CN104007432A - Landmark laying method for checking plane precision of airborne laser radar - Google Patents

Abstract

The invention discloses a landmark laying method for checking the plane precision of an airborne laser radar, relates to quality evaluation of laser radar scanning data and belongs to the technical field of airborne laser radars. Due to the feature that semi-random discrete scanning is conducted on the ground surface by the laser radar, ground surface feature points are difficult to collect directly, and a mature and practical method does not exist in quality control and evaluation of airborne LIDAR data in the current stage. According to the landmark laying method for checking the plane precision of the airborne laser radar, based on point cloud data geometrical features, ground surface landmarks are specially designed, a fitting positioning method is used for extracting ground surface mark checking points so as to improve the checking precision of a point cloud plane, and therefore indexes and the method for data quality evaluation based on the statistical method is established.

Description

A kind of terrestrial reference distribution method of inspection machine bone laser radar plane precision

Technical field

The present invention relates to quality assessment and the evaluation of Laser Radar Scanning data, by designing special cross terrestrial reference, based on a cloud strength difference, accurately identify terrestrial reference, and adopt least square center line fitting algorithm to determine terrestrial reference center, thereby realize the accuracy checking of Laser Radar Scanning datum plane and quality assessment, belong to airborne laser radar technical field.

Background technology

Airborne laser radar (LiDAR-light detection and ranging) is a kind of Novel active airborne remote sensing earth observation technology.From function, airborne LiDAR is a kind of set laser, GPS (GPS) and three kinds of technology of Inertial Measurement Unit (IMU) system, the combination of these three kinds of technology, can pin-point accuracy the ground three-dimensional coordinate of Area Objects positioningly.Now, the aviation digital camera of the millions pixel being also equipped with in airborne laser radar system, for obtaining the high-definition digital image on ground, for subsequent production orthography provides basic data.Airborne laser radar technology compare with traditional photogrammetric measurement have fast, efficient, be subject to the advantage that weather effect is little, in fields such as mapping, land resources, forest, mineral products, digital city, city planning, disaster monitoring (riverbank disaster monitoring), evaluation of risk, there is important application.

Airborne LiDAR point cloud refers to that the signal of accepting by airborne laser transmitter processes that obtained expression earth's surface is three-dimensional configuration, discrete, the data point set of Density inhomogeneity, can be with the truth on higher precision reflection earth's surface.Wherein, the error of LiDAR point cloud can be divided into relative error and absolute error.Relative error refers to the relative deviation between the different air strips of LiDAR point cloud, can carry out error evaluation and elimination by the manual calibration of section and air strips adjustment, and domestic and international most of research work at present also concentrates on the elimination aspect of relative error.Absolute error refers to the deviation of LiDAR point cloud and actual geographic position, because LiDAR point cloud distributes irregular, its density is also not exclusively even, while choosing reference mark or tie point in LiDAR point cloud, cannot image height resolution image image feature point be can accurately choose like that and reference mark or tie point made, therefore for the inspection of LiDAR absolute error, can only carry out based on statistical mode.

Airborne LIDAR elevation absolute error can be utilized the ground surface elevation of LiDAR point cloud matching, with the reference mark elevation comparison of laying, carries out Quantitative evaluation.The absolute error inspection of horizontal direction, due to a characteristic for the discrete distribution of cloud, need to design ad hoc surface mark thing.At present the most authoritative horizontal accuracy evaluation method is designed by Csanyi. and Toth in the world, utilizes a kind of special circular terrestrial reference, by measuring circular terrestrial reference virtual center point and putting on a distance at cloud matching center, the plane precision of checkpoint cloud based on ground.The prerequisite of the method energy accurate evaluation horizontal accuracy is to guarantee that abundant some cloud is got to circularly to put on, and what therefore circular terrestrial reference diameter will be very is large, carries inconveniently and with high costs, and the value of popularization is not high.Domestic LiDAR engineering project is in order to control cost, and flying height is greater than 1000 meters conventionally, and some cloud density is lower, and the feasibility of the method is just poorer.Reference precision or spot diameter that at present the horizontal accuracy of LiDAR point cloud usually only provides by equipment vendors estimate, tightness is nowhere near.Along with deepening continuously of LiDAR technology application, the disappearance of LiDAR horizontal accuracy appraisement system, becomes one of further obstacle of promoting of this technology gradually.

Summary of the invention

The present invention is directed to the now methodical problem that is difficult to the airborne LiDAR horizontal accuracy of effective evaluation, provide a kind of by laying ground distinctive mark thing, make it in airborne LiDAR cloud data, as checkpoint, can be known identification and accurately locate, and field operation can accurately be measured as reference mark in measuring on the spot, thus the method checking for airborne laser radar plane precision.

Comprise the following steps:

Step 1: the design of cross terrestrial reference.Terrestrial reference material selects to be set to color and surrounding environment has bigger difference and reflectivity obviously than the material of object height peripherally.Terrestrial reference is shaped as cruciform, by the white varnish board of polylith square, is assembled, the cross live width of terrestrial reference should be a cloud spacing wide more than 1.25 times, it is more than 10 times that cross line length is its live width.The error of fitting in cloud data of this cross terrestrial reference central plane coordinate is better than 5% of equalization point spacing so, i.e. this error of fitting variance D _intend=(0.05 λ) ².

Step 2: the laying of cross terrestrial reference.Laying region is the Image of Flat Ground of continuous reflection attribute, and around without house trees, blocks; Guarantee to have 30 cross terrestrial references at least in calibration field, each cross terrestrial reference spacing is even, and at least 50% cross terrestrial reference is laid in overlapping region, air strips.

Step 3: adopt the centre coordinate of traditional field operation measurement means Accurate Measurement cross terrestrial reference, as reference mark coordinate.

Step 4: half Random Discrete scanning is carried out in earth's surface by airborne LiDAR.

Step 5: the utilize intensive analysis rapid extraction cross point cloud of putting on, the some cloud based in principle of least square matching terrestrial reference, extract the center line of two intersections, the intersecting point coordinate of getting two center lines is for as check point coordinates.

Step 6: carry out plane precision inspection and grade estimation in conjunction with the data of obtaining above;

Advantage of the present invention: the 1. some cloud in cross terrestrial reference is convenient to identification and is extracted, and has higher coupling fitting precision, so the reliability of precision evaluation has obtained guarantee.2. cross terrestrial reference has easy assembling, easily detachable advantage, and traditional circular terrestrial reference compares, and cost is lower, and reusability is better, more meets China's actual conditions.

Accompanying drawing explanation

Accompanying drawing provides based on cross terrestrial reference and carries out illustrating of airborne LiDAR point cloud plane precision inspection.

Fig. 1: the design diagram of cross terrestrial reference;

Fig. 2: the strength characteristic figure of cross terrestrial reference in cloud data;

Fig. 3: cross terrestrial reference matching center schematic diagram in cloud data.

Embodiment

Below in conjunction with accompanying drawing and exemplifying embodiment, the invention will be further described.

Step 1: the design of cross terrestrial reference

1. material is selected

Select white varnish board to be the material that cross terrestrial reference is laid, as shown in table 1: for the infrared band laser of LiDAR, material and the surrounding environment of white varnish board have bigger difference, and its reflectivity is obviously better than atural object around.

The reflection strength statistical form of table 1 cross terrestrial reference and common material near-infrared band

2. size design

For conveniently assemble and disassemble, carry and improve reusability, cross landmark designing is the combination of a polylith square sheet material.Its length and width size, as accompanying drawing 1, is laid terrestrial reference sample on the spot as accompanying drawing 2.

Cross terrestrial reference central point as checkpoint in a cloud can be by some cloud two the intersecting straight lines automatic acquisitions of matching (approximating method refers to step 5) in cross terrestrial reference, therefore this two aspects factor restriction of the size of the main receptor site cloud of the fitting precision of checkpoint equalization point spacing and cross terrestrial reference.Therefore, ground dimensioning needs to fly the some cloud average headway design of measuring according to default boat, thereby reaches the object of exact matching.According to the cloud data density p in line of flight design objective, can convert and obtain a cloud average headway the cross live width that this programme arranges cross terrestrial reference is 1.25 λ, and the cross line length of cross terrestrial reference is 12.5 λ.[CH/T8024-2011 airborne laser radar data obtains technical manual] (in 3: term and definition: 3.6 cloud density)

According to this place dimensioning, carried out the experimental verification of fitting precision, its method summary is: 1. in experiment cloud data, a large amount of simulation terrestrial references are set, target profile scope and centre coordinate are known in analog; 2. the some cloud extracting within the scope of simulation terrestrial reference carries out matching location (approximating method refers to step 5); 3. the error of fitting of digital simulation coordinate and Simulation Center coordinate.

Empirical tests obtains, and the error of fitting in cloud data of cross terrestrial reference central plane coordinate is better than 5% of equalization point spacing, i.e. this error of fitting variance D _intend=(0.05 λ) ².Above-mentioned error of fitting cloud data same place measuring error limit poor 10% with interior (in airborne LiDAR data acquisition technology standard, stipulate: cloud data same place measuring error be no more than equalization point spacing 2/3), meet the quality assessment condition of cloud data horizontal accuracy completely.

Step 2: the laying of cross terrestrial reference

Due to the coupling location of surface mark for the reference mark between many air strips or tie point, the position of terrestrial reference should be determined according to the line of flight and distribution method.Its installation position should be laid in overlapping region, air strips as much as possible; This region should be near-infrared band reflection strength continuously and Image of Flat Ground, as: soil, pitch, thereby contrast, the identification of the reflection strength on increase terrestrial reference and ground; Near terrestrial reference in 20 meters without the blocking of house trees, its general layout should be convenient to traditional field operation and measure, is easy to interpretation.The spatial relationship of terrestrial reference and some cloud as shown in Figure 3.

Surface mark preferably can be laid on local existing leveling point, can directly by leveling point coordinate, find out marker position if so, can improve data accuracy like this.

Step 3: the centre coordinate of Accurate Measurement cross terrestrial reference

After laying by the distribution method in step 2, adopt traditional field operation metering system (as: GPS RTK measurement) to carry out accurate field survey to the central point of cross terrestrial reference, the measurement of coordinates value at this terrestrial reference center can be used as reference mark coordinate (x _control, y _control).As accompanying drawing 3

Step 4: airborne LiDAR carries out half Random Discrete scanning to earth's surface

Detailed process and requirement that airborne LiDAR point cloud obtains, please refer to " People's Republic of China's Surveying and Mapping Industry standard-airborne LiDAR data acquisition technology regulation " that State Bureau of Surveying and Mapping issued in 2011.Repeat no more herein.It is to be noted: the order of step 3 and step 4 can arbitrarily exchange.

Step 5: in cloud data, terrestrial reference extracts and accurately locates

According to a cloud reflection strength feature extraction the cloud data put on

Due to the special substance of cross terrestrial reference, the reflection strength of the laser point cloud (near-infrared band) that scans cross and put on is much larger than the reflection strength of common ground, as table 1.Therefore, according to the reflection strength characteristic of cross terrestrial reference, extract reflection strength value and be greater than the 200 (cloud datas that obtained by LiDAR equipment, its reflection strength value is between 0～255, the higher >40% of reflectivity of white varnish board, corresponding reflection strength count value >200) high intensity points cloud is as the some cloud that accurately scans cross and put on.

2. by two central line l of the some cloud matching cross terrestrial reference in cross terrestrial reference ₁, l ₂, obtain two central line l ₁, l ₂the planimetric coordinates of point of crossing.

Straight line l ₁, l ₂be two central line for the treatment of matching, establish l ₁straight-line equation be: y=Ax+B, according to the principle of least square, fitting a straight line with put on square minimum of contiguous high intensity points cloud deviation, numeral expression formula is:

$\mathrm{\φ}(A,B)=\underset{i=1}{\overset{n}{\mathrm{\Σ}}}{(y_i-{\mathrm{Ax}}_i-B)}^2---\left(1\right)$

Determined the value of fitting a straight line two constant A, B, also just determined straight-line equation, therefore using A, B in formula (3) as component, also differentiated respectively, making it is 0:

$\begin{array}{c}\frac{\∂\mathrm{\φ}}{\∂A}=2\underset{i=1}{\overset{n}{\mathrm{\Σ}}}(y_i-{\mathrm{Ax}}_i-B)x_i\\ \frac{\∂\mathrm{\φ}}{\∂B}=2\underset{i=1}{\overset{n}{\mathrm{\Σ}}}(y_i-{\mathrm{Ax}}_i-B)\end{array}---\left(2\right)$

Through arranging:

$\left\{\begin{array}{c}\underset{i=1}{\overset{n}{\mathrm{\Σ}}}{x}_i{y}_i-A\underset{i=1}{\overset{n}{\mathrm{\Σ}}}{x}_i^2-B\underset{i=1}{\overset{n}{\mathrm{\Σ}}}{x}_i=0\\ \underset{i=1}{\overset{n}{\mathrm{\Σ}}}{y}_i-A\underset{i=1}{\overset{n}{\mathrm{\Σ}}}{x}_i-\mathrm{nB}=0\end{array}\right.---\left(3\right)$

X in formula _i, y _ifor a cloud coordinate, n is the number of Points on Straight Line cloud, and substitution above formula calculates can obtain fitting a straight line y=Ax+B, A, and B is straight-line equation coefficient.

In like manner, can calculate and obtain l ₂straight-line equation: y=Cx+D.Finally solve following linear equation in two unknowns group, just can obtain cross and be marked on the matching central plane coordinate (x in a cloud _intend, y _intend), that is:

Calculate

Step 6: plane precision inspection and grade estimation

When obtaining cross by said method, be marked on after the coordinate on a cloud, can be used for follow-up plane precision inspection and grade estimation.

First utilize the field operation measuring center point (x of N the cross terrestrial reference of having laid _{i control}, y _{i control}) (seeing step 3) as reference mark, is marked on the matching central plane coordinate (x in a cloud cross _{i intends}, y _{i intends}) (seeing step 5) as corresponding checkpoint, calculates physical control point and the variance D that puts checkpoint on cloud overall measurement error in the plane _always, computing formula is as follows:

Due to the error of fitting of checkpoint on the field operation measuring error that comprises reference mark in this error, the scanning survey error of putting cloud self and some cloud, known according to law of propagation of errors in adjustment theory:

D _always=D _control+ D _point+ D _intend(7)

In formula: D _alwaysfor the variance of reference mark with the plane overall measurement error of some cloud matching checkpoint, by formula (6), calculate and obtain; D _controlfor the variance of the field operation measuring error at reference mark, according to actual field operation measuring accuracy, obtain D _intendfor this method cross be marked on the variance of the error of fitting of the matching central plane coordinate in a cloud, by step 1, provided; D _pointfor the variance of the scanning survey error of a cloud self, it is an important indicator for cloud quality assessment.

And in actual adjustment, the measured value precision at field operation reference mark is very high, is conventionally considered as true value, i.e. the variance D of field operation measuring error _control=0, formula (7) can be reduced to:

D _always=D _point+ D _intend(8) therefore put the variance D of the scanning survey error of cloud self _pointbe:

D _point=D _always-D _intend(9)

Above formula is inspection, the grade estimation result based on Ground Control Information.Work as D _always> 5D _intendtime, the cross terrestrial reference error of fitting of this method can be ignored, reference mark is larger with the global error of putting cloud matching checkpoint, directly reflected that real some cloud measuring accuracy is lower, illustrate that Gai Ce district cloud data need to further utilize ground control point and matching to check that point-to-point cloud carries out simultaneous adjustment correction and (utilizes dominating pair of vertices survey district to carry out simultaneous adjustment and be corrected as the basic step in Point Cloud Processing, in this programme, repeat no more, this programme is only with providing reference mark and the correspondence proving point thereof resolving for simultaneous adjustment), thereby further improve the measuring accuracy of some cloud, work as D _always≤ 5D _intendtime, illustrate that some cloud measuring accuracy are higher, without the correction of ground control point.

Claims (4)

1. a terrestrial reference distribution method for inspection machine bone laser radar plane precision, is characterized in that, comprises the following steps:

Step 1, select white varnish board as cross terrestrial reference;

According to cloud data density p, convert and obtain a cloud average headway the cross live width that cross terrestrial reference is set is 1.25 λ, and the cross line length of cross terrestrial reference is 12.5 λ; The error of fitting in cloud data of this cross terrestrial reference central plane coordinate is better than 5% of equalization point spacing so, i.e. this error of fitting variance D _intend=(0.05 λ) ²;

Be marked on the continuous and smooth earth's surface of near-infrared band reflection strength in overlapping region, air strips step 2, layout cross;

The field operation measuring center coordinate of step 3, measurement cross coordinate is as reference mark coordinate x _control, y _control;

Step 4, utilize airborne LiDAR to scan and obtain LiDAR cloud data earth's surface;

Step 5, according to a cloud reflection strength feature extraction cross the cloud data put on;

Two central line l of step 6, the Points cloud Fitting cross coordinate line by cross terrestrial reference ₁, l ₂, obtain two central line l ₁, l ₂the planimetric coordinates x of point of crossing _intend, y _intend;

The field operation measuring center point x of N the cross terrestrial reference that step 7, utilization have been laid _{i control}, y _{i control}as reference mark, i ∈ 1～N is marked on the matching central plane coordinate x in a cloud cross _{i intends}, y _{i intends}as corresponding checkpoint, the overall measurement error variance D in the plane of the checkpoint on calculating physical control point and some cloud _always, computing formula is as follows:

Step 8, work as D _always> 5D _intendtime, with the reference mark on ground and the point-to-point cloud of the inspection of matching, carry out simultaneous adjustment correction; Work as D _always≤ 5D _intendtime, without the correction of ground control point.

2. the terrestrial reference distribution method of a kind of inspection machine bone laser radar plane precision according to claim 1, is characterized in that, in described step 2, cross terrestrial reference is laid on existing leveling point.

3. the terrestrial reference distribution method of a kind of inspection machine bone laser radar plane precision according to claim 1, it is characterized in that, described step 5 comprises the following steps, according to the reflection strength characteristic of cross terrestrial reference, extract reflection strength value in cloud data be greater than 200 some cloud as cross the some cloud put on.

4. the terrestrial reference distribution method of a kind of inspection machine bone laser radar plane precision according to claim 1, is characterized in that, described step 6 comprises the following steps:

Set straight line l ₁/ l ₂straight-line equation be: y=Ax+B, according to following formula, try to achieve parameter A and B

$\left\{\begin{array}{c}\underset{i=1}{\overset{n}{\mathrm{\Σ}}}{x}_i{y}_i-A\underset{i=1}{\overset{n}{\mathrm{\Σ}}}{x}_i^2-B\underset{i=1}{\overset{n}{\mathrm{\Σ}}}{x}_i=0\\ \underset{i=1}{\overset{n}{\mathrm{\Σ}}}{y}_i-A\underset{i=1}{\overset{n}{\mathrm{\Σ}}}{x}_i-\mathrm{nB}=0\end{array}\right.$

X in formula _i, y _ifor straight line l ₁/ l ₂upper some cloud coordinate, n is straight line l ₁/ l ₂the number of upper some cloud, calculates substitution above formula can obtain fitting a straight line y=Ax+B, A, and B is straight-line equation coefficient.