Distinguished Professor, Founding Director
Sign in to view more
Patentee: Shenzhen UniversityHui Huang,Kai Xu
A method of scene reconstruction includes: a depth image taken by a camera mounted on a robot is acquired, and a current tensor field key frame is generated based on the depth image; spatial-temporal interpolation is performed on the current tensor field key frame and a previous tensor field key frame to obtain an interpolated tensor field key frame; a robot planning path is acquired by particle advection algorithm according to the current tensor field key frame, the previous tensor field key frame, and the interpolated tensor field key frame; a camera planning path is acquired according to the robot planning path and an energy equation of the camera; and a scene is reconstructed according to the depth images captured by the camera.
Publication Patent Date: 20190730Application Date: 20170922Patent type: 发明专利Legal Status: ActivePublication Patent Number: US10,366,503
Patentee: Shenzhen UniversityHui Huang,Kai Xu,Junyu Zhang
This inventor provides a kind of recognition method and device, including the following stages: S1: get the first feature vector by putting the depth image of current best view of the object to be recognized to feature extractor; S2: get the output of pooling layer by putting the first feature vector to first hidden layer, then get the classification by putting the output of pooling layer to a hierarchical classifier; S3: generate the partial region of current view according to the output of pooling payer and view observation vector; S4: get the second feature vector by putting the partial region to second hidden layer, and the second feature vector includes the output of pooling layer and information of current view; S5: generate the next best view using a fully connected layer and the second feature vector; S6: repeat S1 to S5 until reach the leaf node of the hierarchical classifier. This inventor scans and recognizes the unknown scene, while solving the problem of next best view.
Publication Patent Date: 20190917Application Date: 20170620Patent type: 发明专利Legal Status: ActivePublication Patent Number: US10,417,526
Patentee: Shenzhen UniversityHui Huang,Yang Zhou,Huajie Shi
The present invention provides a method, apparatus, storage medium and device for controlled synthesis of inhomogeneous textures, which relates to the field of texture synthesis technique in image processing. The method includes: obtaining an input texture exemplar from user, and analyzing the features of the exemplar, so as to extract the source guidance channels; herein the guidance channels comprise a progression map and an orientation field of the texture exemplar; obtaining a progression map and an orientation field from the user specified target, which are used as the target guidance channels; comparing the differences between the source guidance channels and the target guidance channels, so as to form the progression constraint and orientation constraint; determining the distance measure between the texture exemplar and target image, and adding the progression constraint and orientation constraint into the distance measure; the target image produced by minimizing the sum of distances over all patches with the constrained distance measure, is the final output of controlled synthesis.
Publication Patent Date: 20190101Application Date: 20170525Patent type: 发明专利Legal Status: ActivePublication Patent Number: US10,169,908
Patentee: Shenzhen Institute of Advanced Technology, CASHui Huang,Zhuming Hao,Minglun Gong,Dani Lischinski,Daniel Cohen-Or
The present disclosure provides a navigation method based on a three-dimensional scene, comprising: calculating an interest value of a viewpoint of a camera based on heights, volumes, irregularities and uniqueness of buildings in a scene; generating trajectory parameters of the camera according to the interest value of the viewpoint, so as to navigate according to the trajectory parameters. The navigation method based on a three-dimensional scene of the present disclosure obtains a reasonable interest value of the viewpoint based on heights, volumes, irregularities and uniqueness of the buildings, thereby achieving a high-quality navigation.
Publication Patent Date: 20180904Application Date: 20161223Patent type: 发明专利Legal Status: ActivePublication Patent Number: US10,066,956
Patentee: Shenzhen Institute of Advanced TechnologyHui Huang,Shihao Wu,Minglun Gong,Matthias Zwicker,Daniel Cohen-or
The present disclosure provides a three-dimensional point cloud model reconstruction method and a device. The method comprises: 1) sampling and WLPO-consolidating an input point set to generate an initial surface point set, copying the initial surface point set as an initial position of an interior skeleton point set, to establish a correspondence relation between surface points and skeleton points; 2) moving points in the interior skeleton point set inwards along a direction opposite to a normal vector thereof, to generate interior points; 3) using a self-adaptive anisotropic neighborhood as a regularization term to perform an optimization of the interior points, and generating skeleton points; 4) performing a consolidation and completion of the initial surface point set using the skeleton points, to generate consolidated surface points; 5) reconstructing a three-dimensional point cloud model according to the skeleton points, the surface points and the correspondence relation between the surface points and the skeleton points.
Publication Patent Date: 20181002Application Date: 20151230Patent type: 发明专利Legal Status: ActivePublication Patent Number: US10/089,781
Bin Wang,Hui Huang,Longhua Wu
The present disclosure discloses a method and a device for elastic object deformation modeling. The method comprises: acquiring a static point cloud of the elastic object and dynamic point cloud sequences; establishing a simulation tetrahedral mesh model; driving the simulation tetrahedral mesh model to track the dynamic point cloud sequences, to obtain track deformation sequences; iteratively estimating material property coefficients and corresponding reference shapes of the elastic object; performing the following operations in each iteration: obtaining a reference shape corresponding to a current material property coefficient; driving the simulation tetrahedral mesh model to simulate the deformation from the same initial deformation according to the coefficient and the reference shape to obtain a simulation deformation sequences; calculating a positional deviation between the simulation deformation sequences and the track deformation sequences; and updating the material property coefficients in a direction in which the positional deviation is decreased; establishing an elastic object deformation model according to a material property coefficient under a minimum positional deviation and corresponding reference shape. The technical solution can establish a vivid elastic object deformation model.
Publication Patent Date: 20190827Application Date: 20150826Patent type: 发明专利Legal Status: ActivePublication Patent Number: US10,394,979
Patentee: Shenzhen Institute of Advanced TechnologyHui Huang,Kangxue Yin,Daniel Cohen-Or
A method for reconstructing surface from a point cloud includes following steps: (a) extracting skeletal curves from an input point cloud; (b) editing the extracted skeletal curves, and assigning sweeping path; (c) obtaining sliced point clouds along the edited skeletal curves, and fitting a closed NURBS curve according to the sliced point clouds; (d) reconstructing the point cloud to get generalized cylinders along the assigned sweeping path, according to the closed NURBS curves; (e) merging the generalized cylinders into a single surface, and smoothing intersections of the generalized cylinders so as to reconstruct surface from the point cloud. The invention further relates to a system for reconstructing surface from a point cloud. The invention can reconstruct the surface with high accuracy by the minimum interactions, and can deal with point cloud data having missing region caused by occlusion. In addition, the invention can achieve high reconstruction quality and fine controllability.
Publication Patent Date: 20181106Application Date: 20140624Patent type: 发明专利Legal Status: ActivePublication Patent Number: US10,121,283
Patentee: Shenzhen Institute of Advanced TechnologyHui Huang,Shihao Wu,Wei Sun,Pinxin Long,Minglun Gong,Daniel Cohen-Or,Oliver Deussen
A method for reconstructing a three-dimensional model of point clouds includes following steps: a, scanning to obtain point clouds of an object required for a three-dimensional model reconstruction; b, analyzing quality of the obtained point clouds; c, computing a new scanning view based on the analyzed point clouds; d, scanning according to the new scanning view and updating the point clouds of step a based on point clouds obtained by the scanning according to the new scanning view in real time; and e, reconstructing a three-dimensional model according to the point clouds updated in real time. The invention further relates to a system for reconstructing a three-dimensional model of point clouds. The invention can realize full automatic reconstruction of a three-dimensional model and create a model of point clouds with high quality. In addition, the invention is easy to implement and can achieve high efficiency.
Publication Patent Date: 20180828Application Date: 20140613Patent type: 发明专利Legal Status: ActivePublication Patent Number: US10,062,207
Kangxue Yin,Hui Huang
Provided is a method for extracting an image salient curve. The method comprises the following steps: drawing an approximate curve along a salient edge of an image from which a salient curve is to be extracted; obtaining short edges in the image; calculating a harmonic vector field by using the drawn curve as a boundary condition; filtering the short edges in the image by using the harmonic vector field; updating the vector field by using the short edges left in the image as boundary conditions; and obtaining an optimal salient curve of the image by using the energy of a minimized spline curve in the vector field. Also provided is a system for extracting an image salient curve. The image salient curve can ensure the smoothness and a bending characteristic.
Publication Patent Date: 20170718Application Date: 20130926Patent type: 发明专利Legal Status: ActivePublication Patent Number: US9,710,725
Disclosed is a method for automatically optimizing point cloud data quality, including the following steps of: acquiring initial point cloud data for a target to be reconstructed, to obtain an initial discrete point cloud; performing preliminary data cleaning on the obtained initial discrete point cloud to obtain a Locally Optimal Projection operator (LOP) sampling model; obtaining a Possion reconstruction point cloud model by using a Possion surface reconstruction method on the obtained initial discrete point cloud; performing iterative closest point algorithm registration on the obtained Possion reconstruction point cloud model and the obtained initial discrete point cloud; and for each point on a currently registered model, calculating a weight of a surrounding point within a certain radius distance region of a position corresponding to the point for the point on the obtained LOP sampling model, and comparing the weight with a threshold, to determine whether a region where the point is located requires repeated scanning. Further disclosed is a system for automatically optimizing point cloud data quality.
Publication Patent Date: 20170620Application Date: 20130917Patent type: 发明专利Legal Status: ActivePublication Patent Number: US9,684,817
Patentee: Shenzhen Institute of Advanced TechnologyHui Huang,Shihao Wu,Baoquan Chen,Liangliang Nan
A method for extracting a skeleton form a point cloud includes: obtaining inputted point cloud sampling data; contracting the point cloud using an iterative formula and obtaining skeleton branches.
Publication Patent Date: 20160712Application Date: 20130913Patent type: 发明专利Legal Status: ActivePublication Patent Number: US9,390,552
Patentee: Shenzhen Institute of Advanced TechnologyHui Huang,Kangxue Yin,Minglun Gong,Baoquan Chen,Yunhai Wang
Provided are an image restoration method and device. The method comprises: processing image blocks, which are initially registered, to acquire connection curves among the image blocks; constructing an ambient field of images to be restored by means of the connection curve; by minimizing energy of the connection curve in the ambient field, registering the image blocks; and performing image filling on the registered image blocks to acquire a restored image. The device comprises: a processing unit used for processing the image blocks, which are initially registered, to acquire the connection curve among the image blocks; an ambient field construction unit used for constructing the ambient field of the image to be restored by means of the connection curve; a registering unit used for registering the image blocks by minimizing the energy of the connection curve in the ambient field; and a filling unit used for performing image filling on the registered image blocks to acquire the restored image. The present invention can be applied to restore any damaged image and improve the accuracy.
Publication Patent Date: 20170117Application Date: 20130419Patent type: 发明专利Legal Status: ActivePublication Patent Number: US9,547,885