52 results about "Texture reconstruction" patented technology

A method for texture reconstruction associated with a three-dimensional scan of an object includes scanning, at a processor, a sequence of image frames captured by an image capture device at different three-dimensional viewpoints. The method also includes generating a composite confidence map based on the sequence of image frames. The composite confidence map includes pixel values for scanned pixels in the sequence of image frames. The method further includes identifying one or more holes of a three-dimensional model based on the composite confidence map.

The invention discloses a cuboid automatic detection and three-dimensional reconstruction system and method. The system comprises a driving displacement module, a laser scanning module, a texture acquisition module, and a central processing module. The driving displacement module comprises a multi-axis stepping motor driver, an X-axis translation table, a Y-axis translation table, a Z-axis translation table, a rotation table, and an objective table. The laser scanning module is disposed over the objective table and is used for acquiring the laser point cloud of the surface of a to-be-tested cuboid. The texture acquisition module is disposed on the Z-axis translation table and is used for acquiring the texture image of the surface of the to-be-tested cuboid. The central processing module is connected with the multi-axis stepping motor driver, the laser scanning module, and the texture acquisition module. The cuboid automatic detection and three-dimensional reconstruction system and method achieves automation in laser point cloud acquisition analsysis, image capturing and processing, and three-dimensional model reconstruction, may rapidly achieve the automatic detection and three-dimensional texture reconstruction of the cuboid, and solves low efficiency and large work intensity in the prior art.

A method is provided, including the following method operations: using a robot having a plurality of sensors to acquire sensor data about a local environment; processing the sensor data to generate a spatial model of the local environment, the spatial model defining virtual surfaces that correspond to real surfaces in the local environment; further processing the sensor data to generate texture information that is associated to the virtual surfaces defined by the spatial model; tracking a location and orientation of a head-mounted display (HMD) in the local environment; using the spatial model, the texture information, and the tracked location and orientation of the HMD to render a view of a virtual space that corresponds to the local environment; presenting the view of the virtual environment through the HMD.

The invention provides a human body real-time modeling system and a modeling method thereof adopting a dual kinect and a rotating platform, comprising a geometric reconstruction module; a camera calibration module; a texture mapping module, etc. Through human body scanning, filtering and denoising, camera tracking and calibration, point cloud fusion and texture mapping, the final three-dimensionalhuman body model is obtained. The invention has the following advantages: the traditional ICP algorithm is improved, the speed of three-dimensional model reconstruction is accelerated, and the requirement of computer configuration is lowered; the real-timeness of human body reconstruction system is kept, the human body model is reconstructed in a short time, the requirement for stationary remaining for a long time on people is loweed, and no special clothes or props are needed; a cheap, fully automatic human body reconstruction system is built and developed, breaking through the traditionalhuman body reconstruction program funding, accuracy constraints; On the basis of geometric reconstruction, the high-quality texture is reconstructed by using the high-fidelity color image provided byKinect.

The invention discloses a face model construction method, a face recognition method, a face model construction device and face recognition equipment. The face model construction method comprises the following steps: firstly, reconstructing a three-dimensional face shape through a group of cascaded regression models or nonlinear regression models obtained by training by taking feature points of a two-dimensional image as input; then, providing a three-dimensional face texture reconstruction algorithm based on a convolution pyramid, reconstructing texture on the reconstructed three-dimensional face shape, wherein an algorithm comprises two stages of texture mapping and texture fusion, so that the illumination difference between multi-surface illumination can be well eliminated, vivid three-dimensional face texture information can be recovered, and a three-dimensional face model with complete textures is obtained. The multi-view face image is obtained by rendering the three-dimensional face model, the face database can be constructed according to the multi-view face image, face recognition is achieved through the face database, and the recognition rate is greatly increased.

The invention relates to a video coding system based on sparse sampling and texture reconstruction in the filed of image processing technology; the system comprises an input module, a frame selection module, a first pretreatment module, a coder and decoder module, a second pretreatment module, an aftertreatment module and a rank ordering output module; wherein the frame selection module comprises a frame extraction unit and a stacking unit; the pretreatment module comprises a visual sense attention submodule and a non-linear filtering submodule; the aftertreatment module comprises an index creating submodule, a searching submodule and a reconstruction submodule; in the invention, the coding efficiency is greatly improved, 20 percent of the highest coding gain can be obtained on the H.264 coding standard; the system has strong adaptability, good space-time consistency and good reconstruction quality, the coding efficiency can be improved under the condition of not reducing the video subjective effect.

The invention discloses a real-time facial expression reconstruction method. The method comprises the following steps of artificially manufacturing 29 facial form models and 46 expression models; carrying out face key point detection on one or more input pictures or videos; utilizing key points obtained by input photo or video detection and key points of a pre-calibrated human face three-dimensional model to perform face shape fitting on the human face key points through an error minimization criterion, and extracting human face textures through photos or videos to reconstruct an expression-free human face three-dimensional model. For a subsequently input video, face key points in the video are detected in real time, and face expression parameters are fitted in a mode of minimizing errorsbetween the key points and model key points, and the facial expression parameters are applied to the reconstructed expression face model to realize real-time reconstruction of the facial expression. According to the method, the facial expression of a single person can be reconstructed, and the facial expression of one person can also be applied to the reconstruction of another face, so that real-time face replacement is realized.

The invention discloses an aerial image moving object detection and elimination method for high-resolution texture generation. The method comprises the steps of detecting a suspicious object from theaerial image; screening the suspicious object to obtain a moving object and masking the moving object; removing the moving object by carrying out image restoration on the mask region, so that the problems of image super-resolution texture reconstruction effect caused by position change of moving objects such as vehicles or pedestrians in the aerial photography process and model distortion and texture error mapping in three-dimensional modeling can be effectively improved and solved. Moreover, the appropriate texture pictures are selected for eliminating the moving object according to differentconditions, so that the image after the moving object is eliminated is closer to the real condition of the ground; besides, according to the aerial image moving object detection and elimination method provided by the embodiment of the invention, the moving object under the dynamic background can be effectively detected, and the method is not limited to the adjacent frame images with high overlapping degree, such as videos, and is especially suitable for the condition of lower overlapping degree, such as aerial images, etc.

A method for texture reconstruction associated with a three-dimensional scan of an object includes scanning, at a processor, a sequence of image frames captured by an image capture device at different three-dimensional viewpoints. The method also includes generating a composite confidence map based on the sequence of image frames. The composite confidence map includes pixel values for scanned pixels in the sequence of image frames. The method further includes identifying one or more holes of a three-dimensional model based on the composite confidence map.

The embodiment of the invention discloses a system, a method and equipment for realizing three-dimensional augmented reality of multi-channel video fusion. According to the system and the method, video streams are generated for images shot by multiple paths of video acquisition devices; data synchronization is carried out on the returned video streams; after the video frame texture obtained by resolving the video stream is mapped to the three-dimensional scene, texture reconstruction is carried out on the texture of the overlapping region of the texture mapping according to the weight condition contributed by the texture of the projectors, so that the overlapping condition of the projection regions among the plurality of projectors is reduced, and the display effect of the rendered three-dimensional scene is improved.

The embodiment of the invention provides a three-dimensional human body reconstruction method and device, equipment and a storage medium. The method comprises the steps: carrying out the geometric reconstruction of a human body based on a single human body image of a target human body, and obtaining a three-dimensional grid model of the target human body; based on the single human body image, performing local geometric reconstruction on a local part of the target human body to obtain a three-dimensional grid model of the local part; fusing the three-dimensional grid model of the local part with the three-dimensional grid model of the target human body to obtain an initial three-dimensional model; and performing human body texture reconstruction according to the initial three-dimensional model and the single human body image to obtain a three-dimensional human body model of the target human body. According to the embodiment of the invention, the local part in the three-dimensional grid model of the target human body is clearer and more accurate, and the reconstruction effect of the local part is improved.

The invention discloses a texture fusion method for RGB-D camera real-time three-dimensional reconstruction. The method includes: processing the RGB-D data stream to obtain the definition of a color image, selecting a key frame to extract a foreground, performing filtering and denoising on a depth image, calculating a normal vector of a depth image point cloud, and establishing a reconstructed data stream; quantitatively establishing an adaptive weight field of the color image in a mode of combining a probability method and a heuristic method, and taking the adaptive weight field as confidencecoefficient distribution of a real-time frame for describing color data; and by comparing the confidence coefficient weight in the adaptive weight field of the real-time frame with the latest confidence coefficient weight of the reference point cloud, selecting an operation from three operations of replacement, fusion and reservation to update a texture result, thereby realizing texture fusion applied to three-dimensional reconstruction. According to the method, high-quality data can be extracted, blurring of texture fusion can be effectively reduced, a clear texture reconstruction result isrealized, and the method is embedded into an RGB-D reconstruction framework with relatively low calculation cost, so that the texture reconstruction precision is remarkably improved.

A method and apparatus for texture filtering is provide wherein a filter select module is adapted to select a filtering mode based upon a sampling rate of polygon and texture data. The filter mode is selected by determining the filter characteristics of the selected filtering mode based upon the sampling rate and a degree of warping per texture coordinate. A texture reconstruction filter characteristic is morphed based upon the input polygon and texture data so that, after subsamples are aggregated, an effective filter characteristic matches the texture reconstruction filter characteristic of a texture reconstruction filter used for coarse sampling. Subsequently, a texel blending module computes texel blending factors based on the filtering mode determined by the filter select module.

The invention discloses a three-dimensional face modeling method based on a double-tributary network, and the method comprises the following steps: 1) constructing a face shape reconstruction tributary based on an auto-encoder structure, and obtaining an identity space diagram and an expression space diagram; 2) respectively expressing an identity attribute component and an expression attribute component by utilizing the identity space diagram and the expression space diagram; 3) constructing a facial texture reconstruction branch based on a semantic region to obtain a facial texture map withhigh truth; and 4) constructing a fusion module: correspondingly combining the reconstructed face shape model and the reconstructed face texture map according to a predefined three-dimensional vertextopological relation to obtain a final reconstructed three-dimensional face model. According to the invention, decoupling learning is carried out on the identity and expression attributes of the humanface, and processing of the face shape and the face texture is carried out by using the double-tributary network, so that accurate three-dimensional human face reconstruction with large-amplitude expressions is achieved.

The invention relates to a high-quality 3D texture reconstruction method. According to the high-quality 3D texture reconstruction method, a fuzzy detection module and a dynamic color weight distribution function are additionally adopted in texture reconstruction; a fuzzy detection algorithm is adopted to set a certain fuzzy threshold value, so that fuzzy pictures inputted into a system are removed; and the dynamic color weight distribution function is designed so as to dynamically distribute color weights for newly acquired color pictures, so that color information stored in voxels can be updated. With the high-quality 3D texture reconstruction method adopted, the problem of color edge ghosting can be solved, and the texture quality of a reconstructed 3D model can be greatly improved to a certain extent.

The invention provides a multi-view real-time human body motion, gesture, expression and texture reconstruction system, which comprises the following steps of: defining a capture area by a plurality of camera frames, and calibrating camera internal parameters and camera external parameters of a plurality of cameras by a camera calibration method; collecting a human body image through the pluralityof calibrated cameras, processing the human body image to transcode the human body image into an RGB image, and then completing single-purpose human body posture estimation; obtaining a hot spot mapand joint affinity of each joint of the human body through single-purpose human body posture estimation, and performing non-maximum suppression on the joint hot spot map to obtain coordinates of eachjoint; therefore, human body three-dimensional joint coordinates are obtained, and a human body three-dimensional reconstruction model is obtained. According to the system, deep learning is utilized to complete estimation of human body postures, and fitting and rendering can be performed on human body models of multiple persons in real time in a test environment.

The invention relates to a three-dimensional shape and texture reconstruction method and system based on a single image and a storage medium, and solves the problem that three-dimensional face data cannot be efficiently and accurately obtained. The method comprises the following steps: acquiring face image information; analyzing the face image information through a normalized coordinate coding prediction network, and mapping the analyzed data information to a UV space to form UV space texture mapping information; analyzing the UV spatial texture mapping information through a filling network toobtain complete texture map information; analyzing the complete texture map information through a shape network to obtain UV position map information; searching face three-dimensional shape coordinate information corresponding to the current UV position map information from a preset corresponding relationship between reference UV position map information and reference face three-dimensional shapecoordinate information; and forming three-dimensional texture face information according to the searched face three-dimensional shape coordinate information. According to the invention, the requiredreconstructed three-dimensional face data can be efficiently and accurately obtained.

The invention belongs to the field of image processing, and discloses a texture surface defect detection method and system. The method comprises an offline training stage and an online detection stageI; in the offline training stage, multi-channel texture priori of an input texture image is extracted through a priori extraction step; a texture background of the input image is accurately reconstructed through a texture reconstruction step under the guidance of the extracted priori, more accurate texture features are encoded in a hidden space of a texture reconstruction module, and defects are inhibited from being reconstructed in the texture background; and finally, through pixel-level adversarial learning, the texture reconstruction precision is further improved. During detection, the defects can be detected only by subtracting the reconstructed texture background image from the input image. The method has high detection precision for defects with different sizes and different contrastratios on different texture surfaces.

The embodiment of the invention provides a three-dimensional human body reconstruction method and device, equipment and a storage medium, and the method can comprise the steps: carrying out the fitting of human body parameters of a parameterized human body template based on a human body image of a target human body, and obtaining a first three-dimensional grid model of the target human body; performing three-dimensional human body reconstruction based on image features extracted from the human body image, and determining a second three-dimensional grid model of the target human body; fusing the first three-dimensional mesh model and the second three-dimensional mesh model to obtain an initial three-dimensional model; and performing human body texture reconstruction according to the initial three-dimensional model and the human body image to obtain a three-dimensional human body model with textures. According to the embodiment of the invention, the three-dimensional human body model obtained by reconstruction not only ensures the robustness of the human body shape, but also improves the authenticity and accuracy of the human body area.

The invention provides a human body scene image eigen-decomposition and relighting method and device, and the method comprises the steps: making a human body model through employing Clo3d, carrying out the light tracing through employing blender, and carrying out the rendering, and obtaining a training data set; pre-training a normal prediction network by using a single image; taking the original image and the normal graph obtained through prediction as input of an image eigen-decomposition network, and obtaining a reflectivity graph, a shadow graph and spherical harmonic illumination parameters through prediction; re-illuminating the original image by using the new spherical harmonic illumination parameter; and finally, decomposing a single picture into a reflectivity picture and a shadow picture, so the application of picture relighting, 3D model synthesis, model surface texture reconstruction and the like can be achieved. The shadow of the input picture can be removed in the field of three-dimensional reconstruction, and the surface texture quality of a three-dimensional reconstruction model is improved.

The invention discloses a method and a device for reconstructing a three-dimensional scene in real time. The method comprises a geometric reconstruction process and a texture reconstruction process, wherein the geometric reconstruction process comprises the steps of inputting an RGB image and a depth image of a scene to be reconstructed in real time, and dividing the RGB image and the depth imageinto a key frame and a non-key frame for processing according to frame similarity, and the processing of the key frame comprises the steps of storing feature points, performing loopback detection to find similar frames, and performing global pose optimization and registration; for the non-key frame, registering with a relative pose relative to the key frame; after TSDF field fusion, grids are extracted through a moving cubic algorithm to generate geometry of a three-dimensional scene model; wherein the texture reconstruction process comprises the steps of restoring and realizing texture optimization through a texture mapping mode to reconstruct color information of a three-dimensional scene model, selecting images under a plurality of visual angles from an input RGB image sequence, extracting texture pieces, splicing the texture pieces into a complete texture, and mapping the texture pieces. According to the method, the three-dimensional model of the high-quality appearance can be reconstructed in real time.

The embodiment of the invention provides a 3D model reconstruction method and device and electronic equipment, and belongs to the technical field of data processing, and the method comprises the steps: carrying out the segmentation of a target object from an input image containing the target object, so as to obtain a texture map of the target object on the input image; executing skeleton detectionon the target object to form a 3D skeleton model of the target object; acquiring an edge part of the texture map and a depth value of the edge part on a triangular patch in the 3D skeleton model; when the depth value is greater than a preset threshold, predicting the edge texture of the edge part by using a preset network to form an edge prediction texture map; and forming a 3D reconstruction model of the target object based on the edge prediction texture map. Through the scheme of the invention, the mapping texture can be generated based on the plurality of images, and the texture reconstruction accuracy of the 3D model is improved.

The invention discloses an image processing method, device and equipment, a storage medium and a program product, and relates to the technical field of image processing. The specific implementation scheme is as follows: acquiring a to-be-processed image of at least one part of a target object from an image comprising the target object; taking the to-be-processed image of at least one part as a node element, and constructing an association graph of all parts of the target object by utilizing the node element; and inputting the association graph into a graph neural network model, and outputting texture complemented images of all parts of the target object by utilizing the graph neural network model. According to the embodiment of the invention, the high-quality complete three-dimensional texture map can be generated for the target object, the cost of three-dimensional texture reconstruction can be remarkably reduced, and the omnibearing simulation rendering of the target object is realized.

The invention discloses a laser radar and camera fused irregular object measurement method. The measurement method comprises the following steps: step 1, carrying out joint calibration on a laser radar and a monocular camera module; step 2, the laser radar and the monocular camera perform synchronous execution, respectively acquire laser point cloud data and sequence images of the irregular object, and respectively generate a laser point cloud model and image point cloud data of the irregular object through respective data fusion processing; 3, performing registration fusion of multi-source point cloud data on the laser point cloud model and the image point cloud data to generate a multi-source point cloud model; 4, performing geometric and texture reconstruction on the multi-source point cloud model to generate a three-dimensional reconstruction model; and 5, carrying out volume measurement on the three-dimensional reconstruction model. According to the measurement method, the laser radar point cloud data and the camera shooting point cloud data are fused, the defects of independent use of the laser radar point cloud data and the camera shooting point cloud data are overcome, and the three-dimensional reconstruction effect and the measurement precision are improved.

The invention relates to the field of image processing, and discloses an image texture reconstruction method, device and equipment and a storage medium. The method comprises the steps of acquiring a three-dimensional image to be reconstructed; performing weight setting processing on the three-dimensional image to obtain a pixel coordinate weight set corresponding to the three-dimensional image; performing occlusion judgment processing on the three-dimensional image to obtain an occlusion area and a non-occlusion area corresponding to the three-dimensional image; constructing a triangular topological graph in the occlusion area; matching a reconstruction point with the same pixel value of the feature vertexes in the non-occlusion area; taking the reconstruction point as a vertex, taking thetriangular topological graph as a shape, one side being parallel to a diagonal side of the feature vertex in the triangular topological graph, and constructing a collection triangle in the non-occlusion area; and acquiring pixel value distribution in the acquired triangle, replacing and modifying the pixel value distribution in the triangular topological graph with the pixel value distribution, and determining an area corresponding to the triangular topological graph as the non-occlusion area.

The invention discloses an annular high-precision measuring device and method based on speckle structured light. The annular high-precision measuring device comprises a supporting structure, an annular guide rail connected to one end of the supporting structure, a cantilever rotationally connected to the lower end of the annular guide rail, a depth camera module assembly which is vertically connected to the cantilever, and a servo system, wherein the servo system is arranged above the annular guide rail and fixed to the supporting structure, and an output end of the servo system is connected with the cantilever. According to the annular high-precision measuring device and the method, annular scanning can be carried out in the horizontal direction, linear scanning can be carried out in thevertical direction, texture reconstruction and appearance reconstruction can be carried out on the target object in a short time, authenticity and integrity of external texture features of a reconstruction model are improved, and application scenes are expanded.

The invention discloses a cuboid automatic detection and three-dimensional reconstruction system and method. The system comprises a driving displacement module, a laser scanning module, a texture acquisition module, and a central processing module. The driving displacement module comprises a multi-axis stepping motor driver, an X-axis translation table, a Y-axis translation table, a Z-axis translation table, a rotation table, and an objective table. The laser scanning module is disposed over the objective table and is used for acquiring the laser point cloud of the surface of a to-be-tested cuboid. The texture acquisition module is disposed on the Z-axis translation table and is used for acquiring the texture image of the surface of the to-be-tested cuboid. The central processing module is connected with the multi-axis stepping motor driver, the laser scanning module, and the texture acquisition module. The cuboid automatic detection and three-dimensional reconstruction system and method achieves automation in laser point cloud acquisition analsysis, image capturing and processing, and three-dimensional model reconstruction, may rapidly achieve the automatic detection and three-dimensional texture reconstruction of the cuboid, and solves low efficiency and large work intensity in the prior art.

The invention discloses a construction method for macroscopic and microscopic texture reconstruction of a cement concrete pavement. In the construction method, the cement concrete pavement is grinded by diamond blades with close intervals, a thin layer of 4-6 mm on the cement concrete pavement is grinded, the surface of the grinded cement concrete pavement is provided with grooves with longitudinal textures, and the grooves are arranged in a structure of a lamp wick shape. In the cement concrete pavement prepared by adopting the construction method for the macroscopic and microscopic texture reconstruction of the cement concrete pavement, the surface of the grinded cement concrete pavement is provided with the grooves with longitudinal textures, and the grooves are arranged in the structure of the lamp wick shape. After construction, the longitudinal texture with the lamp wick shape is formed on the pavement, and a zigzag configuration is also formed among the grooves while increasing the macroscopic configuration of the pavement. Through the formed longitudinal texture, the skid resistance performance of the pavement is enhanced, and the contact noise of tires and the pavement is reduced.

The invention provides a super-resolution three-dimensional texture reconstruction method and device and equipment thereof. The method is applied to establish a super-resolution realistic texture model of a target to-be-measured object conforming to a preset measurement volume. The method comprises: acquiring depth data and a texture image of a target to-be-measured object, wherein the depth data are phase parameters of the target to-be-measured object at least comprising five visual angles; generating a three-dimensional model of the to-be-measured object according to the depth data; constructing a mapping energy equation according to the depth data, the texture image and the three-dimensional model; generating super-resolution texture image mapping with optimal energy according to the mapping energy equation; and generating the super-resolution realistic texture model according to the super-resolution texture image mapping. Sampling processing of texture three-dimensional mapping on images is avoided, and the texture resolution of a reconstructed three-dimensional model is effectively improved; and the resolution improvement effect is obvious.

A method is provided, including the following method operations: using a robot having a plurality of sensors to acquire sensor data about a local environment; processing the sensor data to generate a spatial model of the local environment, the spatial model defining virtual surfaces that correspond to real surfaces in the local environment; further processing the sensor data to generate texture information that is associated to the virtual surfaces defined by the spatial model; tracking a location and orientation of a head-mounted display (HMD) in the local environment; using the spatial model, the texture information, and the tracked location and orientation of the HMD to render a view of a virtual space that corresponds to the local environment; presenting the view of the virtual environment through the HMD.