38results about How to "Improve 3D reconstruction accuracy" patented technology

The invention relates to an SFS (Shape From Shading) three-dimensional reconstruction sparse-DEM (Digital Elevation Model) encrypting method considering surface spectral information. The invention creatively provides a coarse-resolution DEM-grid interpolation-encrypting method by combining multi-spectral remote-sensing images with SFS three-dimensional reconstruction; ground-object spectral information is utilized to estimate the reflectivity of different ground-object types, and the influences of surface-vegetation coverage types on the precision of SFS reconstruction are removed; in the invention, multi-spectral images are utilized to replace full-color images, and the precision of three-dimensional reconstruction is improved by applying the assistance of the ground-object spectral information; the influences of the surface-vegetation coverage types and ground-object shadows on establishing a surface three-dimensional model and image spectral-brightness transforming relation can be effectively inhibited; and relative to a commonly-used DEM interpolation algorithm, the precision is markedly improved, and examples validate that the method can be used for improving the DEM spatial resolution to be two times of original data and finishing large-range rapid DEM establishment based on sparse control points.

The invention discloses a binocular visual sense reconstruction method considering three-dimensional distortion, belonging to the computer visual sense measurement technology field. The binocular visual sense reconstruction method considering the three-dimensional distortion introduces a three-dimensional distortion correction function to perform compensation optimization on a visual sense reconstruction result and realizes three-dimensional high accuracy remodeling. The binocular visual sense reconstruction method considering the three-dimensional distortion adopts a traditional binocular visual sense reconstruction method to obtain a three-dimensional reconstruction initial values of measurement points, solves a three-dimensional distortion compensation value on the basis of a relative relation between a standard object and a binocular camera and the reconstruction initial values, thus performs distortion compensation on coordinates of the camera of a tested point which are respectively extracted from a left image and a right image, combines with binocular visual sense theory to reconstruct, and realizes the binocular three-dimensional high accuracy reconstruction. The binocular visual sense reconstruction method considering three-dimensional distortion is high in reconstruction accuracy, can compensate the three-dimensional distortion of the binocular visual sense, realizes form and surface reconstruction of a part having a big curvature and a big size and improves three-dimensional reconstruction accuracy of the binocular visual sense.

The invention discloses a super-high-resolution three dimensional reconstruction method based on a global variation technology. The method comprises the following steps: low-resolution three dimensional geometrical model and a high-resolution surface normal graph of a scene to be constructed is obtained; a global optimization energy function is constructed and obtained according to the difference between the low-resolution geometrical model and the high-resolution normal graph; the minimum value of the global optimization energy function is solved, a depth value of each pixel point is obtained, and three dimensional reconstruction is carried out according to the depth values of all the pixel points. The super-high-resolution three dimensional reconstruction method based on non-local total variation regular constraint condition lowers dependence of a final reconstruction result on an original result effectively and raises the three dimensional reconstruction precision greatly.

The invention discloses a method for correcting kinematic errors in a microscopic vision system, which is applied to the microscopic vision system. The method comprises the following steps of: calibrating a camera through a calibration board, establishing a camera model, tracking angular points of the calibration board, generating a tracking trajectory and an ideal revolving trajectory of the angular points through a least square method, then performing error evaluation, establishing an error correcting model, and correcting the microscopic image by virtue of the error correct model. The method provided by the invention can be used for establishing the error correct model for the microscopic vision system only through one-time error evaluation; and the model can be used for any microscopic movement video collected by the system, reducing the movement errors of the system and improving the three-dimensional reconstruction precision, thus being high in practicability.

A 3D reconstruction method based on multi-field-depth image fusion belongs to the technical field of 3D reconstruction. The problem that the precision of 3D reconstruction is low and the true 3D morphology of a sample under test cannot be obtained is solved. The technical scheme includes the following steps: (1) acquiring a high-resolution original image sequence with different depth of field in the same position of a sample under test; (2) carrying out static wavelet transformation on the original image sequence to obtain corresponding high and low frequency coefficients; (3) designing corresponding fusion rules for the high and low frequency coefficients to obtain high and low frequency fusion coefficients, obtaining a fused image through inverse static wavelet transformation, and obtaining an initial 3D reconstruction result based on the high frequency fusion rule; (4) carrying out Gaussian low-pass filtering on the initial 3D reconstruction result to obtain a final 3D reconstruction result of the sample under test; and (5) mapping the fused image and the final 3D reconstruction result one by one to obtain a true 3D reconstruction result of the sample under test. The 3D reconstruction method of the invention has the advantages of high precision of 3D reconstruction, true reconstruction result, and the like.

The invention discloses a non-contact nondestructive omnibearing three-dimensional modeling method. The method comprises the following steps: calibrating a binocular camera by using a calibration board, and obtaining a stereo image pair; calculating an adaptive cross for each pixel in the image; estimating an adaptive matching window for each pixel according to the adaptive cross; obtaining matching cost by using the adaptive matching window, and respectively building a data item, a refined item, a smooth item and a division item by using the matching cost; then carrying out energy minimization on an energy equation to obtain final parallax distribution; and carrying out three-dimensional reconstruction according to the final parallax distribution. By adopting the non-contact nondestructive omnibearing three-dimensional modeling method, three-dimensional reconstruction is carried out by using a stereo vision method; the damage to an object caused by a contact three-dimensional modeling method is avoided; the three-dimensional reconstruction accuracy is effectively improved by adopting the adaptive matching cost method; the non-contact nondestructive omnibearing three-dimensional modeling method has good practicability.

The invention discloses a single-photon-array-based gating type diffuse reflection winding-angle imaging system and method. The system is composed of a laser emitter, a scanning galvanometer system, an intermediate reflecting surface, a distance gating module, a micro lens group, an SPAD array detector, a time-dependent photon counter and a digital processing unit. Laser emitted by the laser emitter is reflected by the surface of the intermediate reflecting surface, the surface of a hidden target, and the intermediate reflecting surface successively and then the reflected laser passes by the distance gating module; the laser is collected by the SPAD array detector; and the time-dependent photon counter is triggered. A time photon counting histogram is obtained according to the photon number and photon arrival time obtained by statistics by the time-dependent photon counter. And then the digital processing unit processes a plurality of time photon counting histograms according to a time sequential order to complete three-dimensional reconstruction of the hidden target object.

The invention discloses a pyramid Transform-based point cloud reconstruction method, device and equipment, and a medium, and belongs to the field of computer vision. The method comprises the steps that: after point feature data corresponding to each view are obtained, the point feature data are input into a two-channel pyramid network, and point features input in each layer pass through three modules of sampling, a graph neural network and Transformer; when the point feature data passes through the first channel, the updated output of each layer of attention mechanism is input into the next layer, and when the point feature data passes through the second channel, the updated output of each layer of attention mechanism and the updated output of the attention mechanism of the corresponding layer of the first channel located in the same layer as the next layer are spliced and then input into the next layer; and the updated output of the attention mechanism of the bottommost layer of the second channel is converted into a point cloud coordinate to obtain a point cloud result of each view, and an intersection of the point cloud results of each view is taken to represent a target object after point cloud reconstruction. Therefore, the three-dimensional reconstruction precision of the object is higher.

The invention relates to a three-dimensional reconstruction method and device for a mirror object, computer equipment, a storage medium and a computer program product, is applied to the technical field of three-dimensional reconstruction, and is used for improving the three-dimensional reconstruction effect of the mirror object. The method comprises the following steps: inputting a shot image of a mirror surface object into a rendering information network model, and decoding pixel points in the shot image through the rendering information network model to obtain reconstruction information and first rendering information for the pixel points; inputting the reconstruction information and a preset visual angle of the mirror surface object into an implicit scene model for rendering processing to obtain second rendering information and a preset visual angle image which are output by the implicit scene model and aim at the mirror surface object; and performing three-dimensional reconstruction on the mirror surface object according to the shot image, the preset view angle image, the first rendering information and the second rendering information to obtain a three-dimensional reconstruction model of the mirror surface object.

The invention belongs to the technical field of object surface three-dimensional reconstruction, and provides an endoscopic three-dimensional detection method, which comprises the following steps of: converting a polarization reconstruction dense point cloud and a shadow reconstruction dense point cloud of feature matching points under a pixel coordinate system and a sparse point cloud of stereoscopic vision feature matching points under a world coordinate system to construct a minimum error model; and fusing the three to obtain three-dimensional reconstruction coordinates under a world coordinate system. According to the method, the defect that the complete coordinates of the object cannot be obtained through a traditional three-dimensional endoscopic detection method can be overcome, dense point clouds of all the pixel points are obtained, the obtained reconstruction result is closer to an actual object, meanwhile, a to-be-measured curve in the two-dimensional image is obtained through an edge detection method, the to-be-measured curve is mapped into a world coordinate system through the same method to obtain fused three-dimensional point coordinates, and the length value of the three-dimensional curve is obtained through polynomial fitting and Euclidean distance calculation methods.

The embodiment of the invention relates to a calibration method and a system of a trinocular vision system for three-dimensional reconstruction, and the method comprises the steps: employing a Zhang calibration method to independently calibrate the internal parameters and distortion parameters of each camera in the trinocular vision system; collecting an image through each camera after independentcalibration, and calibrating each pair of cameras in the trinocular vision system in pairs based on the marked control points in the image; and based on the control points in the image, calibrating atrinocular camera by adopting a beam adjustment parameter optimization method for the cameras which finish pairwise calibration. According to the method provided by the invention, the calibration ofthe trinocular vision system can be realized, and the trinocular camera can obtain more abundant information, so that the precision of three-dimensional reconstruction is higher than that of a binocular camera.

The invention discloses a luminosity stereo three-dimensional reconstruction method and a spectrophotometric stereo camera. The luminosity stereo three-dimensional reconstruction method comprises the following steps: simultaneously projecting respective infrared light generated by three infrared light sources to a target object; collecting light reflected by the target object through a lens to obtain a source optical path; extracting optical signals related to the three infrared light sources from the source optical path separately to obtain three-path optical signals; converting the three-path optical signals to corresponding infrared image data respectively to obtain three groups of infrared image data; and utilizing the three groups of infrared image data to carry out three-dimensional reconstruction operation on the target object to obtain three-dimensional reconstruction data of the target object. According to the luminosity stereo three-dimensional reconstruction method and the spectrophotometric stereo camera, the accuracy of subsequent three-dimensional reconstruction can be improved; and in addition, the amount of the lens is only one, so that the difficulty brought by synchronous regulation of multiple lenses can be prevented, and actual application operation of a user is convenient.

The invention belongs to the field of optical measurement, and particularly discloses an iterative phase correction method for a low-quality stripe image. The method comprises the following steps of projecting a phase shift stripe image, shooting and decoding to obtain an initial phase, projecting the initial phase to an image brightness space, and obtaining a phase projection stripe image of thecurrent iteration period on the basis, conducting Gaussian filtering to acquire a Gaussian filtering phase of a current iteration period, judging whether an error between the Gaussian filtering phaseof the current iteration period and the Gaussian filtering phase of the previous iteration period is smaller than or equal to a preset threshold value or not, and if yes, outputting the Gaussian filtering phase, and if not, projecting the phase to an image brightness space for iteration. According to the method, the phase is projected to the image brightness space to obtain the stripe image underthe uniform reflectivity, the influence of Gaussian noise is suppressed, iterative filtering is performed on the stripe image to eliminate the influence of the object surface reflectivity and the light source nonlinear brightness, and thus high-precision phase recovery of the low-quality stripe image is realized.

The invention belongs to the field of building plastering robots and computer data processing, and particularly relates to an indoor plastering robot high-precision three-dimensional reconstruction system based on LCOS (reflective projection display). Patterns written in advance are projected on a wall surface needing plastering through an LCOS projection technology, wall surface projection is captured through a binocular camera, then acquired pictures are subjected to data processing through a PC, and three-dimensional information reconstruction of the wall surface needing plastering is completed; the PC is connected with a robot master control module through a TCP to send data to the plastering robot, and the plastering robot master control module controls a plastering machine to be adjusted to a designated operation position for plastering. The active three-dimensional reconstruction projection technology and the building plastering robot are combined, and the problems that the precision of laser in the field of building robots is not enough, and the effect is not good are solved.

The invention relates to a three-dimensional reconstruction and reproduction system and method for oblique photography, and the system comprises a data collection module which is used for collecting image data of different-angle oblique scenes; the data preprocessing module is used for preprocessing the image data to obtain a preprocessed image; the aerial triangulation processing module is used for acquiring the position and posture of the preprocessed image in the three-dimensional space, and optimizing and aligning the preprocessed image based on the position and posture of the preprocessed image in the three-dimensional space; the point cloud generation module is used for calculating and optimizing depth information of the aligned image, generating an initial point cloud, optimizing the initial point cloud and generating a dense point cloud; the three-dimensional model reconstruction module is used for optimizing the dense point cloud, converting the optimized dense point cloud into a continuous three-dimensional surface, generating a triangular mesh model, extracting textures of the image data, mapping the textures to the triangular mesh model, and reconstructing a three-dimensional model; and the scene reproduction module is used for performing scene reproduction by using the reconstructed three-dimensional model.

A high-frequency region enhanced photometric three-dimensional reconstruction method based on deep learning comprises the steps of shooting a plurality of images of an object to be reconstructed by using a photometric three-dimensional system, and outputting accurate surface normal three-dimensional reconstruction by using a deep learning algorithm, designing a surface normal generation network to generate a surface normal of an object needing to be reconstructed from an image and illumination; enabling the attention weight generation network to generate an attention weight map of an object needing to be reconstructed from the image; processing the attention weight loss function pixel by pixel; and using the trained network to carry out surface normal reconstruction of the photometric stereo image. According to the method, the surface normal information and the high-frequency information are learned respectively through the proposed surface normal generation network and the attention weight generation network, training is carried out by utilizing the proposed attention weight loss, and the reconstruction precision of the surface of the high-frequency region such as the wrinkle edge can be improved. Compared with a previous traditional photometric stereo method, the method improves the three-dimensional reconstruction precision, especially the details of the surface of the object to be reconstructed.

The invention relates to the technical field of image processing, in particular to a human tissue ultrasonic modeling method, ultrasonic equipment and a storage medium. The human tissue ultrasonic modeling method comprises the steps of enabling a probe to conduct ultrasonic scanning on a to-be-modeled tissue in the preset direction to acquire an ultrasonic image of each section of the to-be-modeled tissue; acquiring six-degree-of-freedom parameters corresponding to ultrasonic images of different sections scanned by the probe; and inputting the ultrasonic image of each tangent plane and the corresponding six-degree-of-freedom parameters into a trained deep neural network model to obtain a three-dimensional ultrasonic model of the to-be-modeled tissue. The ultrasound images of the sections of the to-be-modeled tissue can be reconstructed through the trained deep neural network model, the reconstruction accuracy is high, and the efficiency of medical staff for searching the standard section is improved.

The invention relates to an SFS (Shape From Shading) three-dimensional reconstruction sparse-DEM (Digital Elevation Model) encrypting method considering surface spectral information. The invention creatively provides a coarse-resolution DEM-grid interpolation-encrypting method by combining multi-spectral remote-sensing images with SFS three-dimensional reconstruction; ground-object spectral information is utilized to estimate the reflectivity of different ground-object types, and the influences of surface-vegetation coverage types on the precision of SFS reconstruction are removed; in the invention, multi-spectral images are utilized to replace full-color images, and the precision of three-dimensional reconstruction is improved by applying the assistance of the ground-object spectral information; the influences of the surface-vegetation coverage types and ground-object shadows on establishing a surface three-dimensional model and image spectral-brightness transforming relation can be effectively inhibited; and relative to a commonly-used DEM interpolation algorithm, the precision is markedly improved, and examples validate that the method can be used for improving the DEM spatial resolution to be two times of original data and finishing large-range rapid DEM establishment based on sparse control points.

The application provides a workpiece surface three-dimensional reconstruction method and device, and belongs to the technical field of machine vision. The method comprises the following steps: collecting a test workpiece surface image after a projected fringe pattern and performing phase unwrapping to obtain an absolute phase image of the test workpiece; based on a preset phase height model of each pixel, calculating an initial predicted height and a variance of each pixel in the test workpiece surface image; according to the variance of each pixel in the test workpiece surface image, calculating a height weight coefficient of each pixel; and according to the variance and the height weight coefficient of each pixel in the test workpiece surface image, correcting the initial predicted height of the pixel to obtain a height measurement result of each pixel, so as to realize three-dimensional reconstruction of the test workpiece. The application can effectively reduce the three-dimensional reconstruction error of the complex workpiece surface caused by factors such as imaging system noise, nonlinear error of the projector, phase unwrapping error and the like, and effectively improve the three-dimensional reconstruction precision of the complex workpiece surface.

The invention provides a laser processing head capable of automatically controlling processing focus length. The laser processing head comprises a Z-axis lifting skateboard, a laser processing head base plate and a rotation ranging mechanism. The Z-axis lifting skateboard is mounted on a laser processing machine and is driven by the same to move. A light guiding barrel and a focusing lens are arranged on the laser processing head base plate. The rotation ranging mechanism comprises a rotating disk, a driven gear, a driving gear, a servo motor and a wireless laser displacement sensor. The wireless laser displacement sensor is fixedly mounted on the bottom surface of the rotating disk with the mounting location enabling offset to be generated between measurement points of the wireless laser displacement sensor and processing focuses. The laser processing head can be intelligently applied to multi-assortment, multiple-batch and automated production occasions without shutdown adjustment and manual intervention. In addition, the invention further provides a laser processing method of the automatically controlling processing focus length.

The present application relates to a method, device, computer equipment, storage medium and computer program product for three-dimensional reconstruction of mirror objects, which are applied in the technical field of three-dimensional reconstruction and used to improve the effect of three-dimensional reconstruction of mirror objects. The method includes: inputting the photographed image of the mirror object into the rendering information network model, decoding the pixels in the photographed image through the rendering information network model, and obtaining the reconstruction information and the second Rendering information; input the reconstruction information and the preset viewing angle of the specular object into the implicit scene model for rendering processing, and obtain the second rendering information and the preset viewing angle for the specular object output by the implicit scene model A perspective image is set; according to the captured image, the preset perspective image, the first rendering information, and the second rendering information, perform three-dimensional reconstruction on the specular object to obtain a three-dimensional reconstruction model of the specular object.

This invention discloses a geological disaster monitoring system based on a three-dimensional structured quantum dot imaging device, comprising a MEMS structured light projection module, a short-wave infrared quantum dot imaging module, a synchronization control and signal processing module, and a data processing and disaster early warning module. The MEMS structured light projection module uses a dual-axis MEMS micromirror to perform two-dimensional scanning of short-wave infrared laser light to form a structured light pattern. The short-wave infrared quantum dot imaging module employs a heterogeneous structure with a three-dimensional structured electron transport layer conformally coated with a lead sulfide quantum dot photosensitive layer to enhance short-wave infrared light absorption efficiency and carrier separation efficiency. The synchronization control and signal processing module achieves nanosecond-level time synchronization. The data processing and disaster early warning module monitors surface cracks, landslides, and subsidence areas through three-dimensional point cloud reconstruction and multi-temporal deformation inversion. This invention has the advantages of wide spectral response, high sensitivity, strong low-light imaging capability, and high three-dimensional reconstruction accuracy.

The invention belongs to the field of optical measurement and visual sense detection, and particularly relates to a camera internal image sensor installation error separation method based on visual sense measurement. The method comprises steps of analyzing a deviation included angle error, a deviation distance error and a deviation rotation angle error between the actual installation position and the ideal installation position of a camera internal image sensor, and establishing a three coordinate difference equation of an actual imaging point and an ideal imaging point in the image sensor; drawing a relation three-dimensional graph between the deviation included angle error, the deviation distance error and the deviation rotation angle error and the coordinate difference; then carrying out camera calibration, and calculating a coordinate deviation amount; calculating a lens distortion error at the same time; and finally finding the deviation included angle optimal solution, a deviation distance optimal solution and a deviation rotation angle optimal solution of the image sensor. The method comprehensively considers the lens distortion error and the camera internal image sensor installation error, analyzes and verifies errors caused by the deviation between the actual installation position and the ideal installation position of the image sensor, and further provides the three-dimensional reconstruction precision.

The invention discloses a three-dimensional reconstruction method, device and equipment for a target road section. The method comprises the following steps: acquiring multiple travel data of a vehicle on the target road section; constructing a plurality of three-dimensional point clouds of the target road section based on a target image frame sequence and vehicle pose data in the multiple travel data, and resolving a camera pose set under a global geographic coordinate system; performing three-dimensional reconstruction based on the target image frame sequence, the plurality of three-dimensional point clouds and the camera attitude set to obtain a three-dimensional model corresponding to the target road section; and based on the three-dimensional model, determining a static background model belonging to a static background in the target road section. According to the scheme, it can be ensured that the reconstructed three-dimensional model is more complete, the geometric structure is more consistent, and the problems of cavities and inconsistency caused by single data collection are effectively solved.

The invention belongs to the field of optical measurement, and specifically discloses an iterative phase correction method for low-quality fringe images. The method includes: projecting a phase-shifted fringe image, shooting and decoding to obtain an initial phase; projecting the initial phase into the image brightness space, and obtaining a phase-projected fringe image of the current iterative cycle; performing Gaussian filtering on it to obtain a Gaussian filter of the current iterative cycle Phase; determine whether the error between the Gaussian filter phase of the current iteration cycle and the Gaussian filter phase of the previous iteration cycle is less than or equal to the preset threshold, if so, output it; if not, project it into the image brightness space for iteration . The present invention projects the phase into the image brightness space to obtain a fringe image with uniform reflectivity, and suppresses the influence of Gaussian noise. Iterative filtering for the fringe image will eliminate the influence of the surface reflectivity of the object and the nonlinear brightness of the light source, thereby Achieve high-precision phase recovery of low-quality fringe images.

The invention discloses a single-photon-array-based gating type diffuse reflection winding-angle imaging system and method. The system is composed of a laser emitter, a scanning galvanometer system, an intermediate reflecting surface, a distance gating module, a micro lens group, an SPAD array detector, a time-dependent photon counter and a digital processing unit. Laser emitted by the laser emitter is reflected by the surface of the intermediate reflecting surface, the surface of a hidden target, and the intermediate reflecting surface successively and then the reflected laser passes by the distance gating module; the laser is collected by the SPAD array detector; and the time-dependent photon counter is triggered. A time photon counting histogram is obtained according to the photon number and photon arrival time obtained by statistics by the time-dependent photon counter. And then the digital processing unit processes a plurality of time photon counting histograms according to a time sequential order to complete three-dimensional reconstruction of the hidden target object.

The present invention provides a UAV-based stereoscopic imaging method and device, the method comprising: acquiring at least two target images of a target object from different positions by a UAV, wherein at least two of the target images are The target objects are at least partially overlapped; and a stereoscopic image of the target object is generated according to at least two of the target images. The two cameras in the binocular camera are replaced by a drone, and at least two target images of the target object are acquired from different positions, thereby obtaining a stereoscopic image of the target object, and the distance between the phase acquisition points is increased in a low-cost manner. The distance reduces the demand for imaging resolution and improves the accuracy of 3D reconstruction.

The invention discloses a calibration method of a phase-height mapping system based on segmented polynomial fitting, that is, polynomial fitting is used to perform segmented calibration of the phase-height mapping system. According to the method proposed in this case, the measurement range Carry out interval division, and use different fitting orders in different height ranges to perform polynomial fitting calibration, so that the relative error of each position within the measurement range is minimized. In this case, the 3D reconstruction accuracy under the segmental polynomial fitting calibration Compared with the 1-6 order polynomial fitting calibration, the three-dimensional reconstruction accuracy is higher, and the process is more simplified.

The present application discloses a photometric three-dimensional reconstruction method and a spectroscopic photometric stereo camera. The method includes: simultaneously projecting the infrared light generated by each of the three infrared light sources to the target object; and collecting the light reflected by the target object through a lens. , to obtain the source optical path; extract the optical signals related to the above three infrared light sources from the source optical path to obtain three optical signals; respectively convert the above three optical signals into corresponding infrared image data to obtain three groups of infrared images Data: using the above three sets of infrared image data, carry out a three-dimensional reconstruction operation on the target object, and obtain the three-dimensional reconstruction data of the target object. This application can improve the subsequent three-dimensional reconstruction accuracy, and the number of lenses in this application is only one, which can avoid the difficulty caused by synchronous adjustment of multiple lenses, and is convenient for users to operate in practice.