CN101840517B - Control point image database matching method based on image registration - Google Patents

Abstract

The invention relates to a control point image database matching method based on image registration, belonging to the field of surveying science and technique. First, the general range of regions to be matched in images to be matched is determined on the basis of control point image geographical coordinate information, and geometric distortion correction is conducted to the regions to be matched to obtain regions to be registered. Then, step matching strategies are adopted for matching, characteristic points are matched through a relaxation method, image registration is conducted to control point images and the regions to be registered, and the information of the whole control point images is used to constrain the matching of the control point images. In registration results, the matching results of the characteristic points closest to the centers of the control point images are used as final matching results. By adopting the method for matching the control points in the control point image database, good matching success rate can be obtained.

Description

A kind of image database for control point matching process based on Image registration

Technical field

The invention belongs to Surveying Science and Technology field, relate to a kind of image database for control point matching process based on Image registration.The method, based on reference mark image geographic coordinate information, is eliminated image geometry distortion; Adopt classification plan rate, the method for the lax coupling of unique point, carries out Image registration; Finally utilize distance condition constraint to obtain final matching results.The feature that the method has is simple and practical, matching probability is high.

Background technology

Geometric correction is the basic process that image is processed, and no matter is collinearity equation method or polynomial method, all needs to utilize ground control point to resolve transition matrix to realize geometric correction.Inefficiency is chosen at traditional reference mark, inorganizable and management.Therefore, set up image database for control point just very necessary to realize automatically choosing of reference mark.

Zhang Jixian etc. have early studied principle and the method for setting up large-scale reference mark database, have introduced how from database, to extract reference mark and carry out relevant matches.The people such as Chen Qihao have studied the inquiry mode at design, institutional framework and the reference mark of image database for control point.Chen Gong etc. have proposed to utilize image database for control point to carry out image Auto-matching, then according to related coefficient, evaluate the reliability of match point, and adopt some position prediction and pyramid strategy to improve search speed, considered the rotational deformation of image blocks, by introducing least square coupling, made registration reach subpixel accuracy.The cooperation of the France geographical slip-stick artist of institute cloth Mr. Lan HeSEP company, digital acquisition reference mark sub-image and the method for setting up Image Database in VIPS system, have been developed, by map, choose reference mark image, and carry out geographic coordinate measurement and obtain coordinate information, the coordinate range of the image that then can correct according to need recalls corresponding reference mark sub-image by automatically relevant, finds the position at reference mark on new satellite image.

Although the reference mark based on image database for control point is automatically chosen and has been carried out many-sided research, in actual applications Shortcomings still.The design of most of image database for control point does not have to consider to have the image of rotation and dimensional variation, lacks the adaptability to different scale and orientation image; Image Matching strategy based on single-point (part) only depends on the image information at isolated point place, is difficult for eliminating many couplings or false matching problem; Tradition correlation coefficient process cannot be applied to the inconsistent image of radiation characteristic or some position.

Summary of the invention

Problem to be solved by this invention is: a kind of image database for control point matching process based on Image registration is provided.First the method determines the approximate range in the region to be matched in image to be matched based on reference mark image geographic coordinate information, and treats matching area and carry out geometric distortion correction, obtains region subject to registration.Then adopt classification matching strategy, unique point relaxation method coupling, carries out Image registration to reference mark image and region subject to registration, utilizes the information of view picture reference mark image to retrain the coupling of reference mark image.In registration results, image center immediate Feature Points Matching result in chosen distance reference mark is as final matching results.Use the method to carry out the reference mark coupling in image database for control point, obtained and be matched to preferably power.

An image database for control point matching process based on Image registration, is characterized in that, comprises the following steps:

Step 1, take reference mark image as eliminate the geometric distortion of image to be matched with reference to image by geometric distortion cancellation module, obtains the region subject to registration of pyramid image coupling;

Step 2, the region subject to registration after by pyramid image matching module, reference mark image being completed with step 1 carries out that classification is lax mates;

Step 3, the matching result point by matching result acquisition module after step 2 completes concentrates image center nearest matching result in chosen distance reference mark as final matching results.

At above-mentioned a kind of image database for control point matching process based on Image registration, in described step 1, concrete operation steps is as follows:

Step 1.1, parameter acquiring unit reads image coordinate and resolves function and elevation datum: parameter acquiring unit obtains reference mark image sensor imaging geometry model coordinate normal solution function f by image parameters _c, image sensor imaging geometry model coordinate normal solution function f to be matched _p, reference mark image sensor imaging geometry model coordinate counter separate function f ' _c, image sensor imaging geometry model coordinate to be matched counter separate function f ' _p, and image dispersed elevation value Z to be matched _p, make Z _pfor elevation datum.

Step 1.2 is by the image coordinate (x of four angle points of region acquiring unit Reading Control Point image subject to registration _ci, y _ci), the region P in four angle points besieged city _c; By region P _cexpand k pixel, obtain a region P ' _c, be used for calculating the regional extent to be matched in image to be matched, P ' _cfour angular coordinates are (x _c1-k, y _c1-k), (x _c2+ k, y _c2-k), (x _c3-k, y _c3+ k), (x _c4+ k, y _c4+ k);

Step 1.3, by perspective transform parameter acquiring unit by the region P ' in step 1.2 _cfour angular coordinate (x _c1-k, y _c1-k), (x _c2+ k, y _c2-k), (x _c3-k, y _c3+ k), (x _c4+ k, y _c4+ k) the height value benchmark face amount Z and in step 1.1 _pthe geometric model coordinate normal solution function f of reference mark image in substitution step 1.1 _cground coordinate (the X that four angular coordinates of middle calculating are corresponding _pi, Y _pi), i.e. (X _pi, Y _pi)=f _c(x _ci, y _ci, Z _p), obtain the ground region P covering by after the expansion of reference mark image;

By the geometric model coordinate of four angle point ground coordinate substitution images to be matched of the earth coverage area territory P trying to achieve counter separate function f ' _pcalculate the image coordinate of their correspondences on image to be matched, (x _pi, y _pi)=f ' _p(X _pi, X _pi, Z _r), the region to be matched described in above-mentioned steps 1.2 is (x _pi, y _pi) region that surrounds, then by (x _c1-k, y _c1-k), (x _c2+ k, y _c2-k), (x _c3-k, y _c3+ k), (x _c4+ k, y _c4+ k) and (x _pi, y _pi) substitution perspective transform formula

$\begin{array}{c}{x}^{\′}=\frac{l_{1}x+l_{2}y+l_3}{l_{7}x+l_{8}y+1}\\ y^{\′}=\frac{l_{4}x+l_{5}y+l_6}{l_{7}x+l_{8}y+1}\end{array},$ Wherein $\left|\begin{array}{ccc}{l}_1& l_2& l_3\\ l_4& l_5& l_6\\ l_7& l_8& 1\end{array}\right|\≠0$

Obtain l ₁l l ₈8 perspective transform parameters, i=1 wherein, 2,3,4;

Step 1.4, by eight perspective transform parameters that obtain in step 1.3, to overlay area, ground P, the respective regions on image to be matched carries out perspective transform to region subject to registration acquiring unit, to non-integer point, adopt B-spline function method to resample, eliminate geometric distortion, imagery zone after being converted, is the region subject to registration that pyramid image mates.

At above-mentioned a kind of image database for control point matching process based on Image registration, in described step 2, concrete steps are as follows:

Step 2.1, obtains pyramid image matching parameter by matching parameter acquiring unit, and pyramid image matching parameter comprises scaling and pyramid progression, and concrete grammar is: pyramid image interlayer scaling P is set _s=1/2 ²ⁿ, n=1～3, first order pyramid image size S _min>=64 * 64 pixels, first order pyramid image is the minimum image of top layer resolution in pyramid structure; Sampling number

integral part, wherein L and H are length and the height of reference mark image; Pyramid image coupling progression is N _r+ 1;

Step 2.2, carries out pyramid first order coupling by first order matching unit in step 1 in the region subject to registration being completed;

Step 2.3, carries out pyramid second level coupling in the region subject to registration that second level matching unit completes in step 1;

Step 2.4, carries out pyramid N level coupling in the region subject to registration that N level matching unit completes in step 1, method, with in step 2.3, is carried out pyramid image coupling step by step, until afterbody pyramid image;

Step 2.5, records unique point and its matching result that afterbody pyramid mates.

At above-mentioned a kind of image database for control point matching process based on Image registration, in described step 2.2, concrete steps are as follows:

Step 2.21, according to scaling

, to reference mark image and area resample subject to registration, generate reference mark image first order pyramid image and region subject to registration first order pyramid image;

Step 2.22, arranges match window W _m1size is length of side W _a1the square window of=3 or 5 or 7 or 9 or 11 pixels, hunting zone S _s=S _lpixel * S _hpixel, region subject to registration first order pyramid image is long is PL ₁pixel height is PH ₁pixel, reference mark image first order pyramid image is long is OL ₁pixel height is OH ₁pixel, S _l=PL ₁-OL ₁, S _h=PH ₁-OH ₁;

Step 2.23 arranges parallactic grid and reliability graticule mesh on the image first order pyramid image of reference mark, and parallactic grid and reliability graticule mesh form by square net, with coordinate (W _a1/ 2+1, W _a1/ 2+1) be starting point, the summit of square net is grid points, and the square net length of side is W _a1, square net number=(OL ₁/ W _a1) * (OH ₁/ W _a1), by each graticule mesh point value assignment of parallactic grid, be 0;

Step 2.24, is used Shen Jun operator extract minutiae on the first order pyramid image of reference mark image; And calculate the gradient information on raw video of each unique point: comprise Grad $\mathrm{grad}=\sqrt{{(g(x+1,y)-g(x,y))}^2+{(g(x,y+1)-g(x,y))}^2},$ Gradient direction angle gradDir=arctan ((g (x, y+1)-g (x, y))/(g (x+1, y)-g (x, y)))/π, wherein g (x, y) be the gray-scale value that image mid point (x, y) is located, judge whether gradient direction angle g _de> 0.7 radian or g _de< 0.3 radian is this unique point reliability value R _b=1, otherwise this unique point reliability value R _b=0;

Step 2.25, calculates the first candidate matches peak point;

Step 2.26, unique point adds reliability graticule mesh, that is: according to unique point coordinate, within the scope of each grid of reliability graticule mesh, reliability value R _bthe unique point of > 0 adds the net point array of this grid, and net point array is recorded in the coordinate of unique point within the scope of this grid;

Step 2.27, is used existing all reliability value R _bthe unique point of > 0 builds the Delaunay triangulation network;

Step 2.28 adopts relaxation method feature matching method to all reliability value R _bthe unique point of > 0 coupling that relaxes, the concrete grammar of lax coupling is: the first candidate matches peak point of again optimizing each unique point, and the parallax parameter (dx, dy) of optimizing unique point makes it equal the parallax parameter (md of the first candidate matches peak point _x, md _y); If the facies relationship numerical value of the first candidate matches peak point of unique point is greater than 0.75, make this unique point reliability value R _b=2;

Step 2.29, multinomial model is optimized the first candidate matches peak point of unique point and is preserved parallactic grid at the corresponding levels.

At above-mentioned a kind of image database for control point matching process based on Image registration, in described step 2.25, the concrete steps of calculated candidate match peak point are as follows:

Step 2.251, screening unique point: the unique point to extracting, judges whether reliability value R _b> 0, is to carry out process steps 2.253, otherwise carries out process steps 2.252;

Step 2.252, rejects this unique point, chooses next unique point, goes to step 2.51;

Step 2.253, calculates related coefficient, and concrete steps are as follows:

Step a, on the image first order pyramid image of reference mark centered by unique point coordinate (x, y), according to match window size W _m1, calculate target area, target area is and match window square of the same size that its upper left angle point is (x-W _a1/ 2, y-W _a1/ 2) bottom right angle point is (x+W _a1/ 2, y+W _a1/ 2);

Step b, on the first order pyramid image of region subject to registration centered by unique point coordinate (x, y), according to described search range S _s, calculate region of search, region of search is a rectangle, its upper left angle point is (x-S _l/ 2, y-S _h/ 2), bottom right angle point is (x+S _l/ 2, y+S _h/ 2);

Step c, in region of search, the position that overlaps with upper left, region of search angle point with target area starts, and moving object region calculates correlation coefficient ρ (c, r):

$\mathrm{\&rho;}(c,r)=\frac{\underset{i=x-w_{a1}/2}{\overset{w_{a1}}{\mathrm{\&Sigma;}}}\underset{j=y-w_{a1}/2}{\overset{w_{a1}}{\mathrm{\&Sigma;}}}(g_{i,j}-\stackrel{\&OverBar;}{g})(g_{i+r,j+c}^{\&prime;}-{\stackrel{\&OverBar;}{g}}^{\&prime;})}{\sqrt{\underset{i=x-w_{a1}/2}{\overset{w_a}{\mathrm{\&Sigma;}}}\underset{j=y-w_{a1}/2}{\overset{w_a}{\mathrm{\&Sigma;}}}{(g_{i,j}-\stackrel{\&OverBar;}{g})}^2\&times;{(g_{i+r,j+c}^{\&prime;}-{\stackrel{\&OverBar;}{g_{r,c}}}^{\&prime;})}^2}}$

Every movement is once a match window, records the facies relationship numerical value of each match window;

Row, column displacement when wherein c, r are moving object region on region of search, g _{i, j}for the gray-scale value of (i, j) point in target area,

for target area average gray value, g ' _{i+r, j+c}for the gray-scale value of (i+r, j+c) point in region of search,

for in the region of search that participates in while having moved displacement (c, r) on region of search calculating when target area with the average gray value in the big or small corresponding region of match window;

Step 2.254, calculates parallax parameter;

Matching result parallax parameter (the md of each match window _x, md _y) be:

md _x＝x _c-x，

md _y＝y _c-y，

Wherein, (x _c, y _c) be this match window center point coordinate, (x, y) is current unique point coordinate;

Related coefficient and parallax parameter are charged to matching result array BP _cnds, the matching result of each match window is described by facies relationship numerical value and parallax parameter;

Step 2.255, determines the first candidate matches peak point of unique point;

In region of search, the facies relationship numerical value that compares the match window of each match window and its eight neighborhood, judge whether this match window related coefficient is all greater than its eight neighborhood matching window-related coefficient, using this match window central point as candidate matches peak point, carry out process steps 2.56,, otherwise current unique point reliability value R _bbe made as 0, turn process steps 2.52;

Step 2.256, all candidate matches peak points according to related coefficient size sequence, the maximum first candidate matches peak point of classifying as;

Step 2.257, chooses next unique point, turns as step 2.251, complete if all unique points are all calculated, and enters above-mentioned steps 2.26.

At above-mentioned a kind of image database for control point matching process based on Image registration, in described step 2.29, the concrete steps of the first candidate matches peak point of multinomial model optimization unique point are as follows:

Step 2.291, digital simulation multinomial coefficient, the affine Transform Model using in calculating is as follows:

md _x＝a ₁+a ₂×xd+a ₃×yd，

md _y＝b ₁+b ₂×xd+b ₃×yd，

Wherein, md _x, md _yfor the parallax parameter of the first candidate matches peak point of unique point, xd=x-xc, yd=y-yc, the coordinate that x, y are unique point, xc, yc are parallactic grid center point coordinate; By all R in feature point set _bthe unique point substitution affine Transform Model of > 1, each feature point range one prescription journey, calculates a by least square method ₁, a ₂, a ₃, b ₁, b ₂, b ₃six affined transformation coefficients;

Step 2.292, optimize the first candidate matches peak point:

Steps A, calculated characteristics is put the deviate of the first candidate matches peak point, and concrete grammar is as follows:

Sign f=0 is upgraded in order; Judge whether current unique point R in feature point set _b> 1, otherwise goes to step F, is unique point substitution polynomial fitting to be calculated:

px＝a ₁+a ₂×xd+a ₃×yd，

py＝b ₁+b ₂×xd+b ₃×yd，

Wherein, px, the py parallax value for calculating in polynomial fitting, xd=x-xc, yd=y-yc, the coordinate that x, y are unique point, xc, yc are parallactic grid center point coordinate;

Calculation deviation value t again _x, t _y:

R _x=| px-md _x|, t _y=| py-md _y|; Wherein, md _x, md _yparallax parameter for the first candidate matches peak point of unique point;

Step B, judgment bias value: judgement, t _x> 12 and t _y> 12, are to carry out step C, proceed to if not step F;

Step C, calculates the deviation of next candidate matches peak point: judging whether next candidate matches peak point of this unique point exists, is to go to step D; Otherwise go to step E;

Step D, upgrades the first candidate matches peak point; Concrete grammar is: sign f=1 is upgraded in order; According to the parallax parameter of this candidate matches peak point, calculate new deviate t _xi, t _yi;

Judge whether t _xi< t _xand t _yi< t _y, otherwise rotor step C is to make t _x=t _xi, t _y=t _yi, using this candidate matches peak point as the first candidate matches peak point, its parallax parameter value is assigned to the parallax parameter (dx, dy) of unique point, and makes the reliability value R of this unique point _b=1, go to step C;

Step e, judges whether to upgrade sign f=1, is to proceed to step F; Otherwise make this unique point reliability value R _b=0, go to step F;

Step F, judging characteristic point concentrates whether there is next unique point, is to go to step A, otherwise goes to step 2.293;

Step 2.293 replacement parallactic grid grid value: each graticule mesh point value (mx, my) of parallactic grid is:

Mx=a ₁+ a ₂* Sx+a ₃* Sy, my=b ₁+ b ₂* Sx+b ₃* Sy, wherein, Sx, Sy are the distance of this grid points to graticule mesh center.

At above-mentioned a kind of image database for control point matching process based on Image registration, in described step 3, final matching results determines that method is as follows:

If the unique point reliability value R of final entry in step 2.5 _b> 1, calculates its image coordinate (x, y) to the Euclidean distance m of reference mark image center (x ', y '),

travel through all unique points, obtain minimum m value, i.e. the unique point nearest apart from reference mark image center, as reference mark matching result.

The bright technique effect of the party is embodied in: first based on reference mark image geographic coordinate information, determine the region to be matched approximate range in image to be matched, and treat matching area and carry out geometric distortion correction, obtain region subject to registration.Then adopt classification matching strategy, unique point relaxation method coupling, carries out Image registration to reference mark image and region subject to registration, utilizes the information of view picture reference mark image to retrain the coupling of reference mark image.In registration results, image center immediate Feature Points Matching result in chosen distance reference mark is as final matching results.Use the method to carry out the reference mark coupling in image database for control point, obtained and be matched to preferably power.

Accompanying drawing explanation

Fig. 1 is operational flowchart of the present invention;

Fig. 2 is image geometry elimination of the distortion method schematic diagram of the present invention;

Fig. 3 is parallactic grid schematic diagram of the present invention;

Fig. 4 is reliability graticule mesh schematic diagram of the present invention;

Fig. 5 is apparatus structure schematic diagram of the present invention.

Embodiment

Below in conjunction with accompanying drawing, the present invention is described in further detail, this method flow process is shown in Fig. 1, sums up, and the enforcement of this method can be divided into three steps:

1. eliminate image geometry distortion

1.1 read image coordinate resolves function and elevation datum

The image parameters of Reading Control Point image and image to be matched.In embodiment, reference mark image is with Geographical Coordinates Parameter, and image to be matched is with RPC parameter.

By image parameters, obtain reference mark image sensor imaging geometry model coordinate normal solution function f _c, image sensor imaging geometry model coordinate normal solution function f to be matched _p, reference mark image sensor imaging geometry model coordinate counter separate function f ' _c, image sensor imaging geometry model coordinate to be matched counter separate function f ' _p, and image dispersed elevation value Z to be matched _p, make Z _pfor elevation datum.

1.2 calculate image resampling parameter to be matched

Image coordinate (the x of four angle points of Reading Control Point image _ci, y _ci), because reference mark image is regular rectangular shape, the rectangular area P in four angle points besieged city _c.

This rectangular area is extended out to 100 pixels, obtain a larger rectangular area P ' _c, be used for calculating the region to be matched in image to be matched.Four angular coordinates of this rectangular area are (x _c1-100, y _c1-100), (x _c2+ 100, y _c2-100), (x _c3-100, y _c3+ 100), (x _c4+ 100, y _c4+ 100).

These four coordinates and height value benchmark face amount Z _pthe geometric model coordinate normal solution function f of substitution reference mark image _ccalculate the ground coordinate (X of these four some correspondences _pi, Y _pi), i.e. (X _pi, Y _pi)=f _c(x _ci, y _ci, Z _p).Obtain the ground region P covering by after the expansion of reference mark image.

The geometric model coordinate of four angle point ground coordinate substitution images to be matched of the earth coverage area territory P trying to achieve is counter separate function f ' _pcalculate the image coordinate of their correspondences on right image, i.e. (x _pi, y _pi)=f ' _p(X _pi, X _pi, Z _r).I=1 wherein, 2,3,4.

(x _c1-100, y _c1-100), (x _c2+ 100, y _c2-100), (x _c3-100, y _c3+ 100), (x _c4+ 100, y _c4+ 100) and (x _pi, y _pi) substitution perspective transform formula

$\begin{array}{c}{x}^{\&prime;}=\frac{l_{1}x+l_{2}y+l_3}{l_{7}x+l_{8}y+1}\\ y^{\&prime;}=\frac{l_{4}x+l_{5}y+l_6}{l_{7}x+l_{8}y+1}\end{array},$ Wherein $\left|\begin{array}{ccc}{l}_1& l_2& l_3\\ l_4& l_5& l_6\\ l_7& l_8& 1\end{array}\right|\&NotEqual;0$

Every group of coordinate can 2 equations of row, and four groups of coordinates altogether can 8 equations of row, calculate and solve l ₁l l ₈8 perspective transform parameters.

To overlay area, ground P, the respective regions on image to be matched carries out perspective transform to 1.3 eight perspective transform parameters that calculated by 1.2 steps, to non-integer point, adopt B-spline function method to resample, eliminate geometric distortion, imagery zone after being converted, the region subject to registration while being pyramid coupling.

Eliminate geometric distortion, obtain region subject to registration method see Fig. 2.

2 reference mark Image registrations

Reference mark Image registration adopts gold tower matching strategy

2.1 calculate pyramid image matching parameter

In embodiment, image size in reference mark is 512 pixel * 512 pixels, and in image to be matched, region of search size is (512 pixel+200 pixel) * (512 pixel+200 pixel)

It is 1/4 that scaling between pyramidal layer is set, sampling number

pyramid image coupling progression is N _r+ 1=4;

2.2 pyramid classification couplings

2.2.1 the first order is mated

(1) according to scaling 1/4 ³, to reference mark image and area resample subject to registration, generate reference mark image first order pyramid image and first order pyramid image subject to registration;

(2) match window size being set is 5 pixel * 5 pixels, and hunting zone is long is (512 pixel+200 pixel)/2 ³-512 pixel/2 ³=25 pixels, height is (512 pixel+200 pixel)/2 ³-512 pixel/2 ³=25 pixels;

(3) parallactic grid and reliability graticule mesh are set;

On the image first order pyramid image of reference mark, parallactic grid and reliability graticule mesh are set, all with (3,3) be starting point, be spaced apart match window size 5 pixel * 5 pixels, meshes number=(64 pixel/5 pixel) * (64 pixel/5 pixel)=12 * 12, see Fig. 3 and Fig. 4, wherein in Fig. 3, circle represents grid points;

The graticule mesh point value initialize of parallactic grid is 0;

(4) extract minutiae

(4.1) use Shen Jun operator extract minutiae collection on the image first order pyramid image of reference mark;

(4.2) calculated characteristics point gradient direction angle g _dewith Grad g _d;

$g_d=\sqrt{{(g(x+1,y)-g(x,y))}^2+{(g(x,y+1)-g(x,y))}^2},$

g _de＝arctan((g(x，y+1)-g(x，y))/(g(x+1，y)-g(x，y)))/π，

Wherein g (x, y) is the gray-scale value of the unique point (x, y) in the image first order pyramid image of reference mark;

(4.3) judge whether gradient direction angle g _de> 0.7 radian or g _de< 0.3 radian is this unique point reliability value R _b=1, otherwise this unique point reliability value R _b=0;

(5) calculate the first candidate matches peak point,

(5.1) screening unique point;

Unique point to extracting, judges whether reliability value R _b> 0, is the process of carrying out (5.3), otherwise the process of carrying out (5.2);

(5.2) reject this unique point, in feature point set, choose next unique point, turn over journey (5.1); It should be noted in the discussion above that the present invention be only applicable to this step and reject after at least surplus next one meet R _bthe situation of the unique point of > 0, if all unique points are all disallowable, this method lost efficacy;

(5.3) calculate related coefficient;

On the image first order pyramid image of reference mark with unique point coordinate (x, y) centered by, calculate target area, target area is the square consistent with match window size (5 pixel * 5 pixel), its upper left angle point is (x-2, y-2) bottom right angle point is (x+2, y+2);

On the first order pyramid image of region subject to registration with unique point coordinate (x, y) centered by, calculate region of search, region of search is the rectangle consistent with search range (25 pixel * 25 pixel), its upper left angle point is (x-12, y-12) bottom right angle point is (x+12, y+12);

In region of search, the position that overlaps with upper left, region of search angle point with target area starts, and moving object region calculates correlation coefficient ρ (c, r):

$\mathrm{\&rho;}(c,r)=\frac{\underset{i=x-w_{a1}/2}{\overset{w_{a1}}{\mathrm{\&Sigma;}}}\underset{j=y-w_{a1}/2}{\overset{w_{a1}}{\mathrm{\&Sigma;}}}(g_{i,j}-\stackrel{\&OverBar;}{g})(g_{i+r,j+c}^{\&prime;}-{\stackrel{\&OverBar;}{g}}^{\&prime;})}{\sqrt{\underset{i=x-w_{a1}/2}{\overset{w_a}{\mathrm{\&Sigma;}}}\underset{j=y-w_{a1}/2}{\overset{w_a}{\mathrm{\&Sigma;}}}{(g_{i,j}-\stackrel{\&OverBar;}{g})}^2\&times;{(g_{i+r,j+c}^{\&prime;}-{\stackrel{\&OverBar;}{g_{r,c}}}^{\&prime;})}^2}}$

Every movement is once a match window, records the facies relationship numerical value of each match window;

Row, column displacement when wherein c, r are moving object region on region of search, g _{i, j}for the gray-scale value of (i, j) point in target area,

for target area average gray value, g ' _{i+r, j+c}for the gray-scale value of (i+r, j+c) point in region of search,

for in the region of search that participates in while having moved displacement (c, r) on region of search calculating when target area with the average gray value in the big or small corresponding region of match window;

(5.4) calculate parallax parameter;

Matching result parallax parameter (the md of each match window _x, md _y) be:

md _x＝x _c-x，md _y＝y _c-y，

Wherein, x, y are current unique point coordinate, x _c, y _cfor this match window center point coordinate;

The matching result of each match window is described by facies relationship numerical value and parallax parameter, and the related coefficient of each match window and parallax parameter are charged to matching result array BP _cnds;

(5.5) determine the first candidate matches peak point of unique point;

In region of search, the facies relationship numerical value that compares the match window of each match window and its eight neighborhood, judging whether this match window related coefficient is all greater than its eight neighborhood matching window-related coefficient, is using this match window central point as candidate matches peak point;

Traveling through whole match windows, judge whether to exist candidate matches peak point, is the process of carrying out (5.6), otherwise current unique point reliability value R _bbe made as 0, turn over journey (5.2);

(5.6) by all candidate matches peak points according to related coefficient size sequence, the maximum first candidate matches peak point of classifying as;

(5.7) choose next unique point, turn over journey (5.1), complete if all unique points are all calculated, enter sub-step (6);

(6) unique point adds reliability graticule mesh;

According to unique point coordinate, within the scope of each grid of reliability graticule mesh, reliability value R _bthe unique point of > 0 adds the net point array of this grid, and net point array is recorded in the coordinate of unique point within the scope of this grid;

(7) use existing all reliability value R _bthe unique point of > 0 builds the Delaunay triangulation network;

(8) by document < < aviation image characteristic matching research > > (Jiang Wanshou etc., Wuhan University Journal information science version, 28 (5), 2003) the relaxation method feature matching method of introducing in is to all reliability value P _bthe unique point of > 0 coupling that relaxes,

The corresponding candidate matches peak value point set of the present invention of candidate matches point set wherein;

Again optimize the first candidate matches peak point of each unique point, and the parallax parameter (dx, dy) of optimizing unique point makes it equal the parallax parameter (md of the first candidate matches peak point _x, md _y); If the facies relationship numerical value of the first candidate matches peak point of unique point is greater than 0.75, make this unique point reliability value R _b=2.

(9) multinomial model is optimized its first candidate matches peak point of unique point

(9.1) digital simulation multinomial coefficient

Use affine Transform Model:

md _x＝a ₁+a ₂×xd+a ₃×yd，

md _y＝b ₁+b ₂×xd+b ₃×yd，

Wherein, md _x, md _yfor the parallax parameter of the first candidate matches peak point of unique point, xd=x-xc, yd=y-yc, the coordinate that x, y are unique point, xc, yc are parallactic grid center point coordinate,

By all R in feature point set _bthe unique point substitution affine Transform Model of > 1, each feature point range one prescription journey, calculates a by least square method ₁, a ₂, a ₃, b ₁, b ₂, b ₃six affined transformation coefficients;

(9.2) optimize the first candidate matches peak point

(9.2.1) calculated characteristics is put the deviate of the first candidate matches peak point

Sign f=0 is upgraded in order;

Judge whether current unique point R in feature point set _b> 1, otherwise rotor process (9.2.6) is unique point substitution polynomial fitting to be calculated:

px＝a ₁+a ₂×xd+a ₃×yd，

py＝b ₁+b ₂×xd+b ₃×yd，

Wherein, px, the py parallax value for calculating in polynomial fitting, xd=x-xc, yd=y-yc, the coordinate that x, y are unique point, xc, yc are parallactic grid center point coordinate;

Calculation deviation value t again _x, t _y:

t _x＝|px-md _x|，t _y＝|py-md _y|；

Wherein, md _x, md _yparallax parameter for the first candidate matches peak point of unique point;

(9.2.2) judgment bias value

Judgement, t _x> 12 and t _y> 12, are the processes of carrying out (9.2.3), proceed to if not process (9.2.6);

(9.2.3) calculate the deviation of next candidate matches peak point

Judging whether next candidate matches peak point of this unique point exists, is rotor process (9.2.4); Otherwise rotor process (9.2.5);

(9.2.4) upgrade the first candidate matches peak point;

Sign f=1 is upgraded in order;

According to the parallax parameter of this candidate matches peak point, calculate new deviate t _xi, t _yi;

Judge whether t _xi< t _xand t _yi< t _y, otherwise rotor process (9.2.3) is to make t _x=t _xi, t _y=t _yi, using this candidate matches peak point as the first candidate matches peak point, its parallax parameter value is assigned to the parallax parameter (dx, dy) of unique point, and makes the reliability value R of this unique point _b=1, rotor process (9.2.3);

(9.2.5) judge whether to upgrade sign f=1, be rotor process (9.2.6), otherwise make this unique point reliability value R _b=0, rotor process (9.2.6);

(9.2.6) judging characteristic point is concentrated and whether to be had next unique point, is rotor process (9.2.1), otherwise turns over journey (9.3);

(9.3) replacement parallactic grid grid value

Each graticule mesh point value (mx, my) of parallactic grid is:

mx＝a ₁+a ₂×Sx+a ₃×Sy，

my＝b ₁+b ₂×Sx+b ₃×Sy，

Wherein, Sx, Sy are the distance of this grid points to graticule mesh center;

(10) preserve parallactic grid at the corresponding levels;

2.2.2 mate the second level

(1) according to scaling

to reference mark image and area resample subject to registration, generate reference mark image second level pyramid image and the second level, region subject to registration pyramid image, the image second level, reference mark pyramid image is long is OL ₂pixel height is OH ₂pixel, in the present embodiment, according to zoom ratio 1/4 ²to reference mark image and area resample subject to registration, generate reference mark image second level pyramid image and the second level, region subject to registration pyramid image, the image second level, reference mark pyramid image size is (512 pixel/4) * (512 pixel/4)=128 pixel * 128 pixels, and pyramid image size in the second level, region subject to registration is (712 pixel/4) * (712 pixel/4)=178 pixel * 178 pixels;

(2) match window W is set _m2size is length of side W _a2the square window of=3,5 or 7 pixels, hunting zone is length of side S _a2=7～15 square area; In the present embodiment, it is 5 pixel * 5 pixels that match window size is set, and hunting zone is 11 pixel * 11 pixels;

(3) second level parallactic grid and second level reliability graticule mesh are set on the pyramid image of the image second level, reference mark, second level parallactic grid is identical with second level reliability grid structure, forms, all with coordinate (W by square net _a2/ 2+1, W _a2/ 2+1) be starting point, the summit of square net is grid points, and the square net length of side is W _a2, square net number=(OL ₂/ W _a2) * (OH ₂/ W _a2); In the present embodiment, second level parallactic grid and second level reliability graticule mesh are set on the pyramid image of the image second level, reference mark, all with coordinate (3,3) be starting point, interval 5 pixel * 5 pixels, meshes number=(128 pixel/5 pixel) * (128 pixel/5 pixel)=25 * 25;

The grid points coordinate of second level parallactic grid is multiplied by scaling 1/4, and substitution first order parallactic grid is calculated the graticule mesh point value of second level parallactic grid by each graticule mesh point value bilinear interpolation of first order parallactic grid;

(4) according to sub-step in 2.2.1 (4) extract minutiae collection on the pyramid image of the image second level, reference mark, and unique point dependability parameter R is set _bvalue;

(5) by the parallactic grid of the unique point coordinate substitution second level, in the graticule mesh falling in unique point, bilinear interpolation is calculated the parallax value in its x direction and y direction, as the parallax parameter (dx, dy) of this unique point;

(6) determine the region of search of each Feature Points Matching;

On the pyramid image of the image subject to registration second level with unique point coordinate (x, y) centered by, calculate region of search, region of search is the square consistent with the described search range of sub-step (2), coordinate (x, y) with unique point deducts after the parallax parameter (dx, dy) of unique point, centered by, calculate region of search:

x _from＝x-dx-S _a2/2，y _from＝y-dx-S _a2/2；

x _to＝x-dx+S _a2/2，y _to＝y-dy+S _a2/2；

In the present embodiment, calculating region of search is:

x _from＝x-dx-5，y _from＝y-dx-5；

x _to＝x-dx+5，y _to＝y-dy+5；

(x wherein _from, y _from) be upper left, region of search angle point, (x _to, y _to) be bottom right, region of search angle point;

(7) according to sub-step in 2.2.1 (5), calculated characteristics point is concentrated the candidate matches peak point of each unique point, and wherein match window is revised as W _m2, it is S that the length of side is revised as in hunting zone _a2square area; In the present embodiment, match window is revised as 5 pixel * 5 pixels, and 11 pixel * 11 pixels are revised as in hunting zone;

(8) according to 2.2.1 (6)-(10) step, carry out the lax coupling of second level pyramid, and preserve second level parallactic grid;

2.2.3 the third level mates

In the present embodiment, the concrete steps of third level coupling are as follows:

(1) according to zoom ratio 1/4, to reference mark image and area resample subject to registration, generate reference mark image third level pyramid image and region subject to registration third level pyramid image, reference mark image third level pyramid image size is (512 pixel/2) * (512 pixel/2)=256 pixel * 256 pixels, and the region subject to registration third level and word tower image size are (712 pixel/2) * (712 pixel/2)=356 pixel * 356 pixels;

(2) match window size being set is 5 pixel * 5 pixels, and hunting zone is 11 pixel * 11 pixels

(3) according to (3) step in 2.2.2, on the image third level pyramid image of reference mark, parallactic grid and reliability graticule mesh are set, and calculate each graticule mesh point value in parallactic grid;

(4) according to (4) in 2.2.2-(8) step, carry out the lax coupling of third level pyramid, and preserve parallactic grid at the corresponding levels;

2.4 fourth stage couplings, in the present embodiment, the concrete steps of fourth stage coupling are as follows:

(1) reference mark image fourth stage pyramid image size is 512 * 512, and image fourth stage pyramid image size subject to registration is 712 * 712;

(2) match window size being set is 5 pixel * 5 pixels, and hunting zone is 11 pixel * 11 pixels;

(3) according to (3) step in 2.2.2, be located on the image fourth stage pyramid image of reference mark parallactic grid and reliability graticule mesh are set, and calculate each graticule mesh point value in parallactic grid;

(4) according to (4) in 2.2.1-(8) step, carry out the lax coupling of third level pyramid, and preserve unique point and the matching result thereof of pyramid coupling at the corresponding levels;

3. optimize and select matching result

If the unique point reliability value R of final entry _b> 1, calculates its image coordinate (x, y) to the Euclidean distance m of reference mark image center (x ', y '), wherein, travel through all unique points, obtain minimum m value, i.e. the unique point nearest apart from reference mark image center, as reference mark matching result.

In the present embodiment, the result of being mated by the present invention is as following table

Reference mark sum	Correct coupling	Erroneous matching	Be matched to power
				40	32	8	80％

Claims (1)

1. the image database for control point matching process based on Image registration, is characterized in that, comprises the following steps:

Step 1, take reference mark image as eliminate the geometric distortion of image to be matched with reference to image by geometric distortion cancellation module, obtains the region subject to registration of pyramid image coupling; Concrete operation steps is as follows:

Step 1.1, parameter acquiring unit reads image coordinate and resolves function and elevation datum: parameter acquiring unit obtains reference mark image sensor imaging geometry model coordinate normal solution function f by image parameters _c, image sensor imaging geometry model coordinate normal solution function f to be matched _p, the anti-function f of separating of reference mark image sensor imaging geometry model coordinate _c', the anti-function f of separating of image sensor imaging geometry model coordinate to be matched _p', and image dispersed elevation value Z to be matched _p, make Z _pfor elevation datum;

Step 1.2 is by the image coordinate (x of four angle points of region acquiring unit Reading Control Point image subject to registration _ci, y _ci), the region p in four angle points besieged city _c; By region p _cexpand k pixel, obtain a region p _c', be used for calculating the regional extent to be matched in image to be matched, p _c' tetra-angular coordinates are (x _c1-k, y _c1-k), (x _c2+ k, y _c2-k), (x _c3-k, y _c3+ k), (x _c4+ k, y _c4+ k);

Step 1.3, by perspective transform parameter acquiring unit by the region p in step 1.2 _c' four angular coordinate (x _c1-k, y _c1-k), (x _c2+ k, y _c2-k), (x _c3-k, y _c3+ k), (x _c4+ k, y _c4+ k) the height value benchmark face amount Z and in step 1.1 _pthe geometric model coordinate normal solution function f of reference mark image in substitution step 1.1 _cground coordinate (the X that four angular coordinates of middle calculating are corresponding _pi, Y _pi), i.e. (X _pi, y _pi)=f _c(x _ci, y _ci, Z _p), obtain the ground region P covering by after the expansion of reference mark image;

By the geometric model coordinate of four angle point ground coordinate substitution images to be matched of the earth coverage area territory P trying to achieve counter separate function f ' _pcalculate the image coordinate of their correspondences on image to be matched, (x _p, y _pi)=f _p' (X _pi, X _pi, Z _r), the region to be matched described in above-mentioned steps 1.2 is (x _pi, y _pi) region that surrounds, then by (x _c1-k, y _c1-k), (x _c2+ k, y _c2-k), (x _c3-, y _c3+ k), (x _c4+ k, y _c4+ k) and (x _pi, y _pi) substitution perspective transform formula

$\begin{array}{c}{x}^{\text{'}}=\frac{l_{1}x+l_{2}y+l_3}{l_{7}x+l_{8}y+1}\\ y^{\text{'}}=\frac{l_{4}x+l_{5}y+l_6}{l_{7}x+l_{8}y+1}\end{array},$ Wherein $\left|\begin{array}{ccc}{l}_1& l_2& l_3\\ l_4& l_5& l_6\\ l_7& l_8& 1\end{array}\right|\&NotEqual;0$

Obtain 1 ₁l ₈8 perspective transform parameters, i=1 wherein, 2,3,4;

Step 1.4, by eight perspective transform parameters that obtain in step 1.3, to overlay area, ground P, the respective regions on image to be matched carries out perspective transform to region subject to registration acquiring unit, to non-integer point, adopt B-spline function method to resample, eliminate geometric distortion, imagery zone after being converted, is the region subject to registration that pyramid image mates;

Step 2, the region subject to registration after by pyramid image matching module, reference mark image being completed with step 1 carries out that classification is lax mates;

Step 3, the matching result point by matching result acquisition module after step 2 completes concentrates image center nearest matching result in chosen distance reference mark as final matching results.

2. a kind of image database for control point matching process based on Image registration according to claim 1, is characterized in that, in described step 2, concrete steps are as follows:

Step 2.1, obtains pyramid image matching parameter by matching parameter acquiring unit, and pyramid image matching parameter comprises scaling and pyramid progression, and concrete grammar is: pyramid image interlayer scaling P is set _s=1/2 ²ⁿ, n=1～3, first order pyramid image size S _min>=64 * 64 pixels, first order pyramid image is the minimum image of top layer resolution in pyramid structure; Sampling number is

integral part, wherein L and H are length and the height of reference mark image; Pyramid image coupling progression is N _r+ 1;

Step 2.2, carries out pyramid first order coupling by first order matching unit in step 1 in the region subject to registration being completed;

Step 2.3, carries out pyramid second level coupling in the region subject to registration that second level matching unit completes in step 1;

Step 2.4, carries out pyramid N level coupling in the region subject to registration that N level matching unit completes in step 1, method, with in step 2.3, is carried out pyramid image coupling step by step, until afterbody pyramid image;

Step 2.5, records unique point and its matching result that afterbody pyramid mates.

3. a kind of image database for control point matching process based on Image registration according to claim 2, is characterized in that, in described step 2.2, concrete steps are as follows:

Step 2.21, according to scaling

, to reference mark image and area resample subject to registration, generate reference mark image first order pyramid image and region subject to registration first order pyramid image;

Step 2.22, arranges match window W _m1size is length of side W _a1the square window of=3 or 5 or 7 or 9 or 11 pixels, hunting zone S _s=S _lpixel * S _hpixel, region subject to registration first order pyramid image is long is PL ₁pixel height is PH ₁pixel, reference mark image first order pyramid image is long is OL ₁pixel height is OH ₁pixel, S _l=PL ₁-OL ₁, S _h=PH ₁-OH ₁;

Step 2.23 arranges parallactic grid and reliability graticule mesh on the image first order pyramid image of reference mark, and parallactic grid and reliability graticule mesh form by square net, with coordinate (W _a1/ 2+1, W _a1/ 2+1) be starting point, the summit of square net is grid points, and the square net length of side is W _a1, square net number=(OL ₁/ W _a1) * (OH ₁/ W _a1), by each graticule mesh point value assignment of parallactic grid, be 0;

Step 2.24, is used Shen Jun operator extract minutiae on the first order pyramid image of reference mark image; And calculate the gradient information on raw video of each unique point: comprise Grad $\mathrm{grad}=\sqrt{{(g(x+1,y)-g(x,y))}^2+{(g(x,y+1)-g(x,y))}^2},$ Gradient direction angle gradDir=arctan ((g (x, y+1)-g (x, y))/(g (x+1, y)-g (x, y)))/π, wherein g (x, y) be the gray-scale value that image mid point (x, y) is located, judge whether gradient direction angle g _de>07 radian or g _de<03 radian is this unique point reliability value R _b=1, otherwise this unique point reliability value R _b=0;

Step 2.25, calculates the first candidate matches peak point;

Step 2.26, unique point adds reliability graticule mesh, that is: according to unique point coordinate, within the scope of each grid of reliability graticule mesh, reliability value R _bthe unique point of >0 adds the net point array of this grid, and net point array is recorded in the coordinate of unique point within the scope of this grid;

Step 2.27, is used existing all reliability value R _bthe unique point of >0 builds the Delaunay triangulation network;

Step 2.28 adopts relaxation method feature matching method to all reliability value R _bthe unique point of the >0 coupling that relaxes, the concrete grammar of lax coupling is: the first candidate matches peak point of again optimizing each unique point, and the parallax parameter (dx, dy) of optimizing unique point makes it equal the parallax parameter (md of the first candidate matches peak point _x, md _y); If the facies relationship numerical value of the first candidate matches peak point of unique point is greater than 0.75, make this unique point reliability value R _b=2;

Step 2.29, multinomial model is optimized the first candidate matches peak point of unique point and is preserved parallactic grid at the corresponding levels.

4. a kind of image database for control point matching process based on Image registration according to claim 3, is characterized in that, in described step 2.25, the concrete steps of calculated candidate match peak point are as follows:

Step 2.251, screening unique point: the unique point to extracting, judges whether reliability value R _b>0, is to carry out process steps 2.253, otherwise carries out process steps 2.252;

Step 2.252, rejects this unique point, chooses next unique point, goes to step 2.51;

Step 2.253, calculates related coefficient, and concrete steps are as follows:

Step a, on the image first order pyramid image of reference mark centered by unique point coordinate (x, y), according to match window size W _m1, calculate target area, target area is and match window square of the same size that its upper left angle point is (x-W _a1/ 2, y-W _a1/ 2) bottom right angle point is (x+W _a1/ 2, y+W _a1/ 2);

Step b, on the first order pyramid image of region subject to registration centered by unique point coordinate (x, y), according to described search range S _s, calculate region of search, region of search is a rectangle, its upper left angle point is (x-S _l/ 2, y-S _h/ 2), bottom right angle point is (x+S _l/ 2, y+S _h/ 2);

Step c, in region of search, the position that overlaps with upper left, region of search angle point with target area starts, and moving object region calculates correlation coefficient ρ (c, r):

$\mathrm{\&rho;}(c,r)=\frac{\underset{i=x-w_{a1}/2}{\overset{w_{a1}}{\mathrm{\&Sigma;}}}\underset{j=y-w_{a1}/2}{\overset{w_{a1}}{\mathrm{\&Sigma;}}}(g_{i,j}-\stackrel{\&OverBar;}{g})(g_{i+r,j+c}^{\&prime;}-{\stackrel{\&OverBar;}{g}}^{\&prime;})}{\sqrt{\underset{i=x-w_{a1}/2}{\overset{w_a}{\mathrm{\&Sigma;}}}\underset{j=y-w_{a1}/2}{\overset{w_a}{\mathrm{\&Sigma;}}}{(g_{i,j}-\stackrel{\&OverBar;}{g})}^2\&times;{(g_{i+r,j+c}^{\&prime;}-{{\stackrel{\&OverBar;}{g}}_{r,c}}^{\&prime;})}^2}}$

Every movement is once a match window, records the facies relationship numerical value of each match window;

Row, column displacement when wherein c, r are moving object region on region of search, g _{i, j}for the gray-scale value of (i, j) point in target area,

for target area average gray value, g ' _{i+r, j+c}for the gray-scale value of (i+r, j+c) point in region of search,

for in the region of search that participates in while having moved displacement (c, r) on region of search calculating when target area with the average gray value in the big or small corresponding region of match window;

Step 2.254, calculates parallax parameter;

Matching result parallax parameter (the md of each match window _x, md _y) be:

md _x=x _c-x，

md _y=y _c-y，

Wherein, (x _c, y _c) be this match window center point coordinate, (x, y) is current unique point coordinate; Related coefficient and parallax parameter are charged to matching result array BP _cnds, the matching result of each match window is described by facies relationship numerical value and parallax parameter;

Step 2.255, determines the first candidate matches peak point of unique point;

In region of search, the facies relationship numerical value that compares the match window of each match window and its eight neighborhood, judge whether this match window related coefficient is all greater than its eight neighborhood matching window-related coefficient, using this match window central point as candidate matches peak point, carry out process steps 2.56, otherwise current unique point reliability value R _bbe made as 0, turn process steps 2.52;

Step 2.256, all candidate matches peak points according to related coefficient size sequence, the maximum first candidate matches peak point of classifying as;

Step 2.257, chooses next unique point, turns as step 2.251, complete if all unique points are all calculated, and enters above-mentioned steps 2.26.

5. a kind of image database for control point matching process based on Image registration according to claim 4, is characterized in that, in described step 2.29, the concrete steps of the first candidate matches peak point of multinomial model optimization unique point are as follows:

Step 2.291, digital simulation multinomial coefficient, the affine Transform Model using in calculating is as follows:

md _x=a ₁+a ₂×xd+a ₃×yd，

md _y=b ₁+b ₂×xd+b ₃×yd，

Wherein, md _x, md _yfor the parallax parameter of the first candidate matches peak point of unique point, xd=x-xc, yd=y-yc, the coordinate that x, y are unique point, xc, yc are parallactic grid center point coordinate; By all R in feature point set _bthe unique point substitution affine Transform Model of >1, each feature point range one prescription journey, calculates a by least square method ₁, a ₂, a ₃, b ₁, b ₂, b ₃six affined transformation coefficients;

Step 2.292, optimize the first candidate matches peak point:

Steps A, calculated characteristics is put the deviate of the first candidate matches peak point, and concrete grammar is as follows:

Sign f=0 is upgraded in order; Judge whether current unique point R in feature point set _b>1, otherwise go to step F, be unique point substitution polynomial fitting to be calculated:

px=a ₁+a ₂×xd+a ₃×yd，

py=b ₁+b ₂×xd+b ₃×yd，

Wherein, px, the py parallax value for calculating in polynomial fitting, xd=x-xc, yd=y-yc, the coordinate that x, y are unique point, xc, yc are parallactic grid center point coordinate;

Calculation deviation value t again _x, t _y:

T _x=| px-md _x|, t _y=| py-md _y|; Wherein, md _x, md _yparallax parameter for the first candidate matches peak point of unique point;

Step B, judgment bias value: judgement, t _x>12 and t _y>12, is to carry out step C, proceeds to if not step F;

Step C, calculates the deviation of next candidate matches peak point: judging whether next candidate matches peak point of this unique point exists, is to go to step D; Otherwise go to step E;

Step D, upgrades the first candidate matches peak point; Concrete grammar is: sign f=1 is upgraded in order; According to the parallax parameter of this candidate matches peak point, calculate new deviate t _xi, t _yi;

Judge whether t _xi<t _xand t _yi<t _y, otherwise rotor step C is to make t _x=t _xi, t _y=t _yi, using this candidate matches peak point as the first candidate matches peak point, its parallax parameter value is assigned to the parallax parameter (dx, dy) of unique point, and makes the reliability value R of this unique point _b=1, go to step C;

Step e, judges whether to upgrade sign f=1, is to proceed to step F; Otherwise make this unique point reliability value R _b=0, go to step F;

Step F, judging characteristic point concentrates whether there is next unique point, is to go to step A, otherwise goes to step 2.293;

Step 2.293 replacement parallactic grid grid value: each graticule mesh point value (mx, my) of parallactic grid is:

Mx=a ₁+ a ₂* Sx+a ₃* Sy, my=b ₁+ b ₂* Sx+b ₃* Sy, wherein, Sx, Sy are the distance of this grid points to graticule mesh center.

6. a kind of image database for control point matching process based on Image registration according to claim 3, is characterized in that, in described step 3, final matching results determines that method is as follows:

If the unique point reliability value R of final entry in step 2.5 _b>1, calculates its image coordinate (x, y) to the Euclidean distance m of reference mark image center (x ', y '),

travel through all unique points, obtain minimum m value, i.e. the unique point nearest apart from reference mark image center, as reference mark matching result.

Images