness of 4D-CBCT dynamic reconstruction are improved, and complex respiratory movement can be accurately captured.