CN105571598A - Satellite laser altimeter footprint camera pose measuring method - Google Patents

Abstract

The invention discloses a satellite pose measuring method based on a footprint camera. The method comprises the basic steps that 1, according to camera calibration parameters, adopted satellite pose data and DEM product data, footprint camera geometry coarse correction is carried out; 2, evenly-distributed candidate control point areas with the small relief amplitude are selected according to a DEM, and a small window is selected on a footprint camera coarse correction image in the corresponding area to be used as a matched window image; 3, a geographical constraint template matching method is used for finding the position on a treated three-dimensional linear array image, and a vertical control point is obtained through the DEM product data; 4, the elements of exterior orientation of the image are calculated, and the orientation of the footprint camera is obtained. The accurate pose data of a satellite is obtained by accurately solving the accurate orientation of the stereomapping satellite footprint camera.

Description

A kind of assay method of laser satellite altitude gauge footmark camera attitude

Technical field

The invention belongs to Global observation by remote sensing field, relate to the assay method of the laser satellite altitude gauge footmark camera attitude under a kind of auxiliary data.

Background technology

In order to improve the precision of stereoplotting satellite, needing research badly and solving the series of key techniques problem faced in spaceborne attitude orientation process, wherein an important content is exactly the accurate sensing obtaining laser using laser coaxial footmark camera.

Realize footmark camera under different terrain, under satellite jitter conditions, under stereoscopic camera image and footmark camera image visual angle different situations each time can both accurately, Rapid matching, higher requirement is all proposed to the preconditioning technique, the selection technique at reference mark, the quick exact matching technology of image, Three dimensions control point acquiring technology, reference mark adjustment Technique etc. of image.Original measurement attitude due to satellite has a deviation, and DEM also not exclusively accurately, when the picture point using this attitude and DEM to carry out camera image calculates, the elevation estimation of each picture point there will be mistake, image after then correcting there will be irregular distortion, with the reference mark that the mode of images match obtains, probably reach several pixel rank error; The visual angle of footmark camera and the visual angle of stereoscopic camera are also inconsistent, even if stereoscopic camera image did ortho-rectification, distort comparatively greatly, have considerable influence to the accuracy of mating under the distortion change difference that it is caused by visual angle probably causes complicated landform.

The precision of selection to reference mark of position, reference mark is most important, not all place is all applicable to selecting reference mark, even if flat country, when calculating picture point, planar offset occurs, the change of its elevation is also little, also plane distortion is little for image after correction of a final proof, and the precision of final coupling can be higher.And under complicated landform, less plane dislocation can cause the vertical error of each picture point in this region comparatively large, finally cause the irregular deformation of image, easily error hiding occurs.Even if coupling accurately, also exists certain irregular variation, be difficult to constraint.

Design a kind of high-precision Control point extraction method, and the position at reference mark of making rational planning for, suppress or eliminate the inherent error source of Control point extraction, overcome current matching process to a certain extent and cannot evade the inherent error that complicated landform causes, the low inferior problem of matching efficiency, meet the actual needs of the accurate acquisition of high resolving power stereoplotting satellite footmark camera attitude, extracted by high-precision control point and obtain attitude measurement result accurately, laser instrument as high resolving power stereoplotting satellite system provides and points to parameter accurately, this has very important meaning for lifting earth observation satellite data geometric manipulations precision and product quality.

Summary of the invention

Problem to be solved by this invention is: provide a kind of map-making satellite footmark camera attitude measurement method.

Technical scheme provided by the invention adopts following steps:

(1) under the pixel distortion parameter of transient posture data, ccd image, dem data are auxiliary, geometry correction is carried out to ccd image; Concrete steps are as follows:

(1-1) obtain the transient posture data of satellite platform during ccd image imaging, the pixel distortion parameter of ccd image, obtain the dem data surveying district;

(1-2) direct location obtains the floor coordinate points of each pixel of image, and under transferring coordinate points to WGS84 coordinate system.Then super-resolution resampling is carried out to image;

(1-3) use bilinear interpolation, according to the WGS84 coordinate of each picture point, carry out resampling to image, the resolution after resampling is identical with the resolution of stereoscopic camera image;

(2) calculate topographic relief amplitude according to DEM, in image range, extract candidate reference mark; Concrete steps are as follows:

(2-1) intercepting dem data, obtain the region consistent with ccd image after resampling, and carry out resampling to dem data, is the resolution result with CCD geometry correction data consistent by its resolution adjustment;

(2-2) calculate the topographic relief amplitude of DEM, for each picture point of dem data after resampling, centered by it, intercept 128*128 pixel around, calculate the standard deviation of these pixels, using the result of calculation of this standard deviation as the topographic relief amplitude of this pixel;

(2-3) from image, identify the region of landform relatively flat, dem data after resampling is evenly divided into the pane of 64*64 size, in each pane, calculate the pixel that topographic relief amplitude is minimum, this pixel is labeled as candidate's pixel;

(2-4) according to the geographic coordinate of each picture point of raw CCD images, raw CCD images find from the nearest picture point of candidate's cell coordinate, then using this CCD original image as original image reference mark;

(2-5) find reference mark on original image and correspond to reference mark on resampling image: pixel calculates its picpointed coordinate on CCD resampling image, using this picture point of CCD resampling image as correcting image reference mark according to the coordinate figure of original image reference mark pixel;

(2-6) for any one correcting image reference mark, centered by this point, the window of 128*128 size is around chosen, and with the image of this window subgraph as next step template matches;

(3) the Three dimensions control point data that resampling ccd image and twin-line array images match obtain ccd image is extracted; Concrete steps are as follows:

(3-1) to any one subgraph of correcting image, obtain the position that it corresponds to stereoscopic camera image, stereoscopic camera image extracts the image of 1024*1024 size;

(3-2) carry out template matches with the subgraph of correcting image and the subgraph of stereoscopic camera image, obtain its position on stereoscopic camera image;

(3-3) according to the position of correcting image subgraph on stereoscopic camera image that template matches obtains, the position of subgraph central point on stereoscopic camera image can be obtained, calculate the plane geographic coordinate on stereoscopic camera picture position, then can calculate the plane geographic coordinate that on correcting image, picture point is corresponding, and then plane geographic coordinate corresponding to original image reference mark pixel can be obtained;

(3-4) according to plane geographic coordinate, obtain the height value of corresponding pixel on DEM, using this value as the elevation of this planimetric coordinates, obtain Three dimensions control point;

(3-5) RANSAC algorithm excluding gross error is used;

(4) the reference mark data obtained are used to resolve the elements of exterior orientation of CCD camera, according to the elements of interior orientation of raw CCD images reference mark pixel, and by this step (3) gained Three dimensions control point data, calculated the elements of exterior orientation of image by least square adjustment.

(4) use the reference mark data obtained to resolve the attitude data of CCD camera, concrete steps are as follows:

(4-1) according to the elements of interior orientation of raw CCD images reference mark pixel, and by this step (3) gained Three dimensions control point data, the adjustment formula of elements of exterior orientation (attitude data) is set up.

(4-2) by least square adjustment, iterative computation goes out the rotation matrix M of footmark camera.

(4-3) by the angle of setting of the rotation matrix of footmark camera, footmark camera and satellite, the attitude data of satellite is calculated.Angle of setting <r between satellite and footmark camera, p, h>, then the rotation matrix M of satellite _starfor:

$M_{Star}=M\&CenterDot;C_L^P$

Wherein, for angle of setting rotation matrix, tried to achieve by following formula:

$C_L^P=\left\{\begin{array}{ccc}\cos p\&CenterDot;\cosh & \cos p\&CenterDot;\sinh & -\sin p\\ \sin r\&CenterDot;\sin p\&CenterDot;\cosh -\cos r\&CenterDot;\sinh & \sin r\&CenterDot;\sin p\&CenterDot;\sinh +\cos r\&CenterDot;\cosh & \sin r\&CenterDot;\cos p\\ \cos r\&CenterDot;\sin p\&CenterDot;\cosh +\sin r\&CenterDot;\cosh & \cos r\&CenterDot;\sin p\&CenterDot;\sinh -\sin r\&CenterDot;\cosh & \cos r\&CenterDot;\cos p\end{array}\right\}$

The rotation matrix M of satellite is expressed as by elements of exterior orientation:

$M_{Star}=\left\{\begin{array}{ccc}\cos \mathrm{\&alpha;}\&CenterDot;\cos \mathrm{\&gamma;}-\sin \mathrm{\&alpha;}\&CenterDot;\sin \mathrm{\&beta;}\&CenterDot;\sin \mathrm{\&gamma;}& -\cos \mathrm{\&alpha;}\&CenterDot;\sin \mathrm{\&gamma;}-\sin \mathrm{\&alpha;}\&CenterDot;\sin \mathrm{\&beta;}\&CenterDot;\cos \mathrm{\&gamma;}& -\sin \mathrm{\&alpha;}\&CenterDot;\cos \mathrm{\&beta;}\\ \cos \mathrm{\&beta;}\&CenterDot;\sin \mathrm{\&gamma;}& \cos \mathrm{\&beta;}\&CenterDot;\cos \mathrm{\&gamma;}& -\sin \mathrm{\&beta;}\\ \sin \mathrm{\&alpha;}\&CenterDot;\cos \mathrm{\&gamma;}+\cos \mathrm{\&alpha;}\&CenterDot;\sin \mathrm{\&beta;}\&CenterDot;\sin \mathrm{\&gamma;}& -\sin \mathrm{\&alpha;}\&CenterDot;\sin \mathrm{\&gamma;}+\cos \mathrm{\&alpha;}\&CenterDot;\sin \mathrm{\&beta;}\&CenterDot;\cos \mathrm{\&gamma;}& \cos \mathrm{\&alpha;}\&CenterDot;\cos \mathrm{\&beta;}\end{array}\right\}---\left(4\right)$

Then eventually through above formula, solve the elements of exterior orientation of satellite.

The present invention is with good expansibility, and to be applicable to lack on accurate star in inertial navigation situation, realize spaceborne laser high-precision fixed to this business demand of accurate acquisition, and location can be pointed to accurately for other satellite borne equipments provide.

Accompanying drawing explanation

Fig. 1 is high-resolution optical satellite high-precision Control point extraction and attitude measurement whole process schematic diagram.

Fig. 2 is the process flow diagram calculating picture point plane geographic coordinate.

Fig. 3 is the flow process using dem data to calculate topographic relief amplitude in each picture point areas imaging.

Fig. 4 is for selecting the process flow diagram of robust control point from raw CCD images according to DEM.

Fig. 5 is the process flow diagram using matching process to obtain the Three dimensions control point coordinate at CCD reference mark.

Embodiment

For setting forth technical scheme of the present invention and advantage better, below in conjunction with accompanying drawing, implementation process of the present invention is further described.

The first step of the present invention relates to the thick correction of ccd image, and Fig. 2 is that the present invention uses collinearity equation, according to the schematic diagram of the planimetric coordinates of elements of interior orientation, the attitude of satellite, camera angle of setting, initial this pixel of grid DEM.Comprise the following steps:

From DEM, height value is read, the initial height value using this height value as this picture point according to the horizontal coordinate time of exposure that attitude measurement equipment obtains;

Use new height value Calculation Plane coordinate and repeat preceding step, until new elevation is identical with the elevation that previous step iteration obtains.

Sensor picpointed coordinate and topocentric coordinates meet collinearity equation:

$\left\{\begin{array}{c}X-X_S\\ Y-Y_S\\ Z-Z_S\end{array}\right\}=\mathrm{\&lambda;}M\left\{\begin{array}{c}x\\ y\\ z\end{array}\right\}$

In formula: (X, Y, Z) is the object coordinates under drawing coordinate system, (X _s, Y _s, Z _s) be the position of sensor projection centre.M is rotation matrix wherein be into figure coordinate and be tied to geocentric coordinate system rotation matrix, that geocentric coordinate is tied to local coordinate system rotation matrix; that local coordinate is tied to IMU coordinate system rotation matrix, that IMU is to sensor coordinate system rotation matrix; λ is photographic scale.Then have according to above formula:

$\left\{\begin{array}{c}X=X_S+(Z-Z_S)\frac{P\&lsqb;0\&rsqb;}{P\&lsqb;2\&rsqb;}\\ Y=Y_S+(Z-Z_S)\frac{P\&lsqb;1\&rsqb;}{P\&lsqb;2\&rsqb;}\end{array}\right.$

First initial X, Y is determined according to initial elevation, then new elevation is determined according to the elevation of DEM, from DEM, height value is read, the new height value using this height value as this picture point, X, Y that iterative is final according to the horizontal coordinate time of exposure that attitude measurement equipment obtains.

Under transferring terrestrial coordinate point to WGS84 coordinate system.

Use bilinear interpolation, according to the WGS84 coordinate of each picture point, carry out resampling to image, the resolution after resampling is identical with the resolution of stereoscopic camera image.

Second step of the present invention is the extracting method in region, candidate reference mark, first carries out topographic relief amplitude calculating, then obtains candidate reference mark.Fig. 3 is the schematic diagram using dem data to carry out the calculating of candidate's topographic relief amplitude.

As shown in Figure 3, these topographic relief amplitude computing method comprise the following steps:

Calculate the method first according to the geographic range of previous step resampling image, intercept dem data, obtain the region consistent with ccd image after resampling;

Carrying out resampling to dem data, is the resolution result with CCD geometry correction data consistent by its resolution adjustment;

To each picture point of dem data after resampling, centered by it, intercept 128*128 pixel around, calculate the standard deviation of these pixels, using the result of calculation of this standard deviation as the topographic relief amplitude of this pixel.

Inaccurate due to attitude of satellite data, when using geometric correction method to carry out geometry correction to image in a first step, landform causes the irregular deformation of image, and irregular deformation can cause coupling to occur deviation, brings impact to attitude measurement.Therefore, less topographic relief amplitude, the pixel random deformation in this region is less.Select the less region of topographic relief amplitude to be conducive to improving the precision of coupling as region, reference mark, and then improve the precision of attitude measurement.

Fig. 4 is the schematic diagram using candidate reference mark and carry out reference mark extracted region with reference to image, and this method of carrying out reference mark extracted region based on candidate reference mark and reference image comprises the following steps:

Dem data after resampling is evenly divided into the pane of 64*64 size, in each pane, calculates the pixel that topographic relief amplitude is minimum, this pixel is labeled as candidate's pixel;

Picpointed coordinate reference mark is selected, and according to the geographic coordinate of each picture point of raw CCD images, raw CCD images finds from the nearest picture point of candidate's cell coordinate, then using this CCD original image as original image reference mark.The coordinate figure of original image reference mark pixel calculates its picpointed coordinate on CCD resampling image, using this picture point of CCD resampling image as correcting image reference mark.

The stability of attitude measurement can be improved in equally distributed reference mark, method of the present invention is used can effectively to avoid the reference mark selected all to concentrate some or certain several flat site in the picture, algorithm ensure that the even distribution that reference mark is chosen, even if some accidental errors appear in some point, be also difficult to affect overall precision.

The present invention the three or four step is the method for Control point extraction, and Fig. 5 is the schematic diagram using dem data to carry out the calculating of candidate's topographic relief amplitude.

As shown in Figure 5, use matching process to obtain the Three dimensions control point coordinate at CCD reference mark, and the method obtaining final carriage measurement data comprises the following steps:

For any one correcting image reference mark, centered by this point, around choose the window of 128*128 size, and with the image of this window subgraph as next step template matches;

To any one subgraph of correcting image, obtain the position that it corresponds to stereoscopic camera image, stereoscopic camera image extracts the image of a certain size (1024*1024) size;

Carry out template matches with the subgraph of correcting image and the subgraph of stereoscopic camera image, obtain its position on stereoscopic camera image;

According to the position of correcting image subgraph on stereoscopic camera image that template matches obtains, the position of subgraph central point on stereoscopic camera image can be obtained, calculate the plane geographic coordinate on stereoscopic camera picture position, then can calculate the plane geographic coordinate that the picture point of correcting image is corresponding, and then plane geographic coordinate corresponding to original image reference mark pixel can be obtained;

According to plane geographic coordinate, obtain the height value of corresponding pixel on DEM, using this value as the elevation of this planimetric coordinates, obtain Three dimensions control point.

Use the reference mark data after elimination of rough difference obtained in the previous step, according to the elements of interior orientation of raw CCD images reference mark pixel, calculated the elements of exterior orientation of image by least square adjustment.

The present invention uses DEM to originate as the elevation at reference mark, avoids the adverse effect solving twin-line array data elevation deficiency and bring.

Because the optical property of CCD and twin-line array sensor is inconsistent, CCD and twin-line array sensor visual angle is inconsistent, geometry correction brings ccd image irregular deformation, twin-line array image local the factor such as distortion affect, rough error may be contained in the reference mark using images match to obtain, and therefore needs to reject with the precision improving reference mark further.The present invention uses RANSAC algorithm excluding gross error, and this algorithm mature and reliable, stable performance, efficiency are higher, can effectively be used for carrying out reference mark of the present invention elimination of rough difference completely.

The plane geometry precision of twin-line array image is other at 10 meter levels, but herein by equally distributed infilling control point, the precision at each reference mark is all about 0.1 pixel of footmark camera image, pass through adjustment processing, azimuthal precision that algorithm obtains can exceed the precision at reference mark, and the precision of net result is equivalent to plane precision and is better than 3m, meets and exceedes the accuracy requirement of laser guide, for stereoplotting satellite provides accurate reference mark data, ensure that the high precision of final charting results.

Claims (1)

1. an assay method for laser satellite altitude gauge footmark camera attitude, is characterized in that comprising the following steps:

(1) under the pixel distortion parameter of transient posture data, ccd image, dem data are auxiliary, carry out geometry correction to ccd image, concrete steps are as follows:

(1-1) obtain the transient posture data of satellite platform during ccd image imaging, the pixel distortion parameter of ccd image, obtain the dem data surveying district;

(1-2) direct location obtains the floor coordinate points of each pixel of image, and under transferring coordinate points to WGS84 coordinate system.Then super-resolution resampling is carried out to image;

(1-3) use bilinear interpolation, according to the WGS84 coordinate of each picture point, carry out resampling to image, the resolution after resampling is identical with the resolution of stereoscopic camera image;

(2) calculate topographic relief amplitude according to DEM, in image range, extract candidate reference mark; Concrete steps are as follows:

(2-1) intercepting dem data, obtain the region consistent with ccd image after resampling, and carry out resampling to dem data, is the resolution result with CCD geometry correction data consistent by its resolution adjustment;

(2-2) calculate the topographic relief amplitude of DEM, for each picture point of dem data after resampling, centered by it, intercept 128*128 pixel around, calculate the standard deviation of these pixels, using the result of calculation of this standard deviation as the topographic relief amplitude of this pixel;

(2-3) from image, identify the region of landform relatively flat, dem data after resampling is evenly divided into the pane of 64*64 size, in each pane, calculate the pixel that topographic relief amplitude is minimum, this pixel is labeled as candidate's pixel;

(2-4) according to the geographic coordinate of each picture point of raw CCD images, raw CCD images find from the nearest picture point of candidate's cell coordinate, then using this CCD original image as original image reference mark;

(2-5) find reference mark on original image and correspond to reference mark on resampling image: pixel calculates its picpointed coordinate on CCD resampling image, using this picture point of CCD resampling image as correcting image reference mark according to the coordinate figure of original image reference mark pixel;

(2-6) for any one correcting image reference mark, centered by this point, the window of 128*128 size is around chosen, and with the image of this window subgraph as next step template matches;

(3) the Three dimensions control point data that resampling ccd image and twin-line array images match obtain ccd image is extracted; Concrete steps are as follows:

(3-1) to any one subgraph of correcting image, obtain the position that it corresponds to stereoscopic camera image, stereoscopic camera image extracts the image of 1024*1024 size;

(3-2) carry out template matches with the subgraph of correcting image and the subgraph of stereoscopic camera image, obtain its position on stereoscopic camera image;

(3-3) according to the position of correcting image subgraph on stereoscopic camera image that template matches obtains, the position of subgraph central point on stereoscopic camera image can be obtained, calculate the plane geographic coordinate on stereoscopic camera picture position, then can calculate the plane geographic coordinate that on correcting image, picture point is corresponding, and then plane geographic coordinate corresponding to original image reference mark pixel can be obtained;

(3-4) according to plane geographic coordinate, obtain the height value of corresponding pixel on DEM, using this value as the elevation of this planimetric coordinates, obtain Three dimensions control point;

(3-5) RANSAC algorithm excluding gross error is used;

(4) use the reference mark data obtained to resolve the attitude data of CCD camera, concrete steps are as follows:

(4-1) according to the elements of interior orientation of raw CCD images reference mark pixel, and by this step (3) gained Three dimensions control point data, the adjustment formula of elements of exterior orientation (attitude data) is set up.

(4-2) by least square adjustment, iterative computation goes out the rotation matrix of footmark camera.

(4-3) by the angle of setting of the rotation matrix of footmark camera, footmark camera and satellite, the attitude data of satellite is calculated.