CN101038153A - Three-point scaling measuring method - Google Patents

Abstract

The method is a measuring method with three-point to determine the frame of axes. The method includes: a frame of axes is established by seeing the car body as the center; a frame of axes is established by seeing the vidicon as the center; the direction that the front of the car body faces is the shaft Y, and the direction perpendicular to the ground is the shaft Z and the direction perpendicular to the lateral of the car body is the shaft X; two beelines are positioned on the road surface, wherein one beeline is parallel with the cross shaft of the car body frame of axes and the other beeline is parallel with the ordinate axis of the axis of coordinate of the car body; the intersecting point of the two beelines is seemed as the first given scenery point of the three point, and the second given scenery point and the third given scenery point are selected from the other parts of the two beelines; The reference coordinate of the car body frame of axes and the world coordinate of the scenery points corresponding to the pixels are calculated according to the three special scenery points, the coordinate relationship among the three pixels projected on the corresponding vidicon, and the coordinate of the disappeared point.

Description

A kind of three-point scaling measuring method

One, technical field:

The invention belongs to optical imagery measurement and positioning field, is a kind of methods such as environment sensing of vision guided navigation and military unmanned platform of intelligent automobile specifically.

Two, background technology:

In all vision navigation systems and the environment sensing system based on vision, camera calibration and vision measurement all are the most basic, also are one of the most key technology.Camera calibration refers to some combination of obtaining inside and outside parameter or these parameters of video camera by some technological means, and sets up the contrary projection model of video camera.Vision measurement then refers to according to the pixel coordinate in the image of video camera picked-up, calculates world coordinates or bodywork reference frame reference coordinate with the corresponding scene point of pixel.Camera marking method commonly used at present at first will be chosen a large amount of " scene point---pixel " two tuples in road surface and computer screen image, substitution camera model then, find the solution with least square method at last, the contrary projection formula of derivation video camera, vision measuring method at present commonly used then is in the process that intelligent automobile or military unmanned platform are executed the task, computer real-time ground is the (lane line for example of the feature on the identification road surface in screen picture, barrier, the ditch hole, the water mark, stone, buildings shade etc.), then with the contrary projection formula of the computer screen image coordinate substitution video camera of pixel on these features, calculate the bodywork reference frame coordinate of the road surface scene point corresponding, thereby draw the horizontal and fore-and-aft distance between these features and the car body with the feature pixel.

Camera marking method commonly used at present is not easy to the maintenance of camera interior and exterior parameter.At first will choose a large amount of road surface scene point and corresponding pixel thereof, improve the calibrated and calculated amount, secondly the least square method computation complexity is also very big, and these all are unfavorable for the real-time calibration of video camera.Vision measuring method commonly used has at present then been used the contrary projection formula of video camera of the camera interior and exterior parameter foundation that solves in the camera calibration, and the general type of these formula is as follows:

$\left\{\begin{array}{c}{x}_v=\frac{H\×c\×\mathrm{dc}}{r\×\mathrm{dr}\×\cos \mathrm{\α}+f\sin \mathrm{\α}}\\ y_v=\frac{H(f\cos \mathrm{\α}-r\×\mathrm{dr}\×\sin \mathrm{\α})}{r\×\mathrm{dr}\×\cos \mathrm{\α}+f\sin \mathrm{\α}}\end{array}\right.---\left(1\right)$

In the formula 1, comprised a large amount of trigonometric function operations, and in the internal calculation of computing machine, trigonometric function is the most complicated form of calculation, finding the solution all of each trigonometric function must realize through the additive operation of tens of steps even step computer-internal up to a hundred.Therefore, come the real-time implementation camera calibration in vision measurement if utilization is commonly used at present in embedded computer based on the method for a large amount of road surfaces scene point, just necessarily require bigger calculator memory reservoir capacity and higher CPU computing velocity, thereby improved the cost of intelligent automobile vision navigation system and military unmanned platform environment sensory perceptual system.

Three, summary of the invention:

The present invention to the contrary projection formula between the scene point, realizes that gamma camera demarcation and vision measurement calculate according to the coordinate relation derivation pixel between it by select three specific scene point and three corresponding pixels on the road surface.

A kind of optimal technical scheme of the present invention is a kind of three-point scaling measuring method, described measuring method comprises with the car body being that bodywork reference frame is set up at the center, with the video camera is that camera coordinate system is set up at the center, it is that the direction of the vertical vehicle body side of Z axle is an X-axis that the car body front is oriented Y-axis vertical ground direction, on the road surface, draw two straight lines, wherein straight line and bodywork reference frame transverse axis, another straight line is parallel with the car body coordinate axis longitudinal axis, the intersection point of getting described straight line is as first particular scene point in the three-point scaling, each is selected a bit as second particular scene point in the three-point scaling and the 3rd particular scene point, according to three particular scene points and corresponding coordinate relation and the coordinate Calculation of end point and the world coordinates and the bodywork reference frame reference coordinate of the corresponding scene point of pixel between three pixels of projection on the pick-up lens in other positions of described two straight lines respectively.

The pass that further a kind of optimal technical scheme of the present invention is camera coordinate system and bodywork reference frame is X _v, Y _v, Z _v, camera coordinate system X _c, Y _c, Z _c, described bodywork reference frame and camera coordinate system close and are

x _c＝x _v

y _c=y _vCos α+Hsin α, the horizontal ordinate of end point is decided by the slope of parallel lines:

z _c＝-Hcosα+y _vsinα

$x_i=(\frac{B_{v1}\cos \mathrm{\α}}{H}-K_v\sin \mathrm{\α})y_i+\frac{B_{v1}f\sin \mathrm{\α}}{H}+K_{v}f\cos \mathrm{\α}$

$x_i=(\frac{{\mathrm{<figure-callout\; id="2"\; label="B\; v"\; filenames="A20071000080700122.png"\; state="\{\{state\}\}">B</figure-callout>}}_v\cos \mathrm{\&alpha;}}{H}-K_v\sin \mathrm{\&alpha;})y_i+\frac{B_{v2}f\sin \mathrm{\&alpha;}}{H}+K_{v}f\cos \mathrm{\&alpha;}$ And the coordinate of record end point, with the coordinate substitution of end point

$y=\frac{(r_0-r)(r-{\mathrm{VP}}_r)}{(r_0-r_1)(r_1-{\mathrm{VP}}_r)}\&times;\mathrm{dy}+{\mathrm{<figure-callout\; id="0"\; label="y"\; filenames="A20071000080700121.png,A20071000080700122.png"\; state="\{\{state\}\}">y</figure-callout>}}_0.$

$x=\frac{r-{\mathrm{VP}}_r}{r_0-{\mathrm{VP}}_r}\&times;c\&times;\mathrm{dx}\left(r_0\right).$

Four, goal of the invention:

The objective of the invention is in order to solve the expensive problem of present method, make every effort on the road surface, choose the least possible scene point and realize camera calibration, and only comprise in the contrary projection formula of the video camera of making every effort to finally derive add, subtract, the multiplication and division arithmetic, thereby reduce requirement, reduce the cost of intelligent automobile vision navigation system and military unmanned platform environment sensory perceptual system calculator memory reservoir capacity and CPU computing velocity.

Five, accompanying drawing:

1, Fig. 1 is a front projection pinhole imaging system camera model;

2, Fig. 2 is camera coordinate system and bodywork reference frame;

3, Fig. 3 is the fore-and-aft distance instrumentation plan;

4, Fig. 4 calculates synoptic diagram for lateral separation;

5, Fig. 5 is camera calibration and the vision measurement synoptic diagram based on three-point scaling;

6, Fig. 6 is the camera calibration program flow diagram;

7, Fig. 7 is the vision measurement process flow diagram.

Six, embodiment:

Auxiliary facility is used auxiliary facility of the present invention and is comprised, common camera, be used to absorb road image, the road image of video camera picked-up is input to image pick-up card, image pick-up card is installed in the computing machine, it is the part of computer hardware system, the pavement image that is used for the acquisition camera picked-up, computing machine and visual processes software systems thereof, be used to handle the road image that image pick-up card is gathered, the camera calibration software module and the vision measurement software for calculation module of establishment are contained in the computer vision process software system according to the present invention.Software systems comprise independently software module of two of camera calibration and vision measurements, and these two modules are embedded in the visual processes computer software.When carrying out camera calibration, according to the camera calibration processing flow chart, computer software is used the camera calibration software module and is handled.In case camera calibration finishes, can be according to the vision measurement calculation process, computer software is used the vision measurement computing module and is carried out the vision measurement computing.The core of software systems is two computing formula among the present invention:

$y=\frac{(r_0-r)(r-{\mathrm{VP}}_r)}{(r_0-r_1)(r_1-{\mathrm{VP}}_r)}\&times;\mathrm{dy}+{\mathrm{<figure-callout\; id="0"\; label="y"\; filenames="A20071000080700121.png,A20071000080700122.png"\; state="\{\{state\}\}">y</figure-callout>}}_0$

$x=\frac{r-{\mathrm{VP}}_r}{r_0-{\mathrm{VP}}_r}\&times;c\&times;\mathrm{dx}\left(r_0\right)$

The present invention adopts common front projection pinhole imaging system camera model, and as shown in Figure 1, wherein pixel planes is represented computer screen image, and the pixel planes true origin is positioned at the computer screen image center.Coordinate system C represents camera coordinate system.The object point P that finds a view in the 3-D space, its coordinate is (x _c, y _c, z _c), the pixel corresponding with it is P ', coordinate be (c, r), then according to space geometry, (x _c, y _c, z _c) with (c, r) relation between can be by formula 2 expressions.

$\left\{\begin{array}{c}\frac{c\&times;\mathrm{dc}}{f}=\frac{x_c}{y_c}\\ \frac{r\&times;\mathrm{dr}}{f}=-\frac{z_c}{y_c}\end{array}\right.---\left(2\right)$

Wherein, dc represents the width in the plane of delineation of each pixel correspondence, and dr represents the height in each pixel correspondence image plane.Make x _i=c * dc and y _i=r * dr, and substitution formula 2 must look like in the plane of delineation

$\left\{\begin{array}{c}\frac{x_i}{f}=\frac{x_c}{y_c}\\ \frac{y_i}{f}=-\frac{z_c}{y_c}\end{array}\right.---\left(3\right)$

Vegetarian refreshments (x _i, y _i) and (x _c, y _c, z _c) between corresponding relation:

Video camera is installed on the car body top center line, and the distance between the ground is H, camera coordinate system as shown in Figure 1, the angle between camera optical axis and the ground is α, camera optical axis is parallel with the axis of car body in the projection on ground.The initial point of bodywork reference frame is located at the projection on the ground of camera coordinate system initial point, and the bodywork reference frame Y-axis overlaps with the projection of camera optical axis on ground.Relation between camera coordinate system and the bodywork reference frame as shown in Figure 2.

Get 1 P on the road surface, the coordinate of P in camera coordinate system is (x _v, y _v, 0), the coordinate in camera coordinate system is (x _c, y _c, z _c), (x then _v, y _v, 0) and (x _c, y _c, z _c) between have a following relation:

$\left\{\begin{array}{c}{x}_c=x_v\\ y_c=y_v\cos \mathrm{\&alpha;}+H\sin \mathrm{\&alpha;}\\ z_c=-H\cos \mathrm{\&alpha;}+y_v\sin \mathrm{\&alpha;}\end{array}\right.---\left(4\right)$

With formula 4 substitution formula 3, get the coordinate (x of road surface scene point in bodywork reference frame _v, y _v, 0) and respective pixel coordinate (x in plane of delineation coordinate system _i, y _i) corresponding relation:

$\left\{\begin{array}{c}{x}_i=\frac{{\mathrm{fx}}_v}{y_v\cos \mathrm{\&alpha;}+H\sin \mathrm{\&alpha;}}\\ y_i=\frac{f(H\cos \mathrm{\&alpha;}-y_v\sin \mathrm{\&alpha;})}{y_v\cos \mathrm{\&alpha;}+H\sin \mathrm{\&alpha;}}\end{array}\right.---\left(5\right)$

With the line translation of going forward side by side of formula 2,1 substitution formula 5:

$\left\{\begin{array}{c}{x}_v=\frac{H\&times;c\&times;\mathrm{dc}}{r\&times;\mathrm{dr}\&times;\cos \mathrm{\&alpha;}+f\sin \mathrm{\&alpha;}}\\ y_v=\frac{H(f\cos \mathrm{\&alpha;}-r\&times;\mathrm{dr}\&times;\sin \mathrm{\&alpha;})}{r\&times;\mathrm{dr}\&times;\cos \mathrm{\&alpha;}+f\sin \mathrm{\&alpha;}}\end{array}\right.---\left(6\right)$

What camera calibration commonly used at present and vision measuring method were used is exactly shape such as formula 6 computing formula.The present invention chooses and avoids a large amount of trigonometric function operations in the formula 6 for fear of a large amount of scene point, will progressively design new method from next step.

Camera perspective projection has following character: the linear projection on the road surface remains straight line to the plane of delineation and computer screen

If the angle of pitch of video camera is not equal to 90 °, the parallel lines on the road surface project to the plane of delineation and the computer screen backstage can converge at a bit, i.e. end point, and the slope of parallel lines is relevant on the coordinate of end point and the road surface.

Through deriving, can draw that slope is k on the road surface _vParallel lines end point coordinate (formula 8) in end point coordinate (formula 7) and the computer screen image coordinate system in the camera pixel plane coordinate system.

$\left\{\begin{array}{c}{\mathrm{VP}}_{\mathrm{xi}}=\frac{K_{v}f}{\cos \mathrm{\&alpha;}}\\ {\mathrm{VP}}_{\mathrm{yi}}=-f\tan \mathrm{\&alpha;}\end{array}\right.---\left(7\right)$

The fore-and-aft distance measurement refers to measures on the road surface a bit and the fore-and-aft distance between the bodywork reference frame X-axis

$\left\{\begin{array}{c}{\mathrm{VP}}_c=\frac{K_{v}f}{\mathrm{dc}\&times;\cos \mathrm{\&alpha;}}\\ {\mathrm{VP}}_r=-\frac{f\tan \mathrm{\&alpha;}}{\mathrm{dr}}\end{array}\right.---\left(8\right)$

As shown in Figure 3, in bodywork reference frame, get 2 P ₀And P ₁, its coordinate is respectively (x ₀, y ₀) and (x ₀, y ₁), with P ₀And P ₁Corresponding pixel is P ₀', P ₁', its coordinate is respectively (c ₀, r ₀) and (c ₀, r ₁).Take up an official post on the road surface and to get 1 P, its coordinate be (x, y), the coordinate of the pixel P ' corresponding with it be (c, r).Learn P through measuring ₀And P ₁Between fore-and-aft distance be dy, P and P set up an office ₀Between fore-and-aft distance be k * dy, then have:

With formula 6,8 substitution formula 9 and simplification, can get:

$k=\frac{y-y_0}{y_1-y_0}---\left(9\right)$

In the formula 10, Y ₀Fore-and-aft distance on the expression road surface between scene point P and the car body, Y are represented the fore-and-aft distance between any point P and car body on the road surface.

$Y=\frac{(r_0-r)(r-{\mathrm{VP}}_r)}{(r_0-r_1)(r_1-{\mathrm{VP}}_r)}\&times;\mathrm{dy}+Y_0---\left(10\right)$

The lateral separation measurement refers to measures on the road surface a bit and the lateral separation between the bodywork reference frame Y-axis.

In order to design new lateral separation measuring method, the present invention at first proposes a theorem and proof.

Theorem: arbitrary horizontal line is taken up an official post and is got two pixel P in the plane of delineation ₀' and P ₂', two points on the Dui Ying road surface are P with it ₀And P ₂, P then ₀And P ₂Between lateral separation and P ₀' and P ₂' between the ratio of lateral separation be one and P ₀' and P ₂' the irrelevant parameter of horizontal ordinate.

Proof: establish P ₀And P ₂Coordinate be (x ₀, y ₀) and (x ₂, y ₂), P ₀' and P ₂' coordinate be (c ₀, r ₀) and (c ₂, r ₂), then have according to formula 6:

$\frac{x_2-x_0}{c_2-{\mathrm{<figure-callout\; id="0"\; label="c"\; filenames="A20071000080700121.png,A20071000080700122.png"\; state="\{\{state\}\}">c</figure-callout>}}_0}=\frac{\frac{H\&times;\mathrm{dc}\&times;(c_2-c_0)}{r\&times;\mathrm{dr}\&times;\cos \mathrm{\&alpha;}+f\sin \mathrm{\&alpha;}}}{c_2-{\mathrm{<figure-callout\; id="0"\; label="c"\; filenames="A20071000080700121.png,A20071000080700122.png"\; state="\{\{state\}\}">c</figure-callout>}}_0}=\frac{H\&times;\mathrm{dc}}{r\&times;\mathrm{dr}\&times;\cos \mathrm{\&alpha;}+f\sin \mathrm{\&alpha;}}---\left(11\right)$

By formula 11 as can be known, P ₀And P ₂Between lateral separation and P ₀' and P ₂' between the ratio and the P of lateral separation ₀' and P ₂' horizontal ordinate irrelevant, but one and P ₀' and P ₂' the relevant parameter of ordinate.And P ₀' and P ₂' ordinate big more, this parameter is more little.The practical significance of this parameter is that making it is dx (r), then has at the pairing actual width of roadway of capable each pixel of r:

$\mathrm{dx}\left(r\right)=\frac{H\&times;\mathrm{dc}}{r\&times;\mathrm{dr}\&times;\cos \mathrm{\&alpha;}+f\sin \mathrm{\&alpha;}}---\left(12\right)$

In the practical application,, can calculate easily according to the horizontal ordinate of the horizontal ordinate of these two pixels and the scene point corresponding with it as long as certain delegation in the suitably selected computer screen image takes up an official post in this delegation then and get two pixels.

As shown in Figure 4, if suitably choose delegation in advance as the reference line in screen picture, its equation is r=r ₀, and calculate dx (r with the method for introducing above ₀).After this in screen picture, appoint and get a pixel P ', its coordinate be (c, r), the scene point corresponding with pixel P ' is P, its coordinate be (x, y).So, if connect parallel perpendicular line the end point VP in the plane of delineation parallel on the road surface with the bodywork reference frame Y-axis ₀With pixel P ' work one straight line, then the scene point straight line on the road surface of this pixel straight line correspondence also is the perpendicular line parallel with the bodywork reference frame Y-axis, and promptly the scene point of arbitrary pixel correspondence all equates with lateral separation between the car body on this pixel straight line.

As shown in Figure 4, cross end point VP ₀The coordinate that with pixel P ' work one straight line, the intersection point of this straight line and reference line is P ", P sets up an office " for (c ', r ₀), have:

$\frac{c^{\&prime;}}{c}=\frac{r-{\mathrm{VP}}_{r0}}{r_0-{\mathrm{<figure-callout\; id="0"\; label="VP\; r"\; filenames="A20071000080700121.png,A20071000080700122.png"\; state="\{\{state\}\}">VP</figure-callout>}}_r}\&DoubleRightArrow;c^{\&prime;}=\frac{r-{\mathrm{VP}}_{r0}}{r_0-{\mathrm{<figure-callout\; id="0"\; label="VP\; r"\; filenames="A20071000080700121.png,A20071000080700122.png"\; state="\{\{state\}\}">VP</figure-callout>}}_r}\&times;c---\left(13\right)$

Scene point that pixel P ' is corresponding and the transfer between the car body Y-axis are from being:

$X=c^{\&prime;}\&times;\mathrm{dx}\left(r_0\right)=\frac{r-{\mathrm{VP}}_{r0}}{r_0-{\mathrm{<figure-callout\; id="0"\; label="VP\; r"\; filenames="A20071000080700121.png,A20071000080700122.png"\; state="\{\{state\}\}">VP</figure-callout>}}_r}\&times;c\&times;\mathrm{dx}\left(r_0\right)---\left(14\right)$

After the present invention provides relation between bodywork reference frame and the camera coordinate system, make three specific straight lines on the preceding road surface of car body, straight line 1 and straight line 2 are parallel to the bodywork reference frame Y-axis, and straight line 3 is parallel to the bodywork reference frame X-axis, get three specified points as shown in Figure 5, its mid point P ₀(x ₀, y ₀) be the intersection point of straight line 1 and straight line 3, some P ₂(x ₂, y ₂) be the intersection point of straight line 2 and straight line 3, some P ₁(x ₁, y ₁) be positioned on the straight line 1.On the road surface

$\left\{\begin{array}{c}{x}_1=x_0\\ {\mathrm{<figure-callout\; id="2"\; label="y"\; filenames="A20071000080700122.png"\; state="\{\{state\}\}">y</figure-callout>}}_2=y_0\end{array}\right.---\left(15\right)$

Have following relation between the coordinate of three scene point:

Three pixels in the computer screen image coordinate system corresponding with three scene point on the road surface are respectively P ₀' (c ₀, r ₀), P ₁' (c ₁, r ₁) and P ₂' (c ₂, r ₂).Have following relation between the coordinate of these three pixels:

In computer screen image, on the road surface projected pixel line of straight line 1 and straight line 3 will intersect at end point (0, VP _r), will put P ₀(x ₀, y ₀), P ₁(x ₁, y ₁), P ₀' (c ₀, r ₀), P ₁' (c ₁, r ₁) and end point (0, VP _r) coordinate substitution formula 10, the measurement computing formula of fore-and-aft distance:

$y=\frac{(r_0-r)(r-{\mathrm{VP}}_r)}{(r_0-r_1)(r_1-{\mathrm{VP}}_r)}\&times;\mathrm{dy}+y_0---\left(17\right)$

In the formula 17, r is the ordinate of any one pixel in the computer screen image, and y represents the ordinate of scene point on the road surface corresponding with this pixel.

To put P ₀(x ₀, y ₀), P ₂(x ₂, y ₂), P ₀' (c ₀, r ₀), P ₂' (c ₂, r ₂) and end point (0, VP _r) coordinate substitution formula 14:

$x=\frac{r-{\mathrm{VP}}_r}{r_0-{\mathrm{VP}}_r}\&times;c\&times;\mathrm{dx}\left(r_0\right)---\left(18\right)$

In the formula 18, c, r are respectively the horizontal ordinate and the ordinate of any one pixel in the computer screen image, and x is the horizontal ordinate of scene point on the road surface of this pixel correspondence.

By formula 17,18 as can be known, when carrying out camera calibration, only need on the road surface, to make three three straight lines as shown in Figure 5, get three specified points, these three some P ₀(x ₀, y ₀), P ₁(x ₁, y ₁), P ₂(x ₂, y ₂) and with corresponding three the pixel P of these three points ₀' (c ₀, r ₀), P ₁' (c ₁, r ₁), P ₂' (c ₂, r ₂) coordinate be known, and the ordinate VP of end point _rAlso known.Therefore, the camera marking method of the present invention design is only based on three specified points on the road surface, can calibrate the parameter in the contrary projection formula 17,18 that the present invention derives, and compares with traditional scaling method, and is convenient and swift.

In case after camera calibration is finished, can use contrary projection formula 17,18 and carry out vision measurement.The coordinate figure substitution that is about to any one pixel in the computer screen image is the formula 17,18 demarcated of parameter, just can calculate horizontal stroke, the ordinate of road surface scene point in bodywork reference frame of this pixel correspondence, thereby draw horizontal, fore-and-aft distance between this scene point and the car body reference origin.

Contrary projection formula 17,18 that contrast the present invention derives and the contrary projection formula 6 that uses always at present are as can be known, the fixed parameter of representative all is the coordinate of three specified points and the pixel corresponding with it on the road surface in the formula 17,18, the demarcation mode is simple, and does not have trigonometric function to calculate in the formula.Compare with vision measuring method with traditional camera calibration, the present invention all has reduction significantly to the internal storage capacity of embedded computer and the requirement of CPU speed, helps the real-time implementation in the embedded computer system.

Claims (4)

1, a kind of three-point scaling measuring method, described measuring method comprises with the car body being that coordinate system is set up at the center, with the video camera is that coordinate system is set up at the center, it is that the direction of the vertical vehicle body side of Z axle is an X-axis that the car body front is oriented Y-axis vertical ground direction, it is characterized in that: on the road surface, draw two straight lines, wherein straight line and bodywork reference frame transverse axis, another straight line is parallel with the car body coordinate axis longitudinal axis, the intersection point of getting described straight line is as first particular scene point in the three-point scaling, each is selected a bit as second particular scene point in the three-point scaling and the 3rd particular scene point, according to three particular scene points and corresponding coordinate relation and the coordinate Calculation of end point and the world coordinates and the bodywork reference frame reference coordinate of the corresponding scene point of pixel between three pixels of projection on the pick-up lens in other positions of described two straight lines respectively.

2, a kind of three-point scaling measuring method according to claim 1, it is characterized in that: the pass of camera coordinate system and bodywork reference frame is X _v, Y _v, Z _v, camera coordinate system X _c, Y _c, Z _c, described car body coordinate

x _c=x _vIt is y that system closes with camera coordinate system _c=y _vCos α+Hsin α.

z _c＝-Hcosα+y _vsinα

3, a kind of three-point scaling measuring method according to claim 1 is characterized in that: the horizontal ordinate of end point is decided by the slope of parallel lines: $\begin{array}{c}{x}_i=(\frac{B_{v1}\cos \mathrm{\&alpha;}}{H}-K_v\sin \mathrm{\&alpha;})y_i+\frac{B_{v1}f\sin \mathrm{\&alpha;}}{H}+K_{v}f\cos \mathrm{\&alpha;}\\ x_i=(\frac{B_{v2}\cos \mathrm{\&alpha;}}{H}-K_v\sin \mathrm{\&alpha;})y_i+\frac{B_{v2}f\sin \mathrm{\&alpha;}}{H}+K_{v}f\cos \mathrm{\&alpha;}\end{array}$

And the coordinate of record end point.

4, a kind of three-point scaling measuring method according to claim 1 is characterized in that: with the coordinate substitution of end point $\begin{array}{c}y=\frac{(r_0-r)(r-{\mathrm{VP}}_r)}{(r_0-r_1)(r_1-{\mathrm{VP}}_r)}\&times;\mathrm{dy}+y_0.\\ x=\frac{r-{\mathrm{VP}}_r}{r_0-{\mathrm{VP}}_r}\&times;c\&times;\mathrm{dx}\left(r_0\right).\end{array}$

Images