CN104574519A - Threshold-free automatic robust matching method for multi-source residence surface elements - Google Patents

Abstract

The invention relates to a threshold-free automatic robust matching method for multi-source residence surface elements and aims to effectively realize the fact that a 1:1 matching relation, a 1:N matching relation and an M:N matching relation of same-name surface elements can be acquired accurately while accurate approach of positions of two groups of to-be-matched surface elements is not required and large positional deviation and even nonuniform positional deviation are allowed. The method includes: adopting a computer to respectively read two groups of to-be-matched residence data, denoting one group of residences which is a reference surface element combination as R, and denoting the other group of residences which is a target surface element combination as T; generating candidate matching pairs; judging the matching relation of the same-name surface elements; detecting and eliminating conflicts of matching results. The method has the advantages of stability, reliability, easiness in operation and utilization, avoidance of requiring accurate approach of the positions of the two groups of to-be-matched surface elements, allowance of large positional deviation, avoidance of threshold setup, reduction of preprocessing workload and effective applicability to vector map conflation, map updating and change detection.

Description

Multi-source resident terrain element exempt from threshold value automatically sane matching process

Technical field

The present invention relates to map, mainly refer to data integration and the production of map, what refer to a kind of multi-source resident terrain element especially exempts from threshold value sane matching process automatically.

Background technology

Along with the fast development of space data collection technology, people have accumulated abundant spatial data, even if cover same area, different department needs also repeatedly to acquire the data differed from one another for difference.Such as, house property department and Mapping departments all acquire settlement place data, but the attribute information (as ownership, area etc.) of settlement place data is more valued by house property department, and the precision of data is more valued by Mapping departments, if the data of different department can be carried out effective integration, just can integrate the advantage of various version space data, thus can save a large amount of field operations repeat spending.Integrated, multiplexing and shared accurately in order to carry out multi-source Spatial Data, its core and prerequisite steps are exactly automatic identification and the coupling of wanting to realize separate sources spatial data key element of the same name.For this reason, the present invention is especially for the invention work that face factor data (mainly for the resident's terrain element) Develop Data of separate sources in map vector mates.

In various key element matching algorithms, commonly simplified the complicacy of face key element matching problem by dimension-reduction treatment, determined the matching relationship of face key element by the position relationship comparing distance, barycenter and face key element between face key element character pair point or barycenter; Or the profile of the face of extraction key element, matching relationship is determined by calculating the proximity of outline line in distance, the distance measure of normal employing has Fr é chet distance, Hausdorff distance etc., or calculating outline line similarity degree in shape determines matching relationship, its similarity measure is as shape letter of transfer number, Fourier descriptors etc., minority algorithm extracts the skeleton line of key element inside, face, is determined the matching relationship of face key element by the shape similarity contrasting skeleton line.Also some algorithms are had to utilize the interior zone of face key element to mate, common characteristic index is the overlapping area between the key element of calculating face, the ratio that two its overlapping areas of face key element to be matched account for respective area is larger, then they are that the possibility of key element of the same name is also larger.Certainly, some matching algorithms are not used alone above some characteristic indexs, but various index can be carried out conbined usage, but its difficult point is on accurately determining of weights.Also have some algorithms over there in key element matching process, not only can consider the coupling similarity degree of to be matched key element, also can consider the match condition of other face key elements adjacent, further increase the ability that matching algorithm adapts to complex situations.

Although obtained plentiful and substantial progress in the key element matching problem of face, existing algorithm has not still better solved some difficulties, the first, is difficult to the M:N matching relationship existed between Obtaining Accurate of the same name key element; Such as those dimension-reduction algorithms and the method based on outline line, owing to lacking the utilization of opposite key element interior zone, when there is M:N matching relationship, obtain combinatorial surface key element outline line on the whole and will become time-consuming and difficult, these algorithms all cannot process the M:N matching relationship of complexity automatically, can only 1:N be processed, or even the match condition of 1:1.It two is that require that of the same name key element is close as far as possible in position, the normal service range of many algorithms is as matching threshold; Although some algorithms can obtain the M:N matching relationship between of the same name key element, but these class methods have an important prerequisite, namely comparatively close in position between two groups of face key elements to be matched, or clearly meet map drawing norm, but in practice, due to popularity and the complicacy of Data Source, between face key element to be matched, conforming position deviation may be there is, namely some local general deviation is large, and some place deviation is little.Even if corrected by the coordinate system of globality, still can not accuracy registration between two groups of face key elements, these algorithms also will be difficult to prove effective in these cases, and therefore, its improvement and bring new ideas is imperative.

Summary of the invention

For above-mentioned situation, for overcoming the defect of prior art, what the object of the present invention was just to provide a kind of multi-source resident terrain element exempts from threshold value sane matching process automatically, effectively can solve 1:1 matching relationship, 1:N matching relationship and the M:N matching relationship that can either exist between Obtaining Accurate of the same name key element, do not require that again two groups of to be matched key elements are accurately close in position, allow to there is larger position deviation, or even the problem of nonconforming position deviation.

The technical scheme that the present invention solves comprises the following steps:

Step one, data acquisition:

Read two groups of settlement place data to be matched respectively with computing machine, one group of settlement place is reference surface factor combination, is designated as R, and another group settlement place is target face factor combination, is designated as T;

Step 2, generation candidate matches pair:

For reference surface factor combination R to be matched and target face factor combination T, judged by the crossover of locus overlapping relation, candidate matches face factor combination is selected successively from reference surface factor combination and target face factor combination, and generating candidate matches pair, candidate matches is to comprising the various matching relationships that may exist between of the same name the key element such as 1:1,1:N, M:1 and M:N;

The judgement of step 3, of the same name key element matching relationship:

Adopt the matching similarity of two-way calculating face key element to determine the matching relationship of of the same name key element, namely for reference surface factor combination R and target face factor combination T, when according to calculated matching similarity being mutually the optimum matching face factor combination of the other side, and the two coupling corresponding relation is set up;

In order to effectively process various complex match corresponding relation, its calculating thinking of described matching similarity is to judge the matching degree between the key element of face by weighing overlapped degree between the key element of separate sources face; Even if in order to ensure when of the same name key element has larger position deviation, still effectively mate, then first extract the enclosure rectangle of reference surface factor combination and the global shape feature of individual features point, the coupling corresponding relation of unique point of the same name in automatic acquisition of the same name key element, position skew is carried out with reference to face factor combination R, making two groups of to be matched factor combinations in position can registration further, then passing judgment on the matching degree of reference surface factor combination R and target face factor combination T again by weighing overlapping degree;

The calculating of face key element matching similarity is carried out respectively from both direction, right to obtain coupling, method is, first with reference surface factor combination R for benchmark, according to calculated matching similarity, optimum matching target face factor combination T is found from target face key element, then again with target face factor combination T for benchmark, according to calculated matching similarity, optimum matching reference surface factor combination is found from reference surface factor combination, when this optimum matching face factor combination is reference surface factor combination R, then reference surface factor combination R becomes one group of optimum matching face factor combination each other with target face factor combination T-shaped,

The computing method of described matching similarity are: by utilizing overlapped degree between the key element of separate sources face to weigh the matching degree of face key element, if the area of reference surface factor combination R is A _r, the area of target face factor combination T is A _t, the overlapping area of reference surface factor combination R and target face factor combination T is A _o, then overlapping similarity Sim is:

$\mathrm{Sim}=\frac{A_o{A}_o}{A_r{A}_t}---\left(1\right)$

By calculate overlapping similarity judge candidate matches between matching degree, extract the enclosure rectangle of reference surface factor combination and the global shape feature of individual features point, utilize the coupling corresponding relation of unique point of the same name in of the same name key element, position skew is carried out with reference to face factor combination R, make two groups of to be matched factor combinations in position can registration further, the matching degree of reference surface factor combination R and target face factor combination T is passed judgment on by weighing overlapping degree, to each the face key element in target face factor combination T respectively computing reference face factor combination R move front and reference surface factor combination R move after overlapping similarity, average, be designated as matching similarity ρ, formula is as follows:

$\mathrm{\ρ}=(\frac{A_p{A}_p}{A_r{A}_t}+\frac{A_q{A}_q}{A_r{A}_t})/2---\left(2\right)$

In formula (2), A _rfor the area of reference surface factor combination, A _tfor the area of target face factor combination, A _pfor the lap area of R translation the former two, A _qfor the latter two lap area of R translation;

The conflict probe of step 4, matching result and elimination:

According to matching similarity value size, the matching result clashed is given up or retained, namely retain matching similarity value large, and it is less to give up matching similarity, thus the contradiction realizing matching result is eliminated.

The inventive method is reliable and stable, easy to operate, can be applicable to of the same name key element Auto-matching algorithm under the multiple dimensioned situation of multi-source, 1 ︰ 1 matching relationship, 1 ︰ N matching relationship and the M ︰ N matching relationship that can either exist between Obtaining Accurate of the same name key element, do not require that again two groups of to be matched key elements are accurately close in position, allow to there is larger position deviation, or even nonconforming position deviation, threshold value need not be set user, more complicated situation can be adapted to, therefore can be referred to as a kind of sane matching process (algorithm).The method has expanded the applicability of face, map conflation field key element matching technique, it is made to reduce further the requirement of data qualification and restriction, reduce pretreated workload, promote the integrated, shared and multiplexing of spatial data, be effective to map vector merging, map rejuvenation and change detect, have very strong practicality, economic and social benefit is huge.

Accompanying drawing explanation

Fig. 1 is the generation schematic diagram of candidate matches of the present invention to combination.

Fig. 2 is the schematic diagram of determination optimum matching face of the present invention factor combination.

Fig. 3 is the contradiction elimination figure of matching result of the present invention.

Fig. 4 is the present invention two groups face factor data collection schematic diagram to be matched.

The matching result schematic diagram of Fig. 5 to be matching relationship of the present invention be 1:1.

The matching result schematic diagram of Fig. 6 to be matching relationship of the present invention be 1:N.

The matching result schematic diagram of Fig. 7 to be matching relationship of the present invention be M:1.

The matching result schematic diagram of Fig. 8 to be matching relationship of the present invention be M:N.

Fig. 9 is trying hard to of algorithm of the present invention opposing face elements position deviation.

Embodiment

Below in conjunction with concrete condition, this bright embodiment is described further.

The present invention, in concrete enforcement, can be realized by following steps:

Step one, data acquisition:

Read two groups of settlement place data to be matched respectively with computing machine, one group of settlement place is reference surface factor combination, is designated as R, and another group settlement place is target face factor combination, is designated as T;

Step 2, generation candidate matches pair:

Multiple matching relationship may be there is between of the same name key element, comprise 1:1, 1:N, M:1 and M:N, in order to various matching relationship between of the same name key element correctly can be obtained, first potential candidate matches is generated to set, and this set should be able to as much as possible by all couplings correctly to being included, for to be matched two groups of face factor data set, one group is referred to as reference surface elements combination, another group is referred to as target face elements combination, algorithm selects candidate matches face factor combination alternately successively from reference surface elements combination and target face elements combination, and form candidate matches to (as shown in Figure 1), reference surface key element R has three face key elements, i.e. the first reference surface key element R ₁, the second reference surface key element R ₂with the 3rd reference surface key element R ₃, have three face key elements in target face factor combination T, i.e. first object face key element T ₁, the second target face key element T ₂with the 3rd target face key element T ₃, order chooses each face key element in reference surface elements combination, when in target face elements combination with the first reference surface key element R ₁that there is overlapping relation is first object face key element T ₁, first group of candidate matches can be obtained to (R ₁): (T ₁), and then find and first object face key element T in reference surface elements combination ₁the reference surface key element that there is overlapping relation is the first reference surface key element R ₁, the second reference surface key element R ₂, then can obtain the second group again and be paired into (R ₁, R ₂): (T ₁), obtain and the first reference surface key element R further ₁, the second reference surface key element R ₂the target face key element intersected, can obtain the 3rd group of candidate matches to being (R ₁, R ₂): (T ₁, T ₂, T ₃), obtain the 4th candidate matches until last to being (R ₁, R ₂, R ₃): (T ₁, T ₂, T ₃), when meeting two conditions, the generation that candidate matches is right stops, and these two conditions are: one is cannot obtain candidate matches further to (as shown in Figure 2), in acquisition the 4th group of candidate matches to rear, each face key element participates in computing all, new candidate matches pair of cannot having reentried, two is that the candidate matches of preparation generation is to generating before this,

But it is final matching result that the generation that candidate matches is right stops not representing, if the 4th group of candidate matches is to being (R ₁, R ₂, R ₃): (T ₁, T ₂, T ₃), also need to extract and (R from target face factor combination further ₁, R ₂, R ₃) combine (T the most similar ₁, T ₂) combination, be only final correct matching result;

The judgement of step 3, of the same name key element matching relationship:

The calculating of face key element matching similarity is carried out respectively from both direction, right to obtain coupling, method is, first with reference surface factor combination R for benchmark, according to calculated matching similarity, optimum matching target face factor combination T is found from target face key element, then again with target face factor combination T for benchmark, according to calculated matching similarity, optimum matching reference surface factor combination is found from reference surface factor combination, when this optimum matching face factor combination is reference surface factor combination R, then reference surface factor combination R becomes one group of optimum matching face factor combination each other with target face factor combination T-shaped,

Obtaining candidate matches to afterwards, need to weigh the matching degree between reference surface factor combination and target face factor combination, and determine best matching result from candidate matches centering accordingly, in order to make full use of border and the internal information of face key element, degree overlapped between the key element of employing face weighs the matching degree of face key element, as above set reference surface factor combination as R, target face factor combination is T, if the area of reference surface factor combination R is A _r, the area of target face factor combination T is A _t, the overlapping area of reference surface factor combination R and target face factor combination T is A _o, then overlapping similarity Sim is:

$\mathrm{Sim}=\frac{A_o{A}_o}{A_r{A}_t}---\left(1\right)$

By calculate overlapping similarity judge candidate matches between matching degree, its precondition is that the two can coincide enough good in position, but in a practical situation, the two may have larger position deviation, a kind of extreme case as shown in Figure 2, suppose there is a candidate matches pair, its reference surface factor combination is (R ₁, R ₂), target face factor combination is (T ₁-T ₆), deviation is very large in position for the two, now cannot reflect and for this reason, needs matching degree therebetween to solve in the larger situation of position deviation, how to use the problem of overlapping similarity determination optimum matching face factor combination by weighing overlapping degree;

As shown in Figure 2, when by T ₁to T ₃when combining, its shape and (R ₁, R ₂) rather similar, the possibility that they mate each other is maximum; But present problem is how could correctly by T ₁to T ₃combine, for (R ₁, R ₂), T ₁-T ₆in certain or several face key elements combination all with (R ₁, R ₂) M:N matching relationship may be there is, when listing the combination of its all candidate matches faces key element according to the method for exhaustion, its combined number has individual, obviously, therefrom find out and (R according to the method for exhaustion ₁, R ₂) coupling face factor combination the most similar will be very consuming time, therefore, for carrying out the Auto-matching of face key element, first must find a kind of method fast and accurately can determining optimum matching face factor combination;

Notice that, for of the same name key element, their global shape feature will be rather similar, determine candidate matches face factor combination quickly by following four steps thus:

(1) the global shape feature of reference surface factor combination R is extracted: global shape feature refers to enclosure rectangle and the individual features point of R, as shown in Figure 2, first finds spacing two summits farthest on each summit in reference surface factor combination R, is respectively r ₁, r ₃, then respectively at line segment r ₁, r ₃left side and right side find distance line segment r ₁, r ₃having the point of maximum distance, is r ₂and r ₄, then by r ₁, r ₂, r ₃and r ₄the rectangle of four unique point compositions is obviously the enclosure rectangle of reference surface factor combination R, a limit of enclosure rectangle and r ₁r ₃line parallel, another limit then with r ₁r ₃line is vertical, and this enclosure rectangle has the feature irrelevant with similarity transformation, and when reference surface factor combination R entirety offsets, rotates and convergent-divergent, the enclosure rectangle extracted is still by r ₁, r ₂, r ₃and r ₄four unique point compositions;

(2) for target face factor combination T, find and r ₁, r ₂, r ₃and r ₄optimum matching corresponding vertex: method is, order chooses each face key element in target face factor combination T, then order chooses each summit of this face key element, if the summit selected is t ₁(as shown in Figure 2), in T, find a summit p, make by a t ₁the vectorial U formed to p with by a r ₁to r ₂the vector similarity value of the vectorial V formed is maximum, and namely similar between vector, the computing formula of vector similarity α is as follows herein:

$\mathrm{\α}=\frac{\min \left(\right|\left|U\right||,|\left|V\right|\left|\right)}{\max \left(\right|\left|U\right||,|\left|V\right|\left|\right)}\×\frac{[U\·V]}{\left|\right|U\left|\right|V\left|\right|}---\left(2\right)$

This summit can be found in fig. 2 to be t ₂, summit t can also be found respectively according to same steps ₃and t ₄, because vectorial t ₁t ₃with vectorial r ₁r ₃the most similar, vectorial t ₁t ₄with vectorial r ₁r ₄the most similar.

According to the method described above, according to each summit of face key element in T, other three summits can be found, and successive computations three maximum vector similarities, the mean value getting three maximum vector similarities as resultant vector similarity, when resultant vector similarity value is maximum, namely obtain reference surface factor combination and target face factor combination the optimal matching points relation of global shape feature, as shown in Figure 2, r ₁, r ₂, r ₃and r ₄four unique points respectively with t ₁, t ₂, t ₃and t ₄there is coupling corresponding relation in four unique points;

(3) calculate matching similarity: after obtaining optimal matching points by global shape feature, calculate r ₁, r ₂, r ₃and r ₄the mean place of four unique points, is designated as r, in like manner also calculates t ₁, t ₂, t ₃and t ₄the mean place of four unique points, be designated as t, R is moved integrally to t place, position from position r point, now will achieve registration preferably between R and T, so the matching degree of R and T can be passed judgment on by weighing overlapping degree, to each the face key element in T calculate respectively R move front and R move after overlapping similarity, average, be designated as matching similarity ρ, formula is as follows:

$\mathrm{\ρ}=(\frac{A_p{A}_p}{A_r{A}_t}+\frac{A_q{A}_q}{A_r{A}_t})/2---\left(3\right)$

In above formula, A _rfor the area of reference surface factor combination, A _tfor the area of target face factor combination, A _pfor the lap area of R translation the former two, A _qfor the latter two lap area of R translation;

(4) matching similarity is calculated to each the face key element in target face factor combination T, and be that ordered pair matching similarity sorts with size: each the face key element in target face factor combination T is chosen successively according to descending, when reference surface factor combination and its matching similarity value of target face factor combination chosen diminish, then stop the face key element added in T, otherwise then continue the face key element added in T, then formed face factor combination is optimum matching face factor combination; As shown in Figure 2, when moving to empty frame position place with reference to face factor combination R, calculate single key element T respectively ₁to T ₆with the matching similarity of R, suppose that matching similarity is according to T ₁to T ₆order reduce successively, then first choose T ₁face key element, matching similarity is now assumed to be ρ 1, then adds T again in target face factor combination ₂face key element, found that and work as T ₁with T ₂combine, then calculate and the matching similarity of R time, this value increases than ρ 1, then accept T ₁with T ₂combination and continue to add T ₃face key element, found that matching similarity increases again further, then continue to add T ₄face key element, found that matching similarity reduces, then stop adding any key element, finally can select T successively ₁, T ₂and T ₃three face key elements, and they are combined and become R ₁and R ₂optimum matching face factor combination;

The conflict probe of step 4 matching result and elimination:

When candidate matches is to after carrying out the calculating of matching similarity according to formula (3), matching result may be caused to occur conflicting phenomenon, and as shown in Figure 3, can there are six groups of matching relationships, be (R respectively ₁): (T ₁) matching similarity 0.82, (R ₂): (T ₂) matching similarity 0.82, (R ₃): (T ₃) matching similarity 0.85, (R ₁, R ₂): (T ₁, T ₂) matching similarity 0.9, (R ₁, R ₃): (T ₁, T ₃) matching similarity 0.82, (R ₁, R ₂, R ₃): (T ₁, T ₂, T ₃) matching similarity 0.85, be present in phenomenon that is conflicting and conflict between obvious matching result, for this reason, can according to matching similarity value size, give up matching result or retain, the matching result that should first be retained in Fig. 3 is (R ₁, R ₂): (T ₁, T ₂), because its matching similarity is maximum, then retain matching result (R again ₃): (T ₃), adopt the method can solve contradiction and the conflict of matching result preferably, thus realize the sane coupling of multi-source multiple dimensioned key element.

The inventive method is by practical application, and achieve very satisfied Advantageous Effects, the embodiment provided for Fig. 4, concrete condition is as follows:

The one group of separate sources provided by Fig. 4 a, 4b, resident's terrain element of Different periods verify algorithm of the present invention, and wherein, with reference to residential feature class See Figure 4a, total 703 key elements, target residential feature class is shown in Fig. 4 b, has 820 face key elements.Although totally look that two groups of data are comparatively similar, in fact have many faces key element not only to have larger difference in shape, also there is nonconforming range deviation in position, some local deviation is comparatively large, and some place is then less.

(1) ability of algorithm process complex match relation

Algorithm can to 1:1,1:N, M:N, the various matching relationship such as 1:0,0:1 carries out the Auto-matching of face key element, and table 1 has added up match condition and the matching similarity of various matching relationship in this example, to this example, reference data is concentrated and is had 7 face key elements and there occurs error hiding, and 10 face key elements there occurs Lous coupling, 12 its match-type identification errors of face key element.Target data is concentrated and is had 10 face key elements and there occurs error hiding, and 16 face key elements there occurs Lous coupling, 13 its final match-type mistakes of face key element.Total matching accuracy rate is 1-(7+10+12+10+16+13)/(703+820)=94.9%.

Fig. 5 illustrates reference surface factor data collection when matching relationship is 1:1 and target face factor data collection, Fig. 6 illustrates reference surface factor data collection when matching relationship is 1:N and target face factor data collection, Fig. 7 illustrates reference surface factor data collection when matching relationship is M:1 and target face factor data collection, and Fig. 8 illustrates reference surface factor data collection when matching relationship is M:N and target face factor data collection.

The statistics of the various match-type match condition of table 1 and matching similarity

(2) ability of algorithm opposing face elements position deviation

The inventive method does not require that two groups of face factor datas are very close in position, even if there is nonuniformity position deviation phenomenon in be matched two groups of face key elements, namely very little in some local position deviation, very large in some local position deviation, still can obtain correct matching result.Fig. 9 is a partial enlarged drawing in Fig. 4, and the matching result at this place is in table 2, and matching result has 1:1,1:N and M:N tri-kinds of match-types.Observe Fig. 9 can find, No. 13 key elements (see black matrix mark) of reference surface factor data collection (see solid line polygon) are very large with No. 58 key elements (marking see the italic) range deviation in target face factor data collection (see dashed polygon), equally, No. 197 key element range deviations that No. 338 key elements of reference surface factor data collection and target face factor data are concentrated are also quite large, but check the matching result of table 2, can find out that algorithm still correctly can mate this two example.As shown in Table 2, for this example, no matter less than normal of the same name elements position algorithm is, or of the same name elements position deviation is larger, all can obtain correct matching result, this just shows that algorithm can resist the nonuniformity position deviation between of the same name key element preferably, reduces the requirement to data and restriction, makes algorithm have application scenario widely.

The list of table 2 matching result

Source key element	Target component	Match-type	Matching similarity
				29	191	1:1	0.58
27	207	1:1	0.72
				338	197	1:1	0.41
117	183	1:1	0.66
				26	214	1:1	0.65
119	165	1:1	0.56

13	58,59,60,61,62	1:N	0.54
				52	82,85	1:N	0.37
1,2,32	8,12	M:N	0.57
				25,27,28	13,14,15,16,18,20	M:N	0.52

Can clearly be found out by above-mentioned, Advantageous Effects of the present invention is:

1, the various matching relationship such as 1:1,1:N and M:N between of the same name key element can be obtained;

Even if 2 when of the same name key element exists larger position deviation or even nonconforming position deviation, still effectively can mate of the same name key element;

3, unlike additive method, need to arrange threshold value, and the present invention need not the setting of threshold value, the present invention is by carrying out bi-directional matching, when reference surface key element and target face key element can the optimum matching object of the other side each other each other time, just determine their matching relationship, avoiding problems the setting of threshold value.Whole process completes automatically.

In sum, the applicability of face, map conflation field key element matching technique has been expanded in this invention, it is made to reduce further the requirement of data qualification and restriction, reduce pretreated workload, promote the integrated, shared and multiplexing of spatial data, be effective to map vector merging, map rejuvenation and change detect, there is very strong practicality.

Claims (2)

1. multi-source resident terrain element exempt from a threshold value automatically sane matching process, it is characterized in that, comprise the following steps:

Step one, data acquisition:

Read two groups of settlement place data to be matched respectively with computing machine, one group of resident's terrain element is reference surface factor combination, is designated as R, and another group resident terrain element is target face factor combination, is designated as T;

Step 2, generation candidate matches pair:

For reference surface factor combination R to be matched and target face factor combination T, judged by the crossover of locus overlapping relation, candidate matches face factor combination is selected successively from reference surface factor combination and target face factor combination, generate candidate matches pair, candidate matches is to comprising the matching relationship existed between 1:1,1:N, M:1 and M:N of the same name key element;

The judgement of step 3, of the same name key element matching relationship:

Adopt the matching similarity of two-way calculating face key element to determine the matching relationship of of the same name key element, namely for reference surface factor combination R and target face factor combination T, when according to calculated matching similarity being mutually the optimum matching face factor combination of the other side, and the two coupling corresponding relation is set up;

In order to effectively process various complex match corresponding relation, its calculating thinking of described matching similarity is to judge the matching degree between the key element of face by weighing overlapped degree between the key element of separate sources face; Even if in order to ensure when of the same name key element has larger position deviation, still effectively mate, then first extract the enclosure rectangle of reference surface factor combination and the global shape feature of individual features point, the coupling corresponding relation of unique point of the same name in automatic acquisition of the same name key element, position skew is carried out with reference to face factor combination R, making two groups of to be matched factor combinations in position can registration further, then passing judgment on the matching degree of reference surface factor combination R and target face factor combination T again by weighing overlapping degree;

The conflict probe of step 4, matching result and elimination:

According to matching similarity value size, the matching result clashed is given up or retained, namely retain matching similarity value large, and it is less to give up matching similarity, thus the contradiction realizing matching result is eliminated.

2. multi-source resident terrain element according to claim 1 exempt from threshold value automatically sane matching process, it is characterized in that, realized by following concrete steps:

Step one, data acquisition:

Read two groups of settlement place data to be matched respectively with computing machine, one group of settlement place parameter is reference surface factor combination, is designated as R, and another group settlement place parameter is target face factor combination, is designated as T;

Step 2, generation candidate matches pair:

For various matching relationship between of the same name key element correctly can be obtained, first potential candidate matches is generated to set, this set by all couplings correctly to being included, for to be matched two groups of face factor data set, one group is referred to as reference surface elements combination, another group is referred to as target face elements combination, crossover selects candidate matches face factor combination successively from reference surface elements combination and target face elements combination, form candidate matches pair, reference surface key element R has three face key elements, i.e. the first reference surface key element R ₁, the second reference surface key element R ₂with the 3rd reference surface key element R ₃, have three face key elements in target face factor combination T, i.e. first object face key element T ₁, the second target face key element T ₂with the 3rd target face key element T ₃; Order chooses each face key element in reference surface elements combination, when in target face elements combination with the first reference surface key element R ₁that there is overlapping relation is first object face key element T ₁, obtain first group of candidate matches to (R ₁): (T ₁), and then find and first object face key element T in reference surface elements combination ₁the reference surface key element that there is overlapping relation is the first reference surface key element R ₁, the second reference surface key element R ₂, obtain the second group and be paired into (R ₁, R ₂) ︰ (T ₁), then obtain and the first reference surface key element R ₁, the second reference surface key element R ₂the target face key element intersected, obtains the 3rd group of candidate matches to being (R ₁, R ₂): (T ₁, T ₂, T ₃), obtain the 4th candidate matches until last to being (R ₁, R ₂, R ₃) ︰ (T ₁, T ₂, T ₃); When meeting two conditions, the generation that candidate matches is right stops, and these two conditions are: one is cannot obtain candidate matches pair further, and in acquisition the 4th group of candidate matches to rear, each face key element participates in computing all, new candidate matches pair of cannot having reentried; Two is that the candidate matches of preparation generation is to generating before this;

When the 4th group of candidate matches is to being (R ₁, R ₂, R ₃) ︰ (T ₁, T ₂, T ₃), extract and (R from target face factor combination further ₁, R ₂, R ₃) combine (T the most similar ₁, T ₂) combination, be finally correct matching result;

The judgement of step 3, of the same name key element matching relationship:

The calculating of face key element matching similarity is carried out respectively from both direction, right to obtain coupling, method is, first with reference surface factor combination R for benchmark, according to calculated matching similarity, optimum matching target face factor combination T is found from target face key element, then again with target face factor combination T for benchmark, according to calculated matching similarity, optimum matching reference surface factor combination is found from reference surface factor combination, when this optimum matching face factor combination is reference surface factor combination R, then reference surface factor combination R becomes one group of optimum matching face factor combination each other with target face factor combination T-shaped,

In acquisition candidate matches to afterwards, need to weigh the matching degree between reference surface factor combination and target face factor combination, and determine best matching result from candidate matches centering accordingly, degree overlapped between the key element of employing face weighs the matching degree of face key element, according to the reference surface factor combination set above as R, target face factor combination is T, if the area of reference surface factor combination R is A _r, the area of target face factor combination T is A _t, the overlapping area of reference surface factor combination R and target face factor combination T is A _o, then overlapping similarity Sim is:

$\mathrm{Sim}=\frac{A_o{A}_o}{A_r{A}_t}---\left(1\right)$

By calculate overlapping similarity judge candidate matches between matching degree, its precondition be the two can coincide in position, when there being a candidate matches pair, its reference surface factor combination is (R ₁, R ₂), target face factor combination is (T ₁-T ₆), deviation is very large in position for the two, now cannot reflect matching degree therebetween, in the larger situation of position deviation, by following steps determination candidate matches face factor combination by weighing overlapping degree:

(1) the global shape feature of reference surface factor combination R is extracted: global shape feature refers to enclosure rectangle and the individual features point of R, first finds spacing two summits farthest on each summit in reference surface factor combination R, is respectively r ₁, r ₃, then respectively at line segment r ₁, r ₃left side and right side find distance line segment r ₁, r ₃having the point of maximum distance, is r ₂and r ₄, then by r ₁, r ₂, r ₃and r ₄the enclosure rectangle of four unique point composition reference surface factor combination R, a limit of enclosure rectangle and r ₁r ₃line parallel, another limit then with r ₁r ₃line is vertical, and this enclosure rectangle has the feature irrelevant with similarity transformation, and when reference surface factor combination R entirety offsets, rotates and convergent-divergent, the enclosure rectangle extracted is still by r ₁, r ₂, r ₃and r ₄four unique point compositions;

(2) for target face factor combination T, find and r ₁, r ₂, r ₃and r ₄optimum matching corresponding vertex: method is, order chooses each face key element in target face factor combination T, then order chooses each summit of this face key element, if the summit selected is t ₁, in T, find a summit p, make by a t ₁the vectorial U formed to p with by a r ₁to r ₂the vector similarity value of the vectorial V formed is maximum, and namely similar between vector, the computing formula of vector similarity α is as follows:

$\mathrm{\α}=\frac{\min \left(\right|\left|U\right||,|\left|V\right|\left|\right)}{\max \left(\right|\left|U\right||,|\left|V\right|\left|\right)}\×\frac{[U\·V]}{\left|\right|U\left|\right|\left|\right|V\left|\right|}---\left(2\right)$

This summit is found to be t ₂, find summit t respectively according to same steps ₃and t ₄;

According to the method described above, according to each summit of face key element in T, other three summits can be found, and successive computations three maximum vector similarities, get the mean value of three maximum vector similarities as resultant vector similarity, when resultant vector similarity value is maximum, namely obtain reference surface factor combination and target face factor combination the optimal matching points relation of global shape feature, r ₁, r ₂, r ₃and r ₄four unique points respectively with t ₁, t ₂, t ₃and t ₄there is coupling corresponding relation in four unique points;

(3) calculate matching similarity: after obtaining optimal matching points by global shape feature, calculate r ₁, r ₂, r ₃and r ₄the mean place of four unique points, is designated as r, calculates t equally ₁, t ₂, t ₃and t ₄the mean place of four unique points, be designated as t, R is moved integrally to t place, position from position r point, now will realize registration preferably between R and T, pass judgment on the matching degree of R and T by weighing overlapping degree, to each the face key element in T calculate respectively R move front and R move after overlapping similarity, average, be designated as matching similarity ρ, formula is as follows:

$\mathrm{\ρ}=(\frac{A_p{A}_p}{A_r{A}_t}+\frac{A_q{A}_q}{A_r{A}_t})/2---\left(3\right)$

Wherein, A _rfor the area of reference surface factor combination, A _tfor the area of target face factor combination, A _pfor the lap area of R translation the former two, A _qfor the latter two lap area of R translation;

(4) matching similarity is calculated to each the face key element in target face factor combination T, and be that ordered pair matching similarity sorts with size: each the face key element in target face factor combination T is chosen successively according to descending, when reference surface factor combination and its matching similarity value of target face factor combination chosen diminish, then stop the face key element added in T, otherwise then continue the face key element added in T, the face factor combination formed is optimum matching face factor combination; When moving to empty frame position place with reference to face factor combination R, calculate single key element T respectively ₁to T ₆with the matching similarity of R, if matching similarity is according to T ₁to T ₆order reduce successively, then first choose T ₁face key element, matching similarity is now ρ 1, then adds T again in target face factor combination ₂face key element, by T ₁with T ₂combine, when calculating the matching similarity with R, this value increases than ρ 1, then accept T ₁with T ₂combination and continue to add T ₃face key element, when matching similarity increases further, then continues to add T ₄face key element, matching similarity reduces, then stop adding any key element, finally select T successively ₁, T ₂and T ₃three face key elements, combine and become R ₁and R ₂optimum matching face factor combination;

The conflict probe of step 4 matching result and elimination:

When candidate matches is to after carrying out the calculating of matching similarity according to formula (3), matching result may be caused to occur conflicting phenomenon, and as shown in Figure 3, can there are six groups of matching relationships, be (R respectively ₁): (T ₁) matching similarity 0.82, (R ₂): (T ₂) matching similarity 0.82, (R ₃): (T ₃) matching similarity 0.85, (R ₁, R ₂): (T ₁, T ₂) matching similarity 0.9, (R ₁, R ₃): (T ₁, T ₃) matching similarity 0.82, (R ₁, R ₂, R ₃): (T ₁, T ₂, T ₃) matching similarity 0.85, be present in phenomenon that is conflicting and conflict between obvious matching result, for this reason, can according to matching similarity value size, give up matching result or retain, the matching result that should first be retained in Fig. 3 is (R ₁, R ₂): (T ₁, T ₂), because its matching similarity is maximum, then retain matching result (R again ₃): (T ₃), adopt the method can solve contradiction and the conflict of matching result preferably, thus realize the sane coupling of multi-source multiple dimensioned key element.