CN101509782A - Small-sized ground marker capturing and positioning method - Google Patents
Small-sized ground marker capturing and positioning method Download PDFInfo
- Publication number
- CN101509782A CN101509782A CNA2009100610938A CN200910061093A CN101509782A CN 101509782 A CN101509782 A CN 101509782A CN A2009100610938 A CNA2009100610938 A CN A2009100610938A CN 200910061093 A CN200910061093 A CN 200910061093A CN 101509782 A CN101509782 A CN 101509782A
- Authority
- CN
- China
- Prior art keywords
- small
- conspicuous
- sized ground
- ground marker
- terrain feature
- 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.)
- Granted
Links
Images
Landscapes
- Image Processing (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a method for capturing and positioning a small-scale landmark, which pertains to the field of automatic target recognition and navigation guidance and positioning, and comprises the steps of: 1) establishing space constraint relation between the small-scale landmark and a scene distinct object and a reference diagram of the distinct object; 2) detecting the scene distinct object according to the obtained real-time image of the distinct object; 3) limiting the matching search region of the real-time image of the distinct object according to the space constraint relation between the small-scale landmark and the scene distinct object, carrying out matching to the small-scale landmark in the matching search region, and obtaining a rough matching point of the small-scale landmark; 4) in the real-time image of the distinct object, determining the interested search region of the small-scale landmark by taking the rough matching point as the center; and 5) positioning the small-scale landmark accurately in the interested search region. The method accurately captures and positions the small-scale landmark in complex ground scene, thus carrying out navigation guidance position updating accurately to aircrafts in real time.
Description
Technical field
The invention belongs to Target Recognition and navigational guidance field of locating technology, be specifically related to the automatic capturing and positioning method of a kind of small-sized ground marker.
Background technology
The navigation accuracy that improves constantly aircraft is the important subject of field of aerospace technology always.The accumulated error of inertial navigation system increased gradually along with the time, so rely on inertial navigation system can not satisfy the requirement of high precision navigation separately.Can utilize aircraft to carry the photoelectric platform imaging system and obtain extraneous reference information, terrestrial reference be caught the location, thereby aircraft guidance system is carried out error correction, realize the precision navigation location of aircraft.
Small-sized ground marker is meant that the shared picture size of terrestrial reference feature is at the stable atural object of the feature of 3 * 39 * 9 pixel coverages.Small-sized ground marker is a kind of atural object of extensive existence, and its feature generally is not easy to change.If can excavate the potentiality that small-sized ground marker is applied to navigator fix, then the independence of aircraft guidance system will significantly improve, and will be significant to national economy and national security.Small-sized ground marker not only is meant the stable atural object that those physical dimensions own are little, and closely size is bigger also to comprise those, but the smaller atural object of size during remote imaging, thus small-sized ground marker to catch position application extensive.Small-sized ground marker also is a kind of target, for the small-sized ground marker in the complex background under the moving platform condition, it is discerned the location have a lot of difficulties: at first the ratio that accounts in whole background image of small-sized ground marker is very little, and the background image feature is not remarkable relatively; Secondly background is complicated and constantly change (, the disappearance of local atural object or appearance drowning as rising of river), exists a large amount of with the similar pattern of landmark image in the background again; Moving platform itself is unstable once more, and the inertial navigation information of aircraft has error.Therefore, under the complex background small-sized ground marker catch the location be a difficult point problem.
Moon Y S, Zhang Tianxu, Zuo Zhengrong is at " Detection of sea surface smalltargets in infrared images based on multi-level filter and minimum risk bayestest ", International Journal of Pattern Recognition and ArtificialIntelligence, 2000,14 (7): proposed a kind of sea infrared small target detection algorithm among the 907-918 based on multiple-stage filtering, this algorithm is to background, little target and noise spectrum are analyzed, the filtering template of utilizing basic filtering template cascade to constitute different bandwidth is carried out filtering to infrared image, thus the target of detecting.The problem of this method mainly is: for the small-sized ground marker in the complicated ground background, because the existence of a large amount of icotypes is used this method and may be caused a large amount of false-alarms.
Gianhca Marsiglia, Luea Fortunato, Aurora Ondini is at " Template matchingtechniques for automatic IR target recognition in real and simulatedscenarios:tests and evaluations ", Proceeding of SPIE 2003, infrared identification algorithm based on template matches has been proposed among the 5094:159-16, this method adopts the structural drawing of target or texture maps to make the target reference map, mate recognition objective with the real-time figure of sensor output, be applicable to large-scale Target Recognition.The problem of this method mainly is: for the small-sized ground marker in the complex background, terrestrial reference itself is with respect to complex scene around it, and feature is not remarkable, like this when utilizing reference diagram to mate, because the existence of icotype causes the coupling identification probability low.
Even identification location for the conspicuous terrain feature characteristic portion, certain point of crossing as bridge and river, traditional method is the coupling that adopts based on reference diagram, promptly at first prepare two-value or three value reference diagrams according to defending sheet on ground, it is deposited in the moving platform computing machine, according to the inertial navigation information of moving platform, the reference diagram in the computing machine is carried out perspective transform then, with reference subgraph after the conversion and photoelectric imaging sensor realtime graphic coupling, thereby catch the location feature position.But this method is also unreliable, because may there be a plurality of point of crossing in bridge and river in complex background, and may exist and the similar regional area in point of crossing.When the identification location, just may navigate to these zone similarities like this, rather than the navigator fix of appointment point.And if the detection method that directly adopts precision target is discerned the location feature position, then because exist a large amount of in the complex background with the similar zone of this navigator fix unique point, cause a large amount of similar area of appearance in the image, can not accurately navigate to the landmark point of appointment.Therefore, traditional matching process can not reach the reliable navigator fix requirement of aircraft.
The ratio that precision target in the complex background accounts in whole background is very little, with respect to whole background its characteristic remarkable a little less than.Adopt traditional full figure to detect or full figure mates to come small-sized ground marker in the detection and location complex background, owing to have a large amount of parallel patterns in the complex background, this accuracy that can cause catching the location is low, can not reach the reliable accurately positioning requirements of aircraft navigation.
Summary of the invention
The invention provides a kind of small-sized ground marker capturing and positioning method, accurately catch the precision target in the complicated ground scene of location, thereby accurately implement the correction of aircraft real-time navigation guide position.
The automatic capturing and positioning method of a kind of small-sized ground marker, carry out according to following steps:
(1) space constraint of setting up between small-sized ground marker and conspicuous terrain feature concerns and the conspicuous terrain feature reference diagram;
(2) detect scene distinct object according to the conspicuous terrain feature realtime graphic of taking;
(3) according to the match search zone of the relation of the space constraint between small-sized ground marker and conspicuous terrain feature qualification conspicuous terrain feature realtime graphic, in this match search zone, small-sized ground marker is mated, obtain the thick match point of small-sized ground marker;
(4) in the conspicuous terrain feature realtime graphic, be the region of search interested that small-sized ground marker is determined at the center with thick match point;
(5) carry out the accurate location of small-sized ground marker in region of search interested.
Described space constraint relation comprises relation of inclusion, neighbouring relations and the spacing of small-sized ground marker and conspicuous terrain feature.
The region of search interested of described step (4) is the center with thick match point, and zone length and width are by m* (h/tan θ) decision, and h is a flying height, and θ is the angle of pitch, and constant m span is [0,0.1].
The present invention is directed to the defective of existing navigator fix, Matching Location, according to the characteristics of complex background small-sized ground marker, propose a kind of automatic capturing and positioning method of moving platform small-sized ground marker based on the space constraint relation, the technique effect of this method is embodied in:
(1) the present invention sets up the space constraint relational knowledge base between small-sized ground marker and the conspicuous terrain feature in advance on ground, when small-sized ground marker is located, at first detects conspicuous terrain feature, more further according to space constraint relation location small-sized ground marker.The recursion capturing and positioning method of small-sized ground marker has good robustness and precision under this moving platform condition.This is because the correct verification and measurement ratio height of conspicuous terrain feature, although precision is not enough, yet based on the coupling reliability height of conspicuous terrain feature reference diagram, more than one of the restriction relation of small-sized ground marker and each conspicuous terrain feature, and reliably.
(2) set up the geometric relationship between the conspicuous terrain feature in terrestrial reference and the scene on ground according to defending sheet, such as occurring the airport in the scene, bridge during atural objects such as bulk waters, is set up the geometric relationship between these atural objects and the terrestrial reference, obtains N restriction relation.To realtime graphic, detect the conspicuous terrain feature in the image successively during flight.According to this N restriction relation, determine to carry out the match search district of conspicuous terrain feature realtime graphic.So, the field of search has been a full figure no longer just, but the regional area under N restriction relation restriction, so just obviously reduced ambiguous influence, improved the coupling accuracy simultaneously.
(3) when small-sized ground marker is accurately located, not accurately to locate, but carry out the feature templates coupling at the area-of-interest of the thick match point of reference diagram at full figure, so just further reduced ambiguous influence, improved the detection accuracy simultaneously.
Description of drawings
Fig. 1 is a schematic flow sheet of the present invention;
Fig. 2 is banded zone testing process figure;
Fig. 3 is that band detects template figure, and Fig. 3 (a) is a horizontal shuttering, and Fig. 3 (b) is a vertical formwork, and Fig. 3 (c) is 45 degree templates, and Fig. 3 (d) is 135 degree templates;
Fig. 4 is Perspective transformation model figure;
Fig. 5 be contain bridge, river, land defend sheet figure;
Fig. 6 is small-sized ground marker (bridge and river confluence) and river, bridge, the space constraint graph of a relation on land;
Fig. 7 is a bridge band reference diagram;
Fig. 8 contains bridge, river, the reference diagram of conspicuous terrain features such as water channel, land;
Fig. 9 is the real-time figure example in bridge zone that photoelectric imaging sensor obtains on the moving platform;
Figure 10 is the bridge zone binary map from this figure extraction in real time;
Figure 11 is that bridge straight-line segment part Radon conversion is with reference to point diagram;
Figure 12 is the banded zone figure that bridge straight-line segment part Radon conversion obtains;
Figure 13 is the ribbon field of search figure that carries out the thick coupling of conspicuous terrain feature;
Figure 14 is the sub-reference diagram (carrying out perspective transform according to flight inertial navigation parameter) that is used for thick coupling in the conspicuous terrain feature reference diagram;
Figure 15 is thick Matching Location figure as a result;
Figure 16 is the region of interest figure of search terrestrial reference;
Figure 17 is small-sized ground marker (bridge/river confluence) feature templates figure;
Figure 18 is bridge/river confluence at the accurate Matching Location of region of interest figure as a result;
Figure 19 is the satellite mapping that contains ait, loke shore, band atural object;
Figure 20 is a small-sized ground marker (ait) and the space constraint graph of a relation of loke shore, band atural object;
Figure 21 is the reference diagram that contains ait, loke shore, band atural object;
Figure 22 is the real-time figure example that contains ait that photoelectric imaging sensor obtains on the moving platform;
Figure 23 is band atural object testing result figure;
Figure 24 is that band atural object Radon conversion is with reference to point diagram;
Figure 25 is the rectilinear after the band atural object Radon conversion;
Figure 26 is the ribbon match search area schematic of carrying out the thick coupling of conspicuous terrain feature;
Figure 27 is used for the thick template synoptic diagram that mates (according to flight inertia in the conspicuous terrain feature reference diagram
Navigational parameter carries out perspective transform);
Figure 28 is thick Matching Location figure as a result;
Figure 29 is the region of search interested synoptic diagram of ait;
Figure 30 is the two-value feature templates figure of small-sized ground marker (ait);
Figure 31 is that ait is in the accurate Matching Location result schematic diagram of region of interest.
Embodiment
The present invention is further detailed explanation below in conjunction with accompanying drawing and two examples.
Example 1:
The satellite photo that contains bridge, river, land with Fig. 5 is an example, landmark point is a circle part among the figure, i.e. certain point of crossing in bridge and river, and this point of crossing and bridge and river have tangible space constraint relation as can be seen, be it both on the bridge band, locate on the riverbank again.
The flight pre-treatment step is:
(1) prepare on ground
Set up the relation of the space geometry between the conspicuous terrain feature in small-sized ground marker and the background on ground according to defending sheet, conspicuous terrain feature mainly comprises three kinds: remarkable plane class atural object, significantly lines class atural object and marking area class atural object.Bridge is remarkable lines class atural object among Fig. 5, and landmark point is certain point of crossing in bridge and river, sets up landmark point and bridge according to defending sheet, and the space geometry relation between the river as shown in Figure 6.The landmark point coordinate is among Fig. 6
Two end points coordinates of bridge are respectively
With
And landmark point
Residing river region is R.The space geometry relational expression of setting up is:
(x
0, y
0) ∈ l and (x
0, y
0) ∈ R,
Wherein bridge place straight line l is by (x
1, y
1) and (x
2, y
2) determine
The reference diagram for preparing remarkable lines class atural object.The preparation of reference diagram is closely to link to each other with the kind of the classification of real image and target, according to different actual type images and different target, prepares different reference diagrams.For optical imagery, to containing terrestrial reference, the visible images of target and surrounding enviroment is done two-value and is cut apart, manually find a suitable thresholding, follow-up work is to remove black, white noise, adjust respective edges and zone according to original image then, make the terrestrial reference that need mate and target signature more significantly, and remove corresponding noise and interference according to the characteristics of optical imagery image as far as possible.
According to the characteristics of defending sheet and infrared imaging sensor that contain bridge, river, land, the bridge band reference diagram for preparing on ground contains bridge, river as shown in Figure 7, and the reference diagram of conspicuous terrain features such as water channel, land as shown in Figure 8.Simultaneously, consider that this conspicuous terrain feature is a lines class atural object,, be convenient to projection in-flight (Radon) conversion, choose the Radon change point on ground in order to determine the band straight line awing better.On reference diagram, find out two some a and the b of bridge on rectangular part on ground, and write down the coordinate (x of these two points on reference diagram
a, y
a) and (x
b, y
b), it is deposited in the aircraft computer, as shown in figure 11 (circle is two some a and the b for looking on ground partly).
Treatment step is in-flight:
(2) detect remarkable lines class atural object
The real-time figure example in bridge zone that Fig. 9 obtains for photoelectric imaging sensor on the moving platform.At first detect the bridge band among the figure, the flow process of detection bridge banded zone as shown in Figure 2.
The process that detects the bridge straight line among Fig. 9 is as follows: at first utilize 4 templates (as shown in Figure 3) that realtime graphic is done convolution algorithm, can obtain 4 width of cloth convolution results figure.4 width of cloth convolution results images are got maximal value in 4 width of cloth images by pixel, obtain comprehensive 4 directional informations and have the comprehensive convolution results figure of anti-certain rotatory power.Calculate average μ and the standard deviation δ of comprehensive convolution results figure, get threshold value G=μ+K σ, get K=2 and make thresholding, comprehensive convolution results figure is carried out two-value cut apart.Bianry image after traversal is cut apart obtains the connected region among the figure, the pixel that constitutes same connected region is noted with the form of chained list, and given different values of statistical indicant to each connected region, obtains a series of area-of-interests.To these a series of area-of-interests gray average of the pixel on the zone not in the boundary rectangle of zoning respectively,, then be judged to be bridge if gray average and waters average are more or less the same.If a plurality of qualified bridges zone is arranged, then get the elder of length and be the bridge zone.The bridge zone binary map that obtains as shown in figure 10.
In order to guarantee to detect the bridge zone that obtains is strip, and Figure 10 is carried out the Radon conversion.The expression formula of two dimension Radon conversion is:
D is whole realtime graphic rectangular coordinate plane; (x y) is picture point (x, gray-scale value y) to f; ρ is the distance that true origin arrives straight line ρ=xcos θ+ysin θ; θ is the angle of straight line ρ=xcos θ+ysin θ and x axle.It makes f (x, y) carry out integration along straight line ρ=xcos θ+ysin θ, the Radon conversion can be understood as the projection of image in ρ-θ space, the every bit correspondence image space straight line in ρ-θ space, and the Radon conversion is the integration of image slices vegetarian refreshments on each bar straight line.According to the combination of different ρ and θ, image is carried out the Radon conversion, find out maximum value, ρ that it is corresponding and θ are exactly the straight line parameter that need look for.
When carrying out the Radon conversion, be not all to carry out conversion, and just within certain angular range, carry out, consider the error of being used to survey combination in all angles of θ (0～2 π), this angular range is measured composite set by the moving platform inertial navigation and is provided, and can take into account speed and accuracy rate like this.
Navigation information (the course angle α of target seeker according to aircraft, pitching angle theta and height h), just can carry out perspective transform to a and the b in the remarkable lines class atural object reference picture shown in Figure 11, obtain a and b pixel a ' and the b ' in real-time front view, its coordinate is respectively (x
A ', y
A ') and x
B ', y
B ').Perspective transformation model as shown in Figure 4, wherein: φ is the vertical imaging viewing field of imager angle,
Be the horizontal angle of image of imager, the forward sight picturedeep is ROW, and picturewide is COL, and α is the position angle, and θ is the angle of pitch, and h is a flying height.As shown in Figure 4: T
0(x
0, y
0) be the optical axis loca, T
1(x
1, y
1) be certain point on ground, T in the front view picture that photoelectric sensor obtains then
0The pixel position be that (COL/2 ROW/2), establishes T
1Pixel position in the front view picture that photoelectric sensor obtains is (T
1_ COL, T
1_ ROW), then calculate T
1_ COL and T
1The process of _ ROW is as follows:
OT
0＝h/tanθ
tan(∠OMP)＝h/OM
T
1_ROW＝(∠OMP-θ)*ROW/φ
Wherein, OT
0Be optical axis and the earth intersection point T
0Be projected to the distance that the earth O is ordered with imager, the M point is T
1Spot projection is to optical axis longitudinal direction and OT
0The intersection point of straight line, OM then are the distance of some O to a M.
For two known in reference diagram point (x
a, y
a) and (x
b, y
b), their positions on the realtime graphic that photoelectric imaging sensor obtains are x
A ', y
A ') and x
B ', y
B '), then can pass through Perspective transformation model, according to the azimuth angle alpha of navigational system, pitching angle theta, height h calculates x
A ', y
A ', x
B ', y
B 'Following (coordinate difference substitution Perspective transformation model formula is got final product):
Known two points just can be determined straight line, calculate the slope of strip in the realtime graphic
σ promptly is the bridge band straight line parameter under the accurately errorless situation of aircraft inertial navigation information, that is to say that if the inertial navigation information of aircraft is accurate, then ω is exactly a realtime graphic cathetus slope.But because also there is error in the inertial navigation information of aircraft, the definition error is k=Δ α Δ θ Δ h, Δ α, and Δ θ, Δ h is respectively course angle α, pitching angle theta, the error of height h.Therefore, need carry out the Radon conversion, determine realtime graphic cathetus parameter, the angle of R α don conversion should change [σ-ck, σ+ck] in a scope that with σ (being converted into the radian form) is the center, wherein c is a constant, the span of c is [0,100], and present embodiment is taken as 20, in order to take into account accuracy rate and arithmetic speed, the step-length of R α don conversion is set at 2.Obtaining straight-line equation after the Radon conversion is:
ρ＝xcosω+ysinω
White pixel points all among Figure 10 is carried out following processing: equal 255 point for all gray-scale values, calculate the distance that these put straight line ρ=xcos ω+ysin ω, if distance is greater than Δ d (as get Δ d be 3), then will be changed to 0, otherwise remain unchanged apart from gray-scale value greater than the point of Δ d.The banded zone that the Radon transformation results obtains as shown in figure 12, comprising this regional minimum rectangle is exactly the match search zone, as shown in figure 13.
(3) the conspicuous terrain feature reference diagram of bar region direction along the line coupling
Limit the match search zone of conspicuous terrain feature realtime graphic according to the relation of the space constraint between small-sized ground marker and conspicuous terrain feature, in this match search zone, small-sized ground marker is carried out coarse positioning, obtain the thick match point of small-sized ground marker, with this thick match point is the center, determine a suitably regional area of size, in this regional area, detect the terrestrial reference feature and carry out fine positioning.
For realtime graphic, referring to Fig. 9, the attitude parameter of target seeker is: pitching angle theta, course angle α, height h.According to attitude parameter, remarkable lines class atural object reference diagram is carried out perspective transform, obtain the forward sight conversion figure of reference diagram.Conspicuous terrain feature reference diagram after conversion is chosen square matching template, and the center of matching template is a small-sized ground marker, and the length of template and width are determined by θ h μ, wherein θ is the angle of pitch, and h is a height, and μ is a constant, the span of μ is [0,0.01], and present embodiment is taken as 0.005.With the matching template the chosen white ribbon zone and realtime graphic relevant matches in Figure 13.
The conspicuous terrain feature reference diagram subgraph template of choosing as shown in figure 14.According to obtain small-sized ground marker and the space constraint between conspicuous terrain feature relation on every side in the ground preparatory stage, can know that small-sized ground marker to be detected is positioned at certain infall in bridge and river course.The method that direct employing full figure detects or full figure mates detects small-sized ground marker can be very difficult, because the crossing in bridge and river course has a lot, and this small-sized ground marker feature is not remarkable.But small-sized ground marker that obtains according to the ground preparatory stage and the relation of the space constraint between the conspicuous terrain feature on every side, can detect significant bridge atural object earlier, carry out mating along the bridge strip direction then at the reference diagram of small-sized ground marker, not only reduced calculated amount, and can obviously improve like this and detect accuracy rate.The bridge banded zone has been arrived in previous step detection and location, can know that small-sized ground marker point is just on this banded zone, but also do not know the particular location of this small-sized ground marker on band, so according to space constraint relation (small-sized ground marker is on bridge), the conspicuous terrain feature reference diagram that contains small-sized ground marker along strip direction mates, and the field of search of mating is shown in the white ribbon zone among Figure 13.If full figure search, then can be owing to the zone of a large amount of similar small-sized ground marker that exists in the image causes detection inaccurate, and adopt based on the small-sized ground marker and the coupling of the space constraint relation of conspicuous terrain feature on every side, then define match search district scope, thereby improved the detection accuracy rate.
The tolerance that coupling adopts is for going average Normalized Grey Level simple crosscorrelation, and go average Normalized Grey Level simple crosscorrelation matching algorithm to be defined as follows: establishing real-time figure is G
r, its size is M
r* N
r, reference diagram is G
s, its size is M
s* N
s, and M
s＜M
r, N
s＜N
rThen in real time among the figure so that (u v) be that the upper left corner, size are M
s* N
sSubgraph G
r(u, v) with reference to figure G
sBetween go average normalized crosscorrelation tolerance ρ (u v) is:
Wherein
With
Be respectively G
r(u is v) with G
sGray average.(u, v) the correlation matrix of Gou Chenging is exactly a correlation surface by ρ.(u chooses extreme point in v) and obtains the Matching Location point, and the operand of simple crosscorrelation matching algorithm mainly concentrates on correlation surface data ρ, and (u is in the calculating v) from the coupling correlation surface data ρ that calculates again.
Matching result as shown in figure 15.
So just obtained the position of small-sized ground marker roughly, but also out of true is the center with this thick match point, and (area size is determined by m (h/tan θ) to choose a regional area, h is a flying height, θ is the angle of pitch, and m is a constant, and span is [0,0.1], present embodiment is taken as 0.01), this regional area is exactly region of interest (the potential site zone of small-sized ground marker), as shown in figure 16.
(4) detect and accurately locate small-sized ground marker
Through the step of front, determined the area-of-interest at small-sized ground marker place, but can't accurately locate the particular location of small-sized ground marker in area-of-interest.Therefore also need in this region of interest, carry out the accurate location of small-sized ground marker.The method that present embodiment adopts is in the perspective transform of previous step process, is the front view of conspicuous terrain feature reference diagram with the perspective transform of conspicuous terrain feature reference diagram.In the front view of conspicuous terrain feature reference diagram, be that a little template of feature is chosen at the center, carry out characteristic matching with this template and the region of interest of figure in real time, to obtain the exact position of small-sized ground marker with the small-sized ground marker point.The little template size of choosing is according to θ hs decision, and wherein θ is the angle of pitch, and h is a height, and s is a constant, and the span of s is [0,0.005], and present embodiment is taken as 0.0025.The terrestrial reference reference template example of choosing as shown in figure 17, terrestrial reference detection and location result is as shown in figure 18.
Example 2:
Satellite photo with as shown in figure 19 certain ait is an example, the island that landmark point comprises for circle among the figure, the terrestrial reference island is contained in the lake, can see the island next door by a tangible band atural object, has remarkable space constraint relation between ait and this band atural object and the loke shore.
The flight pre-treatment step is:
(1) prepare on ground
Set up the relation of the space geometry between the conspicuous terrain feature in small-sized ground marker and the scene on ground according to defending sheet, conspicuous terrain feature mainly comprises three kinds: remarkable plane class atural object, significantly lines class atural object and marking area class atural object.One bridge block is arranged near the island among Figure 19, and island and loke shore have tangible geometric relationship, the space constraint relation of setting up island as shown in figure 20.
According to the characteristics of ait satellite photo and optical imagery, the conspicuous terrain feature two-value reference diagram for preparing on ground as shown in figure 21.Simultaneously, consider that this conspicuous terrain feature is a lines class atural object,, be convenient to Radon conversion in-flight, choose the Radon change point on ground in order to determine the band straight line awing better.On reference diagram, find out two some α and the b of bridge on rectangular part on ground, and write down the coordinate (x of these two points on reference diagram
a, y
a) and (x
b, y
b), it is deposited in the aircraft computer, as shown in figure 24.
Treatment step is in-flight:
(2) detect remarkable lines class atural object
Near the realtime graphic of the ait that Figure 22 obtains for photoelectric imaging sensor, island ribbon bridge characters of ground object is very remarkable, and it is carried out detection and location.
At first utilize 4 templates (as shown in Figure 3) that image is done convolution algorithm, can obtain 4 width of cloth convolution results figure.4 width of cloth convolution results images are got maximal value in 4 width of cloth images by pixel, obtain comprehensive 4 directional informations and have the comprehensive convolution results figure of anti-certain rotatory power.Calculate average μ and the standard deviation δ of comprehensive convolution results figure, get threshold value G=μ+K σ, get K=2 and make thresholding, comprehensive convolution results figure is carried out two-value cut apart.Bianry image after traversal is cut apart obtains the connected region among the figure, the pixel that constitutes same connected region is noted with the form of chained list, and given different values of statistical indicant to each connected region, obtains a series of area-of-interests.To these a series of area-of-interests gray average of the pixel on the zone not in the boundary rectangle of zoning respectively,, then be judged to be band if gray average and waters average are more or less the same.If a plurality of qualified banded zones are arranged, then get the elder of length and be banded zone.The banded zone that obtains as shown in figure 23.
In order to obtain straight-line equation, Figure 23 is carried out the Radon conversion.(pitching angle theta, course angle α, height h) carries out conversion to some a among Figure 23 and b respectively according to real-time parameter, obtains the coordinate position (x of corresponding real-time figure mid point
A ', y
A ') and (x
B ', y
B ').By
Calculate the slope of band atural object in the realtime graphic, σ promptly is the band atural object parameter under the accurately errorless situation of aircraft navigation information.But because there is error in the navigation information of aircraft, the definition error is k=Δ α Δ θ Δ h, Δ α, and Δ θ, Δ h is course angle respectively, the angle of pitch, the error of height.Therefore, the angle of Radon conversion should change [σ-ck, σ+ck] in a scope that with σ (being converted into the radian form) is the center, and wherein c is a constant, and present embodiment is taken as 30, and in order to take into account accuracy rate and arithmetic speed, the step-length of Radon conversion is set at 2.White pixel among figure point is carried out the Radon conversion
Obtain the polar coordinates expression formula of straight line after the conversion:
ρ＝xcosθ+ysinθ
The straight line r that obtains after the Radon conversion as shown in figure 25.
(3) the conspicuous terrain feature reference diagram of bar region direction along the line coupling
(pitching angle theta, course angle α, height h) carries out perspective transform to reference diagram according to flight parameter, obtains the real-time forward sight conversion figure of reference diagram.Choose a certain size reference subgraph (choose size and determine that by θ h μ wherein θ is the angle of pitch, h is a flying height, and μ is a constant, and present embodiment μ is taken as 0.005), the reference subgraph of choosing as shown in figure 27.
According to ait and the remarkable relation of the space constraint between the band atural object, can know between ait and the remarkable band atural object to have certain geometric relationship that ait but has certain distance with band atural object not on band atural object.In ait as shown in figure 20 and the space constraint between the conspicuous terrain feature relation,, obtain the l ' in the realtime graphic to carrying out perspective transform apart from l between ait and the band bridge atural object.So can know the ait among the real-time figure should be no more than in the banded zone of l ' in the distance with remarkable band atural object.So the region of search of definition ait reference diagram coupling is:
D
xr≤l′
D wherein
XrDistance between the straight line r that pixel x obtains to the Radon conversion in the expression realtime graphic.Not not that full figure has been searched for so just, but utilized the space constraint relation between ait and the remarkable lines atural object, define the match search district greatly, improved the detection accuracy rate.Figure 26 is the match search district.
The matching process that present embodiment adopts is for removing average Normalized Grey Level cross correlation algorithm:
Carry out after the ait reference diagram coupling the result as shown in figure 28.
So just obtained the position of ait roughly, but go back out of true, with the match point is the center, (area size is by m (h/tan θ) decision, and h is a flying height, and θ is the angle of pitch to choose a regional area, m is a constant, it is 0.01 that present embodiment is got m), this regional area is exactly region of interest (the potential site zone of ait), as shown in figure 29.
(4) detect small-sized ground marker
Through the step of front, determined the area-of-interest of ait, but can't accurately locate the particular location of ait in area-of-interest.Therefore also need in this region of interest, carry out the accurate location of ait.The method that present embodiment adopts is in the perspective transform of previous step process, is the front view of conspicuous terrain feature reference diagram with the perspective transform of conspicuous terrain feature reference diagram.In the front view of conspicuous terrain feature reference diagram, be that a little template is chosen at the center, mate with this template and the region of interest of figure in real time, to obtain the exact position of small-sized ground marker with small-sized ground marker point (ait).The little template size of choosing is according to θ hs decision, and wherein θ is the angle of pitch, and h is a height, and the span of constant s is [0,0.005], and present embodiment is taken as 0.0025.The terrestrial reference reference template example of choosing as shown in figure 30, terrestrial reference detection and location result is as shown in figure 31.
Claims (5)
1, the automatic capturing and positioning method of a kind of small-sized ground marker, carry out according to following steps:
(1) space constraint of setting up between small-sized ground marker and conspicuous terrain feature concerns and the conspicuous terrain feature reference diagram;
(2) detect scene distinct object according to the conspicuous terrain feature realtime graphic of taking;
(3) according to the match search zone of the relation of the space constraint between small-sized ground marker and conspicuous terrain feature qualification conspicuous terrain feature realtime graphic, in this match search zone, small-sized ground marker is mated, obtain the thick match point of small-sized ground marker;
(4) in the conspicuous terrain feature realtime graphic, be the region of search interested that small-sized ground marker is determined at the center with thick match point;
(5) carry out the accurate location of small-sized ground marker in region of search interested.
2, the automatic capturing and positioning method of a kind of small-sized ground marker according to claim 1 is characterized in that, described space constraint relation comprises relation of inclusion, neighbouring relations and the spacing of small-sized ground marker and conspicuous terrain feature.
3, the automatic capturing and positioning method of a kind of small-sized ground marker according to claim 1, it is characterized in that, the region of search interested of described step (4) is the center with thick match point, zone length and width are determined by m* (h/tan θ), h is a flying height, θ is the angle of pitch, and constant m span is [0,0.1].
4, according to claim 1 or the automatic capturing and positioning method of 2 or 3 described a kind of small-sized ground markers, it is characterized in that, described step (3) is before the coupling to small-sized ground marker, determine first matching template of conspicuous terrain feature reference diagram at first in the following manner: the conspicuous terrain feature reference diagram is obtained front view do perspective transform, in front view figure, choose first matching template, first matching template is to be the center with the small-sized ground marker position, template length and width are determined by θ * h* μ, wherein θ is the angle of pitch, h is a flying height, the span of constant μ is [0,0.01].
5, according to claim 1 or the automatic capturing and positioning method of 2 or 3 described a kind of small-sized ground markers, it is characterized in that, described step (5) is before accurately locating, determine second matching template of conspicuous terrain feature reference diagram at first in the following manner: the conspicuous terrain feature reference diagram is obtained front view do perspective transform, in front view figure, choose second matching template, second matching template is to be the center with the small-sized ground marker position, template length and width are determined by θ * h*s, wherein θ is the angle of pitch, h is a flying height, the span of constant s is [0,0.005].
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2009100610938A CN101509782B (en) | 2009-03-09 | 2009-03-09 | Small-sized ground marker capturing and positioning method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2009100610938A CN101509782B (en) | 2009-03-09 | 2009-03-09 | Small-sized ground marker capturing and positioning method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN101509782A true CN101509782A (en) | 2009-08-19 |
| CN101509782B CN101509782B (en) | 2011-02-09 |
Family
ID=41002253
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN2009100610938A Expired - Fee Related CN101509782B (en) | 2009-03-09 | 2009-03-09 | Small-sized ground marker capturing and positioning method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN101509782B (en) |
Cited By (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102829785A (en) * | 2012-08-30 | 2012-12-19 | 中国人民解放军国防科学技术大学 | Air vehicle full-parameter navigation method based on sequence image and reference image matching |
| CN102968795A (en) * | 2012-12-03 | 2013-03-13 | 哈尔滨工业大学 | Meteor crater mismatching determination method based on ratio of shaded area to external-contour area |
| CN103456026A (en) * | 2013-07-29 | 2013-12-18 | 华中科技大学 | Method for detecting ground moving object under road landmark constraints |
| CN103699883A (en) * | 2013-12-17 | 2014-04-02 | 华中科技大学 | Method for recognizing and positioning building by village landmark group |
| CN104145173A (en) * | 2012-03-02 | 2014-11-12 | 高通股份有限公司 | Visual ocr for positioning |
| CN104655135A (en) * | 2015-02-09 | 2015-05-27 | 南京邮电大学 | Landmark-recognition-based aircraft visual navigation method |
| CN105913082A (en) * | 2016-04-08 | 2016-08-31 | 北京邦焜威讯网络技术有限公司 | Method and system for classifying objects in image |
| CN107123188A (en) * | 2016-12-20 | 2017-09-01 | 北京联合众为科技发展有限公司 | Ticket of hindering based on template matching algorithm and edge feature is recognized and localization method |
| CN107274441A (en) * | 2017-06-14 | 2017-10-20 | 中国科学院遥感与数字地球研究所 | The wave band calibration method and system of a kind of high spectrum image |
| CN107958454A (en) * | 2017-12-04 | 2018-04-24 | 江苏维普光电科技有限公司 | The fast matching method and system of reticle pattern picture are carried out based on precision surface plate |
| WO2018120735A1 (en) * | 2016-12-29 | 2018-07-05 | 东莞前沿技术研究院 | Locating method and apparatus |
| CN109791414A (en) * | 2016-09-26 | 2019-05-21 | 深圳市大疆创新科技有限公司 | The method and system that view-based access control model lands |
| CN110503679A (en) * | 2019-08-29 | 2019-11-26 | 四川轻化工大学 | A kind of preparation of infrared reference figure and evaluation method |
| CN113160076A (en) * | 2021-04-06 | 2021-07-23 | 中航航空电子有限公司 | Ground object infrared target acquisition method based on target edge neighborhood information |
-
2009
- 2009-03-09 CN CN2009100610938A patent/CN101509782B/en not_active Expired - Fee Related
Cited By (23)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104145173A (en) * | 2012-03-02 | 2014-11-12 | 高通股份有限公司 | Visual ocr for positioning |
| CN102829785B (en) * | 2012-08-30 | 2014-12-31 | 中国人民解放军国防科学技术大学 | Air vehicle full-parameter navigation method based on sequence image and reference image matching |
| CN102829785A (en) * | 2012-08-30 | 2012-12-19 | 中国人民解放军国防科学技术大学 | Air vehicle full-parameter navigation method based on sequence image and reference image matching |
| CN102968795A (en) * | 2012-12-03 | 2013-03-13 | 哈尔滨工业大学 | Meteor crater mismatching determination method based on ratio of shaded area to external-contour area |
| CN102968795B (en) * | 2012-12-03 | 2015-01-21 | 哈尔滨工业大学 | Meteor crater mismatching determination method based on ratio of shaded area to external-contour area |
| CN103456026B (en) * | 2013-07-29 | 2016-11-16 | 华中科技大学 | A kind of Ground moving target detection method under highway terrestrial reference constraint |
| CN103456026A (en) * | 2013-07-29 | 2013-12-18 | 华中科技大学 | Method for detecting ground moving object under road landmark constraints |
| CN103699883A (en) * | 2013-12-17 | 2014-04-02 | 华中科技大学 | Method for recognizing and positioning building by village landmark group |
| CN103699883B (en) * | 2013-12-17 | 2016-04-13 | 华中科技大学 | A kind of method utilizing village landmark group identification to locate buildings |
| CN104655135A (en) * | 2015-02-09 | 2015-05-27 | 南京邮电大学 | Landmark-recognition-based aircraft visual navigation method |
| CN104655135B (en) * | 2015-02-09 | 2018-03-06 | 南京邮电大学 | A kind of aircraft visual navigation method based on terrestrial reference identification |
| CN105913082A (en) * | 2016-04-08 | 2016-08-31 | 北京邦焜威讯网络技术有限公司 | Method and system for classifying objects in image |
| CN105913082B (en) * | 2016-04-08 | 2020-11-27 | 北京邦视科技有限公司 | Method and system for classifying targets in image |
| US11604479B2 (en) | 2016-09-26 | 2023-03-14 | SZ DJI Technology Co., Ltd. | Methods and system for vision-based landing |
| CN109791414A (en) * | 2016-09-26 | 2019-05-21 | 深圳市大疆创新科技有限公司 | The method and system that view-based access control model lands |
| CN107123188A (en) * | 2016-12-20 | 2017-09-01 | 北京联合众为科技发展有限公司 | Ticket of hindering based on template matching algorithm and edge feature is recognized and localization method |
| US11015929B2 (en) | 2016-12-29 | 2021-05-25 | Dongguan Frontier Technology Institute | Positioning method and apparatus |
| WO2018120735A1 (en) * | 2016-12-29 | 2018-07-05 | 东莞前沿技术研究院 | Locating method and apparatus |
| CN107274441B (en) * | 2017-06-14 | 2020-07-03 | 中国科学院遥感与数字地球研究所 | Wave band calibration method and system for hyperspectral image |
| CN107274441A (en) * | 2017-06-14 | 2017-10-20 | 中国科学院遥感与数字地球研究所 | The wave band calibration method and system of a kind of high spectrum image |
| CN107958454A (en) * | 2017-12-04 | 2018-04-24 | 江苏维普光电科技有限公司 | The fast matching method and system of reticle pattern picture are carried out based on precision surface plate |
| CN110503679A (en) * | 2019-08-29 | 2019-11-26 | 四川轻化工大学 | A kind of preparation of infrared reference figure and evaluation method |
| CN113160076A (en) * | 2021-04-06 | 2021-07-23 | 中航航空电子有限公司 | Ground object infrared target acquisition method based on target edge neighborhood information |
Also Published As
| Publication number | Publication date |
|---|---|
| CN101509782B (en) | 2011-02-09 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN101509782B (en) | Small-sized ground marker capturing and positioning method | |
| Tuermer et al. | Airborne vehicle detection in dense urban areas using HoG features and disparity maps | |
| CN101532841B (en) | Method for navigating and positioning aerocraft based on landmark capturing and tracking | |
| CN103954283B (en) | Inertia integrated navigation method based on scene matching aided navigation/vision mileage | |
| CN103149939B (en) | A kind of unmanned plane dynamic target tracking of view-based access control model and localization method | |
| CN103411609B (en) | A kind of aircraft return route planing method based on online composition | |
| CN103175524B (en) | A kind of position of aircraft without view-based access control model under marking environment and attitude determination method | |
| CN113537018B (en) | Water and soil conservation monitoring method based on multi-time-phase satellite remote sensing and unmanned aerial vehicle technology | |
| Qu et al. | Landmark based localization in urban environment | |
| CN107229063A (en) | A kind of pilotless automobile navigation and positioning accuracy antidote merged based on GNSS and visual odometry | |
| CN104268935A (en) | Feature-based airborne laser point cloud and image data fusion system and method | |
| CN106651953A (en) | Vehicle position and gesture estimation method based on traffic sign | |
| CN103822635A (en) | Visual information based real-time calculation method of spatial position of flying unmanned aircraft | |
| CN104536009A (en) | Laser infrared composite ground building recognition and navigation method | |
| CN101620671A (en) | Method for indirectly positioning and identifying three-dimensional buildings by using riverway landmarks | |
| CN108759823A (en) | The positioning of low speed automatic driving vehicle and method for correcting error in particular link based on images match | |
| Dumble et al. | Airborne vision-aided navigation using road intersection features | |
| Chellappa et al. | On the positioning of multisensor imagery for exploitation and target recognition | |
| CN112613437B (en) | Method for identifying illegal buildings | |
| Al-Rawabdeh et al. | A robust registration algorithm for point clouds from UAV images for change detection | |
| Del Pizzo et al. | Reliable vessel attitude estimation by wide angle camera | |
| CN110927765B (en) | Laser radar and satellite navigation fused target online positioning method | |
| CN117470259A (en) | Primary and secondary type space-ground cooperative multi-sensor fusion three-dimensional map building system | |
| Jende et al. | Fully automatic feature-based registration of mobile mapping and aerial nadir images for enabling the adjustment of mobile platform locations in GNSS-denied urban environments | |
| CN103473787A (en) | On-bridge-moving-object detection method based on space geometry relation |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C53 | Correction of patent of invention or patent application | ||
| CB03 | Change of inventor or designer information |
Inventor after: Zhang Tianxu Inventor after: Yang Xiaoyu Inventor after: Li Yaobo Inventor after: Yang Fengjun Inventor after: Sang Nong Inventor after: Wang Ze Inventor after: Gao Shiying Inventor after: Li Mi Inventor after: Yang Weidong Inventor after: Wan Meijun Inventor after: Li Yun Inventor after: Zhong Sheng Inventor after: Yan Luxin Inventor after: He Yonggang Inventor before: Zhang Tianxu Inventor before: Li Yaobo Inventor before: Yang Fengjun Inventor before: Sang Nong Inventor before: Wang Ze Inventor before: Li Mi Inventor before: Yang Weidong Inventor before: Wan Meijun Inventor before: Li Yun Inventor before: Zhong Sheng Inventor before: Yan Luxin Inventor before: He Yonggang Inventor before: Yang Xiaoyu |
|
| COR | Change of bibliographic data |
Free format text: CORRECT: INVENTOR; FROM: ZHANG TIANXU LI MI YANG WEIDONG WAN MEIJUN LI YUN ZHONG SHENG YAN LUXIN HE YONGGANG YANG XIAOYU LI YAOBO YANG FENGJUN SANG NONG WANG ZE TO: ZHANG TIANXU GAO SHIYING LI MI YANG WEIDONG WAN MEIJUN LI YUN ZHONG SHENG YAN LUXIN HE YONGGANG YANG XIAOYU LI YAOBO YANG FENGJUN SANG NONG WANG ZE |
|
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20110209 |
|
| CF01 | Termination of patent right due to non-payment of annual fee |