CN113450269A - Point cloud key point extraction method based on 3D vision - Google Patents
Point cloud key point extraction method based on 3D vision Download PDFInfo
- Publication number
- CN113450269A CN113450269A CN202110567048.0A CN202110567048A CN113450269A CN 113450269 A CN113450269 A CN 113450269A CN 202110567048 A CN202110567048 A CN 202110567048A CN 113450269 A CN113450269 A CN 113450269A
- Authority
- CN
- China
- Prior art keywords
- point
- point cloud
- feature
- key
- points
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000605 extraction Methods 0.000 title claims abstract description 19
- 230000009466 transformation Effects 0.000 claims abstract description 20
- 238000000034 method Methods 0.000 claims abstract description 18
- 239000011159 matrix material Substances 0.000 claims abstract description 17
- 238000004364 calculation method Methods 0.000 claims abstract description 13
- 238000001914 filtration Methods 0.000 claims abstract description 9
- 238000012545 processing Methods 0.000 claims abstract description 4
- 238000007619 statistical method Methods 0.000 claims abstract description 4
- 239000013598 vector Substances 0.000 claims description 33
- 238000000354 decomposition reaction Methods 0.000 claims description 9
- 238000006243 chemical reaction Methods 0.000 claims description 4
- 238000003384 imaging method Methods 0.000 claims description 3
- 238000005070 sampling Methods 0.000 claims description 3
- 230000002194 synthesizing effect Effects 0.000 claims description 3
- 230000008030 elimination Effects 0.000 claims description 2
- 238000003379 elimination reaction Methods 0.000 claims description 2
- 238000004422 calculation algorithm Methods 0.000 abstract description 4
- 230000015572 biosynthetic process Effects 0.000 abstract 1
- 230000002349 favourable effect Effects 0.000 abstract 1
- 238000003786 synthesis reaction Methods 0.000 abstract 1
- 238000010586 diagram Methods 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 238000001514 detection method Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Images
Classifications
-
- G06T5/70—
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/243—Classification techniques relating to the number of classes
- G06F18/2433—Single-class perspective, e.g. one-against-all classification; Novelty detection; Outlier detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/50—Depth or shape recovery
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10028—Range image; Depth image; 3D point clouds
Abstract
The invention discloses a point cloud key point extraction method based on 3D vision, which belongs to the technical field of 3D vision and is characterized in that a point cloud picture and point cloud data are obtained by utilizing the synthesis of a depth map and a color map; carrying out statistical filtering processing, carrying out statistical analysis on the neighborhood of each point in the point cloud, and removing outliers; calculating the feature degree of each point in the point cloud, determining feature points according to the feature degrees, and establishing a rapid feature point histogram of the key points of the target point cloud; and calculating to obtain a coordinate transformation matrix of the key points of the source point cloud and the key points of the target point cloud. The method is scientific and reasonable, is safe and convenient to use, removes noise points irrelevant to the component point cloud by carrying out statistical filtering on the point cloud and taking out outliers, and is favorable for reducing algorithm time and space complexity; the source point cloud key points and the target point cloud key points are selected by utilizing the feature degrees, and a fast feature point histogram of the key points is established, so that feature calculation data are few, the calculation efficiency and the calculation precision are improved, and the anti-noise capability is strong.
Description
Technical Field
The invention relates to the technical field of 3D vision, in particular to a point cloud key point extraction method based on 3D vision.
Background
For some complicated heterogeneous parts, a traditional detection mode of manual comparison such as a sample plate, a mold and the like is adopted, the evaluation precision depends on the experience of a technician, and accurate digital description cannot be obtained. With the powerful development of the 3D visual technology, the point cloud data with high precision and high density can be obtained very quickly. The point cloud data on the surface of the part is analyzed and processed, the characteristics of the part are extracted, and quantitative description of the quality of the part and reverse three-dimensional reconstruction of the characteristics can be achieved.
The research of point cloud feature point extraction and feature line fitting mainly focuses on two aspects, namely feature extraction based on a grid model and feature extraction based on scattered point clouds. The method for extracting the characteristic line of the grid model can be divided into automatic extraction and semi-automatic extraction: usually abrupt points of curvature on the grid are found as feature points and these discrete features are then fitted to a feature line. Compared with the characteristic line extraction of a grid model, the scattered point cloud has no topological information, at present, the research of directly extracting the characteristics from the scattered point cloud is relatively less, the repeatability of extracting the key points is higher, the accuracy degree of the key points cannot be ensured, and the extraction efficiency is lower. Therefore, a method for extracting key points of a point cloud based on 3D vision is needed to solve the above problems.
Disclosure of Invention
The purpose of the invention is as follows: the invention aims to provide a point cloud key point extraction method based on 3D vision to solve the problem of low key point extraction efficiency in the prior art.
The technical scheme is as follows: the invention relates to a point cloud key point extraction method based on 3D vision, which comprises the following steps,
s10, shooting through a depth camera to obtain a depth map and a color map image of the complex heterogeneous piece, and performing coordinate conversion by using a formula; synthesizing by using the depth map and the color map to obtain a point cloud map and point cloud data;
s20, carrying out statistical filtering processing, carrying out statistical analysis on the neighborhood of each point in the point cloud, and calculating the average distance between the point cloud and all the adjacent points; points with the average distance exceeding the standard range are defined as outliers and are removed from the data;
s30, after the point cloud is subjected to statistical filtering, calculating the normal vector change degree of each point in the point cloud, selecting a point with a large normal vector change degree as a characteristic point, wherein the normal vector change degree is the characteristic degree; the larger the change of the characteristic degree is, the larger the surface relief of the component is;
s40, calculating the SPFH value of each feature point, selecting one feature point as a source point cloud key point, matching and continuously repeating the source point cloud key point and the neighbor of the source point cloud key point, and generating a rapid feature point histogram of the source point cloud key point by continuously changing the weight of the neighbor SPFH value; defining the feature points of the complex heterogeneous pieces to be extracted as target point cloud key points, and generating a fast feature point histogram of the target point cloud key points; establishing a feature vector for describing the geometric characteristics of a local area of the source point cloud key point and the target point cloud key point by utilizing the statistical characteristics of the histogram;
s50, calculating Euclidean distance between the source point cloud key point feature vector and the target point cloud key point feature vector; obtaining a source point cloud key point Kd-tree and a target point cloud key point Kd-tree according to the feature point histogram of the source point cloud key point and the feature point histogram of the target point cloud; and calculating to obtain a coordinate transformation matrix of the key points of the source point cloud and the target point cloud based on a key point cloud registration method.
Preferably, in S10, the coordinate transformation formula is as follows:
xw＝zc*(u-u0)*dx(dy)(dz)/f
wherein x iswIs the coordinate of the world coordinate system, u is the coordinate of the center coordinate of the image in the world coordinate system0Is the central coordinate of the image, zcIs the z-axis value of the camera coordinate system (i.e., the distance of the target object from the camera), dx、dy、 dzFor each pixel of the imaging element the corresponding physical width and height are f the camera focal length.
Preferably, in S30, the feature degree calculation formula is as follows:
wherein: thetaijIs a normal vector angle f of a neighboring pointiIs the characteristic value of a certain point in the point cloud, and k is the number of the adjacent points.
Preferably, the S40 includes the following steps:
s4001, a characteristic point is given as a source point cloud key point, an angle difference value after normal decomposition is calculated, and the angle difference value is defined as SPFH;
s4002, calculating a final value of the fast feature point histogram by using the weighted adjacent SPFH:
wherein, wkFor the weights, i.e. the distance between two points, FPFH (P)q) Namely a fast feature histogram of key points, SPFH is a simplified point feature histogram, and k is the number of neighboring points.
Preferably, in S50, the euclidean distance calculation formula is as follows:
Si＝(Si1 Si2 ... Si120)
Tj＝(Tj1 Tj2 ... Tj120)
wherein S isiAs source cloud key point feature vectors, TjIs a feature vector of key points of the target point cloud, d (S)i，Tj) Is the Euclidean distance between two feature vectors, p is the p-th key point characteristicAnd (5) sign vectors.
Preferably, in S50, the calculating a coordinate transformation matrix of the key points of the source point cloud and the target point cloud includes the following steps:
s5001, searching the nearest point and the distance of the key point of the source point cloud in a Kd-tree of the key point of the target point cloud, and if the distance is smaller than a fixed threshold, adding the point into a first pre-corresponding relation set;
s5002, searching the nearest point and the distance of the key point of the target point cloud in the Kd-tree of the source point cloud key point, and if the distance is smaller than a fixed threshold, adding the point pair into a second pre-corresponding relationship set;
s5003, taking an intersection of the first pre-corresponding relationship set and the second pre-corresponding relationship set as an initial corresponding relationship, wherein the intersection is the same corresponding relationship in the first pre-corresponding relationship set and the second pre-corresponding relationship set;
s5004, according to the initial corresponding relation, adopting random sampling consistency to carry out corresponding elimination of error relation, randomly selecting three groups of corresponding relations from the obtained corresponding relations, calculating a coordinate transformation matrix from a source point cloud key point to a target point cloud key point according to the three groups of corresponding relations, calculating the distance deviation between the source point cloud key point and the corresponding point in the target point cloud after coordinate transformation of all the corresponding relations, if the deviation is less than a set threshold value, the corresponding relation belongs to a sample in the model, otherwise, the corresponding relation belongs to a sample outside the model;
s5005, storing all internal samples, repeating the above process, counting the times of each group of corresponding relations belonging to the internal samples until the iteration times reach a set value (the set value depends on the desired precision, the larger the set value is, the higher the precision is), ending the iteration to obtain a final corresponding relation set, finding out three groups of corresponding relations which belong to the internal samples and have the largest number, and calculating the coordinate transformation matrix of the source point cloud key point and the target point cloud key point by using singular value decomposition.
Has the advantages that: according to the method, the noise points irrelevant to the component point cloud are removed by performing statistical filtering on the point cloud and taking out the outliers, so that the algorithm time and the space complexity are reduced; selecting source point cloud key points and target point cloud key points by using the feature degrees, and establishing a key point fast feature point histogram, wherein feature calculation data are few, calculation efficiency and calculation precision are improved, and the anti-noise capability is strong; in addition, according to the key point search with the maximum key point neighborhood curvature mean value, the noise resistance of the algorithm is enhanced, and the repetition of the key points at the same position is reduced.
Drawings
FIG. 1 is a general flowchart of a method for extracting a point cloud key point based on 3D vision according to the present invention;
FIG. 2 is a schematic diagram of angle difference after normal decomposition of a point cloud key point extraction method based on 3D vision according to the present invention;
FIG. 3 is a point cloud diagram of a point cloud key point extraction method based on 3D vision according to the invention;
FIG. 4 is a filtered point cloud image of a 3D vision-based point cloud key point extraction method of the present invention;
FIG. 5 is a feature point histogram of a 3D vision-based point cloud key point extraction method of the present invention;
Detailed Description
The technical solution of the present invention is described in detail below with reference to the accompanying drawings, but the scope of the present invention is not limited to the embodiments.
Example 1: as shown in fig. 1-5, a method for extracting a point cloud key point based on 3D vision includes the following steps,
step 40, calculating the SPFH value of each feature point, selecting one feature point as a source point cloud key point, matching and continuously repeating the source point cloud key point and the neighbor thereof, and generating a rapid feature point histogram of the source point cloud key point by continuously changing the weight of the neighbor SPFH value; defining the feature points of the complex heterogeneous pieces to be extracted as target point cloud key points, and generating a fast feature point histogram of the target point cloud key points; establishing a feature vector for describing the geometric characteristics of a local area of the source point cloud key point and the target point cloud key point by utilizing the statistical characteristics of the histogram;
step 50, calculating Euclidean distance between the source point cloud key point feature vector and the target point cloud key point feature vector; obtaining a source point cloud key point Kd-tree and a target point cloud key point Kd-tree (obtained through a PCL point cloud library open source algorithm, and redundant description is not repeated herein) according to the feature point histogram of the source point cloud key point and the feature point histogram of the target point cloud; and calculating to obtain a coordinate transformation matrix of the key points of the source point cloud and the target point cloud based on a key point cloud registration method. In S10, the coordinate conversion formula is as follows:
xw＝zc*(u-u0)*dx(dy)(dz)/f
wherein x iswIs the coordinate of the world coordinate system, u is the coordinate of the center coordinate of the image in the world coordinate system, u0Is the central coordinate of the image, zcIs the z-axis value of the camera coordinate system (i.e., the distance of the target object from the camera), dx、dy、 dzFor each pixel of the imaging element the corresponding physical width and height are f the camera focal length.
In step 30, the feature degree calculation formula is as follows:
wherein: thetaijIs a normal vector angle f of a neighboring pointiIs the characteristic value of a certain point in the point cloud, and k is the number of the adjacent points.
Step 40 comprises the steps of:
step 4001, a feature point is given as a source point cloud key point, an angle difference value after normal decomposition is calculated, and the angle difference value is defined as SPFH;
the angle differences after normal decomposition are respectively alpha, phi and theta (see the attached figure 2), and the calculation formulas are respectively as follows: α ═ v × nt，
Step 4002, calculate the final value of the fast feature point histogram using the weighted adjacent SPFH:
wherein, wkFor the weights, i.e. the distance between two points, FPFH (P)q) Namely a fast feature histogram of key points, SPFH is a simplified point feature histogram, and k is the number of neighboring points. In this example, k is chosen to be 50.
In step 50, the euclidean distance calculation formula is as follows:
Si＝(Si1 Si2 ... Si120)
Tj＝(Tj1 Tj2 ... Tj120)
wherein S isiAs source cloud key point feature vectors, TjIs a feature vector of key points of the target point cloud, d (S)i，Tj) And p is the Euclidean distance between two feature vectors and is the p-th key point feature vector.
In step 50, calculating a coordinate transformation matrix of the source point cloud key points and the target point cloud key points comprises the following steps:
step 5001, searching the nearest neighbor point of the key point of the source point cloud and the distance thereof in the Kd-tree of the key point of the target point cloud (the distance has an open source program algorithm in PCL, namely the point cloud library), and if the distance is less than a fixed threshold (the threshold is 0.02), adding the point into the first pre-corresponding relationship set;
step 5002, searching the nearest neighbor point of the key point of the target point cloud and the distance thereof in the Kd-tree of the source point cloud key point (the distance has an open source program algorithm in PCL, namely the point cloud library), and if the distance is less than a fixed threshold (the threshold is 0.02), adding the point pair into a second pre-corresponding relationship set;
step 5003, taking an intersection of the first pre-correspondence set and the second pre-correspondence set as an initial correspondence, where the intersection is the same correspondence in the first pre-correspondence set and the second pre-correspondence set;
step 5004, according to the initial corresponding relationship, adopting random sampling consistency to make corresponding removal of error relationship, randomly selecting three groups of corresponding relationships from the obtained corresponding relationships, calculating coordinate transformation matrix from source point cloud key point to target point cloud key point according to these three groups of corresponding relationships, calculating distance deviation between the source point cloud key point and corresponding point in the target point cloud after coordinate transformation of all corresponding relationships, if the deviation is less than set threshold value, said corresponding relationship belongs to sample in the model, otherwise it belongs to sample outside the model;
step 5005, storing all internal samples, repeating the above process, counting the number of times that each group of corresponding relations belong to the internal samples until the number of iterations reaches a set value (the set value depends on the desired accuracy, the larger the set value is, the higher the accuracy is), ending the iterations to obtain a final corresponding relation set, finding the three groups of corresponding relations that belong to the internal samples with the largest number, and calculating the coordinate transformation matrix of the source point cloud key point and the target point cloud key point by using singular value decomposition. The transformation matrix is estimated by using Singular Value Decomposition (SVD), which is more accurate than a QR matrix, and the matrix obtained by the SVD method has the minimum square error, so that the coordinate transformation matrix of the source point cloud key points and the target point cloud key points is more accurate, and the result is more accurate
As above, while the invention has been shown and described with reference to certain preferred embodiments, it is not to be construed as limited thereto. Various changes in form and detail may be made therein without departing from the spirit and scope of the invention as defined by the appended claims.
Claims (6)
1. A point cloud key point extraction method based on 3D vision is characterized by comprising the following steps:
s10, shooting through a depth camera to obtain a depth map and a color map image of the complex heterogeneous piece, and performing coordinate conversion by using a formula; synthesizing by using the depth map and the color map to obtain a point cloud map and point cloud data;
s20, carrying out statistical filtering processing, carrying out statistical analysis on the neighborhood of each point in the point cloud, and calculating the average distance between the point cloud and all the adjacent points; points with the average distance exceeding the standard range are defined as outliers and are removed from the data;
s30, after the point cloud is subjected to statistical filtering, calculating the normal vector change degree of each point in the point cloud, selecting a point with a large normal vector change degree as a characteristic point, wherein the normal vector change degree is the characteristic degree; the larger the change of the characteristic degree is, the larger the surface relief of the component is;
s40, calculating the SPFH value of each feature point, selecting one feature point as a source point cloud key point, matching and continuously repeating the source point cloud key point and the neighbor of the source point cloud key point, and generating a rapid feature point histogram of the source point cloud key point by continuously changing the weight of the neighbor SPFH value; defining the feature points of the complex heterogeneous pieces to be extracted as target point cloud key points, and generating a fast feature point histogram of the target point cloud key points; establishing a feature vector for describing the geometric characteristics of a local area of the source point cloud key point and the target point cloud key point by utilizing the statistical characteristics of the histogram;
s50, calculating Euclidean distance between the source point cloud key point feature vector and the target point cloud key point feature vector; obtaining a source point cloud key point Kd-tree and a target point cloud key point Kd-tree according to the feature point histogram of the source point cloud key point and the feature point histogram of the target point cloud; and calculating to obtain a coordinate transformation matrix of the key points of the source point cloud and the target point cloud based on a key point cloud registration method.
2. The method for extracting point cloud key points based on 3D vision according to claim 1, wherein in S10, the coordinate transformation formula is as follows:
xw＝zc*(u-u0)*dx(dy)(dz)/f
wherein x iswIs the coordinate of the world coordinate system, u is the coordinate of the center coordinate of the image in the world coordinate system, u0Is the central coordinate of the image, zcIs the z-axis value of the camera coordinate system (i.e., the distance of the target object from the camera), dx、dy、dzFor each pixel of the imaging element the corresponding physical width and height are f the camera focal length.
3. The method for extracting point cloud key points based on 3D vision according to claim 1, wherein in S30, the feature degree calculation formula is as follows:
wherein: thetaijIs a normal vector angle f of a neighboring pointiIs the characteristic value of a certain point in the point cloud, and k is the number of the adjacent points.
4. The method for extracting point cloud key points based on 3D vision according to claim 1, wherein the step S40 includes the following steps:
s4001, a characteristic point is given as a source point cloud key point, an angle difference value after normal decomposition is calculated, and the angle difference value is defined as SPFH;
s4002, calculating a final value of the fast feature point histogram by using the weighted adjacent SPFH:
wherein, wkFor the weights, i.e. the distance between two points, FPFH (P)q) Namely a fast feature histogram of key points, SPFH is a simplified point feature histogram, and k is the number of neighboring points.
5. The method for extracting point cloud key points based on 3D vision according to claim 1, wherein in S50, the euclidean distance calculation formula is as follows:
Si＝(Si1 Si2 ... Si120)
Tj＝(Tj1 Tj2 ... Tj120)
wherein S isiAs source cloud key point feature vectors, TjIs a feature vector of key points of the target point cloud, d (S)i，Tj) And p is the Euclidean distance between two feature vectors and is the p-th key point feature vector.
6. The method of claim 5, wherein the step of calculating the coordinate transformation matrix of the cloud key points of the source point and the cloud key points of the target point in the step S50 comprises the steps of:
s5001, searching the nearest point and the distance of the key point of the source point cloud in a Kd-tree of the key point of the target point cloud, and if the distance is smaller than a fixed threshold, adding the point into a first pre-corresponding relation set;
s5002, searching the nearest point and the distance of the key point of the target point cloud in the Kd-tree of the source point cloud key point, and if the distance is smaller than a fixed threshold, adding the point pair into a second pre-corresponding relationship set;
s5003, taking an intersection of the first pre-corresponding relationship set and the second pre-corresponding relationship set as an initial corresponding relationship, wherein the intersection is the same corresponding relationship in the first pre-corresponding relationship set and the second pre-corresponding relationship set;
s5004, according to the initial corresponding relation, adopting random sampling consistency to carry out corresponding elimination of error relation, randomly selecting three groups of corresponding relations from the obtained corresponding relations, calculating a coordinate transformation matrix from a source point cloud key point to a target point cloud key point according to the three groups of corresponding relations, calculating the distance deviation between the source point cloud key point and the corresponding point in the target point cloud after coordinate transformation of all the corresponding relations, if the deviation is less than a set threshold value, the corresponding relation belongs to a sample in the model, otherwise, the corresponding relation belongs to a sample outside the model;
s5005, storing all internal samples, repeating the above process, counting the times of each group of corresponding relations belonging to the internal samples until the iteration times reach a set value (the set value depends on the desired precision, the larger the set value is, the higher the precision is), ending the iteration to obtain a final corresponding relation set, finding out three groups of corresponding relations which belong to the internal samples and have the largest number, and calculating the coordinate transformation matrix of the source point cloud key point and the target point cloud key point by using singular value decomposition.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110567048.0A CN113450269A (en) | 2021-05-24 | 2021-05-24 | Point cloud key point extraction method based on 3D vision |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110567048.0A CN113450269A (en) | 2021-05-24 | 2021-05-24 | Point cloud key point extraction method based on 3D vision |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113450269A true CN113450269A (en) | 2021-09-28 |
Family
ID=77810141
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110567048.0A Pending CN113450269A (en) | 2021-05-24 | 2021-05-24 | Point cloud key point extraction method based on 3D vision |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113450269A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115338874A (en) * | 2022-10-19 | 2022-11-15 | 爱夫迪(沈阳)自动化科技有限公司 | Laser radar-based robot real-time control method |
| CN115797423A (en) * | 2022-12-07 | 2023-03-14 | 中国矿业大学(北京) | Registration method and system for optimizing near iteration of mine point cloud based on descriptor characteristics |
| CN116563561A (en) * | 2023-07-06 | 2023-08-08 | 北京优脑银河科技有限公司 | Point cloud feature extraction method, point cloud registration method and readable storage medium |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104143210A (en) * | 2014-07-31 | 2014-11-12 | 哈尔滨工程大学 | Multi-scale normal feature point cloud registering method |
| CN105046694A (en) * | 2015-07-02 | 2015-11-11 | 哈尔滨工程大学 | Quick point cloud registration method based on curved surface fitting coefficient features |
| CN109887015A (en) * | 2019-03-08 | 2019-06-14 | 哈尔滨工程大学 | A kind of point cloud autoegistration method based on local surface feature histogram |
| CN110490912A (en) * | 2019-07-17 | 2019-11-22 | 哈尔滨工程大学 | 3D-RGB point cloud registration method based on local gray level sequence model descriptor |
-
2021
- 2021-05-24 CN CN202110567048.0A patent/CN113450269A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104143210A (en) * | 2014-07-31 | 2014-11-12 | 哈尔滨工程大学 | Multi-scale normal feature point cloud registering method |
| CN105046694A (en) * | 2015-07-02 | 2015-11-11 | 哈尔滨工程大学 | Quick point cloud registration method based on curved surface fitting coefficient features |
| CN109887015A (en) * | 2019-03-08 | 2019-06-14 | 哈尔滨工程大学 | A kind of point cloud autoegistration method based on local surface feature histogram |
| CN110490912A (en) * | 2019-07-17 | 2019-11-22 | 哈尔滨工程大学 | 3D-RGB point cloud registration method based on local gray level sequence model descriptor |
Non-Patent Citations (2)
| Title |
|---|
| 宋成航: ""利用特征点采样一致性改进ICP算法点云配准方法"", 北京测绘, 25 March 2021 (2021-03-25), pages 317 - 322 * |
| 王永信: "《逆向工程及检测技术与应用》", 31 May 2014, 西安：西安交通大学出版社, pages: 18 - 20 * |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115338874A (en) * | 2022-10-19 | 2022-11-15 | 爱夫迪(沈阳)自动化科技有限公司 | Laser radar-based robot real-time control method |
| CN115338874B (en) * | 2022-10-19 | 2023-01-03 | 爱夫迪(沈阳)自动化科技有限公司 | Real-time robot control method based on laser radar |
| CN115797423A (en) * | 2022-12-07 | 2023-03-14 | 中国矿业大学(北京) | Registration method and system for optimizing near iteration of mine point cloud based on descriptor characteristics |
| CN116563561A (en) * | 2023-07-06 | 2023-08-08 | 北京优脑银河科技有限公司 | Point cloud feature extraction method, point cloud registration method and readable storage medium |
| CN116563561B (en) * | 2023-07-06 | 2023-11-14 | 北京优脑银河科技有限公司 | Point cloud feature extraction method, point cloud registration method and readable storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110322453B (en) | 3D point cloud semantic segmentation method based on position attention and auxiliary network | |
| CN103729643B (en) | The identification of three dimensional object in multi-mode scene and posture are determined | |
| CN113450269A (en) | Point cloud key point extraction method based on 3D vision | |
| JP5677798B2 (en) | 3D object recognition and position and orientation determination method in 3D scene | |
| CN111079685B (en) | 3D target detection method | |
| CN107633226B (en) | Human body motion tracking feature processing method | |
| CN108303037B (en) | Method and device for detecting workpiece surface shape difference based on point cloud analysis | |
| CN110544233B (en) | Depth image quality evaluation method based on face recognition application | |
| CN104616278B (en) | Three-dimensional point cloud interest point detection method and system | |
| CN111028292B (en) | Sub-pixel level image matching navigation positioning method | |
| CN110070567B (en) | Ground laser point cloud registration method | |
| CN107133966B (en) | Three-dimensional sonar image background segmentation method based on sampling consistency algorithm | |
| CN107481274B (en) | Robust reconstruction method of three-dimensional crop point cloud | |
| CN104090972A (en) | Image feature extraction and similarity measurement method used for three-dimensional city model retrieval | |
| CN111027140B (en) | Airplane standard part model rapid reconstruction method based on multi-view point cloud data | |
| CN109993800A (en) | A kind of detection method of workpiece size, device and storage medium | |
| CN113470090A (en) | Multi-solid-state laser radar external reference calibration method based on SIFT-SHOT characteristics | |
| CN111091101B (en) | High-precision pedestrian detection method, system and device based on one-step method | |
| CN111524168B (en) | Point cloud data registration method, system and device and computer storage medium | |
| CN108921864A (en) | A kind of Light stripes center extraction method and device | |
| CN106257497B (en) | Matching method and device for image homonymy points | |
| CN107909018B (en) | Stable multi-mode remote sensing image matching method and system | |
| CN113888461A (en) | Method, system and equipment for detecting defects of hardware parts based on deep learning | |
| CN114067075A (en) | Point cloud completion method and device based on generation of countermeasure network | |
| CN114663373A (en) | Point cloud registration method and device for detecting surface quality of part |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |