CN102853835A - Scale invariant feature transform-based unmanned aerial vehicle scene matching positioning method - Google Patents

Abstract

The invention relates to a scale invariant feature transform-based unmanned aerial vehicle scene matching positioning method. The scale invariant feature transform-based unmanned aerial vehicle scene matching positioning method is characterized by comprising the following steps of 1, extracting feature description vectors of an image of a matching target and a front-lower view of an unmanned aerial vehicle by a scale invariant feature transform algorithm, 2, determining if the front-lower view in the frame and the image of the matching target are matching or not, and 3, if the front-lower view and the image of the matching target are matching, recording coordinates of a matching point in a satellite map comprising the image of the matching target and the matching target, in the front-lower view of the unmanned aerial vehicle, calculating current position coordinates of the unmanned aerial vehicle in the satellite map according to the coordinates of the matching point and carrying out positioning of the unmanned aerial vehicle, and if the front-lower view and the image of the matching target are not matching, reading an unmanned aerial vehicle front-lower view in the next frame and sequentially carrying out matching. The scale invariant feature transform-based unmanned aerial vehicle scene matching positioning method realizes accurate matching of a front-lower view of an unmanned aerial vehicle and a matching target in a satellite map, determination of current position coordinates of the unmanned aerial vehicle according to a built unmanned aerial vehicle front-lower view model, and positioning of the unmanned aerial vehicle.

Description

Unmanned vehicle scene matching aided navigation localization method based on the conversion of yardstick invariant features

Technical field

The present invention relates to a kind of unmanned vehicle scene matching aided navigation localization method based on the conversion of yardstick invariant features, be applied in the unmanned vehicle scene matching aided navigation location.

Background technology

Develop rapidly along with aeronautical technology, unmanned vehicle and correlation technique thereof have become the emphasis that various countries are competitively studied, and unmanned vehicle is with its stronger maneuverability, less weight, better aerodynamic quality, lower cost, better adaptive capacity to environment.At present, American-European countries is except the unmanned vehicle of usefulness, and also in the novel unmanned vehicle technology of research, China also is being engaged in the research of correlation technique.

In numerous technology that relate to unmanned vehicle, the unmanned vehicle location technology is an extremely crucial technology, and this technology is mainly used in unmanned vehicle is implemented accurately location and independent navigation, is the assurance that unmanned vehicle is finished the work.Existing unmanned vehicle location technology mainly is to rely on GPS, but, the estimated accuracy of GPS is relevant with the quality that the number of satellite that participates in the location and signal receive equipment, and signal transduction process easily is subject to radio interference, causes signal errors to enlarge.In addition, complex environment also requires the unmanned vehicle location technology to carry out the Integrated using of multiple location technology, and unmanned vehicle scene matching aided navigation location just is based on these and requires to be born.So-called unmanned vehicle scene matching aided navigation location, refer to utilize on the unmanned vehicle camera collection to realtime graphic mate with satellite image or the image that is stored in advance in the unmanned vehicle, obtain positional information.This technology can be used in the situation of GPS inefficacy, the effective location mode of GPS as an alternative, particularly GPS is subjected to external control, so development low cost, high precision, anti-interference or even location technology alternative GPS seem extremely important.

Chinese scholars has also been carried out the research of some aspects, scene matching aided navigation location, mates to reduce the method for matching error as adopting continuous multiple frames, and the method can improve coupling usefulness, but owing to being the multiframe coupling, can increase coupling consuming time; Also have class methods to mate based on single frames, mate consuming time few, but easily cause matching error, in this case, the method such as image characteristic extracting method and similarity measurement accurately sought just becomes the key point of single frames coupling.So, study that a kind of accurately Scene matching method based is significant to unmanned vehicle scene matching aided navigation location fast again.

Summary of the invention

The technical matters that solves

For fear of the deficiencies in the prior art part, the present invention proposes a kind of unmanned vehicle scene matching aided navigation localization method based on the conversion of yardstick invariant features, can substitute or agps system carries out the method for unmanned vehicle location, reduce unmanned vehicle to the dependence of GPS navigation.

Technical scheme

A kind of unmanned vehicle scene matching aided navigation localization method based on the conversion of yardstick invariant features is characterized in that step is as follows:

Step 1 is extracted the feature description vectors F of coupling target image _i: utilize yardstick invariant features transform method to extract coupling target image I ₁Feature description vectors F _i, i=1,2 ... m; Wherein: F _iPresentation video I ₁I feature description vectors; M presentation video I ₁The number of middle feature description vectors and m ∈ (0,10000);

Step 2 is extracted the feature description vectors F of the front lower view of unmanned vehicle _j: utilize yardstick invariant features transform method to extract the front lower view I that unmanned vehicle is taken ₂Feature description vectors F _j, j=1,2 ... n; Wherein: F _jPresentation video I ₂J feature description vectors; N presentation video I ₂The number of middle feature description vectors and n ∈ (0,10000);

Step 3 is searched the matching characteristic point: at F _iIn find out and F _jThe unique point that Euclidean distance is nearest, when 2 Euclidean distances during less than threshold value Q, then this is F _jMatch point; When the unique point number of coupling reached threshold value W, the match is successful with the coupling target image for the front lower view of this unmanned vehicle; Then respectively recording feature point coordinate (fx, fy) and (rex, rey) in the satellite image at coupling target image and this coupling target image place, if it fails to match, repeating step 2;

Step 4 is calculated unmanned vehicle and is now located the position: utilize the position coordinates (fx on coupling target image and the satellite image, fy) and (rex, rey) calculate unmanned vehicle current position coordinates (lxresult, lyresult), concrete steps are as follows:

Step a: utilize $\mathrm{\&Delta;x}=\left\{\begin{array}{cc}h\&times;\tan (c+b+e)& \mathrm{Lw}-\mathrm{fy}>0\\ h\&times;\tan (c+b-e)& \mathrm{Lw}-\mathrm{fy}<0\\ h\&times;\tan (c+b)& \mathrm{Lw}-\mathrm{fy}=0\end{array}\right.$ Calculate the difference Δ x of the ordinate of match point coordinate (fx, fy) and unmanned vehicle position; Wherein: h represents the height parameter of unmanned vehicle; C=pi/2-a-2b; A represents the unmanned vehicle angle of depression; B represent the unmanned vehicle visual angle half; E=arctan (| Lw-fy|/Lm); || expression takes absolute value; Lw=[h/tan (a)-h * tan (c)] * sin (a)/sin (pi/2+b); Lm=Lw/2/tan (a);

Step b: utilize Δ y=|fx-LL| * Lc/Lcw to calculate the difference Δ y of the horizontal ordinate of match point coordinate (fx, fy) and unmanned vehicle position; Wherein: LL represents to mate half of target image width, LL=L/2; $\mathrm{Lc}=\left\{\begin{array}{cc}h/\cos (b+c+e)& \mathrm{Lw}-\mathrm{fy}>0\\ h/\cos (b+c-e)& \mathrm{Lw}-\mathrm{fy}<0\\ h/\cos (b+c)& \mathrm{Lw}-\mathrm{fy}=0\end{array}\right.;$ Lcw＝Lm/cos(e)；

Step c: adopt following formula to calculate, obtain unmanned vehicle current position coordinates (lxresult, lyresult)

$\mathrm{lxresult}=\left\{\begin{array}{cc}\mathrm{\&Delta;x}\&times;\sin \mathrm{\&theta;}+\mathrm{\&Delta;y}\&times;\cos \mathrm{\&theta;}+\mathrm{rex}& \mathrm{fx}-\mathrm{LL}<0\\ \mathrm{\&Delta;x}\&times;\sin \mathrm{\&theta;}-\mathrm{\&Delta;y}\&times;\cos \mathrm{\&theta;}+\mathrm{rex}& \mathrm{fx}-\mathrm{LL}>0\\ \mathrm{\&Delta;x}\&times;\sin \mathrm{\&theta;}+\mathrm{rex}& \mathrm{fx}-\mathrm{LL}=0\end{array}\right.$

$\mathrm{lyresult}=\left\{\begin{array}{cc}\mathrm{\&Delta;x}\&times;\cos \mathrm{\&theta;}-\mathrm{\&Delta;y}\&times;\sin \mathrm{\&theta;}+\mathrm{rex}& \mathrm{fx}-\mathrm{LL}<0\\ \mathrm{\&Delta;x}\&times;\cos \mathrm{\&theta;}+\mathrm{\&Delta;y}\&times;\sin \mathrm{\&theta;}+\mathrm{rex}& \mathrm{fx}-\mathrm{LL}>0\\ \mathrm{\&Delta;x}\&times;\cos \mathrm{\&theta;}+\mathrm{rex}& \mathrm{fx}-\mathrm{LL}=0\end{array}\right.$

Wherein: lxresult represents the horizontal ordinate of unmanned vehicle on satellite mapping; Lyresult represents the ordinate of unmanned vehicle on satellite mapping; θ is the angle of unmanned vehicle heading and direct north.

Described threshold value Q ∈ (0,1).

Described threshold value W ∈ (1,10000).

Beneficial effect

A kind of unmanned vehicle scene matching aided navigation localization method based on the conversion of yardstick invariant features that the present invention proposes, utilize yardstick invariant features mapping algorithm to extract the front lower Characteristic of Image description vectors of looking of unmanned vehicle, and with satellite image in coupling clarification of objective description vectors mate, in the front lower view model of unmanned vehicle, calculate the unmanned vehicle position coordinates according to matching result, and compare with the pre-set flight coordinate of unmanned vehicle, judge whether flight path is correct.

The unmanned vehicle scene matching aided navigation localization method based on the conversion of yardstick invariant features that the present invention proposes, utilize in the satellite mapping coupling target and the front lower view of unmanned vehicle to carry out the unmanned vehicle location, can be used as substituting and aid of GPS, its bearing accuracy is high, and speed is fast.

Description of drawings

Fig. 1: the basic flow sheet of the inventive method;

Fig. 2: coupling target figure;

Fig. 3: the front lower view model of unmanned vehicle;

Fig. 4: unmanned vehicle vertical view;

Fig. 5: unmanned vehicle side view;

Fig. 6: unmanned vehicle coupling positioning system interface.

Embodiment

Now in conjunction with the embodiments, the invention will be further described for accompanying drawing:

The hardware environment that is used for implementing is: AMD Athlon 64 * 25000+ computing machine, 2GB internal memory, 256M video card, the software environment of operation is: Visual Studio 2008 and Windows 7.We have realized the positioning system that the present invention proposes with Visual Studio 2008 softwares.

Process flow diagram of the present invention as shown in Figure 1, implementation is as follows:

1 extracts coupling clarification of objective description vectors F _i:

Utilize yardstick invariant features transform method to extract the feature description vectors that mates on the target image, the coupling target image as shown in Figure 2.Concrete steps are as follows:

At first to coupling target image I ₁Carry out Gaussian smoothing, wherein choose σ _n=0.5, obtain image , choose different σ=σ ₀2 ^O+s/SWith

Do convolution and formed an image pyramid GSS _σ, s=0 wherein ... S-1, o=0 ... O-1, S=3, O=min (log ₂Row, log ₂Col), σ ₀The number of pixel on the vertical direction of=1.5, row presentation video, the number of pixel on the horizontal direction of col presentation video.Then to adjacent GSS _σAsk difference to obtain DOG _σ, for DOG _σEach pixel respectively with a upper yardstick corresponding pixel points and around eight neighborhood territory pixels point, eight neighborhood territory pixels point around the current yardstick, and next yardstick corresponding pixel points and around eight neighborhood territory pixels point make comparisons, if this pixel is minimal value or maximum point, then this pixel is the image significant point, zone take σ as radius around it is marking area, can obtain thus the coordinate X of a series of image significant point, and its corresponding σ is its corresponding scale size λ.For each image significant point, use

Gradient image and gaussian kernel do convolution and obtain gradient image

, σ wherein _G=1.5 σ, and compute gradient image

In with the direction histogram in the marking area of significant point X, wherein the crest meter in each direction histogram interval adds up to this direction zone inside gradient at last, get the interval number L=36 of direction histogram, choose the direction zone that amplitude surpasses its maximal value 80% from direction histogram, be defined as this characteristic area principal direction γ, if any a plurality of directions zone, then there are a plurality of principal direction γ in this characteristic area.Get at last the marking area of image significant point X, be divided into 16 zones by principal direction and vertical direction thereof, in each zonule, add up respectively direction histogram, wherein the crest meter in each direction histogram interval adds up to this direction zone inside gradient assignment at last, get the interval number L=8 of direction histogram, and with the amplitude quantization of each direction histogram between [0,255], then obtain the description vectors F of one 128 dimension _i, i=1,2 ... m.Wherein: F _iPresentation video I ₁I feature description vectors; M presentation video I ₁The number of middle feature description vectors, m ∈ (0,10000).

2 extract the feature description vectors F of the front lower view of unmanned vehicle _j:

The same with the feature description vectors that extracts the coupling target image in the step 1, after the unmanned vehicle camera photographs front lower view, will carry out the feature description vectors to this image and extract, in order to mate with the feature description vectors that mates target image, concrete steps are as follows:

Front lower view I to the unmanned vehicle shooting ₂Carry out Gaussian smoothing, wherein choose σ _n=0.5, obtain image

, choose different σ=σ ₀2 ^O+s/SWith

Do convolution and formed an image pyramid GSS _σ, s=0 wherein ... S-1, o=0 ... O-1, S=3, O=min (log ₂Row, log ₂Col), σ ₀The number of pixel on the vertical direction of=1.5, row presentation video, the number of pixel on the horizontal direction of col presentation video.Then to adjacent GSS _σAsk difference to obtain DOG _σ, for DOG _σEach pixel respectively with a upper yardstick corresponding pixel points and around eight neighborhood territory pixels point, eight neighborhood territory pixels point around the current yardstick, and next yardstick corresponding pixel points and around eight neighborhood territory pixels point make comparisons, if this pixel is minimal value or maximum point, then this pixel is the image significant point, zone take σ as radius around it is marking area, can obtain thus the coordinate X of a series of image significant point, and its corresponding σ is its corresponding scale size λ.For each image significant point X, make Gradient image and gaussian kernel do convolution and obtain gradient image

, σ wherein _G=1.5 σ, and compute gradient image

In with the direction histogram in the marking area of significant point X, wherein the crest meter in each direction histogram interval adds up to this direction zone inside gradient at last, get the interval number L=36 of direction histogram, choose the direction zone that amplitude surpasses its maximal value 80% from direction histogram, be defined as this characteristic area principal direction γ, if any a plurality of directions zone, then there are a plurality of principal direction γ in this characteristic area.Get at last the marking area of image significant point X, be divided into 16 zones by principal direction and vertical direction thereof, in each zonule, add up respectively direction histogram, wherein the crest meter in each direction histogram interval adds up to this direction zone inside gradient assignment at last, get the interval number L=8 of direction histogram, and with the amplitude quantization of each direction histogram between [0,255], then obtain the description vectors F of one 128 dimension _j, j=1,2 ... n.Wherein: F _jPresentation video I ₂J feature description vectors; N presentation video I ₂The number of middle feature description vectors, n ∈ (0,10000).

3 search the matching characteristic point:

At F _iIn find out and F _jThe unique point that Euclidean distance is nearest, when 2 Euclidean distances during less than threshold value Q, Q ∈ (0,1), then this is F _jMatch point; When the unique point number of coupling reaches threshold value W, W ∈ (1,10000), the match is successful with the coupling target image for the front lower view of this unmanned vehicle, respectively recording feature point coordinate (fx, fy) and (rex, rey) in the satellite image at coupling target image and this coupling target image place, if it fails to match, repeating step 2;

4 calculate unmanned vehicle now locates coordinate:

Utilize the field angle b of unmanned vehicle camera, visual field width L, the height parameter h of angle of depression a and unmanned vehicle sets up the front lower view model of unmanned vehicle, as shown in Figure 3; Position coordinates (fx, fy) and (rex, the rey) of match point calculate unmanned vehicle current position coordinates (lxresult, lyresult) on this model utilization coupling target figure and the satellite mapping.

The coordinate Calculation formula of lxresult and lyresult is as follows:

$\mathrm{lxresult}=\left\{\begin{array}{cc}\mathrm{\&Delta;x}\&times;\sin \mathrm{\&theta;}+\mathrm{\&Delta;y}\&times;\cos \mathrm{\&theta;}+\mathrm{rex}& \mathrm{fx}-\mathrm{LL}<0\\ \mathrm{\&Delta;x}\&times;\sin \mathrm{\&theta;}-\mathrm{\&Delta;y}\&times;\cos \mathrm{\&theta;}+\mathrm{rex}& \mathrm{fx}-\mathrm{LL}>0\\ \mathrm{\&Delta;x}\&times;\sin \mathrm{\&theta;}+\mathrm{rex}& \mathrm{fx}-\mathrm{LL}=0\end{array}\right.$

$\mathrm{lyresult}=\left\{\begin{array}{cc}\mathrm{\&Delta;x}\&times;\cos \mathrm{\&theta;}-\mathrm{\&Delta;y}\&times;\sin \mathrm{\&theta;}+\mathrm{rex}& \mathrm{fx}-\mathrm{LL}<0\\ \mathrm{\&Delta;x}\&times;\cos \mathrm{\&theta;}+\mathrm{\&Delta;y}\&times;\sin \mathrm{\&theta;}+\mathrm{rex}& \mathrm{fx}-\mathrm{LL}>0\\ \mathrm{\&Delta;x}\&times;\cos \mathrm{\&theta;}+\mathrm{rex}& \mathrm{fx}-\mathrm{LL}=0\end{array}\right.$

Wherein, lxresult represents the horizontal ordinate of unmanned vehicle on satellite mapping; Lyresult represents the ordinate of unmanned vehicle on satellite mapping; Δ y represents horizontal ordinate poor of match point and unmanned vehicle position, as shown in Figure 4; Δ x represents ordinate poor of match point and unmanned vehicle position, as shown in Figure 4; θ is the angle of unmanned vehicle heading and direct north, as shown in Figure 4; Rex represents the horizontal ordinate position of match point in satellite mapping; Rey represents the ordinate position of match point in satellite mapping; Fx is illustrated in the horizontal ordinate position of match point among the coupling target figure, as shown in Figure 2; Fy is illustrated in the ordinate position of match point among the coupling target figure, as shown in Figure 2; LL represents to mate half of width of target figure, LL=L/2, as shown in Figure 2;

For Δ x, utilize $\mathrm{\&Delta;x}=\left\{\begin{array}{cc}h\&times;\tan (c+b+e)& \mathrm{Lw}-\mathrm{fy}>0\\ h\&times;\tan (c+b-e)& \mathrm{Lw}-\mathrm{fy}<0\\ h\&times;\tan (c+b)& \mathrm{Lw}-\mathrm{fy}=0\end{array}\right.$ Calculate;

Wherein, h represents the height parameter of unmanned vehicle; C=pi/2-a-2b; A represents the unmanned vehicle angle of depression, as shown in Figure 5; B represent the visual angle half, as shown in Figure 5; E=arctan (| Lw-fy|/Lm), as shown in Figure 5; Wherein, Lw=[h/tan (a)-h * tan (c)] * sin (a)/sin (pi/2+b); Lm=Lw/2/tan (a), as shown in Figure 5;

For Δ y, utilize Δ y=abs (fx-LL) * Lc/Lcw to calculate;

Wherein, $\mathrm{Lc}=\left\{\begin{array}{cc}h/\cos (b+c+e)& \mathrm{Lw}-\mathrm{fy}>0\\ h/\cos (b+c-e)& \mathrm{Lw}-\mathrm{fy}<0\\ h/\cos (b+c)& \mathrm{Lw}-\mathrm{fy}=0\end{array}\right.;$ Lcw＝Lm/cos(e)；

The Euclidean distance of the flight coordinate that calculates this position coordinates and plan in advance judges whether flight path is correct, shows simultaneously position and the track that unmanned vehicle is now located in matching system, is convenient to intuitive judgment, as shown in Figure 6.

The unmanned vehicle scene matching aided navigation positioning system of utilizing this paper to set up can be mated the coupling target in the front lower view of unmanned vehicle and the satellite mapping accurately, and judge position and the flight path that unmanned vehicle is now located according to the front lower view model of the unmanned vehicle of setting up, the result shows, this system can judge the position coordinates of unmanned vehicle exactly in satellite mapping, be no more than 3 pixels with predefined grid deviation.

Claims (3)

1. unmanned vehicle scene matching aided navigation localization method based on the conversion of yardstick invariant features is characterized in that step is as follows:

Step 1 is extracted the feature description vectors F of coupling target image _i: utilize yardstick invariant features transform method to extract coupling target image I ₁Feature description vectors F _i, i=1,2 ... m; Wherein: F _iPresentation video I ₁I feature description vectors; M presentation video I ₁The number of middle feature description vectors and m ∈ (0,10000);

Step 2 is extracted the feature description vectors F of the front lower view of unmanned vehicle _j: utilize yardstick invariant features transform method to extract the front lower view I that unmanned vehicle is taken ₂Feature description vectors F _j, j=1,2 ... n; Wherein: F _jPresentation video I ₂J feature description vectors; N presentation video I ₂The number of middle feature description vectors and n ∈ (0,10000);

Step 3 is searched the matching characteristic point: at F _iIn find out and F _jThe unique point that Euclidean distance is nearest, when 2 Euclidean distances during less than threshold value Q, then this is F _jMatch point; When the unique point number of coupling reached threshold value W, the match is successful with the coupling target image for the front lower view of this unmanned vehicle; Then respectively recording feature point coordinate (fx, fy) and (rex, rey) in the satellite image at coupling target image and this coupling target image place, if it fails to match, repeating step 2;

Step 4 is calculated unmanned vehicle and is now located the position: utilize the position coordinates (fx on coupling target image and the satellite image, fy) and (rex, rey) calculate unmanned vehicle current position coordinates (lxresult, lyresult), concrete steps are as follows:

Step a: utilize $\mathrm{\&Delta;x}=\left\{\begin{array}{cc}h\&times;\tan (c+b+e)& \mathrm{Lw}-\mathrm{fy}>0\\ h\&times;\tan (c+b-e)& \mathrm{Lw}-\mathrm{fy}<0\\ h\&times;\tan (c+b)& \mathrm{Lw}-\mathrm{fy}=0\end{array}\right.$ Calculate the difference Δ x of the ordinate of match point coordinate (fx, fy) and unmanned vehicle position; Wherein: h represents the height parameter of unmanned vehicle; C=pi/2-a-2b; A represents the unmanned vehicle angle of depression; B represent the unmanned vehicle visual angle half; E=arctan (| Lw-fy|/Lm); || expression takes absolute value; Lw=[h/tan (a)-h * tan (c)] * sin (a)/sin (pi/2+b); Lm=Lw/2/tan (a);

Step b: utilize Δ y=|fx-LL| * Lc/Lcw to calculate the difference Δ y of the horizontal ordinate of match point coordinate (fx, fy) and unmanned vehicle position; Wherein: LL represents to mate half of target image width, LL=L/2; $\mathrm{Lc}=\left\{\begin{array}{cc}h/\cos (b+c+e)& \mathrm{Lw}-\mathrm{fy}>0\\ h/\cos (b+c-e)& \mathrm{Lw}-\mathrm{fy}<0\\ h/\cos (b+c)& \mathrm{Lw}-\mathrm{fy}=0\end{array}\right.;$ Lcw＝Lm/cos(e)；

Step c: adopt following formula to calculate, obtain unmanned vehicle current position coordinates (lxresult, lyresult)

$\mathrm{lxresult}=\left\{\begin{array}{cc}\mathrm{\&Delta;x}\&times;\sin \mathrm{\&theta;}+\mathrm{\&Delta;y}\&times;\cos \mathrm{\&theta;}+\mathrm{rex}& \mathrm{fx}-\mathrm{LL}<0\\ \mathrm{\&Delta;x}\&times;\sin \mathrm{\&theta;}-\mathrm{\&Delta;y}\&times;\cos \mathrm{\&theta;}+\mathrm{rex}& \mathrm{fx}-\mathrm{LL}>0\\ \mathrm{\&Delta;x}\&times;\sin \mathrm{\&theta;}+\mathrm{rex}& \mathrm{fx}-\mathrm{LL}=0\end{array}\right.$

$\mathrm{lyresult}=\left\{\begin{array}{cc}\mathrm{\&Delta;x}\&times;\cos \mathrm{\&theta;}-\mathrm{\&Delta;y}\&times;\sin \mathrm{\&theta;}+\mathrm{rex}& \mathrm{fx}-\mathrm{LL}<0\\ \mathrm{\&Delta;x}\&times;\cos \mathrm{\&theta;}+\mathrm{\&Delta;y}\&times;\sin \mathrm{\&theta;}+\mathrm{rex}& \mathrm{fx}-\mathrm{LL}>0\\ \mathrm{\&Delta;x}\&times;\cos \mathrm{\&theta;}+\mathrm{rex}& \mathrm{fx}-\mathrm{LL}=0\end{array}\right.$

Wherein: lxresult represents the horizontal ordinate of unmanned vehicle on satellite mapping; Lyresult represents the ordinate of unmanned vehicle on satellite mapping; θ is the angle of unmanned vehicle heading and direct north.

2. described unmanned vehicle scene matching aided navigation localization method based on the conversion of yardstick invariant features according to claim 1 is characterized in that: described threshold value Q ∈ (0,1).

3. described unmanned vehicle scene matching aided navigation localization method based on the conversion of yardstick invariant features according to claim 1 is characterized in that: described threshold value W ∈ (1,10000).

Images