CN109872281A - A kind of progressive encryption triangulation network LiDAR point cloud filtering method under shape information auxiliary - Google Patents
A kind of progressive encryption triangulation network LiDAR point cloud filtering method under shape information auxiliary Download PDFInfo
- Publication number
- CN109872281A CN109872281A CN201811638223.5A CN201811638223A CN109872281A CN 109872281 A CN109872281 A CN 109872281A CN 201811638223 A CN201811638223 A CN 201811638223A CN 109872281 A CN109872281 A CN 109872281A
- Authority
- CN
- China
- Prior art keywords
- point
- point cloud
- waveform
- laser radar
- triangulation network
- 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
Landscapes
- Optical Radar Systems And Details Thereof (AREA)
Abstract
The invention discloses the progressive encryption triangulation network LiDAR point cloud filtering methods under a kind of shape information auxiliary, it carries out Gauss Decomposition to Full wave shape laser radar data first with LM algorithm, extract waveform correlation parameter, and the filtering of the progressive encryption triangulation network is carried out to laser radar point cloud based on waveform correlation parameter and space geometry information, obtain point cloud filter result.Full wave shape laser radar data Gauss Decomposition algorithm in the present invention based on LM algorithm has global convergence, therefore the accuracy of waveform decomposition result and high reliablity, merging point cloud shape information of the present invention and space geometry information simultaneously, it improves seed point and chooses the reliability judged with ground point, therefore a precision for cloud filtering can be improved in the present invention.
Description
Technical field
The present invention relates to remote sensing technology fields, and in particular to the progressive encryption triangulation network under a kind of shape information auxiliary
LiDAR point cloud filtering method.
Background technique
Laser radar (LiDAR) ranging technology laser radar is a kind of active remote sensing technology, can directly acquire ground
Three-dimensional information, and the data acquisition of round-the-clock round-the-clock may be implemented.Airborne laser radar, which refers to by aircraft, carries laser
Scanner.Compared with digital photogrammetry, airborne laser radar has the advantages that directly to record earth's surface elevation.From digital elevation mould
Type mapping, urban transportation research, arrive conservancy ecological engineering etc., airborne laser radar is gradually popular in the application of various environment
And provide important data.
Become point cloud data using the three-dimensional data that airborne laser radar scanning ground obtains, wherein including truly
The information such as face, vegetation and artificial structure, and have the characteristics that high-precision, high density, discrete irregular.Laser radar point cloud number
Contain geography information abundant in, to obtain these information and need to carry out data processing using effective method.In fact, filter
Wave extracts ground point from point cloud data, is the committed step of Point Cloud Processing.Ground point and non-ground points are distinguished
Later, digital complex demodulation, and then subsequent progress building extraction, ground mapping and three-dimensional reconstruction are formed using ground point
Etc. technologies research.
In recent years, a kind of new Full wave shape laser radar technique receives significant attention, it is conventional laser Radar Technology
Primary innovation.Transmitting signal and echo-signal are obtained all-wave graphic data with the interval sampling of very little by it, than using discrete echo
The terrestrial object information of acquisition is more abundant, available more detailed vertical using Full wave shape laser radar especially for vegetation
Straight directional information.By analyzing shape information, discrete point cloud data not only can be obtained, additionally it is possible to obtain Target scalar
Surface texture and physical attribute information.Full wave shape laser radar technique provides more flexible mode of operation for terminal user,
User can be analyzed and be located accordingly to Wave data according to the application demand (such as mapping, forestry, City Modeling) of oneself
Reason.It therefore, will be with important theory significance and practical application with the data processing of shape information auxiliary laser radar points cloud
Value, still, such research is not yet unfolded.
Summary of the invention
It is an object of the invention to overcome above-mentioned the deficiencies in the prior art, a kind of lower progressive of shape information auxiliary is provided and is added
Close triangulation network LiDAR point cloud filtering method, merging point cloud shape information and space geometry information, improve seed point choose with
The reliability of ground point judgement may finally improve a precision for cloud filtering.
The present invention specifically uses following technical scheme:
Progressive encryption triangulation network LiDAR point cloud filtering method under a kind of shape information auxiliary comprising following steps:
S1, Gauss Decomposition is carried out to Full wave shape laser radar data based on LM (Levenberg-Marquardt) algorithm, and
Extract waveform correlation parameter；
S2, the waveform parameter information and space geometry information extracted in conjunction with step S1 carry out laser radar point cloud progressive
Triangulation network filtering is encrypted, point cloud filter result is obtained.
Further, step S1 is specifically included:
S11, the inflection point for seeking Full wave shape laser radar waveform data and initial crest location, by the two neighboring inflection point of odd even
Determine Gauss Decomposition initial parameter value, the Gauss Decomposition initial parameter includes Gaussian Profile center, half-breadth, amplitude；
S12, the waveform that half-breadth is less than transmitted waveform, peak value is less than Gaussian minimum amplitude is rejected, it is small to adjacent peaks interval
It is merged in the Gauss wave crest of fire pulse width；
S13, the obtained waveform of step S12 is fitted based on LM method, obtains Gauss Decomposition Optimal Parameters value；
S14, it is based on waveform decomposition result, extracts waveform parameter.
Further, the waveform parameter that step S14 is extracted includes echo total degree, echo serial number, waveform widths, wave crest
Position, amplitude of wave form.
Further, the step S2 is specifically included:
S21, laser radar point cloud rough error is rejected using statistics with histogram method；
S22, joint waveform parameter information and space geometry information choose reliable initial ground seed point；
S23, initial sparse irregular triangle network is constructed based on the obtained ground seed point of step S22；
The distance and angle parameter of S24, calculating laser point to place triangular apex, in conjunction with the wave of the step S1 point extracted
Shape parameter information constructs judgment criterion；
S25, when laser point meets judgment criterion, by the point be added ground point set, recycle new ground point set to not
Regular triangular net realizes progressive encryption, always iteration, and until the quantity of ground point set keeps stabilization, final acquisition point cloud filtering is tied
Fruit.
Further, step S22 specifically:
S221, ground point are only possible to be the single and last echo in laser pulse, using echo serial number to step S21 at
Laser radar point cloud after reason carries out preliminary screening, and realization part vegetation filters out with building marginal point；
S222, pulse semi-width threshold value ω is calculatedT,In formula, Ω is system fire pulse width；
If S223, pulse semi-width are less than ωT, then the point belongs to target point set (such as flat ground and flat of flat surfaces
Whole building roof), to this part point set, the higher building part of elevation is filtered out using morphology opening operation, thus
Obtain the ground seed point set of the part；
If S224, pulse semi-width are greater than ωT, then the point belong to uneven surface target (such as vegetation, building edge and tiltedly
Hillside fields shape), to this part point set, point cloud normal is calculated, when the normal direction within the scope of the vertex neighborhood is consistent, then the point belongs to
A possibility that clinoform, is big, if normal variation is irregular, illustrates that the point belongs to vegetation or Architectural fringes part.
After adopting the above technical scheme, compared with the background technology, the present invention, having the advantages that
The present invention carries out Gauss Decomposition to Full wave shape laser radar data using LM algorithm, extracts waveform correlation parameter and sky
Between geological information, and based on these information to laser radar point cloud carry out the filtering of the progressive encryption triangulation network, obtain point cloud filtering knot
Fruit, the Full wave shape laser radar data Gauss Decomposition algorithm based on LM algorithm have global convergence, therefore waveform decomposition result
Accuracy and high reliablity, while merging point cloud shape information of the present invention and space geometry information, improve seed point choose with
The reliability of ground point judgement, therefore a precision for cloud filtering can be improved in the present invention.
Detailed description of the invention
Fig. 1 is overall flow schematic diagram of the present invention；
Fig. 2 is that the present invention is based on the Full wave shape laser radar data Gauss Decomposition flow diagrams of LM algorithm；
Fig. 3 is the triangulation network progressive encryption point cloud filtering algorithm flow diagram under shape information of the present invention auxiliary.
Specific embodiment
In order to make the objectives, technical solutions, and advantages of the present invention clearer, with reference to the accompanying drawings and embodiments, right
The present invention is further elaborated.It should be appreciated that the specific embodiments described herein are merely illustrative of the present invention, and
It is not used in the restriction present invention.
Embodiment
Refering to what is shown in Fig. 1, the invention discloses the progressive encryption triangulation network LiDAR point cloud filters under a kind of shape information auxiliary
Wave method comprising following steps:
S1, Gauss Decomposition is carried out to Full wave shape laser radar data based on LM algorithm, and extracts waveform correlation parameter:
Refering to what is shown in Fig. 2, step S1 is specifically included:
S11, the inflection point for seeking Full wave shape laser radar data and initial crest location are determined by the two neighboring inflection point of odd even
Gauss Decomposition initial parameter value；
According to the waveform principle of waveform laser radar, it is known that return laser beam waveform is represented by the sum of multiple Gaussian functions.
Generally for a standard gaussian distribution function, it is 0 by solving its single order and second-order partial differential coefficient and enabling it, waveform number can be found out
According to inflection point and initial crest location.After seeking knee of curve by above method, it can be solved by two adjacent inflection points of odd even
The parameters such as center, half-breadth, the amplitude of Gaussian Profile are calculated, so that it is determined that the Gaussian representation formula of the wavy curve.
S12, the waveform that half-breadth is less than transmitted waveform, peak value is less than Gaussian minimum amplitude is rejected, it is small to adjacent peaks interval
It is merged in the Gauss wave crest of fire pulse width；
It is largely still that the random fluctuation of ambient noise generates, not in the Gaussian waveform obtained by step S11
It is useful signal, therefore rejects the waveform that half-breadth is less than Gaussian minimum amplitude less than transmitted waveform, peak value, and to adjacent peaks
The Gauss wave crest that interval is less than fire pulse width merges, and then, is ranked up to these Gauss wave crests by area, will most
The wave merging of small area is equal to 6 in away from the maximum wave crest of nearest, area, until the number of Gauss wave crest is less than
It is a.
S13, based on LM algorithm, to step S12, treated that waveform is fitted, and obtains Optimal Parameters value；
S14, it is based on waveform decomposition result, extracts waveform parameter.
The waveform parameter that step S14 is extracted includes echo total degree, echo serial number, waveform widths, crest location, waveform vibration
Width.
After the completion of step S1, subsequent step S2 are as follows: in conjunction with shape information and space geometry information that step S1 is extracted, to sharp
Optical radar point cloud carries out the filtering of the progressive encryption triangulation network, obtains point cloud filter result:
Refering to what is shown in Fig. 3, step S2 is specifically included:
It is aobvious lower than ground point that the laser radar point cloud data that S21, waveform decompose concentrates some to point out, these points are thick
Almost, laser radar point cloud rough error is rejected using statistics with histogram method；
S22, joint waveform parameter information and space geometry information choose reliable initial ground seed point；
Wherein, the specific implementation process of step S22 are as follows:
S221, ground point are only possible to be the single and last echo in laser pulse, using echo serial number to step S21 at
Point cloud after reason carries out preliminary screening, and realization part vegetation filters out with building marginal point；
S222, pulse semi-width have reacted the surfacing degree of irradiation target, propose and set pulse semi-width threshold value for point cloud minute
Two parts are segmented into, pulse semi-width threshold value ω is calculatedT,In formula, Ω is system fire pulse width；
If S223, pulse semi-width are less than ωT, then the point belongs to target point set (such as flat ground and flat of flat surfaces
Whole building roof), to this part point set, the higher building part of elevation is filtered out using morphology opening operation, thus
Obtain the ground seed point set of the part；
If S224, pulse semi-width are greater than ωT, then the point belong to uneven surface target (such as vegetation, building edge and tiltedly
Hillside fields shape), to this part point set, point cloud normal is calculated, when the normal direction within the scope of the vertex neighborhood is consistent, then the point belongs to
A possibility that clinoform, is big, if normal variation is irregular, illustrates that the point belongs to vegetation or Architectural fringes part.
S23, the sparse irregular triangle network based on the building of ground seed point initially；
The distance and angle parameter of S24, calculating laser point to place triangular apex, in conjunction with the wave of the step S1 point extracted
Shape parameter information constructs judgment criterion；
S25, when laser point meets judgment criterion, by the point be added ground point set, recycle new ground point set to not
Regular triangular net realizes progressive encryption, always iteration, and until the quantity of ground point set keeps stabilization, final acquisition point cloud filtering is tied
Fruit.
The foregoing is only a preferred embodiment of the present invention, but scope of protection of the present invention is not limited thereto,
In the technical scope disclosed by the present invention, any changes or substitutions that can be easily thought of by anyone skilled in the art,
It should be covered by the protection scope of the present invention.Therefore, protection scope of the present invention should be with scope of protection of the claims
Subject to.
Claims (5)
1. the progressive encryption triangulation network LiDAR point cloud filtering method under a kind of shape information auxiliary, it is characterised in that: including following
Step:
S1, Gauss Decomposition is carried out to Full wave shape laser radar data based on LM algorithm, and extracts waveform correlation parameter；
S2, the waveform parameter information and space geometry information extracted in conjunction with step S1 carry out progressive encryption to laser radar point cloud
Triangulation network filtering obtains point cloud filter result.
2. the progressive encryption triangulation network LiDAR point cloud filtering method under a kind of shape information auxiliary according to claim 1,
It is characterized by: step S1 is specifically included:
S11, the inflection point for seeking Full wave shape laser radar waveform data and initial crest location are determined by the two neighboring inflection point of odd even
Gauss Decomposition initial parameter value, the Gauss Decomposition initial parameter include Gaussian Profile center, half-breadth, amplitude；
S12, the waveform that half-breadth is less than transmitted waveform, peak value is less than Gaussian minimum amplitude is rejected, hair is less than to adjacent peaks interval
The Gauss wave crest for penetrating pulse width merges；
S13, the obtained waveform of step S12 is fitted based on LM method, obtains Gauss Decomposition Optimal Parameters value；
S14, it is based on waveform decomposition result, extracts waveform parameter.
3. the progressive encryption triangulation network LiDAR point cloud filtering method under a kind of shape information auxiliary according to claim 2,
It is characterized by: the waveform parameter that step S14 is extracted includes echo total degree, echo serial number, waveform widths, crest location, wave
Shape amplitude.
4. the progressive encryption triangulation network LiDAR point cloud filtering method under a kind of shape information auxiliary according to claim 3,
It is characterized by: step S2 is specifically included:
S21, laser radar point cloud rough error is rejected using statistics with histogram method；
S22, joint waveform parameter information and space geometry information choose reliable initial ground seed point；
S23, initial sparse irregular triangle network is constructed based on the obtained ground seed point of step S22；
S24, laser point is calculated to the distance and angle parameter of place triangular apex, join in conjunction with the waveform of the step S1 point extracted
Number information, constructs judgment criterion；
S25, when laser point meets judgment criterion, by the point be added ground point set, recycle new ground point set to irregular
The triangulation network realizes progressive encryption, always iteration, until the quantity of ground point set keeps stabilization, final acquisition point cloud filter result.
5. the progressive encryption triangulation network LiDAR point cloud filtering method under a kind of shape information auxiliary according to claim 4,
It is characterized by: step S22 specifically:
S221, ground point are only possible to be the single and last echo in laser pulse, using echo serial number to laser radar point cloud
Preliminary screening is carried out, realization part vegetation filters out with building marginal point；
S222, pulse semi-width threshold value ω is calculatedT,In formula, Ω is system fire pulse width；
If S223, pulse semi-width are less than ωT, then the point belongs to the target point set of flat surfaces, to this part point set, utilizes form
It learns opening operation to filter out the higher building part of elevation, to obtain the ground seed point set of the part；
If S224, pulse semi-width are greater than ωT, then the point belongs to uneven surface target, to this part point set, point cloud normal is calculated,
When the normal direction within the scope of the vertex neighborhood is consistent, then a possibility that point belongs to clinoform, is big, if normal variation is not advised
Then, illustrate that the point belongs to vegetation or Architectural fringes part.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811638223.5A CN109872281A (en) | 2018-12-29 | 2018-12-29 | A kind of progressive encryption triangulation network LiDAR point cloud filtering method under shape information auxiliary |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811638223.5A CN109872281A (en) | 2018-12-29 | 2018-12-29 | A kind of progressive encryption triangulation network LiDAR point cloud filtering method under shape information auxiliary |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109872281A true CN109872281A (en) | 2019-06-11 |
Family
ID=66917434
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811638223.5A Pending CN109872281A (en) | 2018-12-29 | 2018-12-29 | A kind of progressive encryption triangulation network LiDAR point cloud filtering method under shape information auxiliary |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109872281A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110532975A (en) * | 2019-09-03 | 2019-12-03 | 武汉天擎空间信息技术有限公司 | TIN filter correction method based on Full wave shape airborne laser radar data |
| CN111080536A (en) * | 2019-11-13 | 2020-04-28 | 武汉华中天经通视科技有限公司 | Self-adaptive filtering method for airborne laser radar point cloud |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102506824A (en) * | 2011-10-14 | 2012-06-20 | 航天恒星科技有限公司 | Method for generating digital orthophoto map (DOM) by urban low altitude unmanned aerial vehicle |
| CN102722640A (en) * | 2012-05-24 | 2012-10-10 | 中交第二公路勘察设计研究院有限公司 | Airborne laser waveform data decomposition algorithm considering adjacent waveform information |
| CN103217679A (en) * | 2013-03-22 | 2013-07-24 | 北京航空航天大学 | Full-waveform laser radar echo data gaussian decomposition method based on genetic algorithm |
-
2018
- 2018-12-29 CN CN201811638223.5A patent/CN109872281A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102506824A (en) * | 2011-10-14 | 2012-06-20 | 航天恒星科技有限公司 | Method for generating digital orthophoto map (DOM) by urban low altitude unmanned aerial vehicle |
| CN102722640A (en) * | 2012-05-24 | 2012-10-10 | 中交第二公路勘察设计研究院有限公司 | Airborne laser waveform data decomposition algorithm considering adjacent waveform information |
| CN103217679A (en) * | 2013-03-22 | 2013-07-24 | 北京航空航天大学 | Full-waveform laser radar echo data gaussian decomposition method based on genetic algorithm |
Non-Patent Citations (7)
| Title |
|---|
| PETER AXELSSON: "DEM GENERATION FROM LASER SCANNER DATA USING ADAPTIVE TIN MODELS", 《INTERNATIONAL ARCHIVES OF PHOTOGRAMMETRY AND REMOTE SENSING》 * |
| SHENG NIE 等: "A revised progressive TIN densification for filtering airborne LiDAR data", 《MEASUREMENT》 * |
| 周国清 等: "《面阵激光雷达成像原理、技术及应用》", 31 January 2018, 武汉大学出版社 * |
| 李峰 等: "《机载LiDAR系统原理与点云处理方法》", 31 August 2017, 煤炭工业出版社 * |
| 李鹏程 等: "利用波形信息的加权曲面拟合LiDAR点云滤波", 《武汉大学学报 信息科学版》 * |
| 王滨辉 等: "全波形激光雷达的波形优化分解算法", 《测绘学报》 * |
| 隋立春 等: "基于渐进三角网的机载LiDAR点云数据滤波", 《武汉大学 信息科学版》 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110532975A (en) * | 2019-09-03 | 2019-12-03 | 武汉天擎空间信息技术有限公司 | TIN filter correction method based on Full wave shape airborne laser radar data |
| CN111080536A (en) * | 2019-11-13 | 2020-04-28 | 武汉华中天经通视科技有限公司 | Self-adaptive filtering method for airborne laser radar point cloud |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN103017739B (en) | Manufacturing method of true digital ortho map (TDOM) based on light detection and ranging (LiDAR) point cloud and aerial image | |
| CN104049245B (en) | Urban building change detection method based on LiDAR point cloud spatial difference analysis | |
| Zhang et al. | Filtering airborne LiDAR data by embedding smoothness-constrained segmentation in progressive TIN densification | |
| Houser et al. | Sensitivity of post‐hurricane beach and dune recovery to event frequency | |
| US8754802B2 (en) | Determining root correspondence between previously and newly detected objects | |
| CN110262287A (en) | Canopy height on-line water flushing method for the highly automated control of the harvest machinery ceding of Taiwan | |
| CN101196562B (en) | Method for laser radar waveshape data decomposition based on improved EM algorithm | |
| Yang et al. | Filtering of airborne LiDAR bathymetry based on bidirectional cloth simulation | |
| CN107870355B (en) | A kind of kirchhoff type beam deviation method under MODEL OVER COMPLEX TOPOGRAPHY | |
| Liu et al. | LiDAR data reduction for efficient and high quality DEM generation | |
| CN109872281A (en) | A kind of progressive encryption triangulation network LiDAR point cloud filtering method under shape information auxiliary | |
| CN106408581A (en) | Rapid three-dimensional point cloud linear extraction method | |
| CN115249330B (en) | Method and system for evaluating vegetation connectivity | |
| CN109709527A (en) | The Gauss wave crest method of Gauss Decomposition in a kind of Full wave shape laser-measured height echo signal | |
| Conery et al. | Terrestrial LiDAR monitoring of coastal foredune evolution in managed and unmanaged systems | |
| Kada et al. | Sub-surface growing and boundary generalization for 3D building reconstruction | |
| Li et al. | New methodologies for precise building boundary extraction from LiDAR data and high resolution image | |
| CN101464148B (en) | Three-dimensional image detecting, compiling and reconstructing system | |
| Li et al. | Bathymetric LiDAR and multibeam echo-sounding data registration methodology employing a point cloud model | |
| CN106980139B (en) | Seismic ray tracing method based on direction vector | |
| Papasaika et al. | Fusion of digital elevation models from various data sources | |
| Xu et al. | Methods for the construction of DEMs of artificial slopes considering morphological features and semantic information | |
| Xing et al. | Surface Fitting Filtering of LiDAR Point Cloud with Waveform Information | |
| Wang et al. | High-Quality Contour Line Generation from LiDAR Point Clouds for the Area of Forests | |
| CN104599278B (en) | Shallow sea sand wave information extraction method based on remote sensing image |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190611 |
|
| RJ01 | Rejection of invention patent application after publication |