CN107292899B - Angular point feature extraction method for two-dimensional laser scanner - Google Patents

Angular point feature extraction method for two-dimensional laser scanner Download PDF

Info

Publication number
CN107292899B
CN107292899B CN201710312254.0A CN201710312254A CN107292899B CN 107292899 B CN107292899 B CN 107292899B CN 201710312254 A CN201710312254 A CN 201710312254A CN 107292899 B CN107292899 B CN 107292899B
Authority
CN
China
Prior art keywords
neighborhood
point
points
range
subscript
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.)
Expired - Fee Related
Application number
CN201710312254.0A
Other languages
Chinese (zh)
Other versions
CN107292899A (en
Inventor
于慧敏
黎睿
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Zhejiang University ZJU
Original Assignee
Zhejiang University ZJU
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Zhejiang University ZJU filed Critical Zhejiang University ZJU
Priority to CN201710312254.0A priority Critical patent/CN107292899B/en
Publication of CN107292899A publication Critical patent/CN107292899A/en
Application granted granted Critical
Publication of CN107292899B publication Critical patent/CN107292899B/en
Expired - Fee Related legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/10—Segmentation; Edge detection
    • G06T7/13—Edge detection
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/22—Matching criteria, e.g. proximity measures

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Artificial Intelligence (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Evolutionary Biology (AREA)
  • Evolutionary Computation (AREA)
  • General Engineering & Computer Science (AREA)
  • Image Analysis (AREA)
  • Length Measuring Devices By Optical Means (AREA)

Abstract

The invention discloses an angular point feature extraction method suitable for a two-dimensional laser scanner, and aims to detect angular point features such as buildings and tables widely existing in an actual environment. The method comprises the steps of firstly determining a neighborhood range of each point by combining Euclidean distance between two points and cosine distance between corresponding normal vectors, specifically, determining a rough neighborhood range by using a larger Euclidean distance threshold, and then determining a more accurate neighborhood range by using a smaller cosine distance threshold. Meanwhile, in order to better extract the angular points from the point cloud, the invention provides a novel evaluation function, and the accurate angular points can be effectively detected. The method effectively improves the accuracy of extracting the angular point features of the two-dimensional laser scanner and has better robustness.

Description

Angular point feature extraction method for two-dimensional laser scanner
Technical Field
The invention belongs to the technical field of laser positioning and navigation, and particularly relates to a corner point feature extraction method based on a two-dimensional laser scanner. In the corner feature extraction, a dual-threshold neighborhood judgment method and a scoring method designed aiming at the geometric features of corners are involved.
Background
In the simultaneous localization and mapping (SLAM) problem, considerable research has been done on the extraction of local features of images. However, compared with the mature field of vision research, 2D Lidar is also a common sensor in SLAM, but lacks a related feature extraction method.
Foreign scholars have made much research in this regard. The work that focused on 2D Lidar feature extraction earlier was Boss and Zlot, which established an orientation histogram for each point normal vector and a set of weighted projection histograms for each sub-map. Li and Olson propose rasterizing 2D point cloud data to obtain an image, detecting feature points by using Kanade-Tomasi corner points, and describing the neighborhood of the feature points by combining SIFT. The rasterization has the advantages that the existing feature extraction method in the image field can be utilized, and the disadvantage that the rasterization process introduces noise. Tipaldi and aras propose FLIRT, which is considered as the first feature extraction method designed for 2D Lidar, and the method includes three feature extraction methods, namely, normal vector estimation based on original scan data, normal vector estimation based on scan data, and curvature estimation based on scan data. Kalasi et al propose FALKO and OC. Dividing each point into a left neighborhood and a right neighborhood according to Euclidean distance threshold division by FALKO, and simultaneously giving a scoring function to score the left neighborhood and the right neighborhood respectively and accumulating the scoresThe point with the largest number is extracted as the feature point. The OC obtains a Hough frequency spectrum HS (theta) related to a series of angles theta by carrying out Hough transformation on single-frame scanning data, and obtains a main orientation theta according to the value of the Hough frequency spectrum HS (theta)dRotate all points by-thetadAnd then, a scoring function is proposed for adjacent points within the Euclidean distance threshold, and points with high scores are selected as the characteristic points.
Disclosure of Invention
The invention designs a robust 2D Lidar point cloud corner feature extraction method aiming at the condition of two intersecting edges existing in a real scene in large quantity, and improves the 2D point cloud feature extraction work in two aspects: firstly, the problem that false detection and missing detection are easily caused when the neighborhood is determined by singly depending on a Euclidean distance threshold is improved, and the left and right neighborhoods of each point are judged by adopting a strategy of combining a large Euclidean distance and a small normal vector cosine distance dual threshold; secondly, in order to suppress the influence of abnormal points and detect out the corner points with larger covered neighborhood range and more stability, a novel evaluation function which is more effective for extracting the corner point features is designed.
The technical scheme of the invention is as follows:
according to the angular point feature extraction method for the two-dimensional laser scanner, two-dimensional point cloud data are obtained through the two-dimensional laser scanner, and the two-dimensional point cloud data S are processed in the following steps:
step 1: processing each point in S one by one according to the sequence of laser scanning, firstly determining a neighborhood range by using a Euclidean distance threshold, and in the neighborhood range, obtaining a final neighborhood range by using a cosine distance threshold;
step 2: scoring the candidate points and the neighborhoods thereof, dividing the final neighborhood into a left neighborhood and a right neighborhood, respectively evaluating scores of the left neighborhood and the right neighborhood by using an evaluation function, and taking the sum of the scores of the left neighborhood and the right neighborhood as the final score of the point; and screening out the point with the maximum score in the window range as the corner point by utilizing non-maximum value inhibition.
Further, the neighborhood of each point in the two-dimensional point cloud is obtained through the following steps in sequence:
1.1) according to each scanPoint PiDistance D to scanning origin OiTo determine a range of subscripts R' for points that may be within a neighborhood;
1.2) selecting a Euclidean distance threshold ThbFor points P with subscript in RjWhen it is in contact with PiAt a distance of ThbWhen the neighbor is in the neighborhood, adding the neighbor into the neighborhood, and updating R';
1.3) selecting a cosine distance threshold ThsFor points P with subscript in RjWhen the remaining chord distance is at ThsWhen the user is out, the neighborhood is removed, and R' is updated;
1.4) after obtaining the neighborhood range, respectively carrying out straight line fitting on the updated left neighborhood and the updated right neighborhood, and selecting an angle threshold range ThdWhen the angle between the two fitted straight lines is not within the threshold range ThdWhen the content is too high, the point is abandoned and the next treatment is not carried out.
Further, the neighborhood subscript range is obtained by the following steps in sequence:
1.1.1) determining the radius of a circular window from the distance of a scanning point to a scanning origin
Figure BDA0001287470340000021
a. b are constant terms;
1.1.2) determining the possible subscript range of the neighborhood, and making alpha be the angular resolution of the laser scanner, then P is usediThere may be a point P in the neighborhood within the circular window as the center of the circlejThe subscript ranges of:
Figure BDA0001287470340000022
further, the neighborhood range in 1.3) is obtained by the following method: the normal vectors of two adjacent points in the neighborhood should satisfy
Figure BDA0001287470340000023
Wherein | · | | represents the cosine distance,
Figure BDA0001287470340000024
normal vector arc representing jth pointDegree; for the point P not satisfying the conditionjIt is culled out of the neighborhood.
Further, the criterion for fitting a straight line to the points in the left and right neighborhoods is to minimize the sum of squares of distances from the points in the left neighborhood or the right neighborhood to the fitted straight line.
Further, the score of each point in step 2 is obtained by the following method:
2.1) pairs of candidate feature points PiDiscretizing the angle of the line segment between adjacent points by PiThe circle as the center is equally divided into SiEqually divided sector blocks, and the number { I of the sector block where each point in the left and right neighborhoods is located is calculatedt, left,It +1, left,…,Ii-1, leftAnd { I }i +1, right,Ii +2, right,…,Im, right sideWherein t and m are the minimum and maximum values, I and m, respectively, of R' updated in step 1.3)t, leftDenotes the number of the sector block in which the point with subscript t in the left field, Im, right sideThe number of the sector block where the point with the subscript m in the right field is located is represented;
2.2) for each candidate feature point PiThe left and right neighborhoods are scored separately as follows:
left domain scoring of the candidate feature points
Figure BDA0001287470340000031
Left domain scoring of the candidate feature points
Figure BDA0001287470340000032
Where c is a constant term and mod (·) represents a modulo operation;
2.3) the final score for this point is the sum of the left and right neighborhoods: j (i) ═ JLeft side of(i)+JRight side(i)。
The invention has the beneficial effects that:
(1) the invention designs a method for determining the neighborhood by double thresholds, wherein the larger Euclidean distance threshold avoids missing detection caused by too small Euclidean distance, and the smaller cosine distance threshold is combined to avoid false detection caused by too large Euclidean distance.
(2) A novel evaluation function is provided, accurate angular points can be effectively detected, and better repeatability and robustness are achieved.
Drawings
FIG. 1 is a flow chart of the steps of the method for extracting angular point features of a two-dimensional laser scanner according to the present invention;
FIG. 2 is an example of a two-dimensional point cloud obtained by a two-dimensional laser scanner according to the present invention;
FIG. 3 is a result of extracting corners of the two-dimensional point cloud of FIG. 2 according to the present invention;
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
On the contrary, the invention is intended to cover alternatives, modifications, equivalents and alternatives which may be included within the spirit and scope of the invention as defined by the appended claims. Furthermore, in the following detailed description of the present invention, certain specific details are set forth in order to provide a better understanding of the present invention. It will be apparent to one skilled in the art that the present invention may be practiced without these specific details.
The method aims at detecting the corner features formed by the intersection of two edges, such as building edges, table corners and the like. The two-dimensional point cloud data S is processed by the following steps:
step 1: processing each point in S one by one according to the sequence of laser scanning, firstly determining a rough neighborhood range by using a larger Euclidean distance threshold, and in the neighborhood range, obtaining a final neighborhood range by using a smaller cosine distance threshold between corresponding normal vectors. The neighborhood of each point in the two-dimensional point cloud is obtained through the following steps in sequence:
1.1) according toEach scanning point PiDistance D to scanning origin OiTo determine a range of subscripts R' for points that may be within a neighborhood; the neighborhood subscript range is obtained by the following steps in sequence:
1.1.1) determining the radius of a circular window from the distance of a scanning point to a scanning origin
Figure BDA0001287470340000033
a. b are constant terms;
1.1.2) determining the possible subscript range of the neighborhood, and making alpha be the angular resolution of the laser scanner, then P is usediThere may be a point P in the neighborhood within the circular window as the center of the circlejThe subscript ranges of:
Figure BDA0001287470340000041
1.2) selecting a Euclidean distance threshold ThbFor points P with subscript in RjWhen it is in contact with PiAt a distance of ThbWhen it is inside, add it to the neighborhood, update the neighborhood R '→ R' + { Pj}。；
1.3) selecting a cosine distance threshold ThsFor points P with subscript in RjWhen the remaining chord distance is at ThsIn other times, it is culled out of the neighborhood, and the neighborhood R '→ R' - { P ] is updatedj}. . The neighborhood range in this step is obtained by the following method: point PiThe normal vector of (A) satisfies
Figure BDA0001287470340000042
Figure BDA0001287470340000043
Is a point Pi-1And point PiShould satisfy the normal vector between two adjacent points in the neighborhood
Figure BDA0001287470340000044
Wherein | · | | represents the cosine distance,
Figure BDA0001287470340000045
representing the normal vector radian of the jth point; for the point P not satisfying the conditionjIt is culled out of the neighborhood.
1.4) after obtaining the neighborhood range, respectively performing straight line fitting on the updated left neighborhood and the updated right neighborhood (performing straight line fitting on all points in the left field or the right field), and selecting an angle threshold range ThdWhen the angle between the two fitted straight lines is not within the threshold range ThdWhen the content is in the range, the point is abandoned and the next step of treatment is not carried out; the criterion for fitting a straight line to the points in the left and right neighborhoods is to minimize the sum of the squares of the distances from the points in the left or right neighborhood to the fitted straight line.
Step 2: scoring the candidate points and the neighborhoods thereof which meet the conditions in 1.4), dividing the final neighborhood into a left neighborhood and a right neighborhood, respectively scoring the left neighborhood and the right neighborhood by using an evaluation function, and taking the sum of the scores of the left neighborhood and the right neighborhood as the final score of the point. And screening out points with maximum scores in the window range as corner points by utilizing non-maximum value inhibition.
The score of each point in the step is obtained by the following method:
2.1) pairs of candidate feature points PiDiscretizing the angle of the line segment between adjacent points by PiThe circle as the center is equally divided into SiEqually divided sector blocks, and the number { I of the sector block where each point in the left and right neighborhoods is located is calculatedt, left,It +1, left,…,I1-1, leftAnd { I }i +1, right,Ii +2, right,…,Im, right sideWherein t and m are the minimum and maximum values, I and m, respectively, of R' updated in step 1.3)t, leftDenotes the number of the sector block in which the point with subscript t in the left field, Im, right sideThe number of the sector block where the point with the subscript m in the right field is located is represented;
2.2) for each candidate feature point PiThe left and right neighborhoods are scored separately as follows:
left domain scoring of the candidate feature points
Figure BDA0001287470340000046
Left domain scoring of the candidate feature points
Figure BDA0001287470340000047
Where c is a constant term and mod (·) represents a modulo operation;
2.3) the final score for this point is the sum of the left and right neighborhoods: j (i) ═ JLeft side of(i)+JRight side(i)。
Example 1
Referring to fig. 1, a flowchart of steps of a corner feature extraction method for a two-dimensional laser scanner according to an embodiment of the present invention is shown, and specific steps and methods are as described above and are not repeated.
Acquiring two-dimensional point cloud data through a two-dimensional laser scanner with the resolution ratio of pi/180, and processing the two-dimensional point cloud data through the following steps:
1. two-dimensional point cloud data is obtained using a two-dimensional laser scanner, as shown in fig. 2.
2. Let a equal to 0.2 and b equal to 0.07 according to the distance D from the origin of each scanning pointiDetermining the radius of the circular window as
Figure BDA0001287470340000051
It can be determined that the subscript ranges for points that may be in the neighborhood are
Figure BDA0001287470340000052
3. For a point P within the range of 2 mid-subscriptsjCalculate the sum of PiEuclidean distance of di,jSelecting Euclidean distance threshold value as Thb＝3*riWhen d isi,j<ThbUpdate the neighborhood R->R’+{Pj}。
4. Calculating the updated normal vector of each point in the neighborhood
Figure BDA0001287470340000053
Selecting cosine distance threshold value as ThsPi/3, the normal vectors of two adjacent points should satisfy:
Figure BDA0001287470340000054
for points that do not satisfy the condition, update neighborhood R '→ R' -Pj。
5. After the neighborhood range is obtained, respectively carrying out straight line fitting on the updated left neighborhood and the updated right neighborhood, and selecting an angle threshold ThdAnd pi/6, when the angle between the two fitted straight lines is not within the threshold value, discarding the point and not performing the next processing.
6. And (3) calculating the score of each candidate feature point according to the step (2), taking the window size of the NMS as 0.2m, and selecting the score maximum value in the window size as a corner point. The final corner extraction results for the scan data of fig. 2 are shown in fig. 3.
The above description is only for the purpose of illustrating the preferred embodiments of the present invention and is not to be construed as limiting the invention, and any modifications, equivalents and improvements made within the spirit and principle of the present invention are intended to be included within the scope of the present invention.

Claims (1)

1. The method for extracting the angular point features of the two-dimensional laser scanner is characterized in that two-dimensional point cloud data are obtained through the two-dimensional laser scanner, and the two-dimensional point cloud data S are processed in the following steps:
step 1: processing each point in S one by one according to the sequence of laser scanning, firstly determining a neighborhood range by using a Euclidean distance threshold, and in the neighborhood range, obtaining a final neighborhood range by using a cosine distance threshold;
step 2: scoring the candidate points and the neighborhoods thereof, dividing the final neighborhood into a left neighborhood and a right neighborhood, respectively evaluating scores of the left neighborhood and the right neighborhood by using an evaluation function, and taking the sum of the scores of the left neighborhood and the right neighborhood as the final score of the point; screening out the point with the maximum score in the window range as an angular point by utilizing non-maximum value inhibition;
the neighborhood of each point in the two-dimensional point cloud is obtained through the following steps in sequence:
1.1) according to each scanning point PiDistance D to scanning origin OiTo determine a range of subscripts R' for points that may be within a neighborhood;
1.2) selecting a Euclidean distance thresholdThbFor points P with subscript in RjWhen it is in contact with PiAt a distance of ThbWhen the neighbor is in the neighborhood, adding the neighbor into the neighborhood, and updating R';
1.3) selecting a cosine distance threshold ThsFor points P with subscript in RjWhen the remaining chord distance is at ThsWhen the user is out, the neighborhood is removed, and R' is updated;
1.4) after obtaining the neighborhood range, respectively carrying out straight line fitting on the updated left neighborhood and the updated right neighborhood, and selecting an angle threshold range ThdWhen the angle between the two fitted straight lines is not within the threshold range ThdWhen the content is in the range, the point is abandoned and the next step of treatment is not carried out;
the neighborhood subscript range is obtained by the following steps in sequence:
1.1.1) determining the radius of a circular window from the distance of a scanning point to a scanning origin
Figure FDA0002610886830000011
a. b are constant terms;
1.1.2) determining the possible subscript range of the neighborhood, and making alpha be the angular resolution of the laser scanner, then P is usediThere may be a point P in the neighborhood within the circular window as the center of the circlejThe subscript ranges of:
Figure FDA0002610886830000012
1.3) is obtained by the following method: the normal vectors of two adjacent points in the neighborhood should satisfy
Figure FDA0002610886830000013
Figure FDA0002610886830000014
Wherein |) represents the cosine distance,
Figure FDA0002610886830000015
representing the normal vector radian of the jth point; for unsatisfied conditionsPoint P ofjRejecting out the neighborhood;
the straight line fitting criterion of the points in the left and right neighborhoods is that the sum of squares of distances from the points in the left neighborhood or the right neighborhood to a fitting straight line is minimum;
the score of each point in the step 2 is obtained by the following method:
2.1) pairs of candidate feature points PiDiscretizing the angle of the line segment between adjacent points by PiThe circle as the center is equally divided into SiEqually divided sector blocks, and the number { I of the sector block where each point in the left and right neighborhoods is located is calculatedt, left,It +1, left,…,Ii-1, leftAnd { I }i +1, right,Ii +2, right,…,Im, right sideWherein t and m are the minimum and maximum values, I and m, respectively, of R' updated in step 1.3)t, leftDenotes the number of the sector block in which the point with the index t in the left neighborhood is located, Im, right sideThe number of the sector block where the point with the subscript m in the right neighborhood is located is represented;
2.2) for each candidate feature point PiThe left and right neighborhoods are scored separately as follows:
left neighborhood scoring of the candidate feature point
Figure FDA0002610886830000021
Left neighborhood scoring of the candidate feature point
Figure FDA0002610886830000022
Where c is a constant term and mod (·) represents a modulo operation;
2.3) the final score for this point is the sum of the left and right neighborhoods: j (i) ═ JLeft side of(i)+JRight side(i)。
CN201710312254.0A 2017-05-05 2017-05-05 Angular point feature extraction method for two-dimensional laser scanner Expired - Fee Related CN107292899B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201710312254.0A CN107292899B (en) 2017-05-05 2017-05-05 Angular point feature extraction method for two-dimensional laser scanner

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201710312254.0A CN107292899B (en) 2017-05-05 2017-05-05 Angular point feature extraction method for two-dimensional laser scanner

Publications (2)

Publication Number Publication Date
CN107292899A CN107292899A (en) 2017-10-24
CN107292899B true CN107292899B (en) 2020-12-29

Family

ID=60095260

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201710312254.0A Expired - Fee Related CN107292899B (en) 2017-05-05 2017-05-05 Angular point feature extraction method for two-dimensional laser scanner

Country Status (1)

Country Link
CN (1) CN107292899B (en)

Families Citing this family (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109532021B (en) * 2018-10-10 2020-08-25 浙江大学 3D printing fused deposition defect layer-by-layer detection method based on structural light property abnormal points
CN109919957B (en) * 2019-01-08 2020-11-27 同济大学 Corner detection method based on dynamic vision sensor
CN111696115B (en) * 2020-04-29 2023-04-14 山东省国土测绘院 Zebra line corner extraction method and system based on point cloud scanning line
CN112581426B (en) * 2020-11-06 2023-01-17 上海达适医疗科技有限公司 Method for identifying left leg and right leg of infrared thermal imaging image

Citations (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105547166A (en) * 2015-12-17 2016-05-04 华中科技大学 2D laser scanning based rapid detection method for heading window of heading machine

Family Cites Families (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101794384B (en) * 2010-03-12 2012-04-18 浙江大学 Shooting action identification method based on human body skeleton map extraction and grouping motion diagram inquiry
CN103020342B (en) * 2012-12-04 2015-07-15 南京大学 Method for extracting contour and corner of building from ground LiDAR data
CN103778414A (en) * 2014-01-17 2014-05-07 杭州电子科技大学 Real-time face recognition method based on deep neural network
US9436987B2 (en) * 2014-04-30 2016-09-06 Seiko Epson Corporation Geodesic distance based primitive segmentation and fitting for 3D modeling of non-rigid objects from 2D images
CN104091325A (en) * 2014-06-16 2014-10-08 哈尔滨工业大学 Image ROI positioning method and device based on dimension invariant feature transformation during automobile instrument panel visual detection
CN104463894B (en) * 2014-12-26 2020-03-24 山东理工大学 Multi-view three-dimensional laser point cloud global optimization integral registration method
CN104850857B (en) * 2015-05-15 2017-12-05 浙江大学 Across the video camera pedestrian target matching process of view-based access control model spatial saliency constraint
CN105528580B (en) * 2015-12-04 2019-02-12 杭州电子科技大学 A kind of EO-1 hyperion Curve Matching method based on absorption peak feature
CN106052674B (en) * 2016-05-20 2019-07-26 青岛克路德机器人有限公司 A kind of SLAM method and system of Indoor Robot
CN106097348B (en) * 2016-06-13 2019-03-05 大连理工大学 A kind of fusion method of three-dimensional laser point cloud and two dimensional image

Patent Citations (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105547166A (en) * 2015-12-17 2016-05-04 华中科技大学 2D laser scanning based rapid detection method for heading window of heading machine

Also Published As

Publication number Publication date
CN107292899A (en) 2017-10-24

Similar Documents

Publication Publication Date Title
CN107292899B (en) Angular point feature extraction method for two-dimensional laser scanner
CN110443836B (en) Point cloud data automatic registration method and device based on plane features
CN109035276B (en) Image edge extraction method and device and automatic driving system
CN101887586B (en) Self-adaptive angular-point detection method based on image contour sharpness
CN101295363B (en) Method and system for determining objects poses from range images
CN109767439A (en) A kind of multiple dimensioned difference of self-adapting window and the object detection method of bilateral filtering
CN109035207B (en) Density self-adaptive laser point cloud characteristic detection method
JP5385105B2 (en) Image search method and system
CN108830832A (en) A kind of plastic barrel surface defects detection algorithm based on machine vision
CN115761550A (en) Water surface target detection method based on laser radar point cloud and camera image fusion
CN108550166B (en) Spatial target image matching method
CN108460792B (en) Efficient focusing stereo matching method based on image segmentation
CN105427286B (en) A kind of infrared target detection method split based on gray scale and gradient
CN112017223A (en) Heterologous image registration method based on improved SIFT-Delaunay
CN108389260A (en) A kind of three-dimensional rebuilding method based on Kinect sensor
US20150131853A1 (en) Stereo matching system and method for generating disparity map using same
CN110580705A (en) Method for detecting building edge points based on double-domain image signal filtering
CN114004894A (en) Method for determining space relation between laser radar and binocular camera based on three calibration plates
CN106815851B (en) A kind of grid circle oil level indicator automatic reading method of view-based access control model measurement
CN106355576B (en) SAR image registration method based on MRF image segmentation algorithm
CN108615041B (en) Angular point detection method
CN105243661A (en) Corner detection method based on SUSAN operator
CN112734816A (en) Heterogeneous image registration method based on CSS-Delaunay
CN111160292B (en) Human eye detection method
CN110610503B (en) Three-dimensional information recovery method for electric knife switch based on three-dimensional matching

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
GR01 Patent grant
GR01 Patent grant
CF01 Termination of patent right due to non-payment of annual fee

Granted publication date: 20201229

CF01 Termination of patent right due to non-payment of annual fee