WO2019080488A1 - 一种基于多尺度协方差描述子与局部敏感黎曼核稀疏分类的三维人脸识别方法 - Google Patents
一种基于多尺度协方差描述子与局部敏感黎曼核稀疏分类的三维人脸识别方法Info
- Publication number
- WO2019080488A1 WO2019080488A1 PCT/CN2018/087385 CN2018087385W WO2019080488A1 WO 2019080488 A1 WO2019080488 A1 WO 2019080488A1 CN 2018087385 W CN2018087385 W CN 2018087385W WO 2019080488 A1 WO2019080488 A1 WO 2019080488A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- scale
- face
- riemann
- covariance
- local
- Prior art date
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/60—Type of objects
- G06V20/64—Three-dimensional objects
- G06V20/653—Three-dimensional objects by matching three-dimensional models, e.g. conformal mapping of Riemann surfaces
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
- G06V40/16—Human faces, e.g. facial parts, sketches or expressions
- G06V40/168—Feature extraction; Face representation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
- G06V40/16—Human faces, e.g. facial parts, sketches or expressions
- G06V40/172—Classification, e.g. identification
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/213—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods
- G06F18/2134—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods based on separation criteria, e.g. independent component analysis
Definitions
- the invention relates to the field of digital image processing and pattern recognition, in particular to a three-dimensional face recognition method based on multi-scale covariance descriptor and local sensitive Riemann kernel sparse classification.
- the three-dimensional face data acquired by the three-dimensional face scanner can effectively contain the spatial geometric information inherent to the face. Since the three-dimensional shape data is robust to changes in illumination and view, and unlike the two-dimensional data, the pixel values are susceptible to makeup, etc. These characteristics provide an objective basis for the accurate identification of individual identities. With the evolution of the times, the development of anthropometric techniques and the enhancement of computing power have greatly facilitated the transfer of face recognition methods from a purely two-dimensional image-based approach to a three-dimensional face recognition method using face space shape information. In recent years, 3D face recognition including geometric information has become a research and application hotspot, especially the implementation of Face Recognition Grand Challenge (FRGC), which has greatly promoted 3D face recognition at home and abroad. Research.
- FRGC Face Recognition Grand Challenge
- the technical problem to be solved by the present invention is to provide a three-dimensional face recognition method based on multi-scale covariance descriptors and locally sensitive Riemann kernel sparse classification, and to obtain local covariance descriptors at different scales by using continuously varying scale parameters. It can effectively improve the representation ability of single-scale local covariance descriptors, and locally sensitive Riemann kernel sparse classification can effectively utilize the locality of multi-scale descriptors.
- the present invention provides a three-dimensional face recognition method based on multi-scale covariance descriptors and locally sensitive Riemann kernel sparse classification, including the following steps:
- automatically pre-processing the original G library face model and the P test set face models respectively comprises the following steps:
- the shape index of each point in the point cloud of the face calculates the shape index of each point in the point cloud of the face, and select the connected area composed of points with the shape index in the range of 0.85-1.0 as the initial nose point candidate area.
- calculate the centroid position of the face point cloud at the tip of the nose The candidate region is selected as the nose region by the nearest connected region of the centroid position; the centroid of the nose region is selected as the nose tip; finally, the nose tip is the center of the ball, and the radius of 90 mm is used to cut the three-dimensional face region of interest;
- step (2) the scale space is established according to the library set face model and the test set face model after the automatic pre-processing in step (1), and the multi-scale key point detection and the neighborhood extraction thereof are performed, including the following steps. :
- M refers to the original 3D face mesh
- a d ⁇ d-dimensional local covariance descriptor is extracted for each key point neighborhood of each scale, and multi-scale fusion is constructed by constructing multi-scale covariance of these local covariance descriptors.
- the descriptor includes the following steps:
- n y and n z represent the values of the point normal in the x and z axis directions, respectively; extracting the amplitude characteristic F 3 : The magnitude of
- n x and n y represent the values of the point normals in the x and y directions, respectively, and the gradient features and amplitude features can be used to describe the trend of the geometrical direction of the three-dimensional face surface in a particular direction;
- ⁇ s is the average eigenvector of the region P is
- the covariance matrix C is a symmetric positive definite matrix, its diagonal elements represent the variance of each feature, and the non-diagonal elements represent the correlation between features
- the region P C is the covariance matrix of dimension 3 ⁇ 3 is independently fixed thereto having a magnitude independent;
- s is the number of scales
- ⁇ s is the weight coefficient
- ⁇ s is the ratio of the recognition rate of Rank-1 at each scale to the sum of the Rank-1 recognition rates of all scales
- C i is the i-th key point Multi-scale covariance descriptor.
- mapping the local covariance descriptor to the reproducible Hilbert space, and proposing the local sensitive Riemann kernel sparse representation to classify the three-dimensional face includes the following steps:
- ⁇ is the regularization parameter, Indicates that the corresponding elements of the vector are multiplied.
- p k ⁇ R N represents a local operator for measuring test samples
- the Lagrangian multiplier method can be used to solve the problem.
- the equation is optimized by the Lagrangian coefficient and the objective function is solved.
- the analytical solution is obtained.
- ⁇ g ( ⁇ ) means that only the coefficient corresponding to class g is selected
- r g (Y) is the mean of the reconstructed residual sum of m descriptors of g class, and the face set model and test set with the smallest residual
- the face model can be determined as a three-dimensional face of the same person
- a Log-Euclidean Gaussian kernel capable of accurately measuring the reconstructed residual is used, and its corresponding expression is:
- the beneficial effects of the present invention are as follows: (1) It is proposed to extract different types of effective features based on key point neighborhoods directly on the three-dimensional face mesh, not only a single geometric or spatial feature utilizing the shape region; (2) Continuously changing scale parameters obtain visual processing information at different scales, and deeply explore the essential features of three-dimensional human faces.
- the present invention proposes to use facial features under multiple scale fusions for recognition; (3) by Riemann kernel sparse coding The local constraints are introduced to produce better classification performance. A three-dimensional face recognition method based on local sensitive Riemann kernel sparse classification is proposed.
- Figure 1 is a schematic flow chart of the method of the present invention.
- FIG. 2 is a schematic diagram of a primitive face model of the present invention.
- FIG 3 is a schematic view of a face region model after cutting according to the present invention.
- a three-dimensional face recognition method based on multi-scale covariance descriptors and locally sensitive Riemann kernel sparse classification includes the following steps:
- a three-dimensional face recognition method based on multi-scale covariance descriptor and local sensitive Riemann kernel sparse classification is implemented in the Windows operating system, and the three-dimensional face recognition is realized by the Matlab R2015b programming tool in the Windows operating system. Process.
- the experimental data was from the FRGC v2.0 3D Face Database, which contained 4,007 3D face models for 466 individuals tested.
- Step 1 The specific processing process of automatically pre-processing the original G library face model and P test set face model is:
- Step 1.1 Filling some small holes in the face with the effective neighborhood of the adjacent three-dimensional point cloud coordinates (x, y, z) by bicubic interpolation;
- Step 1.2 Face cutting, determine the position of the nose point according to the Shape Index feature and geometric constraints, point Shape index descriptor through its maximum curvature And minimum curvature Calculated as
- the shape index of each point in the face point cloud is calculated, and the connected region composed of points whose shape index is in the range of (0.85-1.0) is selected as the initial nose point candidate region.
- the centroid position of the face point cloud is calculated, and a connected area closest to the centroid position is selected as the nose tip area in the nose tip candidate area. Select the center of mass of the tip of the nose as the tip of the nose.
- the nose point is the center of the ball, and the 90mm radius is used to make the ball, and the three-dimensional face area of interest is cut;
- Step 1.3 Posture correction, posture correction is performed by using Principal Component Analysis (PCA) on the cut face. Taking the nose point as the coordinate origin, the feature vector corresponding to the largest feature value is taken as the Y axis, and the feature vector corresponding to the smallest feature value is used as the Z axis to establish a new Pose Coordinate System (PCS).
- PCS Pose Coordinate System
- the face area has a frontal pose, and each point is represented by a unique x, y, z coordinate;
- Step 1.4 Smooth denoising, triangulate the face point cloud in the spatial three-dimensional coordinate system, obtain a spatial triangular mesh, and then use the mesh-based smoothing algorithm to smooth and denoise the face region, after 10 iterations, Get a smooth 3D face mesh.
- Step 2 Establish a scale space and perform multi-scale key point detection and neighborhood extraction on the library set face model and the test set face model after step 1 automatic pre-processing.
- the specific processing procedure is:
- Step 2.2 Based on Gaussian smoothing of the mesh surface to establish the scale space, and obtain the grid processing information at different scales by continuously varying scale parameters. Construct an input mesh scale space containing Gaussian smoothing process, as shown in equation (2):
- M refers to the original 3D face mesh
- the Gaussian filter of the 3D face mesh passes through a binomial filter (moving from each mesh vertex V i to V j represents a point in a ring neighborhood N i of V i ), a new three-dimensional face mesh is obtained, and so on.
- the present invention selects the first three scales of the three-dimensional face data and the original face data (the original face scale is marked as 0) for subsequent processing.
- the dimension of the vector, the specific processing is:
- Step 3.1.1 Extracting the geodetic distance feature F 1 :
- F 1 represents the point in the neighborhood region P is The geodesic distance to the center point p i .
- the geodesic distance is the shortest distance between two points on the surface of the three-dimensional human face. It is a kind of feature that is highly discriminative even for deformed faces.
- Step 3.1.2 Extract Gradient Features F 2 : Point
- the gradient feature F 2 is defined as
- n y and n z represent the values of the point normals in the x and z axis directions, respectively;
- Step 3.1.3 Extract the amplitude feature F 3 : The magnitude of
- n x and n y represent the values of the point normals in the x and y axis directions, respectively.
- Gradient features and amplitude features can be used to describe the trend of the geometrical direction of a three-dimensional face surface in a particular direction.
- geodesic distance, shape index, volume, gradient, amplitude, shape diameter function, curvature, and Laplace-Beltrami descriptors can all be used to characterize three-dimensional faces.
- the present invention selects geodesic distance, gradient, and amplitude features for constructing multi-scale covariance descriptors. The selected features are a good reflection of the metrics between the points, depicting the local surface of the face and the changing trend in a particular direction.
- the first geodesic distance feature F 1s and so on.
- a set of 3-dimensional feature vectors representing all points in the region P is .
- a 3 ⁇ 3 covariance matrix C is used to represent a given three-dimensional key point neighborhood P is defined as follows:
- ⁇ s is the average eigenvector of the region P is .
- the covariance matrix C is a symmetric positive definite matrix whose diagonal elements represent the variance of each feature and the non-diagonal elements represent the correlation between the features.
- the covariance matrix C is of the region P is has an independent fixed 3 ⁇ 3 dimension irrespective of its size;
- Step 3.3 Multi-scale fusion of local covariance descriptors:
- s is the number of scales and ⁇ s is the weight coefficient.
- ⁇ s is the ratio of the recognition rate of Rank-1 at each scale to the sum of the Rank-1 recognition rates of all scales.
- C i is the multi-scale covariance descriptor of the ith key point.
- Step 4 According to the multi-scale covariance descriptor extracted in step 3, a local sensitive Riemann kernel sparse representation is proposed to classify and recognize the three-dimensional face.
- the invention proposes a local sensitive Riemann kernel sparse representation to classify and recognize three-dimensional human faces.
- ⁇ is the regularization parameter, Indicates that the corresponding elements of the vector are multiplied.
- p k ⁇ R N represents a local operator for measuring test samples
- This model is a typical optimization problem with equality constraints, which can be solved by Lagrange Multiplier.
- the equation is optimized by combining the equality constraint with the objective function by Lagrangian coefficient.
- ⁇ g ( ⁇ ) means that only the coefficient corresponding to class g is selected
- r g (Y) is the mean of the reconstructed residual sum of m descriptors of g class, and the face set model and test set with the smallest residual
- the face model can be determined as a three-dimensional face of the same person.
- the invention adopts a Log-Euclidean Gaussian kernel capable of accurately measuring the reconstruction residual, and the corresponding expression is:
- ⁇ is a parameter of the kernel function K(X, Y).
- the library set face is an offline processing mode
- the test face is an online processing mode
- Step 6 Identification experiment, the experiment uses R1RR (Rank-one Recognition Rate) as the recognition performance index.
- Step 6.1 Experiment 1, this experiment uses the FRGC v2.0 database, which collects 4,007 person face clouds of 466 objects, including smiles, surprises, anger and other expressions. Three recognition experiments were performed on the database, and each experiment consisted of the first neutral face of each object (a total of 466). (1) Neutral vs. Others, the remaining 3,541 faces constitute the test set; (2) Neutral vs. Neutral, the rest of the neutral face as a test set; (3) Neutral vs. Non-neutral, the remaining non-neutral Face as a test set. The three groups of experiments obtained the Rank-1 recognition rate of 98.3%, 100% and 95.7%, respectively.
- Step 6.2 Experiment 2, this experiment is based on the Bosphorus database, which collects 4666 face cloudes of 105 objects, with rich expressions and large expressions.
- a total of 105 three-dimensional scan data composed of each person's first neutral face was used as a library set, and the remaining neutral faces and expression faces were tested as test sets.
- the test set has a Rank-1 recognition rate of 100% for neutral faces, and the Rank-1 recognition rate for faces with expressions of anger, disgust, fear, happiness, sadness, and surprise is 97.2% and 94.2%, respectively. , 97.1%, 96.2%, 98.5% and 98.6%. It can be seen that the algorithm proposed by the invention has good robustness to expression changes.
Landscapes
- Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Theoretical Computer Science (AREA)
- Oral & Maxillofacial Surgery (AREA)
- General Health & Medical Sciences (AREA)
- Human Computer Interaction (AREA)
- Software Systems (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Image Analysis (AREA)
- Magnetic Resonance Imaging Apparatus (AREA)
Abstract
Description
Claims (5)
- 一种基于多尺度协方差描述子与局部敏感黎曼核稀疏分类的三维人脸识别方法,其特征在于,包括如下步骤:(1)分别对原始的G个库集人脸模型和P个测试集人脸模型进行自动预处理,用来改善三维数据的质量;(2)根据步骤(1)自动预处理之后的库集人脸模型和测试集人脸模型建立尺度空间并进行多尺度关键点检测及其邻域提取;(3)对每个尺度下的每个关键点邻域提取d×d维的局部协方差描述子,并对这些局部协方差描述子进行多尺度融合构建多尺度协方差描述子,d为所提取的特征向量的维数;(4)映射局部协方差描述子到可再生希尔伯特空间,提出局部敏感黎曼核稀疏表示对三维人脸进行分类识别。
- 如权利要求1所述的基于多尺度协方差描述子与局部敏感黎曼核稀疏分类的三维人脸识别方法,其特征在于,步骤(1)中,分别对原始的G个库集人脸模型和P个测试集人脸模型进行自动预处理具体包括如下步骤:(11)对人脸中的一些小孔洞采用其临近三维点云坐标(x,y,z)的有效邻域通过双三次插值进行填补;首先计算人脸点云中每一个点的形状指数,并选取形状指数在0.85-1.0范围内的点组成的连通区域作为初始的鼻尖点候选区域;其次计算人脸点云的质心位置,在鼻尖候选区域选择靠质心位置最近的一个连通区域作为鼻尖区域;选取鼻尖区域的质心作为鼻尖点;最后以鼻尖点为球心,90mm为半径做球,切割感兴趣的三维人脸区域;(13)姿势矫正,通过对切割后的人脸采用主成分分析PCA进行姿态矫正;以鼻尖点为坐标原点,最大的特征值对应的特征向量作为Y轴,最小的特征值对应的特征向量作为Z轴,建立一个新的右手姿势坐标系统PCS;在新的坐标系统中,人脸区域有一个正面姿态,且每个点由唯一的x,y,z坐标表示;(14)平滑去噪,对空间三维坐标系中的人脸点云三角化,得到空间三角网格,然后用基于网格的平滑算法对人脸区域进行平滑去噪,经过10次迭代处理,得到表面平滑的三维人脸网格。
- 如权利要求1所述的基于多尺度协方差描述子与局部敏感黎曼核稀疏分类的三维人脸识别方法,其特征在于,步骤(2)中,根据步骤(1)自动预处理之后的库集人脸模型和测试集人脸模型建立尺度空间并进行多尺度关键点检测及其邻域提取,具体包括如下步骤:(21)通过最远点采样方法对原始人脸均匀采样m个关键点p i0(i=1,…,m),本发明中m=37;(22)基于网格曲面的高斯平滑来建立尺度空间,并通过连续变化的尺度参数获得不同尺度下的网格处理信息,构建一个包含高斯平滑处理过程的输入网格尺度空间,如式(2)所示:其中M指原始三维人脸网格, 表示近似的σ s阶高斯滤波器,并且阶数σ s=2 s/kσ 0以指数形式变化,其中k和s是相应的平滑参数;三维人脸网格的高斯滤波器经过二项式滤波器卷积后,从每个网格顶点V i移动到 V j表示V i的一环邻域N i中的点,得到新的三维人脸网格,依此类推;为了得到平滑曲面,利用离散卷积值逼近期望的指数增长速度,令 表示平均边缘长度,s=0,1,…,n scales+2,本发明选择前3个尺度的三维人脸数据和原始人脸数据用于后续处理,原始人脸尺度记为0,所提取的多尺度关键点为p is(i=1,…,37;s=0,…,3);(23)对于每个尺度s,以关键点p is(i=1,…,37)为中心,以测地距离r=13为半径提取关键点邻域P is(i=1,…,37),三维人脸表面用多尺度局部区域{P is,i=1,…,37;s=0,…,3}来表示。
- 如权利要求1所述的基于多尺度协方差描述子与局部敏感黎曼核稀疏分类的三维人脸识别方法,其特征在于,步骤(3)中,对每个尺度下的每个关键点邻域提取d×d 维的局部协方差描述子,并对这些局部协方差描述子进行多尺度融合构建多尺度协方差描述子,具体包括如下步骤:(31)对每个尺度s下的关键点邻域P is(i=1,…,37)中的点 提取3个不同类型的特征F d(d=1,2,3),m i为邻域P is中点的个数:提取测地距离特征F 1:F 1表示邻域区域P is中的点 到中心点p i的测地距离,测地距离是连接三维人脸曲面上两个点之间的最短距离,是一类即使对形变人脸也具有高判别性的特征;提取梯度特征F 2:点 的梯度特征F 2定义为其中n x和n y分别表示点法线在x和y轴方向上的值,梯度特征和幅度特征可用来描述三维人脸曲面的几何法向上特定方向上的变化趋势;(32)构建每个尺度s下关键点邻域的协方差描述子C is(i=1,…,37):根据步骤(31),对于区域P is内的每一个点 j=1,…,m i,m i为区域P is中的点数,提取3维特征向量 表示点 的第1个测地距离特征F 1s,以此类推; 表示区域P is中的所有点的3维特征向量的集合,用一个3×3的协方差矩阵C is来表示一个给定的三维关键点邻域区域P is,定义如下:μ s为区域P is的平均特征向量,协方差矩阵C is是一个对称正定矩阵,它的对角元素表示的是每个特征的方差,非对角元素表示特征之间的相关性,区域P is的协方差矩阵C is具有与其大小无关的独立固定的3×3的维数;(33)局部协方差描述子的多尺度融合:其中,s为尺度个数,λ s为权重系数,λ s为各尺度下的Rank-1的识别率与所有尺度的Rank-1识别率的总和之比,C i即为第i个关键点的多尺度协方差描述子。
- 如权利要求1所述的基于多尺度协方差描述子与局部敏感黎曼核稀疏分类的三维人脸识别方法,其特征在于,步骤(4)中,映射局部协方差描述子到可再生希尔伯特空间,提出局部敏感黎曼核稀疏表示对三维人脸进行分类识别具体包括如下步骤:(41)库集字典建立,给定包含G个库集人脸的黎曼字典D={D 1,…,D g,…,D G},其中D g=[D g,1,D g,2,…,D g,m],g=[1,…,G],其中 表示第g个人的第m个关键点邻域的d×d维的多尺度协方差描述子,D中包含了L=G·m个协方差描述子;其中λ为正则化参数, 表示矢量对应元素相乘。p k∈R N表示局部算子,用于度量测试样本 和黎曼字典 中各列之间的黎曼距离,即用于测量测试样本和每个训练样本在核特征空间 中的黎曼距离,并赋予基向量不同的自由度;求局部敏感的黎曼核稀疏分类的LASSO模型的解析解,即在等式约束1 Tx k=1下,使目标函数最小;该模型是一类典型的有等式约束的最优化问题,可采用拉格朗日乘数法进行求解,通过拉格朗日系数把等式约束和目标函数进行组合,对该式进行最优化求 解,得到其解析解为相比于 范数约束下的稀疏表示,基于敏感约束下的稀疏表示可以得到更具有判别性的特征和解析解,因此其求解速度比 范数约束下的稀疏表示快得多,通过直接求解方程的系数矢量x k,可以实现局部敏感黎曼核稀疏表示分类,如下式其中δ g(·)表示仅选择类g对应的系数,r g(Y)为第g类m个描述子的重建残差和的均值,其残差最小的库集人脸模型与测试集人脸模型可以判定为同一个人的三维人脸;采用能精准测量重建残差的Log-Euclidean高斯核,其对应表达式为:K(X,Y)=exp(-γ||log(X)-log(Y)|| 2) (11)其中γ为核函数K(X,Y)的参数,本发明的实验参数为λ=10e -3,γ=2×10e -2。
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201711024463.1A CN107748871B (zh) | 2017-10-27 | 2017-10-27 | 一种基于多尺度协方差描述子与局部敏感黎曼核稀疏分类的三维人脸识别方法 |
CN201711024463.1 | 2017-10-27 |
Publications (1)
Publication Number | Publication Date |
---|---|
WO2019080488A1 true WO2019080488A1 (zh) | 2019-05-02 |
Family
ID=61254210
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/CN2018/087385 WO2019080488A1 (zh) | 2017-10-27 | 2018-05-17 | 一种基于多尺度协方差描述子与局部敏感黎曼核稀疏分类的三维人脸识别方法 |
Country Status (2)
Country | Link |
---|---|
CN (1) | CN107748871B (zh) |
WO (1) | WO2019080488A1 (zh) |
Cited By (38)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110276408A (zh) * | 2019-06-27 | 2019-09-24 | 腾讯科技(深圳)有限公司 | 3d图像的分类方法、装置、设备及存储介质 |
CN110287997A (zh) * | 2019-05-28 | 2019-09-27 | 沈阳航空航天大学 | 一种自适应加权局部约束稀疏编码方法 |
CN110490912A (zh) * | 2019-07-17 | 2019-11-22 | 哈尔滨工程大学 | 基于局部灰度顺序模型描述符的3d-rgb点云配准方法 |
CN110555245A (zh) * | 2019-08-13 | 2019-12-10 | 中国航发贵阳发动机设计研究所 | 一种三维尺度精准关注部位提取应力梯度的方法 |
CN110991227A (zh) * | 2019-10-23 | 2020-04-10 | 东北大学 | 一种基于深度类残差网络的三维物体识别和定位方法 |
CN111079684A (zh) * | 2019-12-24 | 2020-04-28 | 河南中原大数据研究院有限公司 | 一种基于粗糙-精细拟合的三维人脸检测方法 |
CN111104749A (zh) * | 2019-12-24 | 2020-05-05 | 山东恒道如一数字传媒有限公司 | 一种基于渐进式外接球结构的碰撞检测算法 |
CN111127658A (zh) * | 2019-12-23 | 2020-05-08 | 北京工商大学 | 一种基于点云重建三角网格曲面的保特征曲面重建方法 |
CN111126246A (zh) * | 2019-12-20 | 2020-05-08 | 河南中原大数据研究院有限公司 | 基于3d点云几何特征的人脸活体检测方法 |
CN111241960A (zh) * | 2020-01-06 | 2020-06-05 | 佛山科学技术学院 | 一种基于维纳滤波与pca的人脸识别方法及系统 |
CN111369610A (zh) * | 2020-03-05 | 2020-07-03 | 山东交通学院 | 基于可信度信息的点云数据粗差定位和剔除方法 |
CN111369458A (zh) * | 2020-02-28 | 2020-07-03 | 中国人民解放军空军工程大学 | 基于多尺度滚动引导滤波平滑的红外弱小目标背景抑制方法 |
CN111444802A (zh) * | 2020-03-18 | 2020-07-24 | 重庆邮电大学 | 一种人脸识别方法、装置及智能终端 |
CN111563959A (zh) * | 2020-05-06 | 2020-08-21 | 厦门美图之家科技有限公司 | 人脸三维可形变模型的更新方法、装置、设备及介质 |
CN111611996A (zh) * | 2020-04-22 | 2020-09-01 | 青岛联合创智科技有限公司 | 一种点云特征点描述子的计算方法 |
CN111768485A (zh) * | 2020-06-28 | 2020-10-13 | 北京百度网讯科技有限公司 | 三维图像的关键点标注方法、装置、电子设备及存储介质 |
CN111814874A (zh) * | 2020-07-08 | 2020-10-23 | 东华大学 | 一种用于点云深度学习的多尺度特征提取增强方法及模块 |
CN111860668A (zh) * | 2020-07-27 | 2020-10-30 | 辽宁工程技术大学 | 一种针对原始3d点云处理的深度卷积网络的点云识别方法 |
CN112002014A (zh) * | 2020-08-31 | 2020-11-27 | 中国科学院自动化研究所 | 面向精细结构的三维人脸重建方法、系统、装置 |
CN112183276A (zh) * | 2020-09-21 | 2021-01-05 | 西安理工大学 | 基于特征描述符的部分遮挡人脸识别方法 |
CN112307809A (zh) * | 2019-07-26 | 2021-02-02 | 中国科学院沈阳自动化研究所 | 一种基于稀疏特征点云的主动目标识别方法 |
CN112733705A (zh) * | 2021-01-07 | 2021-04-30 | 中科魔镜(深圳)科技发展有限公司 | 一种基于人体面部的3d智能分析系统 |
CN112766215A (zh) * | 2021-01-29 | 2021-05-07 | 北京字跳网络技术有限公司 | 人脸融合方法、装置、电子设备及存储介质 |
CN112836582A (zh) * | 2021-01-05 | 2021-05-25 | 北京大学 | 基于动态稀疏子空间的高维流系统结构变点在线检测方法 |
CN113052193A (zh) * | 2019-12-27 | 2021-06-29 | 沈阳新松机器人自动化股份有限公司 | 机器人重定位方法及系统 |
CN113111548A (zh) * | 2021-03-27 | 2021-07-13 | 西北工业大学 | 一种基于周角差值的产品三维特征点提取方法 |
CN113657259A (zh) * | 2021-08-16 | 2021-11-16 | 西安航空学院 | 基于鲁棒特征提取的单样本人脸识别方法 |
CN113674332A (zh) * | 2021-08-19 | 2021-11-19 | 上海应用技术大学 | 一种基于拓扑结构与多尺度特征的点云配准方法 |
CN113724400A (zh) * | 2021-07-26 | 2021-11-30 | 泉州装备制造研究所 | 一种面向倾斜摄影的多属性融合建筑物点云提取方法 |
CN113763274A (zh) * | 2021-09-08 | 2021-12-07 | 湖北工业大学 | 一种联合局部相位锐度定向描述的多源图像匹配方法 |
CN113887529A (zh) * | 2021-11-09 | 2022-01-04 | 天津大学 | 基于运动单元特征分解的三维人脸表情生成系统 |
CN114511911A (zh) * | 2022-02-25 | 2022-05-17 | 支付宝(杭州)信息技术有限公司 | 一种人脸识别方法、装置以及设备 |
CN114842276A (zh) * | 2022-05-18 | 2022-08-02 | 扬州大学 | 一种基于多图融合的典型相关分析的降维方法 |
CN116026528A (zh) * | 2023-01-14 | 2023-04-28 | 慈溪市远辉照明电器有限公司 | 一种高防水安全型三防灯 |
CN116226661A (zh) * | 2023-01-04 | 2023-06-06 | 浙江大邦科技有限公司 | 设备状态运行监测装置及方法 |
CN116561809A (zh) * | 2023-07-10 | 2023-08-08 | 北京中超伟业信息安全技术股份有限公司 | 一种基于点云识别保密介质的销毁方法 |
CN117290732A (zh) * | 2023-11-24 | 2023-12-26 | 山东理工昊明新能源有限公司 | 故障分类模型的构建方法、风电设备故障分类方法及装置 |
CN117789185A (zh) * | 2024-02-28 | 2024-03-29 | 浙江驿公里智能科技有限公司 | 基于深度学习的汽车油孔姿态识别系统及方法 |
Families Citing this family (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107748871B (zh) * | 2017-10-27 | 2021-04-06 | 东南大学 | 一种基于多尺度协方差描述子与局部敏感黎曼核稀疏分类的三维人脸识别方法 |
CN108830888B (zh) * | 2018-05-24 | 2021-09-14 | 中北大学 | 基于改进的多尺度协方差矩阵特征描述子的粗匹配方法 |
CN108764351B (zh) * | 2018-05-30 | 2021-08-31 | 佛山科学技术学院 | 一种基于测地距离的黎曼流形保持核学习方法及装置 |
CN109871818B (zh) * | 2019-02-27 | 2023-05-02 | 东南大学 | 基于法向量分布直方图和协方差描述子的人脸识别方法 |
CN110083715B (zh) * | 2019-03-20 | 2021-05-25 | 杭州电子科技大学 | 一种基于核稀疏表示的三维模型分类检索方法 |
CN112001231B (zh) * | 2020-07-09 | 2023-07-21 | 哈尔滨工业大学(深圳) | 加权多任务稀疏表示的三维人脸识别方法、系统及介质 |
CN111858991A (zh) * | 2020-08-06 | 2020-10-30 | 南京大学 | 一种基于协方差度量的小样本学习算法 |
CN112164098A (zh) * | 2020-09-02 | 2021-01-01 | 武汉大学 | 一种利用车载LiDAR系统预测城市道路局部坍塌的方法 |
CN112733758B (zh) * | 2021-01-15 | 2023-09-01 | 哈尔滨工业大学(深圳) | 黎曼几何不变性下基于曲线的三维人脸识别方法及系统 |
CN113740220A (zh) * | 2021-09-07 | 2021-12-03 | 中国人民解放军国防科技大学 | 基于高分辨气溶胶资料的多尺度三维变分同化方法 |
Citations (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101650777A (zh) * | 2009-09-07 | 2010-02-17 | 东南大学 | 一种基于密集点对应的快速三维人脸识别方法 |
CN101986328A (zh) * | 2010-12-06 | 2011-03-16 | 东南大学 | 一种基于局部描述符的三维人脸识别方法 |
CN102592136A (zh) * | 2011-12-21 | 2012-07-18 | 东南大学 | 基于几何图像中中频信息的三维人脸识别方法 |
CN104091162A (zh) * | 2014-07-17 | 2014-10-08 | 东南大学 | 基于特征点的三维人脸识别方法 |
CN104463111A (zh) * | 2014-11-21 | 2015-03-25 | 天津工业大学 | 多尺度特征区域曲率相融合的三维人脸识别方法 |
CN104598879A (zh) * | 2015-01-07 | 2015-05-06 | 东南大学 | 基于半刚性区域面部轮廓线的三维人脸识别方法 |
CN106022228A (zh) * | 2016-05-11 | 2016-10-12 | 东南大学 | 一种基于网格纵横局部二值模式的三维人脸识别方法 |
CN106096503A (zh) * | 2016-05-30 | 2016-11-09 | 东南大学 | 一种基于关键点和局部特征的三维人脸识别方法 |
CN107748871A (zh) * | 2017-10-27 | 2018-03-02 | 东南大学 | 一种基于多尺度协方差描述子与局部敏感黎曼核稀疏分类的三维人脸识别方法 |
Family Cites Families (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105930790B (zh) * | 2016-04-19 | 2021-02-05 | 电子科技大学 | 基于核稀疏编码的人体行为识别方法 |
CN106530338B (zh) * | 2016-10-31 | 2019-02-05 | 武汉纺织大学 | 生物组织非线性形变前后mr影像特征点匹配方法及系统 |
-
2017
- 2017-10-27 CN CN201711024463.1A patent/CN107748871B/zh active Active
-
2018
- 2018-05-17 WO PCT/CN2018/087385 patent/WO2019080488A1/zh active Application Filing
Patent Citations (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101650777A (zh) * | 2009-09-07 | 2010-02-17 | 东南大学 | 一种基于密集点对应的快速三维人脸识别方法 |
CN101986328A (zh) * | 2010-12-06 | 2011-03-16 | 东南大学 | 一种基于局部描述符的三维人脸识别方法 |
CN102592136A (zh) * | 2011-12-21 | 2012-07-18 | 东南大学 | 基于几何图像中中频信息的三维人脸识别方法 |
CN104091162A (zh) * | 2014-07-17 | 2014-10-08 | 东南大学 | 基于特征点的三维人脸识别方法 |
CN104463111A (zh) * | 2014-11-21 | 2015-03-25 | 天津工业大学 | 多尺度特征区域曲率相融合的三维人脸识别方法 |
CN104598879A (zh) * | 2015-01-07 | 2015-05-06 | 东南大学 | 基于半刚性区域面部轮廓线的三维人脸识别方法 |
CN106022228A (zh) * | 2016-05-11 | 2016-10-12 | 东南大学 | 一种基于网格纵横局部二值模式的三维人脸识别方法 |
CN106096503A (zh) * | 2016-05-30 | 2016-11-09 | 东南大学 | 一种基于关键点和局部特征的三维人脸识别方法 |
CN107748871A (zh) * | 2017-10-27 | 2018-03-02 | 东南大学 | 一种基于多尺度协方差描述子与局部敏感黎曼核稀疏分类的三维人脸识别方法 |
Non-Patent Citations (1)
Title |
---|
HARIRI, W. ET AL.: "3D Face Recognition Using Covariance Based Descriptors", PATTERN RECOGNITION LETTERS, vol. 78, 15 July 2016 (2016-07-15), pages 1 - 7, XP029618307, DOI: doi:10.1016/j.patrec.2016.03.028 * |
Cited By (67)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110287997A (zh) * | 2019-05-28 | 2019-09-27 | 沈阳航空航天大学 | 一种自适应加权局部约束稀疏编码方法 |
CN110287997B (zh) * | 2019-05-28 | 2023-04-18 | 沈阳航空航天大学 | 一种自适应加权局部约束稀疏编码方法 |
CN110276408A (zh) * | 2019-06-27 | 2019-09-24 | 腾讯科技(深圳)有限公司 | 3d图像的分类方法、装置、设备及存储介质 |
CN110276408B (zh) * | 2019-06-27 | 2022-11-22 | 腾讯科技(深圳)有限公司 | 3d图像的分类方法、装置、设备及存储介质 |
CN110490912A (zh) * | 2019-07-17 | 2019-11-22 | 哈尔滨工程大学 | 基于局部灰度顺序模型描述符的3d-rgb点云配准方法 |
CN110490912B (zh) * | 2019-07-17 | 2023-03-31 | 哈尔滨工程大学 | 基于局部灰度顺序模型描述符的3d-rgb点云配准方法 |
CN112307809B (zh) * | 2019-07-26 | 2023-07-25 | 中国科学院沈阳自动化研究所 | 一种基于稀疏特征点云的主动目标识别方法 |
CN112307809A (zh) * | 2019-07-26 | 2021-02-02 | 中国科学院沈阳自动化研究所 | 一种基于稀疏特征点云的主动目标识别方法 |
CN110555245A (zh) * | 2019-08-13 | 2019-12-10 | 中国航发贵阳发动机设计研究所 | 一种三维尺度精准关注部位提取应力梯度的方法 |
CN110555245B (zh) * | 2019-08-13 | 2023-10-24 | 中国航发贵阳发动机设计研究所 | 一种三维尺度精准关注部位提取应力梯度的方法 |
CN110991227A (zh) * | 2019-10-23 | 2020-04-10 | 东北大学 | 一种基于深度类残差网络的三维物体识别和定位方法 |
CN110991227B (zh) * | 2019-10-23 | 2023-06-30 | 东北大学 | 一种基于深度类残差网络的三维物体识别和定位方法 |
CN111126246A (zh) * | 2019-12-20 | 2020-05-08 | 河南中原大数据研究院有限公司 | 基于3d点云几何特征的人脸活体检测方法 |
CN111126246B (zh) * | 2019-12-20 | 2023-04-07 | 陕西西图数联科技有限公司 | 基于3d点云几何特征的人脸活体检测方法 |
CN111127658A (zh) * | 2019-12-23 | 2020-05-08 | 北京工商大学 | 一种基于点云重建三角网格曲面的保特征曲面重建方法 |
CN111104749A (zh) * | 2019-12-24 | 2020-05-05 | 山东恒道如一数字传媒有限公司 | 一种基于渐进式外接球结构的碰撞检测算法 |
CN111079684A (zh) * | 2019-12-24 | 2020-04-28 | 河南中原大数据研究院有限公司 | 一种基于粗糙-精细拟合的三维人脸检测方法 |
CN111104749B (zh) * | 2019-12-24 | 2023-09-15 | 山东恒道如一数字传媒有限公司 | 一种基于渐进式外接球结构的碰撞检测算法 |
CN111079684B (zh) * | 2019-12-24 | 2023-04-07 | 陕西西图数联科技有限公司 | 一种基于粗糙-精细拟合的三维人脸检测方法 |
CN113052193B (zh) * | 2019-12-27 | 2023-07-11 | 沈阳新松机器人自动化股份有限公司 | 机器人重定位方法及系统 |
CN113052193A (zh) * | 2019-12-27 | 2021-06-29 | 沈阳新松机器人自动化股份有限公司 | 机器人重定位方法及系统 |
CN111241960B (zh) * | 2020-01-06 | 2023-05-30 | 佛山科学技术学院 | 一种基于维纳滤波与pca的人脸识别方法及系统 |
CN111241960A (zh) * | 2020-01-06 | 2020-06-05 | 佛山科学技术学院 | 一种基于维纳滤波与pca的人脸识别方法及系统 |
CN111369458A (zh) * | 2020-02-28 | 2020-07-03 | 中国人民解放军空军工程大学 | 基于多尺度滚动引导滤波平滑的红外弱小目标背景抑制方法 |
CN111369458B (zh) * | 2020-02-28 | 2023-04-07 | 中国人民解放军空军工程大学 | 基于多尺度滚动引导滤波平滑的红外弱小目标背景抑制方法 |
CN111369610A (zh) * | 2020-03-05 | 2020-07-03 | 山东交通学院 | 基于可信度信息的点云数据粗差定位和剔除方法 |
CN111369610B (zh) * | 2020-03-05 | 2022-09-06 | 山东交通学院 | 基于可信度信息的点云数据粗差定位和剔除方法 |
CN111444802A (zh) * | 2020-03-18 | 2020-07-24 | 重庆邮电大学 | 一种人脸识别方法、装置及智能终端 |
CN111444802B (zh) * | 2020-03-18 | 2022-05-20 | 重庆邮电大学 | 一种人脸识别方法、装置及智能终端 |
CN111611996B (zh) * | 2020-04-22 | 2023-06-20 | 青岛联合创智科技有限公司 | 一种点云特征点描述子的计算方法 |
CN111611996A (zh) * | 2020-04-22 | 2020-09-01 | 青岛联合创智科技有限公司 | 一种点云特征点描述子的计算方法 |
CN111563959B (zh) * | 2020-05-06 | 2023-04-28 | 厦门美图之家科技有限公司 | 人脸三维可形变模型的更新方法、装置、设备及介质 |
CN111563959A (zh) * | 2020-05-06 | 2020-08-21 | 厦门美图之家科技有限公司 | 人脸三维可形变模型的更新方法、装置、设备及介质 |
CN111768485A (zh) * | 2020-06-28 | 2020-10-13 | 北京百度网讯科技有限公司 | 三维图像的关键点标注方法、装置、电子设备及存储介质 |
CN111768485B (zh) * | 2020-06-28 | 2024-01-12 | 北京百度网讯科技有限公司 | 三维图像的关键点标注方法、装置、电子设备及存储介质 |
CN111814874A (zh) * | 2020-07-08 | 2020-10-23 | 东华大学 | 一种用于点云深度学习的多尺度特征提取增强方法及模块 |
CN111814874B (zh) * | 2020-07-08 | 2024-04-02 | 东华大学 | 一种用于点云深度学习的多尺度特征提取增强方法及系统 |
CN111860668A (zh) * | 2020-07-27 | 2020-10-30 | 辽宁工程技术大学 | 一种针对原始3d点云处理的深度卷积网络的点云识别方法 |
CN111860668B (zh) * | 2020-07-27 | 2024-04-02 | 辽宁工程技术大学 | 一种针对原始3d点云处理的深度卷积网络的点云识别方法 |
CN112002014A (zh) * | 2020-08-31 | 2020-11-27 | 中国科学院自动化研究所 | 面向精细结构的三维人脸重建方法、系统、装置 |
CN112002014B (zh) * | 2020-08-31 | 2023-12-15 | 中国科学院自动化研究所 | 面向精细结构的三维人脸重建方法、系统、装置 |
CN112183276B (zh) * | 2020-09-21 | 2024-02-09 | 西安理工大学 | 基于特征描述符的部分遮挡人脸识别方法 |
CN112183276A (zh) * | 2020-09-21 | 2021-01-05 | 西安理工大学 | 基于特征描述符的部分遮挡人脸识别方法 |
CN112836582B (zh) * | 2021-01-05 | 2023-09-26 | 北京大学 | 基于动态稀疏子空间的高维流系统结构变点在线检测方法 |
CN112836582A (zh) * | 2021-01-05 | 2021-05-25 | 北京大学 | 基于动态稀疏子空间的高维流系统结构变点在线检测方法 |
CN112733705A (zh) * | 2021-01-07 | 2021-04-30 | 中科魔镜(深圳)科技发展有限公司 | 一种基于人体面部的3d智能分析系统 |
CN112766215A (zh) * | 2021-01-29 | 2021-05-07 | 北京字跳网络技术有限公司 | 人脸融合方法、装置、电子设备及存储介质 |
CN113111548A (zh) * | 2021-03-27 | 2021-07-13 | 西北工业大学 | 一种基于周角差值的产品三维特征点提取方法 |
CN113724400A (zh) * | 2021-07-26 | 2021-11-30 | 泉州装备制造研究所 | 一种面向倾斜摄影的多属性融合建筑物点云提取方法 |
CN113657259A (zh) * | 2021-08-16 | 2021-11-16 | 西安航空学院 | 基于鲁棒特征提取的单样本人脸识别方法 |
CN113657259B (zh) * | 2021-08-16 | 2023-07-21 | 西安航空学院 | 基于鲁棒特征提取的单样本人脸识别方法 |
CN113674332A (zh) * | 2021-08-19 | 2021-11-19 | 上海应用技术大学 | 一种基于拓扑结构与多尺度特征的点云配准方法 |
CN113674332B (zh) * | 2021-08-19 | 2024-05-21 | 上海应用技术大学 | 一种基于拓扑结构与多尺度特征的点云配准方法 |
CN113763274B (zh) * | 2021-09-08 | 2023-06-06 | 湖北工业大学 | 一种联合局部相位锐度定向描述的多源图像匹配方法 |
CN113763274A (zh) * | 2021-09-08 | 2021-12-07 | 湖北工业大学 | 一种联合局部相位锐度定向描述的多源图像匹配方法 |
CN113887529A (zh) * | 2021-11-09 | 2022-01-04 | 天津大学 | 基于运动单元特征分解的三维人脸表情生成系统 |
CN114511911A (zh) * | 2022-02-25 | 2022-05-17 | 支付宝(杭州)信息技术有限公司 | 一种人脸识别方法、装置以及设备 |
CN114842276A (zh) * | 2022-05-18 | 2022-08-02 | 扬州大学 | 一种基于多图融合的典型相关分析的降维方法 |
CN114842276B (zh) * | 2022-05-18 | 2024-03-26 | 扬州大学 | 一种基于多图融合的典型相关分析的降维方法 |
CN116226661A (zh) * | 2023-01-04 | 2023-06-06 | 浙江大邦科技有限公司 | 设备状态运行监测装置及方法 |
CN116026528A (zh) * | 2023-01-14 | 2023-04-28 | 慈溪市远辉照明电器有限公司 | 一种高防水安全型三防灯 |
CN116561809B (zh) * | 2023-07-10 | 2023-10-24 | 北京中超伟业信息安全技术股份有限公司 | 一种基于点云识别保密介质的销毁方法 |
CN116561809A (zh) * | 2023-07-10 | 2023-08-08 | 北京中超伟业信息安全技术股份有限公司 | 一种基于点云识别保密介质的销毁方法 |
CN117290732A (zh) * | 2023-11-24 | 2023-12-26 | 山东理工昊明新能源有限公司 | 故障分类模型的构建方法、风电设备故障分类方法及装置 |
CN117290732B (zh) * | 2023-11-24 | 2024-03-01 | 山东理工昊明新能源有限公司 | 故障分类模型的构建方法、风电设备故障分类方法及装置 |
CN117789185A (zh) * | 2024-02-28 | 2024-03-29 | 浙江驿公里智能科技有限公司 | 基于深度学习的汽车油孔姿态识别系统及方法 |
CN117789185B (zh) * | 2024-02-28 | 2024-05-10 | 浙江驿公里智能科技有限公司 | 基于深度学习的汽车油孔姿态识别系统及方法 |
Also Published As
Publication number | Publication date |
---|---|
CN107748871A (zh) | 2018-03-02 |
CN107748871B (zh) | 2021-04-06 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
WO2019080488A1 (zh) | 一种基于多尺度协方差描述子与局部敏感黎曼核稀疏分类的三维人脸识别方法 | |
CN108898606B (zh) | 医学图像的自动分割方法、系统、设备及存储介质 | |
CN107122396B (zh) | 基于深度卷积神经网络的三维模型检索方法 | |
CN107742102B (zh) | 一种基于深度传感器的手势识别方法 | |
CN108052942B (zh) | 一种飞机飞行姿态的视觉图像识别方法 | |
CN101819628B (zh) | 结合形状特征的稀疏表示人脸识别方法 | |
WO2017219391A1 (zh) | 一种基于三维数据的人脸识别系统 | |
CN104200240B (zh) | 一种基于内容自适应哈希编码的草图检索方法 | |
CN101833672B (zh) | 基于约束采样与形状特征的稀疏表示人脸识别方法 | |
CN109684969B (zh) | 凝视位置估计方法、计算机设备及存储介质 | |
JP7135659B2 (ja) | 形状補完装置、形状補完学習装置、方法、及びプログラム | |
CN106778474A (zh) | 3d人体识别方法及设备 | |
CN106980848A (zh) | 基于曲波变换和稀疏学习的人脸表情识别方法 | |
CN107301643B (zh) | 基于鲁棒稀疏表示与拉普拉斯正则项的显著目标检测方法 | |
CN106844620B (zh) | 一种基于视图的特征匹配三维模型检索方法 | |
CN106096517A (zh) | 一种基于低秩矩阵与特征脸的人脸识别方法 | |
CN112784782B (zh) | 一种基于多视角双注意网络的三维物体识别方法 | |
CN113168729B (zh) | 一种基于局部参考坐标系的3d形状匹配方法及装置 | |
Zhou et al. | 2D compressive sensing and multi-feature fusion for effective 3D shape retrieval | |
Zhao et al. | Three-dimensional face reconstruction of static images and computer standardization issues | |
CN116911079A (zh) | 一种不完备模型的自演化建模方法及系统 | |
Deng et al. | Point cloud resampling via hypergraph signal processing | |
Xia et al. | Realpoint3d: Point cloud generation from a single image with complex background | |
CN113723468A (zh) | 一种三维点云的物体检测方法 | |
Ganapathi et al. | Facet-Level Segmentation of 3d Textures on Cultural Heritage Objects |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
121 | Ep: the epo has been informed by wipo that ep was designated in this application |
Ref document number: 18870362 Country of ref document: EP Kind code of ref document: A1 |
|
NENP | Non-entry into the national phase |
Ref country code: DE |
|
122 | Ep: pct application non-entry in european phase |
Ref document number: 18870362 Country of ref document: EP Kind code of ref document: A1 |
|
122 | Ep: pct application non-entry in european phase |
Ref document number: 18870362 Country of ref document: EP Kind code of ref document: A1 |
|
32PN | Ep: public notification in the ep bulletin as address of the adressee cannot be established |
Free format text: NOTING OF LOSS OF RIGHTS (EPO FORM 1205A DATED 17.12.2020) |