CN104408719B - A kind of conllinear three characteristic points monocular vision space-location method - Google Patents

Abstract

The present invention relates to the conllinear three characteristic points monocular vision space-location method of one kind, belong to field of machine vision.The present invention will determine positioning shooting machine system and be operated in perspective model, it is used as characteristic point by choosing three collinear points, and be wherein midpoint, characteristic point space coordinate and the relational expression of projection coordinate are determined according to perspective projection relation, the image for shooting characteristic point using video camera is used as projection coordinate, identified relational expression is resolved according to the position relationship between characteristic point, length and projection coordinate, so as to obtain the three dimensional space coordinate position of characteristic point, the positioning to characteristic point is realized.The localization method of the present invention can be completed using algebraic operation, it is to avoid interative computation, improve sterically defined precision and efficiency.

Description

A kind of conllinear three characteristic points monocular vision space-location method

Technical field

The present invention relates to the conllinear three characteristic points monocular vision space-location method of one kind, belong to technical field of machine vision.

Background technology

Monocular vision positioning is exactly only to complete positioning work with a video camera.It has easy to use and applied widely , also will not be as full side without solving the problems, such as optimal distance and Feature Points Matching in stereoscopic vision between two video cameras etc. specific Position vision sensor produces very big distortion like that.

In machine vision research field, how position is completed under the conditions of monocular vision and the solution of posture has become one Individual important research direction, monocular vision location technology is applied to many aspects, for example：Camera calibration, robot localization, regard Feel servo, target following and monitoring etc..

The researcher of various countries proposes a variety of algorithms to solve the problem of monocular vision is positioned, the spy being directed to Levy a little have it is coplanar also have non-coplanar, the quantity of required characteristic point is also different, and algorithm uses the algorithm of iterative approach mostly, Computational efficiency is low.

The content of the invention

The purpose of the present invention is a kind of conllinear three characteristic points monocular vision space-location method, to solve existing monocular vision Space-location method uses the problem of computational efficiency caused by iterative approach algorithm is low and not accurate enough.

The technical scheme is that：A kind of conllinear three characteristic points monocular vision space-location method, it is characterised in that should Localization method comprises the following steps：

1) three collinear points are chosen as characteristic point, and one of point is midpoint；

2) space coordinate of above three characteristic point and the relational expression of projection coordinate are determined according to projection projection relation；

3) image of positioning shooting machine shooting characteristic point is used as projection coordinate, according to the position between three characteristic points Relation, length and projection coordinate are to step 2) determined by relational expression resolve, obtain characteristic point three dimensional space coordinate position Put, realize the positioning to characteristic point.

The step 2) in determined by relational expression be：

Wherein (x_a,y_a,z_a)、(x_b,y_b,z_b) and (x, y, z) be respectively three characteristic points A, B and C space coordinate, C is Midpoint between AB, (xu_a,yu_a)、(xu_b,yu_b) and (xu, yu) be respectively the throwing of three characteristic points A, B and C on focal plane Shadow coordinate, f is focal length used when shooting the projection coordinate of characteristic point.

The step 3) in positioning shooting machine be operated in perspective projection model.

The step 3) process that is resolved is as follows：

A) according to wherein for midpoint feature to step 2) determined by relational expression progress abbreviation,

z+Δz-a₁*x-a₁* Δ x=0

z-Δz-a₂*x+a₂* Δ x=0

z-a₃* x=0

z+Δz-b₁*y-b₁* Δ y=0

z-Δz-b₂*y+a₂* Δ y=0

z-b₃* y=0

Wherein, a₁=f/xu_a, a₂=f/xu_b, a₃=f/xu, b₁=f/yu_a, b₂=f/yu_b, b₃=f/yu, Δ x=x_a-x =x-x_b, Δ y=y_a- y=y-y_b, Δ z=z_a- z=z-z_b, the focal length that f is used when being shot by video camera, (xu_a,yu_a)、 (xu_b,yu_b), (xu, yu) be the feature point coordinates that arrives of video camera captured in real-time, a₁、a₂、a₃、b₁、b₂、b₃It can calculate in real time；

B) result after abbreviation further derive and can obtained：

Δ x=z/c₁

Δ y=z/c₂

Δ z=z/c₃

Wherein c₁=(a₁-a₂)*a₃/(2a₃-a₁-a₂), c₂=(b₁-b₂)*b₃/(2b₃-b₁-b₂)c₃=(a₁-a₂)*a₃/ (a₁*a₃+a₂*a₃-2a₁*a₂)；

C) length between any two characteristic point is brought into step B) relational expression and step 2 that are derived) pass that determines It is formula, the space coordinate (x, y, z) for solving characteristic point C is：

Wherein L is AC length；

D) according to three characteristic points it is conllinear and one of them be midpoint feature and step C) the midpoint seat that calculates Mark solves the space coordinate of other two characteristic points, so as to realize the space orientation to three collinear feature points.

The beneficial effects of the invention are as follows:The present invention will determine positioning shooting machine system and be operated in perspective model, pass through Three collinear points are chosen as characteristic point, and are wherein midpoint, characteristic point space coordinate is determined according to perspective projection relation With the relational expression of projection coordinate, the image using video camera shooting characteristic point is as projection coordinate, according to the position between characteristic point Put relation, length and projection coordinate to resolve identified relational expression, so as to obtain the three dimensional space coordinate position of characteristic point Put, realize the positioning to characteristic point.The localization method of the present invention can be completed using algebraic operation, it is to avoid interative computation, carried High sterically defined precision and efficiency.

Brief description of the drawings

Fig. 1 is perspective projection model coordinate system schematic diagram；

Fig. 2 is characteristic point spatial relationship schematic diagram.

Embodiment

The embodiment to the present invention is further described below in conjunction with the accompanying drawings.

It is an object of the invention to provide a kind of monocular vision space-location method of conllinear three characteristic point, with quick, accurate Completion characteristic point locus resolving.Current invention assumes that positioning shooting machine system is operated in perspective projection model.

1. choosing 3 collinear points as characteristic point, A (x are designated as respectively_a,y_a,z_a)、B(x_b,y_b,z_b) and C (x, y, z), A, B Respectively two end points, C is the midpoint of AB lines, and AC length is L.

2. characteristic point space coordinate and the relation of projection coordinate are determined according to perspective projection relation.

As shown in figure 1, camera coordinate system is [Xc, Yc, Zc], image coordinate system is [xu, yu], is thrown according to perspective Shadow equation, then have：

I.e. the x coordinate of subpoint is only related to imaging focal length, the x coordinate of characteristic point and z coordinate, and the y-coordinate of subpoint is only It is related to imaging focal length, the y-coordinate of characteristic point and z coordinate.

Three feature point coordinates respectively A ' (xu that video camera captured in real-time is arrived_a,yu_a)、B’(xu_b,yu_b)、C’(xu, Yu), focal length used during shooting is f.Pass between characteristic point space coordinate and projection coordinate is determined according to perspective projection relation It is to be：

3. according to the position relationship between three characteristic points, C is A, B midpoint, it is known that：

x_a=x+ Δ x 2-1

x_b=x- Δ x 2-2

y_a=y+ Δ y 2-3

y_b=y- Δ y 2-4

z_a=z+ Δ z 2-5

z_b=z- Δ z 2-6

4. characteristic point space coordinate and projection according to determined by the relation pair of locus coordinate between three characteristic points Relational expression between coordinate, which becomes, to be turned to：

5. formula is further simplified as：

z+Δz-a₁*x-a₁* Δ x=0 4-1

z-Δz-a₂*x+a₂* Δ x=0 4-2

z-a₃* x=0 4-3

z+Δz-b₁*y-b₁* Δ y=0 4-4

z-Δz-b₂*y+a₂* Δ y=0 4-5

z-b₃* y=0 4-6

Wherein：

a₁=f/xu_a

a₂=f/xu_b

a₃=f/xu

b₁=f/yu_a

b₂=f/yu_b

b₃=f/yu

F is known focal length, (xu_a,yu_a)、(xu_b,yu_b), (xu, yu) be the feature point coordinates that arrives of video camera captured in real-time, Therefore, a₁、a₂、a₃、b₁、b₂、b₃It can also calculate in real time (being the equal of a datum).

Further derive and can obtain 6. formula will be obtained in step 5：

Δ x=z/c₁ 5-1

Δ y=z/c₂ 5-2

Δ z=z/c₃ 5-3

Wherein：

c₁=(a₁-a₂)*a₃/(2a₃-a₁-a₂)

c₂=(b₁-b₂)*b₃/(2b₃-b₁-b₂)

c₃=(a₁-a₂)*a₃/(a₁*a₃+a₂*a₃-2a₁*a₂)

7. because the information in projection process is lost, the space coordinate position of characteristic point can not be only obtained with above-listed formula Put, it is therefore desirable to which extra and information is assisted.

Pass through demarcation in advance to characteristic point, it is known that AC length is L, then has：

Formula 5-1,5-2,5-3 substitution 5-4 can be obtained：

The z that 5-5 is solved, which is substituted into formula 1-3 and 1-6, can try to achieve x, y,

So far midpoint C space coordinate (x, y, z) is obtained.

The z that 5-5 is solved, which is substituted into formula 5-1,5-2 and 5-3, can try to achieve Δ x, Δ y, Δ z, by Δ x, Δ y, Δ z value Substitution formula 2-1 to 2-6 can try to achieve A, B space coordinate coordinate A (x_a,y_a,z_a)、B(x_b,y_b,z_b)。

Claims (2)

1. a kind of conllinear three characteristic points monocular vision space-location method, it is characterised in that the localization method comprises the following steps：

1) three collinear points are chosen as characteristic point, and one of point is midpoint；

2) space coordinate of above three characteristic point and the relational expression of projection coordinate are determined according to projection projection relation；

3) use the image that positioning shooting machine shoots characteristic point as projection coordinate, closed according to the position between three characteristic points System, length and projection coordinate are to step 2) determined by relational expression resolve, obtain the three dimensional space coordinate position of characteristic point, Realize the positioning to characteristic point；

The step 2) in determined by relational expression be：

$z_a=f\frac{x_a}{{\mathrm{xu}}_a}$

$z_b=f\frac{x_b}{{\mathrm{xu}}_b}$

$z=f\frac{x}{xu}$

$z_a=f\frac{y_a}{{\mathrm{yu}}_a}$

$z_b=f\frac{y_b}{{\mathrm{yu}}_b}$

$z=f\frac{y}{yu}$

Wherein (x_a,y_a,z_a)、(x_b,y_b,z_b) and (x, y, z) be respectively three characteristic points A, B and C space coordinate, C be AB it Between midpoint, (xu_a,yu_a)、(xu_b,yu_b) and (xu, yu) be respectively that projection on focal plane of three characteristic points A, B and C is sat Mark, f is focal length used when shooting the projection coordinate of characteristic point；

The step 3) process that is resolved is as follows：

A) according to wherein for midpoint feature to step 2) determined by relational expression progress abbreviation,

z+Δz-a₁*x-a₁* Δ x=0

z-Δz-a₂*x+a₂* Δ x=0

z-a₃* x=0

z+Δz-b₁*y-b₁* Δ y=0

z-Δz-b₂*y+a₂* Δ y=0

z-b₃* y=0

Wherein, a₁=f/xu_a, a₂=f/xu_b, a₃=f/xu, b₁=f/yu_a, b₂=f/yu_b, b₃=f/yu, Δ x=x_a- x=x- x_b, Δ y=y_a- y=y-y_b, Δ z=z_a- z=z-z_b, f is focal length used when shooting the projection coordinate of characteristic point, (xu_a, yu_a)、(xu_b,yu_b), (xu, yu) be the projection coordinate of three characteristic points A, B and C on focal plane, a₁、a₂、a₃、b₁、b₂、b₃Can Calculate in real time；

B) result after abbreviation further derive and can obtained：

Δ x=z/c₁

Δ y=z/c₂

Δ z=z/c₃

Wherein c₁=(a₁-a₂)*a₃/(2a₃-a₁-a₂), c₂=(b₁-b₂)*b₃/(2b₃-b₁-b₂), c₃=(a₁-a₂)*a₃/(a₁*a₃+ a₂*a₃-2a₁*a₂)；

C the length between the length between characteristic point A and characteristic point C or characteristic point B and characteristic point C) is substituted into step B) pushed away The relational expression and step 2 led) relational expression that determines, the space coordinate (x, y, z) for solving characteristic point C is：

$x=\frac{xu*z}{f}$

$y=\frac{yu*z}{f}$

$z=\frac{L}{\sqrt{\frac{1}{{c_1}^2}+\frac{1}{{c_2}^2}+\frac{1}{{c_3}^2}}}$

Wherein L is AC length；

D) according to three characteristic points it is conllinear and one of them by midpoint feature and step C) the middle point coordinates that calculates asks The space coordinate of other two characteristic points is solved, so as to realize the space orientation to three collinear feature points.

2. conllinear three characteristic points monocular vision space-location method according to claim 1, it is characterised in that the step 3) the positioning shooting machine in is operated in perspective projection model.