CN1110775C - Automatic contrast system for planedrawing points - Google Patents
Automatic contrast system for planedrawing points Download PDFInfo
- Publication number
- CN1110775C CN1110775C CN96119761A CN96119761A CN1110775C CN 1110775 C CN1110775 C CN 1110775C CN 96119761 A CN96119761 A CN 96119761A CN 96119761 A CN96119761 A CN 96119761A CN 1110775 C CN1110775 C CN 1110775C
- Authority
- CN
- China
- Prior art keywords
- point
- pairing
- theta
- characteristic direction
- match
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Expired - Lifetime
Links
Images
Landscapes
- Image Analysis (AREA)
Abstract
The present invention provides an automatic contrast system for plane figure points. The method for the system comprises the following steps of roughly contrasting, providing a lowest condition for satisfying the matching operation, calculating the matching probability, using a fuzzy relaxation method to calculate the matching probability of all the points, selecting matching points, namely, the probable matching points are selected according to the matching probability among all the points between the two figures, calculating the degree of approximation between the two figures, calculating the index of the degree of approximation between the two figures and judging whether the two figures are approximate. Whether the two figures are approximate is determined according to a preset critical value.
Description
The present invention relates to the automatic comparison system of a kind of planimetric map form point, particularly the planar point more than two groups set is contrasted its feature, the method and apparatus that whether is similar to the figure that determines this point set representative about a kind of.
Contrasting two sheet of planar graphic materials is important topics of computer video and image processing.After input imagery being done special processing (for example feature extraction), required unique point in the time of can be done Figure recognition according to this.Contrast the identical degree of unique point of two block graphicses, can be used to judge that two picture groups resemble the similarity degree of data.For example, finger print image can obtain the unique point on the finger print image after treatment, utilizes these characteristic point coordinates and characteristic direction, can judge that two fingerprints are the fingerprint of same finger.
When the contrast of processing such as graphic materials such as handwriting, fingerprint, the difficulty of Xie Jueing comprises possibly: (1) even figure to be contrasted from same source, contained feature is counted may be different, it is unpredictable that first vegetarian refreshments of (2) two block graphicses is positioned at the probability of same position (coordinate), and the unique point in (3) set of diagrams shape to distribute may be result etc. after another picture group form point changes through rotation, translation or amplification etc.These characteristics are the greatest difficulties of carrying out the figure contrast.
The automatic compare device and the control methods that have disclosed a kind of planimetric map form point in No. 79109743 Taiwan patent, the control methods that this invention is adopted is " fuzzy relaxation (Fuzzy Relaxation) ", be according to " calculate in two figures one group a bit with any pairing probability of another group the time; under these 2 prerequisites that constitute pairing; calculate the pairing of other points with other points of another group of this group; to the supporting rate of this pairing " calculate between two groups of planimetric map form points one group a bit and the pairing probability of another group between have a few, and utilization is chosen the match point that method is chosen the best to front sequence.Then reach the relation of " zoom factor " four kinds of factors again according to " pairing rate ", " on average match probability ", " distance between average match point ", treat the figure of contrast and make geometric transformation, try to achieve the most similar graphical dots group.In this, so-called " pairing " is meant that when two graphical set are similar one group of contained point should be organized contained point with another and should be overlapping one to one or neighbouring relations in coordinate system.In the present invention, also continue to use it.
The fuzzy relaxation account form that in this patent case, is proposed, can represent by following formula:
S wherein
(r)(pi qj) represents when being adapted to the r time, the supporting rate of pi point and the pairing of qj point, and pi represents a bit in the resolution chart (P) to be contrasted, and qj represents a bit in the reference pattern (Q) to be contrasted,
1 represents 2 distances when same coordinate plane, and among the h representative of graphics P, the point of non-pi, the point of non-qj among the k representative of graphics Q, m are the contained minimum value of counting of P, Q two figures, S
(0)(pi is 0 or 1 qj).
In addition, be to improve processing speed, this invention also provides a kind of " thick pairing comparision ", and screening in advance must not become between the point of pairing and concerns, with saving computing time.Think through comparing can not pairing point, its S
(0)(pi qj) is made as 0, otherwise is 1.
For solving the problem that distortion, distortions such as the rotation of graphic materials unique point, translation or ratio convergent-divergent are produced, this invention more provides a kind of utilization " least mean-square error " value, graphical dots to desire contrast is done " rotation ", " translation " and/or processing such as " proportional zooms ", this picture group form point is adjusted into the situation of the most suitable contrast.
Though above-mentioned No. 79109743 Taiwan patent case provides a kind of high accuracy and the automatic contrast discrimination system of planimetric map form point rapidly, but still has the higher shortcoming of false rejection rate.Contrasting in a large number with 800 pieces of fingerprints, is 0.1% o'clock at false acceptance rate (FAR), and the false rejection rate of this method (FRR) is 25%.Therefore, seek a kind of reduction FRR, and can take into account height and distinguish rate and the automatic control methods of planimetric map form point fast, become and be engaged in the problem that image comparison technical research personnel try hard to overcome.
The object of the present invention is to provide a kind of automatic compare device of planimetric map form point, may the unpredictable graphical dots of ratio different and/or that point repeats do correct contrast to always counting.
Even purpose of the present invention also is to provide a kind of graphical dots generation translation, rotation, the automatic compare device of planimetric map form point that still can correctly contrast.
Another object of the present invention also is to provide a kind of automatic compare device of planimetric map form point of quick contrast.
Another object of the present invention also is to provide the automatic control methods of planimetric map form point of above-mentioned effect.
For addressing the above problem, method proposed by the invention may further comprise the steps:
1. thick contrast: when any point that any point and another figure comprised that proposes to be comprised in the figure formed pairing, the lowest term that institute should satisfy was screening out the point that can not match in two figures, so as to saving computing time.
2. calculate the pairing probability: utilize fuzzy relaxation, calculate the pairing probability of being had a few in have a few and another figure in the figure.
3. choose match point:, choose the point that may match in two figures according to the height of the pairing probability between having a few between two figures.
4. calculate the degree of approximation between two figures: the similarity degree index of calculating two figures.And
5. judge whether two figures are approximate:, determine whether two figures are approximate according to a default critical value.
In the present invention, the condition of thick contrast comprises: the direction condition of (1) unique point, the X-axis translation condition of (2) unique point, and the Y-axis translation condition of (3) unique point.When any point in the figure and any direction and the coordinate characteristic when satisfying this three condition in another figure, think that be possible pairing at these 2; Otherwise think impossible pairing, give mark.
When calculating two figure similarity degree indexs among the present invention, it considers that factor comprises: form counting of pairing between (1) two figure, and (2) reference pattern pairing rate, (3) resolution chart pairing rate, (4) on average match probability, and (5) on average match the degree of approach.The index mark is high more, represents two figures similar more.
In addition, in the present invention, also be included in preliminary pairing after, according to the result of pairing, obtain best how much adjusted values, to adjust resolution chart, reach correct contrast and distinguish the result.
When calculating the pairing probability, the present invention proposes a kind of " fuzzy relaxation " implementation step and mode of novelty, to improve accuracy and the speed that contrast is distinguished.The fuzzy relaxation that the present invention proposes can be represented by following formula:
Above-mentioned and other objects and advantages of the present invention can be by following detailed description, and with reference to accompanying drawing, and clearer:
Fig. 1 represents the automatic compare device's of planimetric map form point of the present invention system diagram.
Fig. 2 represents the automatic control methods process flow diagram of planimetric map form point of the present invention.
Table 1-1 represents reference pattern Pattern A to be contrasted.
Table 1-2 represents reference pattern Pattern B to be contrasted.
Table 1-3 represents the reference pattern Pattern C that another is to be contrasted.
The thick pairing result of table 2 expression Pattern B and Pattern A.
The thick pairing result of table 3 expression Pattern C and Pattern A.
Table 4 is illustrated among Pattern B and the Pattern A, under b1 and the a1 pairing prerequisite, and the possibility of other pairing.
Table 5 expression Pattern B and Pattern A are with the fuzzy relaxation comparing result.
Table 6 expression Pattern C and Pattern A are with the fuzzy relaxation comparing result.
Table 7 expression Pattern B and Pattern A tentatively choose the pairing result.
Table 8 expression Pattern C and Pattern A tentatively choose the pairing result.
Table 9 expression Pattern B through first adjusted coordinate and with the range difference of Pattern A match point.
Table 10 expression Pattern B through adjusted coordinate once again and with the range difference of Pattern A match point.
Table 11 expression Pattern C through first adjusted coordinate and with the range difference of Pattern A match point.
Table 12 expression Pattern C through adjusted coordinate once again and with the range difference of Pattern A match point.
Table 13 expression the present invention and prior art effect comparison sheet.
Consider two groups of planimetric map form points (Planer Point Pattern):
P={p1, p2 ..., pm} is resolution chart (Test Pattern), total m point.
Q={q1, q2 ..., qn} is reference pattern (Reference Pattern), total n point.
Wherein each point is so that (x, y represent that D) (x y) is some coordinate in the plane, and D is the characteristic direction of point.
Employed in the present invention planar graph, normally with graphic materials through after the pre-treatment, obtain its image feature point and get.For example, distinguish in the application that unique point may be meant the end points and the point of crossing of streakline at finger print image.In the application that handwriting is distinguished, may be the end points and the point of crossing of literal line segment.When the contrast of other graphic materials is used, then need determine unique point according to it.
In the present invention, above-mentioned " characteristic direction " is defined as:
This point for its tangential direction that links line segment or segment of curve, represents with angle during for the end points of line segment or segment of curve, and is zero degree with positive horizontal direction, is forward with the counter clockwise direction, and clockwise direction is a negative sense.This point is during for the point of crossing of line segment or segment of curve, is the two decomposition line directions of intersecting the acute angle angle that line segments or segment of curve form, and represent with angle, and be zero degree with positive horizontal direction, with the counter clockwise direction forward, clockwise direction is a negative sense.
When other were used, its characteristic direction can define according to the feature of figure.
If the P figure must be to be rotated counterclockwise the θ angle, (tx ty) afterwards, just can make point in most point and the Q figure overlapping or successfully match along X-axis and Y-axis translation again.We can (θ) function be described the geometric relationship of this pairing for tx, ty.When just seeking best pairing, (tx, ty θ) make the P figure through the rotation of θ angle and (tx, ty) after the translational movement, the contained point of the point that it is contained and Q figure has the lap of maximum or near lap to try to achieve one group.
Under this definition, suppose that 1 pi in the resolution chart and 1 qj in the reference pattern are one when successfully matching, then this pairing must be satisfied following three formulas:
X
qj=tx+x
picosθ-y
pisinθ (1)
Y
qj=ty+x
pisinθ+y
picosθ (2)
D
qj=(D
pi+Dr)+θ (3)
Wherein Xpi and Ypi distinguish the coordinate of representative point pi, D
PiRepresent its characteristic direction; Xqj and Yqj be the coordinate of representative point qj respectively.D
QjRepresent its characteristic direction.Tx, ty, the definition of θ is the same.Dr is change difference.
(θ=0), D ideally when angle without spin
QjWith D
PiShould be identical.When rotation takes place, D
QjShould equal D
PiAdd the anglec of rotation.But actual in extracting some change difference on the characteristic direction, represent this difference with Dr.Dr is defined as: after the P graphical dots is rotated through the θ angle, and the measures of dispersion of the characteristic direction of the characteristic direction of the some pi of P figure and the some qj of Q figure, its unit is an angle.
Below with example the automatic comparison system of planimetric map form point of the present invention is described.Fig. 1 represents the automatic compare device's of planimetric map form point of the present invention system diagram.As shown in the figure, the automatic compare device of planimetric map form point of the present invention comprises a sampler (1), a thick contrast means (2), one pairing probability calculation device (3), one selecting device (4), an adjusting gear (5), a figure similarity calculation element (6) and a judgment means (7).
Following with reference to the action of flowchart text device of the present invention and the step of the inventive method.Fig. 2 represents the process flow diagram of the automatic control methods of planimetric map form point of the present invention.As shown in the figure, in (101) step, sampler (1) is obtained two groups of planimetric map form point data, comprises the coordinate and the characteristic direction thereof of each graphical dots.
[thick pairing]
In (102) step, thick contrast means (2) is slightly matched to the point in this two figure.
The purpose of thick pairing be to remove some pi in the P figure (i=1,2 ..., m) with the Q figure in some qj (j=1,2 ..., combination that can not pairing between n) is to save the processing time.The condition of screening can be derived by (1), (2), (3) formula, comprising:
(a) unique point direction condition
By (3) formula, can get:
D
qj-D
pi=Dr+θ
|D
qj-D
pi|<=|Dr|+|θ|
Make that DR is the maximum change direction of tolerance, θ T is the maximum anglec of rotation of tolerance, then:
|D
qj-D
pi|<=DR+θT (4)
If D
QjWith D
PiDifference greater than (DR+ θ T), then pi<==>qj can not match.
(b) X-axis translation condition
By (1) formula, can get:
x
qj-x
pi=tx+x
pi(cosθ-1)-y
pisinθ
|x
qj-x
pi|≤|tx|+x
pi(cosθ-1)-y
pisinθ|
Make TX be | tx|+|x
Pi| (cos θ-1)-y
PiSin θ | maximal value, then:
|X
qj-X
pi|≤TX (5)
If X
QjWith X
PiDifference greater than TX, then pi<==>qj can not match.
(c) Y-axis translation condition
By (2) formula, can get
y
qj-y
pi=ty+x
pisinθ+y
pi(cosθ-1)
|y
qj-y
pi|≤|ty|+x
pisinθ+y
pi(cosθ-1)|
Make TY be | ty|+|x
PiSin θ+y
Pj(cos θ-1) | maximal value, then:
|y
qj-y
pi|≤TY (6)
If y
QjWith y
PiDifference greater than TY, then pi<==>qj can not match.
With S
(0)(pi qj) represents the initial pairing probability of pi and qj, if pi<==>qj can not match, and then makes S
(0)(pi, qj)=0; If pi<==>qj may match, then make S (pi, qj)=1.Therefore, in pi<==when qj satisfies (4), (5), (6) three conditions, make S (pi, qj)=1; Otherwise make S (pi, qj)=0.This condition inspection can be referred to as thick pairing.
[fuzzy relaxation (Fuzzy Relaxat ion Method)]
If any point qj in any point pi in the P graphical dots and the Q graphical dots " initially match probability " is identical.In thick comparison process, specify one " initial pairing probability " to each to match point: if can not match, then establishing probability is zero; Then be made as 1 if may match.
In (103) step, pairing probability calculation device (3) utilizes fuzzy relaxation to revise the pairing probability of each pairing repeatedly.At this, the initial pairing probability of pi and qj according to ' when pi and qj are correct pairings, other points of P figure (ph, h=1,2 ..., m, other points of h ≠ i) and Q figure (qk, k=1,2 ..., n, k ≠ j) become the possibility of pairing ' revise repeatedly.
When supposing to be adapted to the r time repeatedly, the pairing probability of pi and qj is S
(r)(pi, qj), and S
(0)(pi qj) is ' an initial pairing probability '.S then
(0)(pi, qj) can revise by following formula:
Though do not restrained by any theory, adopt the reason of max computing to be in the following formula: because of pi and qj are 1 pair 1 corresponding, the some ph of P figure only looks for 1 qk in the Q figure, and forms the pairing probability maximum of matching.
In following formula, Cij (h k) is representative ' when pi and qj are correct pairings, the possibility that other match points ph and qk match ':
Δ
s: work as D
s≤ N
sThe time,
Otherwise Δ
s=1; (d) Δ
c: work as D
c≤ N
cThe time,
Otherwise Δ
c=1; (e) Δ
l: work as d
l≤ N
lThe time,
Otherwise Δ
l=1; (f)
Wherein, D
sBe the differential seat angle of initiation feature direction, N
sBe fixed maximum tolerance start angle poor.With N
sDifferential seat angle is normalized between 0 to 1.If differential seat angle D
sGreater than N
s, then establish Δ
s=1.So 0≤Δ
s≤ 1;
D
eBe the differential seat angle of terminal point characteristic direction, N
eBe fixed maximum tolerance terminal point differential seat angle.With N
eDifferential seat angle is normalized between 0 to 1.If differential seat angle D
eGreater than N
e, then establish Δ e=1.So 0≤Δ
e≤ 1;
d
1Be two the pairing between length distance poor, N
1Be the length distance of fixed maximum tolerance poor.With N
1Length difference is normalized between 0 to 1.If range difference d
1Greater than N
1, then establish Δ
1=1.So 0≤Δ
l≤ 1; If emphasize that the influence of length difference is greater than differential seat angle, then if range difference d
1Greater than N
1, then establish Δ l=Δ s=Δ e=1 simultaneously.
In (c) formula, Δ
θIt is Δ
sWith Δ
eAverage, so 0≤Δ
θ≤ 1.
In (b) formula, with weighted number w
1And w
θAdjust Δ
lAnd Δ
θTo the ratio that influences of Δ, and with this measure the difference regular turning between 0 to 1, so 0≤Δ≤1.
Δ and C in (a) formula
Ij(h k) is inversely proportional to.Because 0≤Δ≤1, so 0.5≤C
Ij(h, k)≤1.0.So ' pairing probability ' will successively decrease monotonously by 1 after revising repeatedly.Repeated calculation for several times after, match correct point, the value of its ' pairing probability ' will raise; The point that pairing is wrong, its ' pairing probability ' levels off to 0.
By the aforementioned calculation result, can find out any point in certain picture group form point, with any point in another picture group form point, form the height of pairing possibility.
[tentatively choosing of pairing]
After repeatedly ' pairing probability ' being revised for several times with ' fuzzy relaxation ', the best pairing point between P and the Q just can be selected according to the value of ' pairing probability '.Therefore in (104) step, selecting device (4) with the height of ' pairing probability ' value S as the standard of choosing match point.
When at first order matrix S (i) (j) represents pi and qj pairing, final " pairing probability " S through adjusting (pi, qj), wherein: i=1,2,3 ..., m, j=1,2,3 ..., n; Wherein m is the some number of P, and n is the some number of Q.Then the maximum between two figures pairing number K not can greater than m or n:K≤min (m, n).
The mode that selecting device (4) is chosen match point comprises:
Obtain the pairing probability matrix S (i) (j) of each point between two figures in (104a) step;
Establish k=1 in (104b) step;
Find out pairing pa of maximal value smax and qb in (104c) step in S (i) (j) lining, think that an assembly is right.With op (k) record pa point, oq (k) record qb point, sim (k) writes down the probability smax of this pairing.
Going on foot in S (i) (j) matrix at (104d), is zero with a row and the capable all elements removing of b.
Make k=k+1 in (104e) step;
Whether judge smax=0 in (104f) step,, get back to (104c) as not;
In this way, then in (104g) with selected N assembly to (N=K-1) as choosing the result.
[the thin portion of pairing chooses]
Through tentatively choosing the N assembly selected to being correct pairing.But " pairing probability " high more pairing should be correct more pairing in theory.So go on foot in (105), adjusting gear (5) is selected the pairing of the highest preceding G group of pairing probability according to selecting device (4), find out how much best adjustment amount (tx, ty, θ), after making the p point of this G assembly centering through rotation and translation conversion, and the square distance between its corresponding point q and reach minimum.Just look for one group (tx, ty, θ), make E (tx, ty, θ) minimum, wherein:
With the p point of N assembly centering with the best (tx, ty, θ) do the rotation translation and adjust after, in (106) adjusting gear (5) judge group p point and corresponding some q between the two square distance and whether less than fixed scope d1.Then keep this pairing in this way, otherwise delete this pairing.
It is right to suppose to keep the M assembly through screening back, (107) step engagement positions (5) find out once again right best how much adjustment amounts of this M assembly (tx, ty, θ).Just find out one group (tx, ty, θ) make E (tx, ty are minimum θ), wherein:
With the p point of this M assembly centering with the best (tx, ty, θ) do the adjustment of rotation translation after, in (108) step, adjusting gear (5) judge once again this group p point and corresponding some q distance between the two whether less than fixed scope d2.In this way, then keep this pairing, otherwise also delete this pairing.At this d2<=d1.
After choosing through twice, it is right to obtain the K assembly at last.In (109) step, figure degree of approximation calculation element (6) writes down mean distance between this K pairing with diff, with the average pairing probability of md record match point.
[calculating of two figure similarity degrees]
According to preceding method, can obtain the K that counts that matches between two picture group form points, each pairing probability S (i) (j), and coordinate of each pairing (px ', py ') with (qx, qy).Go on foot in (110), figure degree of approximation calculation element (6) can be according to these data, probability is on average matched in (1) pairing number, (2) reference pattern pairing rate, (3) resolution chart pairing rate, (4) of calculating two picture group form points, and the distance between (5) average match point and.Then calculate the similarity degree of two figures in (111) with utilizing this five group factor again, and represent with index S core.The high more representative similarity degree of index mark is high more; The low more representative similarity degree of mark is low more.
Each factor is defined as follows:
(1) pairing number: K
The pairing number is represented counting that success is matched between two figures.The pairing number is high more, represents that two figures are similar more, so K is directly proportional with Score.
(2) reference pattern pairing rate: S1
Counting=K/n of S1=pairing number/reference pattern.
When reference pattern is counted when a lot, and the match point that finds seldom, represents that two figure similarity degrees are low, so Score is directly proportional with S1.
(3) resolution chart pairing rate: S2
Counting=K/m of S2=pairing number/resolution chart.
When resolution chart is counted when a lot, and the match point that finds seldom, represents that two figure similarity degrees are low, so Score is directly proportional with S2.
(4) on average match probability: md
The mean value of all pairing probability of md=.
Average pairing probability height, then two figure similarity degrees are high more, so Score is directly proportional with md.
(5) degree of approach between average match point: S3
S3=through adjustment distance between the average match point in conversion back and inverse=1.0/ (1.0+diff).
If average distance is zero, the expression match point is overlapping fully after changing, and represents that the similarity degree of two figures is very high.So Score is directly proportional with S3.In following formula in order to prevent that denominator from being that zero situation takes place, and adds 1 on denominator.
By above-mentioned definition and analysis, learn that Score is directly proportional with K, S1, S2, S3 and md, the definable following formula is represented the similarity degree of two figures:
The C=constant.With a square influence that comes this factor pair of lay special stress on Score.
[judgement]
At last, judgment means (7) judges according to a predetermined critical value (Threshold) whether two block graphicses are approximate in (112) step.That is, at Score during, be judged as approximately greater than this critical value, otherwise cannot not be judged as approximately.
[embodiment]
Below with embodiment, illustrate that the automatic control methods of planimetric map form point of the present invention makes the process of figure contrast.
Three picture group form points on the table 1 expression plane are respectively Pattern A (table 1-1), PatternB (table 1-2) and Pattern C (table 1-3).Each block graphics respectively contains 8 points, and characteristic point coordinates and direction all are shown in the table.Wherein Pattern B is that Pattern A is through translation and postrotational approximate figure; Pattern C and Pattern A are not approximate.Represent reference pattern Q at this Pattern A; Pattern B and Pattern C all represent resolution chart P.
One, thick pairing
In this example, TX is decided to be 50, TY is decided to be 50, (DR+ θ T) is decided to be 60 degree.
(1) contrast of Pattern B and Patter A
With a1 point and b1 point is example, illustrates that thick pairing process is as follows:
A1=in the table 1 (24,48,0), b1=(33,36,20).
(a) unique point direction condition
| D
A1-D
B1|=| 0-20|=20<=60 are so condition satisfies.(b) X-axis translation condition
| X
A1-X
B1|=| 24-33|=9≤50 are so condition satisfies.(c) Y-axis translation condition
| y
A1-y
B1|=| 48-36|=12≤50 are so condition satisfies.Because of thick contrast condition all satisfies, make S
(0)(b1) (a1)=1.The thick comparison process of b3 and a1 is as follows: a1=in the table 1 (24,48,0), b3=(76,53,65).(a) unique point direction condition
| D
A1-D
B3|=| 0-65||=65>60 are not so condition satisfies.(b) X-axis translation condition
| x
A1-x
B3|=| 24-76|=52>50 are not so condition satisfies.(c) Y-axis translation condition
| y
A1-y
B3|=| 48-53|=5≤50 are so condition satisfies.
Do not satisfy because of having ready conditions in the thick contrast condition, make S
(0)(b3) (a1)=0.
Contrast between all the other points repeats no more.Table 2 is results of Pattern B and the thick pairing of Pattern A.
(2) contrast of Pattern C and Pattern A
According to Same Way, contrast Pattern C and Pattern A.Table 3 is results of Pattern C and the thick pairing of Pattern A.
After thick pairing, can screen out impossible pairing over half.
Two, fuzzy relaxation
N in this example
sBe decided to be 45 degree, N
eBe decided to be 45 degree, N
1Be decided to be 10, w
1Be decided to be 1, w
θBe decided to be 1.Revise repeatedly 2 times with fuzzy relaxation, i.e. r=2.
(1) contrast of Pattern B and Pattern A
Under the prerequisite of b1 and a1 pairing, calculate the possibility C of b2 and a2 pairing
11(2,2):
So, Δ
l=d
1/ N
1=1/10=0.1 °
So, Δ
s=D
s/ N
s=2/45=0.0444 °
So, Δ
c=D
c/ N
c=2/45=0.0444 °
Under the prerequisite of b1 and a1 pairing, calculate the possibility C of b2 and a3 pairing
11(2,3) are 0.5.
In like manner, under the prerequisite of b1 and a1 pairing, calculate the possibility C of other some pairing
Ij(h, k), i=1 wherein, j=1, h=2 ..., 8, k=2 ... 8.The result is shown in table 4.According to the numerical value of table 4, can (7) formula correction a1 and the pairing probability of b1.After twice correction, the probability that obtains matching is 0.714.
The pairing probability of Pattern B and other points of Pattern A can be according to calculating with method.Table 5 expression Pattern B and Pattern A show the pairing probability between each point according to fuzzy relaxation result of calculation.
(2) contrast of Pattern C and Pattern A
Table 6 expression Pattern C and Pattern A show the pairing probability of each pairing according to fuzzy relaxation result of calculation.
Three, pairing tentatively chooses
It is right to find out the N assembly in regular turn according to the size of pairing probability.The pairing that is not selected is represented its pairing probable value with 0.
(1) contrast of Pattern B and Pattern A
Size according to the pairing probability is found out 8 assembly in regular turn to (N=8).It is as follows that order is chosen in its pairing:
(b1 and a1 pairing) op[1]=b1, oq[1]=a1, sim[1]=0.714, the 1st row and the 1st pairing probability of going are designated as 0 entirely.
(b2 and a2 pairing) op[2]=b2, oq[2]=a1, sim[2]=0.670, the 2nd row and the 2nd pairing probability of going are designated as 0 entirely.
(b3 and a3 pairing) op[3]=b3, oq[3]=a1, sim[3]=0.617, the 3rd row and the 3rd pairing probability of going are designated as 0 entirely.
(b4 and a4 pairing) op[4]=b4, oq[4]=a1, sim[4]=0.601, the 4th row and the 4th pairing probability of going are designated as 0 entirely.
(b5 and a5 pairing) op[5]=b5, oq[5]=a1, sim[5]=0.567, the 5th row and the 5th pairing probability of going are designated as 0 entirely.
(b6 and a6 pairing) op[6]=b6, oq[6]=a1, sim[6]=0.547, the 6th row and the 6th pairing probability of going are designated as 0 entirely.
(b7 and a7 pairing) op[7]=b7, oq[7]=a1, sim[7]=0.547, the 7th row and the 7th pairing probability of going are designated as 0 entirely.
(b8 and a8 pairing) op[8]=b8, oq[8]=a1, sim[8]=0.529, the pairing probability of the 8th row and eighth row is designated as 0 entirely.
Table 7 is results that Pattern B and Pattern A tentatively choose pairing.
(2) contrast of Pattern C and Pattern A
Size according to the pairing probability is found out 6 assembly in regular turn to (N=6).It is as follows that order is chosen in its pairing:
(c8 and a8 pairing) op[1]=b8, oq[1]=c8, sim[1]=0.383, the pairing probability of the 8th row and eighth row is designated as 0 entirely.
(c4 and a5 pairing) op[2]=b4, oq[2]=c5, sim[2]=0.378, the 4th row and the 5th pairing probability of going are designated as 0 entirely.
(c5 and a4 pairing) op[3]=b5, oq[3]=c4, sim[3]=0.370, the 5th row and the 4th pairing probability of going are designated as 0 entirely.
(c3 and a1 pairing) op[4]=b3, oq[4]=c1, sim[4]=0.364, the 3rd row and the 1st pairing probability of going are designated as 0 entirely.
(c6 and a7 pairing) op[5]=b6, oq[5]=c7, sim[5]=0.350, the 6th row and the 7th pairing probability of going are designated as 0 entirely.
(c2 and a2 pairing) op[6]=b2, oq[6]=c2, sim[6]=0.321, the 2nd row and the 2nd pairing probability of going are designated as 0 entirely.
Table 8 is results that Pattern C and Pattern A tentatively choose pairing.
Four, Pei Dui thin portion chooses
In this example, d1 is decided to be 225, and d2 is 150, and G is 4.
(1) contrast of Pattern B and Pattern A
The highest preceding four assembly of pairing probability are right, and (tx, ty θ), make E (tx, ty, θ) minimum with least mean-square error method (LeastMean-Square-Error Method), to try to achieve one.The result draw (tx, ty, θ)=(19.92,27.54 ,-22.5 degree).
After the best adjustment, the coordinate of adjusting back Pattern B is as shown in table 9.
Result as calculated, the range difference that has 8 pairings is less than d1, so M=8
Right with this M assembly, obtain best adjustment amount: (tx, ty, θ)=(19.93,26.68 ,-21.3 degree).After the optimization adjustment, the coordinate that obtains Pattern B is as shown in table 10.
Result as calculated, the range difference that has 8 pairings is less than d2, so K=8.And obtain diff=1.71, md=2.398.
(2) contrast of Pattern C and Pattern A
Right with the highest preceding four assembly of pairing probability, calculate how much best adjustment amounts (tx, ty, θ)=(8.74 ,-0.20,6.48 degree).After the best adjustment, obtain each point coordinate of Pattern C and and the match point of Pattern A between distance, be shown in table 11.After this adjusted, the range difference that 4 pairings are arranged was less than d1, so M=4.
Right with this M assembly, calculate again best adjustment amount (tx, ty, θ)=(23.12,9.24 ,-0.93 degree).After the best adjustment, the range difference that 4 pairings are arranged is less than d2, so K=4.And obtain diff=11.68, md=0.369.Its adjusted table 12 that the results are shown in.
Five, the calculating of two figure similarity degrees
In example, C is decided to be 10000.0.
(1) contrast of Pattern B and Pattern A
K=8,
S1=K/n=8/8=1,
S2=K/m=8/8=1,
md=0.598957,
S3=1/(1+diff)=0.368846。
Score=C·K
2·S
1 2·S
2·S
3 2·md=52151.48。
(2) contrast of Pattern C and Pattern A
K=4,
S1=K/n=4/8=0.5,
S2=K/m=4/8=0.5,
md=0.368672,
S3=1/(1+diff)=0.226394。
Score=C·K
2·S
1 2·S
2·S
3 2·md=377.92
Six, judge
Through the calculating of similarity degree, the degree of approximation index of Pattern B and Pattern A is 52151.48, and mark is similar to so judge B and A two figures much larger than set critical value (Threthold=5000.0).
Through the calculating of similarity degree, the degree of approximation index of Pattern C and Pattern A is 377.92, and mark is much smaller than set critical value (Threthold=5000.0), so judge that C and A two figures are not approximate.
Contrast in a large number with 800 pieces of fingerprints, experimental result also compares with control methods that No. 79109743 the Taiwan patent is carried, can improve the rate of distinguishing really with checking the inventive method.
The fingerprint of experiment is to get 20 times finger print image respectively by 40 fingers under two periods, altogether 800 pieces of fingerprints, and obtain the unique point (characteristic direction that comprises coordinate and fingerprint line) of each fingerprint through pre-treatment.
The identical fingerprints contrast:
The fingerprint of each root finger has 20 pieces, and each piece compares with other 19 pieces of fingerprints, because of 40 fingers are arranged, so identical fingerprints contrast number of times is: and 40* (20*19)=15200, and write down each to the contrast mark.
Different fingerprint contrasts:
Because of 40 fingers are arranged, the fingerprint of each root finger has 20 pieces, and the fingerprint of each piece fingerprint and other root finger compares, so different fingerprint contrast number of times is: and 800 * 780=624000 time, and record contrasts mark each time.
Compare number of times=624000+15200+639200 contrast altogether.
Experimental data and result are as shown in table 13.Experimental data shows that the rate of distinguishing of the present invention exceeds more than the twice than the control methods that No. 79109743 the Taiwan patent is carried.
It more than is explanation to the automatic comparison system of planimetric map form point of the present invention.Skilled person in the present technique field is not difficult by above-mentioned explanation, understands spirit of the present invention and also makes different variations thus and extend.But only otherwise exceed spirit of the present invention, all fall within the following claim scope.
Table 1-1
Table 1-2
Table 1-3
The thick pairing result of table 2:Pattern B and Pattern A
The thick pairing result of table 3:Pattern C and Pattern A
Table 4: under b1 and a1 pairing prerequisite, the possibility of its pairing of base
Table 5:Pattern B and Pattern A are with fuzzy method of relaxation comparison result
Table 6:Pattern C and Pattern A are with fuzzy method of relaxation comparison result
a1 | a2 | a3 | a4 | a5 | a6 | a7 | a8 | |
b1 | 1 | 1 | 1 | 0 | 0 | 0 | 0 | 0 |
b2 | 1 | 1 | 1 | 1 | 0 | 0 | 0 | 0 |
b3 | 0 | 1 | 1 | 1 | 0 | 0 | 0 | 0 |
b4 | 0 | 0 | 1 | 1 | 0 | 0 | 0 | 0 |
b5 | 0 | 0 | 0 | 0 | 1 | 1 | 0 | 0 |
b6 | 0 | 0 | 0 | 0 | 1 | 1 | 1 | 0 |
b7 | 0 | 0 | 0 | 0 | 0 | 1 | 1 | 1 |
b8 | 0 | 0 | 0 | 0 | 0 | 0 | 1 | 1 |
a1 | a2 | a3 | a4 | a5 | a6 | a7 | a8 | |
c1 | 1 | 1 | 0 | 0 | 0 | 0 | 0 | 0 |
c2 | 1 | 1 | 0 | 0 | 0 | 0 | 0 | 0 |
c3 | 1 | 1 | 1 | 1 | 0 | 0 | 0 | 0 |
c4 | 0 | 1 | 1 | 1 | 1 | 0 | 0 | 0 |
c5 | 0 | 0 | 1 | 1 | 0 | 1 | 1 | 0 |
c6 | 0 | 0 | 0 | 1 | 0 | 1 | 1 | 0 |
c7 | 0 | 0 | 0 | 0 | 0 | 0 | 1 | 1 |
c8 | 0 | 0 | 0 | 0 | 0 | 0 | 1 | 1 |
a2 | a3 | a4 | a5 | a6 | a7 | a8 | |
b2 | 0.933 | 0.500 | 0.500 | 0.000 | 0.000 | 0.000 | 0.000 |
b3 | 0.500 | 0.900 | 0.500 | 0.000 | 0.000 | 0.000 | 0.000 |
b4 | 0.000 | 0.500 | 0.978 | 0.000 | 0.000 | 0.000 | 0.000 |
b5 | 0.000 | 0.000 | 0.000 | 0.923 | 0.570 | 0.000 | 0.000 |
b6 | 0.000 | 0.000 | 0.000 | 0.636 | 0.870 | 0.500 | 0.000 |
b7 | 0.000 | 0.000 | 0.000 | 0.000 | 0.500 | 0.989 | 0.500 |
b8 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.500 | 1.000 |
a1 | a2 | a3 | a4 | a5 | a6 | a7 | a8 | |
b1 | 0.714 | 0.499 | 0.448 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 |
b2 | 0.502 | 0.670 | 0.480 | 0.451 | 0.000 | 0.000 | 0.000 | 0.000 |
b3 | 0.000 | 0.467 | 0.617 | 0.496 | 0.000 | 0.000 | 0.000 | 0.000 |
b4 | 0.000 | 0.000 | 0.471 | 0.601 | 0.000 | 0.000 | 0.000 | 0.000 |
b5 | 0.000 | 0.000 | 0.000 | 0.000 | 0.567 | 0.482 | 0.000 | 0.000 |
b6 | 0.000 | 0.000 | 0.000 | 0.000 | 0.497 | 0.547 | 0.473 | 0.000 |
b7 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.477 | 0.547 | 0.523 |
b8 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.477 | 0.529 |
a1 | a2 | a3 | a4 | a5 | a6 | a7 | a8 | |
c1 | 0.360 | 0.319 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 |
c2 | 0.327 | 0.321 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 |
c3 | 0.364 | 0.354 | 0.326 | 0.312 | 0.000 | 0.000 | 0.000 | 0.000 |
c4 | 0.000 | 0.324 | 0.338 | 0.311 | 0.378 | 0.000 | 0.000 | 0.000 |
c5 | 0.000 | 0.000 | 0.343 | 0.370 | 0.000 | 0.335 | 0.321 | 0.000 |
c6 | 0.000 | 0.000 | 0.000 | 0.332 | 0.000 | 0.335 | 0.350 | 0.000 |
c7 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.327 | 0.317 |
c8 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.320 | 0.383 |
Table 7:Pattern B and Pattern A tentatively choose the pairing result
a1 | a2 | a3 | a4 | a5 | a6 | a7 | a8 | ||
b1 | 0.714 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | B1 and a1 pairing |
b2 | 0 | 0.670 | 0 | 0 | 0 | 0 | 0 | 0 | B2 and a2 pairing |
b3 | 0 | 0 | 0.617 | 0 | 0 | 0 | 0 | 0 | B3 and a3 pairing |
b4 | 0 | 0 | 0 | 0.601 | 0 | 0 | 0 | 0 | B4 and a4 pairing |
b5 | 0 | 0 | 0 | 0 | 0.567 | 0 | 0 | 0 | B5 and a5 pairing |
b6 | 0 | 0 | 0 | 0 | 0 | 0.547 | 0 | 0 | B6 and a6 pairing |
b7 | 0 | 0 | 0 | 0 | 0 | 0 | 0.547 | 0 | B7 and a7 pairing |
b8 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0.529 | B8 and a8 pairing |
Table 8:Pattern C and Pattern A tentatively choose the pairing result
Table 9:Pattern B with first adjusted coordinate and with the range difference of Pattern A match point.
a1 | a2 | a3 | a4 | a5 | a6 | a7 | a8 | ||
c1 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | |
c2 | 0 | 0.321 | 0 | 0 | 0 | 0 | 0 | 0 | C2 and a2 pairing |
c3 | 0.364 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | C3 and a1 pairing |
c4 | 0 | 0 | 0 | 0 | 0.378 | 0 | 0 | 0 | C4 and a5 pairing |
c5 | 0 | 0 | 0 | 0.370 | 0 | 0 | 0 | 0 | C5 and a4 pairing |
c6 | 0 | 0 | 0 | 0 | 0 | 0 | 0.350 | 0 | C6 and a7 pairing |
c7 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | |
c8 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0.383 | C8 and a8 pairing |
The PattenB coordinate | Coordinate after the B conversion | The coordinate of A and B match point | The range difference of pairing | |||
The X coordinate | The Y coordinate | The X coordinate | The Y coordinate | The X coordinate | The Y coordinate | |
33 55 76 100 21 64 41 86 | 36 45 53 64 74 90 79 99 | 24.35 48.12 70.58 96.96 27 81 73.66 48.2 97.43 | 48.16 48.05 47.4 48.38 87.86 86.18 84.82 86.07 | 24 48 72 96 24 72 48 96 | 48 48 48 48 88 88 88 88 | 0.15 0.02 2.38 1.07 14.52 6.06 10.13 5.75 |
Table 10:Pattern B through adjusted coordinate once again and with the range difference of Pattern A match point.
The match point pitch difference of table 11:Patern C each point coordinate and Pattern A after first adjustment.
Table 12:Pattern C is range difference through between the match point of each point coordinate and Pattern A after the adjustment once again.
Patten B coordinate | Coordinate after the B conversion | The coordinate of A and B match point | The range difference of pairing | |||
The X coordinate | The Y coordinate | The X coordinate | The Y coordinate | The X coordinate | The Y coordinate | |
33 55 76 100 21 64 41 86 | 36 45 53 64 74 90 79 99 | 23.88 47.65 70.12 96.47 26.48 72.35 46.93 96.12 | 48.26 48.66 48.5 50.05 88.02 87.34 85.43 87.74 | 24 48 72 96 24 72 48 96 | 48 48 48 48 88 88 88 88 | 0.08 0.57 3.8 4.42 6.15 0.57 7.76 0.08 |
Patten C coordinate | Coordinate after the C conversion | The coordinate of A and C match point | The range difference of pairing | |||
The X coordinate | The Y coordinate | The X coordinate | The Y coordinate | The X coordinate | The Y coordinate | |
120 48 88 48 88 16 | 80 80 40 40 80 80 | 101.46 29.92 74.18 34.44 69.67 -1.87 | 92.84 84.71 49.48 44.97 89.23 81.1 | 96 24 96 24 72 48 | 88 88 48 48 88 48 | 53.24 45.88 478.25 118.14 6.96 3582.91 |
The PattenC coordinate | Coordinate after the C conversion | The coordinate of A and C timing point | The range difference of pairing | |||
The X coordinate | The Y coordinate | The X coordinate | The Y coordinate | The X coordinate | The Y coordinate | |
120 48 48 88 | 80 80 40 80 | 98.16 26.17 25.52 66.16 | 87.29 88.45 48.46 87.8 | 96 24 24 72 | 88 88 48 88 | 5.16 4.89 2.51 34.14 |
Table 13: the present invention and prior art effect comparison sheet
Claims (16)
1. the automatic control methods of a planimetric map form point is distributed in order to determine two that point on the plane is gathered formed figure P and whether Q is close, wherein:
P={p1, p2 ..., pm} is reference pattern, comprises the m point, Q={q1, and q2 ... qn} is resolution chart, comprises the n point, wherein each point with (x, y, D) expression, (x y) puts in the plane coordinate for this, and D is the characteristic direction of this point:
This control methods comprises:
Coordinate and characteristic direction according to the contained point of this P figure and this Q figure are the contained point of this Q figure, seek unique match point in this P figure;
According to the result of pairing, calculate the degree of approximation value of two figures; And
Set a critical value, two figure degree of approximation values are higher than this critical value person, be judged as approximate; To figure degree of approximation value subcritical value person, cannot not be judged as approximately;
Wherein, this pairing step comprises: specify a bit (qj) and a bit (pi) in this P figure in this Q figure to match, calculate (qk, k=1,2 of having a few in this Q figure,, (ph, the h=1 of having a few among the n, k ≠ j) and this P figure, 2 ..., m, the total value of the pairing possibility of h ≠ i); And
According to the height of this total value, determine in this P figure with this Q figure in the point of some pairing;
Wherein, this calculate the pairing possibility of being had a few in the have a few and this P figure in this Q figure (total value of Cij (h, k)) (method of S (pi, qj)) comprises following formula:
, wherein,
Representative in this Q figure the qj point and this P figure in the pi point be when pairing, the qk point in this Q figure (k=1,2 ..., ph point among the n, k ≠ j) and P figure (h=1,2 ..., m, the pairing possibility of h ≠ i), wherein,
Δ s representative (difference of the characteristic direction of pi and piph line segment direction) and (difference of the characteristic direction of qj and qjpk line segment direction) both differences; Δ e representative (difference of the characteristic direction of ph and piph line segment direction) and (difference of the characteristic direction of qk and qjqk line segment direction) both differences; Δ 1 is represented the difference apart from piph and qjqk of two pairs of match points, and w1, w θ are constant;
This step of calculating the degree of approximation value (Score) of two figures comprises following formula:
Score=C * K
2* S
1 2* S
2* S
3 2* md, wherein,
C is a constant;
K is the group numbers of two figures through pairing;
S
1Be reference pattern pairing rate, S
1Counting=K/n of=pairing number/reference pattern;
S
2Be resolution chart pairing rate, S
2Counting=K/m of=pairing number/resolution chart;
Md is average pairing support, the pairing possibility mean value of all pairings of md=; And
S
3Be the degree of approach between average match point, S
3=1.0/ (between the average match point of 1.0+ distance and).
2. the method for claim 1, wherein the point that comprises of this reference pattern and this resolution chart is end points or the point of crossing of representing contained line segment of a plan view image or segment of curve; When this point was the end points of line segment or segment of curve, this characteristic direction was located the tangential direction of this line segment or segment of curve for this place; When this point was the point of crossing of line segment or segment of curve, characteristic direction was located the decomposition line direction of the acute angle of two line segments or segment of curve formation for this point of crossing; And this characteristic direction is to represent with the angle of the positive horizontal direction of this direction and planimetric coordinates.
3. method as claimed in claim 1 or 2, also comprise a thick pairing step before pairing, the maximum tolerance amount of the distance value when this thick pairing comprises characteristic direction angle difference when setting the contained point of the contained point of P figure and Q figure becomes pairing and is positioned at same coordinate plane; P figure and Q figure are placed same coordinate plane, and the characteristic direction differential seat angle and the distance value of wantonly two the different graphic points of calculating, and with the characteristic direction angle difference or the distance value when being positioned at same coordinate plane exceeds this maximum tolerance amount, the point that adheres to different graphic separately is set at its Ci, (h, k are 0 to j.
4. method as claimed in claim 1 or 2, the thin portion after pairing that also comprises chooses step, and wherein this thin portion chooses step and comprises:
Look for one group how much adjustment amount (tx, ty, θ) make E (tx, ty, θ) minimum, wherein:
x
qj=tx+x
picosθ-y
pisinθ,
y
Qj=ty+ * x
PiSin θ+y
PiCos θ, and
Dqj=(Dpi+Dr)+θ; Wherein,
Dpi and Dqj represent the characteristic direction of a pi and qj respectively; Dr represents to change difference, is a figure after rotation θ angle, the difference that still exists between the characteristic direction of two match points; And
Wherein,
G is a predetermined value; (tx, ty θ) adjust with this geometry adjustment amount with this reference diagram form point; And
Be the pairings of two figures once again.
5. method as claimed in claim 4, wherein this thin portion to choose step be only this G group match point to be done how much to adjust, and after adjusting again, the spacing that only keeps resolution chart and resolution chart match point less than the pairing of fixed scope, and delete all the other pairings.
6. method as claimed in claim 4 comprises that the thin portion of secondary chooses step at least.
7. method as claimed in claim 5 comprises that the thin portion of secondary chooses step at least.
8. method as claimed in claim 3, wherein this pairing step comprises with following formula and repeats to adjust pairing possibility between this two figures each point:
Wherein, S
(r)(pi, qj) representative is through being adjusted to the r time pairing possibility, S
(0)(pi qj) thinks in this thick pairing result and is made as 1 in the time of may matching, and is made as 0 in thinking in the time of can not matching.
9. the automatic compare device of a planimetric map form point is distributed in order to determine two that point on the plane is gathered formed figure P and whether Q is close, wherein:
P={p1, p2 ..., pm} is reference pattern, comprises the m point, Q={q1, and q2 ..., qn} is resolution chart, comprises the n point, wherein each point is so that (x, y represent that D) (x is this some coordinate in the plane y), and D is the characteristic direction of this point;
This compare device comprises:
One sampler is in order to obtain the data of reference pattern and resolution chart unit vegetarian refreshments;
One contrast means can be the contained point of this Q figure according to the coordinate and the characteristic direction of the contained point of this P figure and this Q figure, seeks unique match point in this Q figure;
One figure degree of approximation calculation element according to the result of pairing, calculates the degree of approximation value of two figures; And
One judgment means is set a critical value, and two figure degree of approximation values are higher than this critical value person, is judged as approximately, and figure degree of approximation value is lower than this critical value person, cannot not be judged as approximately,
Wherein, the pairing processing of this contrast means comprises: specify a bit (qj) and a bit (pi) in this P figure in this Q figure to match, calculate (qk, the k=1 of having a few in this Q figure, 2, (ph, the h=1 of having a few among the n, k ≠ j) and this P figure, 2,, m, the total value of the pairing possibility of h ≠ i); And
According to the height of this total value, determine in this P figure with this Q figure in the point of some pairing;
Wherein, the pairing possibility of being had a few in have a few and this Q figure in this Q figure (total value of Cij (h, k)) (calculating of S (pi, qj)) comprises following formula:
Wherein,
Qj point and the pi point in this P figure of representative in this Q figure is when pairing, the qk point in this Q figure (k=1,2 ... ph point among the n, k ≠ j) and this P figure (h=1,2 ... m, the pairing possibility of h ≠ i), wherein,
Δ s representative (difference of the characteristic direction of pi and piph line segment direction) and (difference of the characteristic direction of qj and qjqk line segment direction) both differences; Δ e representative (difference of the characteristic direction of ph and piph line segment direction) and (difference of the characteristic direction of qk and qjqk line segment direction) both differences; Δ l represents the difference apart from piph and qjqk of two pairs of match points, and w1, w θ are constant; And
The calculating of the degree of approximation value (Score) of this two figure comprises following formula:
Score=C * K
2* S
1 2* S
2* S
3 2* md, wherein,
C is a constant;
K is the group numbers of two figures through pairing;
S
1Be reference pattern pairing rate, S
1Counting=K/n of=pairing number/reference pattern;
S
2Be resolution chart pairing rate, S
2Counting=K/m of=pairing number/resolution chart;
Md is average pairing support, the pairing possibility mean value of all pairings of md=; And
S
3Be the degree of approach between average match point, S
3=1.0/ (between the average match point of 1.0+ distance and).
10. device as claimed in claim 9, wherein the point that comprises of this reference pattern and this resolution chart is end points or the point of crossing of representing contained line segment of a plan view image or segment of curve; When this point was the end points of line segment or segment of curve, this characteristic direction was located the tangential direction of this line segment or segment of curve for this place; When this point was the point of crossing of line segment or segment of curve, characteristic direction was located the angular bisector direction of the acute angle of two line segments or segment of curve formation for this point of crossing; And this characteristic direction is to represent with the angle of the positive horizontal direction of this direction and planimetric coordinates.
11. as claim 9 or 10 described devices, also comprise a thick contrast means, can before pairing, set the contained point of contained point of P figure and Q figure and become when pairing maximum tolerance amount at the distance value of characteristic direction angle difference when being positioned at same coordinate plane, P figure and Q figure are placed same coordinate plane, and the characteristic direction differential seat angle and the distance value of wantonly two the different graphic points of calculating, and with the characteristic direction angle difference or the distance value when being positioned at same coordinate plane exceeds this maximum tolerance amount, the point that adheres to different graphic separately is set its Ci, (h is 0 k) to j.
12. as claim 9 or 10 described devices, this contrast means can be carried out thin portion and chosen after pairing, wherein this thin portion chooses and comprises:
Look for one group how much adjustment amount (tx, ty, θ) make E (tx, ty, θ) minimum, wherein:
x
qj=tx+×x
picosθ-y
pisinθ,
y
Qj=ty+x
PiSin θ+y
PiCos θ, and
Dqj=(Dpi+Dr)+θ; Wherein,
Dpi and Dqj represent the characteristic direction of a qi and qj respectively; Dr represents to change difference, is a figure after rotation θ angle, the difference that still exists between the characteristic direction of two match points; And
Wherein,
G is a predetermined value;
(tx, ty θ) adjust with this geometry adjustment amount with this reference diagram form point; And
Be the pairings of two figures once again.
13. device as claimed in claim 12, wherein this thin portion pairing is only this G group match point to be done how much to adjust, and after adjusting again, the spacing that only keeps resolution chart and resolution chart match point less than the pairing of fixed scope, and delete all the other pairings.
14. device as claimed in claim 12, this contrast means are to carry out at least that the thin portion of secondary chooses step.
15. device as claimed in claim 13, this contrast means are to carry out at least that the thin portion of secondary chooses step.
16. device as claimed in claim 11, wherein this contrast means can following formula repeats to adjust the pairing possibility between this two figures each point:
Wherein, S
(r)(pi, qj) representative is through being adjusted to the r time pairing possibility, S
(0)(pi qj) thinks in this thick pairing result and is made as 1 in the time of may matching, and is made as 0 in thinking in the time of can not matching.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN96119761A CN1110775C (en) | 1996-12-10 | 1996-12-10 | Automatic contrast system for planedrawing points |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN96119761A CN1110775C (en) | 1996-12-10 | 1996-12-10 | Automatic contrast system for planedrawing points |
Publications (2)
Publication Number | Publication Date |
---|---|
CN1184983A CN1184983A (en) | 1998-06-17 |
CN1110775C true CN1110775C (en) | 2003-06-04 |
Family
ID=5125956
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN96119761A Expired - Lifetime CN1110775C (en) | 1996-12-10 | 1996-12-10 | Automatic contrast system for planedrawing points |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN1110775C (en) |
Families Citing this family (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103473413B (en) * | 2013-09-11 | 2016-08-17 | 中国核电工程有限公司 | A kind of comparative approach for various a large amount of Mechanics Calculation results |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US5075896A (en) * | 1989-10-25 | 1991-12-24 | Xerox Corporation | Character and phoneme recognition based on probability clustering |
US5392367A (en) * | 1991-03-28 | 1995-02-21 | Hsu; Wen H. | Automatic planar point pattern matching device and the matching method thereof |
-
1996
- 1996-12-10 CN CN96119761A patent/CN1110775C/en not_active Expired - Lifetime
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US5075896A (en) * | 1989-10-25 | 1991-12-24 | Xerox Corporation | Character and phoneme recognition based on probability clustering |
US5392367A (en) * | 1991-03-28 | 1995-02-21 | Hsu; Wen H. | Automatic planar point pattern matching device and the matching method thereof |
Also Published As
Publication number | Publication date |
---|---|
CN1184983A (en) | 1998-06-17 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN1139039C (en) | Methods for extraction and recognition of pattern in image, method for image abnormality juding, and memory medium with image processing programs | |
CN1235177C (en) | Hand-wirte signature recognition program, method and device | |
CN1818927A (en) | Fingerprint identifying method and system | |
CN1286066C (en) | Paper securities true-false distinguishing method and paper securities input direction distinguishing method | |
CN1599913A (en) | Iris identification system and method, and storage media having program thereof | |
CN1187952C (en) | Equipment and method for correcting input image distortion | |
CN1794266A (en) | Biocharacteristics fusioned identity distinguishing and identification method | |
CN1664846A (en) | On-line hand-written Chinese characters recognition method based on statistic structural features | |
CN1045835C (en) | Character generating method and apparatus | |
CN1945602A (en) | Characteristic selecting method based on artificial nerve network | |
CN1098515C (en) | Character generating apparatus and method of the same | |
CN1384538A (en) | Imperfection detecting method and device for semiconductor ICs | |
CN1197244A (en) | Data hiding method and data extraction method using statistical examination | |
CN101055620A (en) | Shape comparison device and method | |
CN1251130C (en) | Method for identifying multi-font multi-character size print form Tibetan character | |
CN1400807A (en) | Image processing method and equipment, image processing system and storage medium | |
CN1707521A (en) | Image processing device and method, recording medium, and program | |
CN1471055A (en) | Image distortion correction method and apparatus | |
CN1684492A (en) | Image dictionary creating apparatus, coding apparatus, image dictionary creating method | |
CN1379364A (en) | Graph comparing device and graph comparing method | |
CN101030257A (en) | File-image cutting method based on Chinese characteristics | |
CN1200387C (en) | Statistic handwriting identification and verification method based on separate character | |
CN1195830A (en) | Stripe pattern matching system | |
CN1170251C (en) | Beer-bottle raised character fetching-identifying hardware system and processing method | |
CN1110775C (en) | Automatic contrast system for planedrawing points |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
C14 | Grant of patent or utility model | ||
GR01 | Patent grant | ||
CX01 | Expiry of patent term |
Granted publication date: 20030604 |
|
EXPY | Termination of patent right or utility model |