CN102645209A - Joint positioning method for spatial points by means of onboard LiDAR point cloud and high resolution images - Google Patents

Abstract

A joint positioning method for spatial points by means of onboard LiDAR point cloud and high resolution images includes the steps: denoising original LiDAR data; generating a high-precision DEM (digital elevation model); manually supplementing elements of exterior orientation of images acquired by high-precision POS (point-of-sale) data; calculating object-side coordinates (X, Y and Z) of identical points by means of multi-image forward intersection after calculating initial values by means of two-image forward intersection by the aid of two optional points among the selected identical points; and obtaining precise object-side coordinates (X, Y and Z') corresponding to the identical points by the aid of the object-side coordinates of the identical points solved by means of previous multi-image forward intersection. By means of multi-image forward intersection realized on the premise of elevation constraint of the LiDAR point cloud, object-side coordinates of acquired image points are quite high in plane precision and elevation precision.

Description

Airborne LiDAR point cloud and high resolution image carry out the combined positioning-method of spatial point

Technical field

Patent of the present invention relates to airborne LiDAR technical field of data processing, relates in particular to the pinpoint method of associating that a kind of integrated LiDAR cloud data and high resolution image carry out spatial point.

Background technology

Airborne LiDAR is a kind of novel active airborne remote sensing earth observation technology; Make flight just can synchronous acquisition arrive laser point cloud and high resolving power aviation image through carrying high-resolution digital camera; Because the POS system that airborne LiDAR carries has inertial navigation and GPS positioning function; Therefore can directly obtain three-dimensional point cloud information and the high resolving power boat sheet and the elements of exterior orientation thereof of high spatial resolution; Can carry out high-precision direct geo-location in theory; If but the very long and loaded down with trivial details sky three of industry is not handled under the situation of overall adjustment in carrying out; The realistic accuracy of the elements of exterior orientation of the high resolution image that it obtains must receive the combined influence of accidental error and systematic error, makes traditional biplate forward intersection be difficult to reach the target of the direct geo-location of high precision, can't satisfy the demand of current remote sensing mapping.

Summary of the invention

To plane precision and the vertical accuracy not high problem of tradition based on the direct geographic positioning of biplate forward intersection; The airborne LiDAR that utilization of the present invention has a high resolution CCD camera boat fly in the journey can synchronous acquisition to the characteristics of high-precision laser point cloud and high resolving power aviation image; In order to realize obtaining fast the target of high-precision direct geo-location; The elevation information of having introduced laser point cloud carries out the elevation constraint to the adjustment result of the multi-disc forward intersection of high resolution image; Thereby realizing obtaining fast the high precision volume coordinate of picture point, is a kind of integrated high precision of multi-sensor data advantage, fast direct geographic positioning.

For achieving the above object, the present invention has adopted following technical scheme: airborne LiDAR point cloud and high resolution image carry out the combined positioning-method of spatial point,

1, original laser radar (LiDAR) data denoising step, this step adopt k nearest neighbor ball Denoising Algorithm, remove the noise that exists in the some cloud;

2, high accuracy number ground model DEM(digital elevation model) generate step; The laser radar data of this step after to denoising; Adopt triangulation network progressive encryption iteration filtering method to obtain beating laser point on the ground, and interpolation generate high accuracy number ground model dem data;

3, after manually selecting the initial point of same place, be aided with the elements of exterior orientation of high precision image that POS system obtains, look semi-automatic all the other points of selection same place in the aviation image striding air strips more.

4, through in the same place of choosing any 2 utilize biplate forward intersection to calculate initial value after, utilize multi-disc forward intersection method calculate same place object coordinates (X, Y, Z).

The object coordinates of the same place that 5, solves through a last step multi-disc forward intersection; According to its planimetric coordinates (X; Y) value is inserted in second goes on foot in the high accuracy DEM data of obtaining and is obtained accurate height value Z ', accomplishes the elevation constraint, thereby obtains the corresponding accurate object coordinates (X of same place; Y, Z ').

In technique scheme, said k nearest neighbor ball method is carried out cloud data denoising step and is:

At first data point set carries out space lattice and divides, and there is the space ball in imagination, and is the centre of sphere with current measuring point, and radius is got the distance of measuring point six of cube grids to the place respectively;

Get the minimum space ball of radius, in the grid that interferes with it, carry out the K-neighbor search, stop principle, then stop search if satisfy the search of being set up;

Otherwise, thereby get the contiguous ball of K that the space ball of long radius is more set up point to be located; A cloud is being carried out in the process of noise processed, the point in the k nearest neighbor ball of point to be located and foundation judge apart from size whether this point to be located is noise.

In technique scheme, the step that generates high accuracy DEM after the filtering of said triangulation network progressive encryption iteration is:

1. raw data is carried out the utmost point low spot and aerial point in the k nearest neighbor ball Filtering Processing rejecting data;

2. the outsourcing rectangle of construction data, the height value on four summits of this outsourcing rectangle is set according to the arest neighbors criterion, then the outsourcing rectangle is carried out triangulation, and with it as initial ground surface model;

3. data are carried out the graticule mesh tissue, grid should be slightly larger than the size of maximum buildings in the territory, cloud sector, and wherein the minimum point in each grid is millet cake initially, and the millet cake of choosing is initially joined in the TIN;

4. calculate the angle on each leg-of-mutton distance of putting its place and it and an Atria summit, if the value that calculates then joins it in TIN less than the pre-set threshold condition;

5. repeat 4. up to there not being new point to join in the TIN;

6. insert in the ground point that obtains and generate the as far as possible little DEM of graticule mesh size.

The present invention has following advantage: 1) plane precision and the vertical accuracy of the object coordinates of the accessed picture point of the present invention's multi-disc forward intersection of carrying out realizing under the prerequisite of elevation constraint through LiDAR point cloud are very high.2) utilize the present invention can reduce the workload of airborne remote sensing field operation control, translocation in a large number,, shorten aerial survey drafting period and expense greatly aspect the large scale one-tenth figure even can cancel field operation control.3) through accurate directly geographic positioning, save field operation and lay the reference mark, avoid complicated, time-consuming sky three operations of tradition, " short, adaptable and fast " advantage that makes airborne LiDAR obtain achievement can better be brought into play and embodied.

Description of drawings

Fig. 1 is the synoptic diagram that concerns of noise spot and neighborhood on every side among the present invention.

Fig. 2 is an iteration triangulation network filtering synoptic diagram among the present invention.

Fig. 3 is a projection coefficient method synoptic diagram.

Fig. 4 is the pinpoint method flow diagram of associating that a kind of airborne laser cloud data provided by the invention and high resolution image data are carried out spatial point.

D representes distance among Fig. 1, and centre point is current measuring point, and other point is the search point;

Embodiment

A kind of airborne LiDAR point cloud provided by the invention and high resolution image carry out the pinpoint method of associating of spatial point; Its essence is to utilize the advantage in several data source; Promptly be advantage and the plane precision advantage that object coordinates had that the forward intersection of high resolution image multi-disc is obtained of utilizing the vertical accuracy of cloud data; Finally realize the quick multi-disc forward intersection of high precision, thereby reach the purpose of the direct geo-location of high precision based on airborne LiDAR cloud data constraint.

Specify performance of the present invention below in conjunction with accompanying drawing, but they do not constitute to qualification of the present invention, only do for example, simultaneously through explain advantage of the present invention will become clear more with understand easily.

A kind of airborne laser cloud data provided by the invention and high resolution image data are carried out the pinpoint method of associating of spatial point, may further comprise the steps:

(1) original LiDAR (laser radar) data denoising:

If the cloud data existence is starkly lower than or the utmost point low spot of projecting environment and aerial point, can considerable influence post-processing algorithm precision, therefore these noise spots of removal before data processing.This method is surveyed the noise of removing in the some cloud through setting up the k nearest neighbor ball, and at first data point set carries out space lattice and divides, and there is the space ball in imagination, and is the centre of sphere with current measuring point, and radius is got the distance of measuring point six of cube grids to the place respectively.Get the minimum space ball of radius, in the grid that interferes with it, carry out the K-neighbor search, stop principle, then stop search if satisfy the search of being set up; Otherwise, thereby get the contiguous ball of K that the space ball of long radius is more set up point to be located.A cloud is being carried out in the process of noise processed, what mainly depend on point in the k nearest neighbor ball of point to be located and foundation judges apart from size whether this point to be located is noise (as shown in Figure 1).

(2) high accuracy DEM generates

After utilizing the filtering of the iteration triangulation network to obtain the ground point set, obtain Grid DEM to inserting in the ground point collection.Committed step is the filtering of the iteration triangulation network, the steps include: that 1. raw data being carried out k nearest neighbor ball Filtering Processing rejects utmost point low spot (noise spot that elevation is very low) and aerial point (noise spot that elevation is very high) (this step is accomplished in the 1st step) in the data; 2. the outsourcing rectangle of construction data, the height value on four summits of this outsourcing rectangle is set according to the arest neighbors criterion, then the outsourcing rectangle is carried out triangulation, and with its as initial ground surface model (as among Fig. 2 a); 3. data are carried out the graticule mesh tissue, grid should be slightly larger than the size of maximum buildings in the territory, cloud sector, and wherein the minimum point in each grid is millet cake initially, and the millet cake of choosing is initially joined in the TIN (TIN); 4. calculate the angle on each leg-of-mutton distance of putting its place and it and an Atria summit, if the value that calculates then joins (like the b among Fig. 2) in the TIN with it less than the pre-set threshold condition; 5. repeat 4. up to there not being new point to join (like the c among Fig. 2) in the TIN; 6. insert in the ground point that obtains and generate the as far as possible little DEM of graticule mesh size.

(3) same place is chosen

From having the semi-automatic same place of choosing the many seeing images picture of air strips of striding of degree of precision elements of exterior orientation, should guarantee to choose the sub-pixel precision of same place when choosing as far as possible.Because many baselines forward intersection its precision under the certain situation of elements of exterior orientation depends on the measurement accuracy of same place to a great extent.

(4) multi-disc forward intersection

Object point is to obtain with the approximate value substitution collinearity equation of required value in the error equation of multi-disc forward intersection, with the collinearity equation linearization, can get the error equation of multi-disc forward intersection: to each picture point, can list two error equations (like formula 1).If certain point appears in the n width of cloth sequential images, then can list 2n equation, ask so can separate with least square adjustment.Note 2 points when resolving:

1: at first utilize projection coefficient method (like formula 2) to calculate initial value.

2: the angle that before calculating, also needs to obtain any two light in the corresponding image rays through photo centre and selected picture point; If angle is less than 15 ° of (being picture point) then removing in these two light; And then reexamine, make at last and look in the light every angle in twos more and all after 15 °, carry out multi-disc forward intersection again.

$u_x=-\frac{\∂x}{\∂X}\mathrm{dX}-\frac{\∂x}{\∂Y}\mathrm{dY}-\frac{\∂x}{\∂Z}\mathrm{dZ}-L_x$

(formula 1)

$u_y=-\frac{\∂y}{\∂X}\mathrm{dX}-\frac{\∂y}{\∂Y}\mathrm{dY}-\frac{\∂y}{\∂Z}-L_y$

X in the formula 1 and y are picpointed coordinates, and X, Y, Z are the corresponding object point volume coordinates of picture point, and Lx, Ly are the matrix observed readings, u _x, u _yIt is the correction of matrix.

$\left\{\begin{array}{c}X=X_{S1}+N_1{X}_1=X_{S2}+N_2{X}_1\\ Y=Y_{S1}+N_1{Y}_1=Y_{S2}+N_2{Y}_1\\ Z=Z_{S1}+N_1{Z}_1=Z_{S2}+N_2{Z}_1\end{array}\right.$ (formula 2)

The projection coefficient method is as shown in Figure 3.S ₁, S ₂Be respectively the projection centre of left and right sides image, Bx, By, Bz is the photographic base component, m1, m2 are respectively the structure picture of object space point M on the image of the left and right sides, X ₁, Y ₁, Z ₁, X ₂, Y ₂, Z ₂, be m1, the empty auxilliary coordinate of the picture of m2, N ₁, N ₂Expression projects to ground spot projection coefficient with left picture point and right picture point.

In the computation process, picture point will be considered owing to the geometric distortion of lens or the known system error correction of measuring instrument, to weaken their influence.Usually principal point offset, radial distortion difference and deflection photogrammetric distortion parameter are provided by manufacturer, utilize radial distortion formula (formula 3) to come the picture point that collinearity equation is found the solution is compensated in this algorithm.

$\left\{\begin{array}{c}\mathrm{\Δ}{x}_r=(x-x_0)(K_1\·r^2+K_2\·r^4+K_3\·r^6+O[r^8\left]\right)\\ \mathrm{\Δ}{y}_r=(y-y_0)(K_1\·r^2+K_2\·r^4+K_3\·r^6+O[r^8\left]\right)\end{array}\right.$ (formula 3)

R wherein ²=(x-x ₀) ²+ (y-y ₀) ², (x ₀, y ₀) be the principal point coordinate, Ki is a distortion factor, i is 1,2,3.

(5) elevation constraint

The object coordinates of the same place of obtaining through a last step multi-disc forward intersection (X, Y, Z); According to (X, thus Y) insert in coordinate figure carries out in the high accuracy DEM data that the 2nd step generates and obtain more accurate height value Z ', retrain thereby accomplish elevation; Constitute the corresponding new high precision object coordinates (X of same place; Y, Z '), realize the direct geo-location of high precision.

Claims (3)

1. airborne LiDAR point cloud and high resolution image carry out the combined positioning-method of spatial point, it is characterized in that it comprises the steps:

1. original LiDAR data denoising step, this step adopt k nearest neighbor ball Denoising Algorithm, remove the noise that exists in the some cloud;

2. high accuracy number ground model DEM generates step, and the laser radar data of this step after to denoising adopts triangulation network progressive encryption iteration filtering method to obtain beating laser point on the ground, and interior slotting generation high accuracy number ground model dem data;

3. after manually selecting the initial point of same place, be aided with the elements of exterior orientation of high precision image that POS system obtains, look semi-automatic all the other points of selection same place in the aviation image striding air strips more;

4. through in the same place of choosing any 2 utilize biplate forward intersection to calculate initial value after, utilize multi-disc forward intersection method calculate same place object coordinates (X, Y, Z);

The object coordinates of the same place that 5. solves through a last step multi-disc forward intersection, (X slottingly in Y) in the high accuracy DEM data obtained in second step of value obtains accurate height value Z ' according to its planimetric coordinates; The constraint of completion elevation; Thereby obtain the corresponding accurate object coordinates (X, Y, Z ') of same place.

2. airborne LiDAR point cloud according to claim 1 and high resolution image carry out the combined positioning-method of spatial point, it is characterized in that said k nearest neighbor ball method carries out cloud data denoising step and be:

At first data point set carries out space lattice and divides, and there is the space ball in imagination, and is the centre of sphere with current measuring point, and radius is got the distance of measuring point six of cube grids to the place respectively;

Get the minimum space ball of radius, in the grid that interferes with it, carry out the K-neighbor search, stop principle, then stop search if satisfy the search of being set up;

Otherwise, thereby get the contiguous ball of K that the space ball of long radius is more set up point to be located; A cloud is being carried out in the process of noise processed, the point in the k nearest neighbor ball of point to be located and foundation judge apart from size whether this point to be located is noise.

3. airborne LiDAR point cloud according to claim 1 and high resolution image carry out the combined positioning-method of spatial point, it is characterized in that the step that generates high accuracy DEM after the filtering of said triangulation network progressive encryption iteration is:

1. raw data is carried out k nearest neighbor ball Filtering Processing and reject utmost point low spot and aerial point in the data;

2. the outsourcing rectangle of construction data, the height value on four summits of this outsourcing rectangle is set according to the arest neighbors criterion, then the outsourcing rectangle is carried out triangulation, and with it as initial ground surface model;

3. data are carried out the graticule mesh tissue, grid should be slightly larger than the size of maximum buildings in the territory, cloud sector, and wherein the minimum point in each grid is millet cake initially, and the millet cake of choosing is initially joined in the TIN;

4. calculate the angle on each leg-of-mutton distance of putting its place and it and an Atria summit, if the value that calculates then joins it in TIN less than the pre-set threshold condition;

5. repeat 4. up to there not being new point to join in the TIN;

6. insert in the ground point that obtains and generate the as far as possible little DEM of graticule mesh size.

Images