CN103389121B - Based on angle detection device and the method for machine vision - Google Patents

Abstract

The invention discloses a kind of angle detection device based on machine vision, comprise housing and the rotating disk being arranged at enclosure interior, the center of rotating disk is provided with rotating shaft, rotating disk is provided with identical but at least two that the vary in size marks of shape, and described at least two are designated circle that center coincides, parallelogram, equilateral triangle or positive even numbers limit shape; Be provided with at least two encoding strip marks between mark, and every bar encoding strip target center line crosses center, the corresponding coding of every bar encoding strip mark, and the coding difference that every two encoding strip marks are corresponding; Also comprise the image collecting device for absorbing rotating disk image and lighting source.The present invention additionally provides a kind of angle detecting method based on machine vision simultaneously.Angle detection device structure of the present invention is simple, and contamination resistance is strong, and the impact resistance of rotating disk is stronger.Angle detecting method of the present invention, for the metrical error that pollutant causes, reject by image processing techniques identification, accuracy of detection is high.

Description

Based on angle detection device and the method for machine vision

Technical field

The present invention relates to encoder techniques field, particularly a kind of angle detection device based on machine vision and method.

Background technology

Rotary encoder is used to measure the device of rotating speed, converts the mechanical quantity such as the angular displacement of output shaft, angular velocity to corresponding electric pulse and with digital output.Rotary encoder can be divided into optical grating principle scrambler and magnetic magnetic grid principle scrambler.Grating principle rotary encoder comprises the photoelectric code disk that a center is provided with rotating shaft, photoelectric code disk there is the groove that annular is logical, dark, when photoelectric code disk rotates a pitch, under light-emitting component irradiates, light activated element obtains A phase, the B phase pulse signal with 90 degree of phase differential, A phase, B phase pulse signal turn after voltage transformation through frequency, obtain the voltage signal proportional with rotating shaft rotating speed, velocity amplitude and displacement can be recorded.The rotary encoder measuring accuracy of grating principle is high, but shortcoming to be contamination resistance poor.Magnetic grid rotary encoder utilizes the magnetic action of magnetic grid and magnetic head, with circle-shaped alternately magnetic on this magnetic rubber ring, produces S pole and N pole, adopts mistor or hall effect sensor as sensitive element, forms spike train by magnetic force, produce signal.The rotary encoder of magnetic grid principle has stronger antipollution, impact resistance, but shortcoming to be measuring accuracy not high.

Summary of the invention

The object of the invention is to overcome the deficiency that raster rotary coder contamination resistance is poor, magnetic grid rotary encoder measuring accuracy is not high existing in prior art, a kind of angle detection device based on machine vision is provided, a kind of angle detecting method based on machine vision is provided simultaneously, while realization accurately detects, improves contamination resistance, impact resistance.

In order to realize foregoing invention object, the invention provides following technical scheme:

A kind of angle detection device based on machine vision, comprise housing and the rotating disk being arranged at enclosure interior, the center of described rotating disk is provided with rotating shaft, rotating disk is provided with identical but at least two that the vary in size marks of shape, and described at least two are designated circle that center coincides, parallelogram, equilateral triangle or positive even numbers limit shape; Be provided with at least two encoding strip marks between mark, and every bar encoding strip target center line crosses center, the corresponding coding of every bar encoding strip mark, and the coding difference that every two encoding strip marks are corresponding; Also comprise the image collecting device for absorbing rotating disk image and lighting source, described lighting source and digital image acquisition device are fixed by housing.

So-called machine vision, replaces human eye measure and judge with machine exactly.Vision Builder for Automated Inspection refers to by machine vision product (i.e. image-pickup device, point CMOS and CCD two kinds) target will be ingested convert picture signal to, send special image processing system to, according to pixel distribution and the information such as brightness, color, be transformed into digitized signal; Picture system carries out various computing to extract clarification of objective to these signals, and then controls on-the-spot device action according to the result differentiated.

Further, described in be designated two, be respectively the first mark and second mark.Preferably, described first mark and the second mark are circle.

Apply the method that the above-mentioned angle detection device based on machine vision carries out angle detection, comprise the following steps:

(1) uncalibrated image harvester inner parameter matrix K;

(2) utilize image collecting device to absorb rotating disk image, from rotating disk image, be partitioned into the identification images of at least two marks, ask at least two identification image equations, and the image coordinate at the center of knowing according at least two identification image Solving Equations label takings;

(3) at moment t1, picked-up encoding strip logo image, resolves the first encoding strip target coding, and extracts the first encoding strip target first boundary image equation;

(4) at moment t2, picked-up encoding strip logo image, resolves the second encoding strip target coding, and extracts the second encoding strip target the second boundary image equation;

(5) ask for the line equation that goes out of mark place plane, and ask for the vanishing point coordinate of the first place, border rectilinear direction on the line that goes out and the vanishing point coordinate of the second boundary place rectilinear direction on the line that goes out;

(6) according to two intersecting point coordinates and parameter matrix K, ask for t1 moment first encoding strip mark and t2 moment second encoding strip target unit direction vector respectively, ask for the angle of rotating in t1 to the t2 time period according to the first encoding strip mark and the second encoding strip target unit direction vector.

Further, described in step (2), at least two marks are circle.

Further, utilize image collecting device to absorb rotating disk image described in step (2), from rotating disk image, be partitioned into the identification image of at least two marks, the method asking at least two identification image equations is:

If g1 is the image that a circular curve obtains through image acquisition device, extract the border of this circular curve, and on border, choose arbitrarily at least 5 points, seek out at least 5 corresponding image coordinate of point, be expressed as (u _tv _t) t=1,2,3..., then the equation general formula of g1 can be described as: $g(u,v)={\left[\begin{array}{c}u\\ v\\ 1\end{array}\right]}^T\·M\·\left[\begin{array}{c}u\\ v\\ 1\end{array}\right]=0,$ $M=\left[\begin{array}{ccc}A& 0.5C& 0.5D\\ 0.5C& B& 0.5E\\ 0.5D& 0.5E& F\end{array}\right],$ Get A=1, by least 5 corresponding image coordinate (u of point _tv _t) substitute in above formula, have ${\left[\begin{array}{c}{u}_i\\ v_i\\ 1\end{array}\right]}^T\·\left[\begin{array}{ccc}1& 0.5c& 0.5d\\ 0.5c& b& 0.5e\\ 0.5d& 0.5e& f\end{array}\right]\·\left[\begin{array}{c}{u}_i\\ v_i\\ 1\end{array}\right]=0,$ Wherein, b=B/A, c=C/A, d=D/A, e=E/A, f=F/A; Solve b, c, d, e, f, then identification image equation is: G _i(u, v)=A _iu ²+ B _iv ²+ C _iuv+D _iu+E _iv+F _i=0 wherein, A _i=1.

Further, described in step (2) according to the method for the image coordinate at the center of at least two identification image Solving Equations label takings knowledges be:

If the image coordinate at center is (u _ov _o), the image coordinate at circular image equation and its center meets relation $\left[\begin{array}{c}{a}^{\′}\\ b^{\′}\\ c^{\′}\end{array}\right]=\left[\begin{array}{ccc}A& 0.5C& 0.5D\\ 0.5C& B& 0.5E\\ 0.5D& 0.5E& F\end{array}\right]\·\left[\begin{array}{c}{u}_0\\ v_0\\ 1\end{array}\right],$ Wherein, a', b', c' be respectively circle the coefficient of the line equation that goes out in the plane; According to $\left[\begin{array}{c}{a}^{\′}\\ b^{\′}\\ c^{\′}\end{array}\right]=\left[\begin{array}{ccc}{A}_1& 0.5C_1& 0.5D_1\\ 0.5C_1& B_1& 0.5E_1\\ 0.5D_1& 0.5E_1& F_1\end{array}\right]\·\left[\begin{array}{c}{u}_0\\ v_0\\ 1\end{array}\right]=\left[\begin{array}{ccc}{A}_2& 0.5C_2& 0.5D_2\\ 0.5C_2& B_2& 0.5E_2\\ 0.5D_2& 0.5E_2& F_2\end{array}\right]\·\left[\begin{array}{c}{u}_0\\ v_0\\ 1\end{array}\right]$ Image coordinate (the u at center can be drawn _ov _o); Wherein, A ₁, B ₁, C ₁, D ₁, E ₁, F ₁for the coefficient of one of them identification image equation, A ₂, B ₂, C ₂, D ₂, E ₂, F ₂for another knows the coefficient of image equation.

Further, the method asking for the line equation that goes out of mark place plane of described step (5) is:

By center image coordinate (u _ov _o), at least two identification image equations, according to equation $\left[\begin{array}{c}{a}^{\′}\\ b^{\′}\\ c^{\′}\end{array}\right]=\left[\begin{array}{ccc}A& 0.5C& 0.5D\\ 0.5C& B& 0.5E\\ 0.5D& 0.5E& F\end{array}\right]\·\left[\begin{array}{c}{u}_0\\ v_0\\ 1\end{array}\right]$ Solve the value of coefficient a', b', the c' of the line equation that goes out, then the line equation that goes out identifying place plane is: a'u+b'v+c'=0.

Further, according to two intersecting point coordinates and parameter matrix K described in step (6), ask for t1 moment first encoding strip mark and t2 moment second encoding strip target unit direction vector respectively, its method is:

If the unit direction vector of place, encoding strip mark border straight line is then it can by the vanishing point coordinate (u that this side up _j, v _j) substitute into ${\mathrm{\ρ}}_j\·\left[\begin{array}{c}{v}_{xj}\\ v_{yj}\\ v_{zj}\end{array}\right]=K^{-1}\left[\begin{array}{c}{u}_j\\ v_j\\ 1\end{array}\right]$ Draw, wherein, ρ _iit is any non-zero constant; Use vectorial method for normalizing, and be marked on the appearance position in binary encoding bar logo image according to encoding strip, the coordinate of unit direction vector can be determined; Obtain the first encoding strip mark and the second encoding strip target unit direction vector thus respectively.

Further, the angle of rotating in t1 to the t2 time period is asked for according to the first encoding strip mark and the second encoding strip target unit direction vector described in step (6), its method is: α=β ± θ, wherein, α is the angle of rotating in t1 to the t2 time period, β is the angle between t1 moment first encoding strip mark and t2 moment second encoding strip target unit direction vector, and θ is the angle between the first encoding strip mark and the second encoding strip mark.

Compared with prior art, beneficial effect of the present invention:

The present invention is based on the angle detection device of machine vision, rotational angle, velocity of rotation, the increment type of rotation acceleration or absolute type can be realized and measure.Due to contactless between image collecting device and rotating disk, angle detection device contamination resistance of the present invention is strong.Because rotating disk does not need to possess transmittancy, therefore rotating disk can adopt the stiff materials such as alloy to make, and impact resistance is stronger.

The angle detection device structure that the present invention is based on machine vision is simple, and the scale (encoding strip mark) of rotating disk is drawn simpler compared to the making of optical grating, and cost is lower.

The present invention is based on the angle detecting method of machine vision, use image interior in a big way on rotating disk to originate as measurement data, coordinate image processing techniques, reduce the impact that image local distortion produces whole detection precision; And if pollutant identifies with rotating disk and differs greatly in visual characteristic, reject by image processing techniques identification, can not have an impact to final measurement, avoid the problem that accuracy of detection that pollutant causes is lower, improve accuracy of detection, measure relative error by experiment lower than 0.5%.

Angle detecting method of the present invention uses flexibly, and the device realizing angle detection can independently make, and also can draw encoding strip mark on existing, then combining image harvester can realize angle detection.

Accompanying drawing illustrates:

Fig. 1 is the angle detection device structural representation that the present invention is based on machine vision.

Fig. 2 is encoding strip target structural representation.

Fig. 3 is the process flow diagram of the angle detecting method that the present invention is based on machine vision.

Mark in figure: 1-rotating shaft, 2-encoding strip mark, 3-cylindrical identifies, and 4-inner circle identifies, 5-image collecting device, 6-rotating disk, 7-housing, 8-lighting source, 201-bits of coded.

Embodiment

Below in conjunction with test example and embodiment, the present invention is described in further detail.But this should be interpreted as that the scope of the above-mentioned theme of the present invention is only limitted to following embodiment, all technology realized based on content of the present invention all belong to scope of the present invention.

With reference to figure 1, angle detection device based on machine vision provided by the invention, comprise the rotating disk 6 that a center is provided with rotating shaft 1, rotating disk 6 is provided with concentric but the cylindrical of different-diameter mark 3 and inner circle mark 4, the diameter of cylindrical mark 3 is greater than inner circle mark 4, be provided with at least two encoding strip marks 2 between cylindrical mark 3 and inner circle mark 4, and the center line of every bar encoding strip mark 2 crosses the center of circle, namely the center of circle of rotating disk 6 is on the center line of encoding strip mark 2.When encoding strip mark 2 is two or more, the distance (i.e. angle) often between adjacent two encoding strip marks 2 can be arranged arbitrarily, but angle between adjacent two encoding strip marks 2 is known.With reference to figure 2, encoding strip mark 2 is provided with several bits of coded 201, and bits of coded represents different encoded radios after being colored as different colors.The bits of coded number that all encoding strips on same rotating disk are put on is identical, and coded system is identical, namely carries out coding/decoding (namely according to equidirectional, reading the encoded radio that encoding strip puts on each bits of coded successively) according to equidirectional, obtains coding.If adopt m kind Color pair bits of coded to carry out painted, an encoding strip is put on and is arranged n bits of coded, so can obtain m ⁿthe encoding strip mark of bar different coding, although the encoding strip mark quantity that rotating disk is arranged can be arranged arbitrarily, the every bar encoding strip target coding on same rotating disk must be different.Because the encoding strip mark on rotating disk is without the need to arranging too much (at least one can realize), for the ease of coding/decoding, preferably adopt two kinds of colors presentation code value 1 and encoded radio 0 respectively, every bar encoding strip mark arranges 4-6 bits of coded.In the present embodiment, black is adopted to represent encoded radio 0, white presentation code value 1, every bar encoding strip mark arranges 6 bits of coded, as shown in Figure 2, this encoding strip target is encoded to 011011 (according to reading successively from top to bottom, namely reading successively from closing on center of circle end to away from center of circle end).Encoding strip is put on and is arranged several bits of coded, and bits of coded is colored as different colours to represent different coding value, such object makes the corresponding coding of an encoding strip mark, certain, other embodiments (such as, directly putting on label coding at encoding strip) can be taked to realize the corresponding coding of an encoding strip mark.Rotating disk 6 is arranged at the inside of housing 7, and one end of rotating shaft 1 passes housing 7, so that be connected with engine.The angle detection device that the present invention is based on machine vision also comprises the lighting source 8 and digital image acquisition device 5 that are arranged at enclosure interior, lighting source 8 and digital image acquisition device 5 are fixed by housing 7, the installation site of lighting source 8 and digital image acquisition device 5 is not particularly limited, but must meet and can obtain image clearly, and at least comprise in the image obtained inner circle mark 4, cylindrical mark 3 part or all of, and at least one complete encoding strip mark 2.Be configured with data processing unit in digital image acquisition device 5, digital image acquisition device 5 directly can adopt digital camera.

It should be noted that, the shape of encoding strip mark 2 can be triangle, the arbitrary shape such as trapezoidal, is not limited only to the rectangle shown in Fig. 2.The cylindrical mark 3 that rotating disk 6 is arranged and inner circle mark 4 are not limited only to two, can arrange multiple concentric circles mark, but be at least two; Cylindrical mark 3 and inner circle mark 4 (concentric circles marks) are also not limited only to circle, cylindrical mark 3 and inner circle mark 4 also can be parallelogram, equilateral triangle and any positive even numbers limit shape, but the center of cylindrical mark 3 and inner circle mark 4 coincides, and size is unequal.

With reference to figure 3, utilize the above-mentioned angle detection device based on machine vision to carry out the method for angle detection, comprise the following steps:

(1) demarcate digital camera (i.e. digital image acquisition device) inner parameter matrix K, and be kept in its supporting data processing unit.The scaling method of digital camera inner parameter matrix K belongs to prior art, does not do too much elaboration herein.For the demarcation of digital camera inner parameter matrix K, again digital camera is installed in the housing of angle detection device after can first demarcating K; Also first digital camera can be installed in the housing of angle detection device, and then adopt self calibration algorithm automatically to complete the demarcation of parameter matrix K, but need increase and calibrate mark accordingly.

(2) digital camera is utilized to absorb rotating disk image (comprising inner circle identification image and cylindrical identification image in rotating disk image), inner circle identification image and cylindrical identification image is partitioned into from rotating disk image, and ask for inner circle identification image equation and cylindrical identification image equation, then ask for its center of circle image coordinate according to inner circle identification image equation and cylindrical identification image equation.

(2-1) inner circle identification image equation and cylindrical identification image equation is asked for.

If g1 is the image that a circular curve collects through digital camera, extract the border of this circular curve, and on border, choose arbitrarily at least 5 points, seek out at least 5 corresponding image coordinate of point, be expressed as (u _tv _t) t=1,2,3..., then the equation general formula of g1 can be described by following equation:

$g(u,v)={\left[\begin{array}{c}u\\ v\\ 1\end{array}\right]}^T\·M\·\left[\begin{array}{c}u\\ v\\ 1\end{array}\right]=0,M=\left[\begin{array}{ccc}A& 0.5C& 0.5D\\ 0.5C& B& 0.5E\\ 0.5D& 0.5E& F\end{array}\right]$

Because the element in Metzler matrix is not 0 entirely, therefore by Metzler matrix divided by A, then Metzler matrix can be expressed as:

$M=\left[\begin{array}{ccc}1& 0.5c& 0.5d\\ 0.5c& b& 0.5e\\ 0.5d& 0.5e& f\end{array}\right]$

Wherein, b=B/A, c=C/A, d=D/A, e=E/A, f=F/A.From above formula, the independent variable in Metzler matrix becomes 5, and (that is, b, c, d, e, f), therefore the seat of at least 5 of known g1 points

Mark (u _tv _t) substitute into $g(u,v)={\left[\begin{array}{c}u\\ v\\ 1\end{array}\right]}^T\·M\·\left[\begin{array}{c}u\\ v\\ 1\end{array}\right]=0,$ At least 5 prescription journeys can be obtained:

${\left[\begin{array}{c}{u}_i\\ v_i\\ 1\end{array}\right]}^T\·\left[\begin{array}{ccc}1& 0.5c& 0.5d\\ 0.5c& b& 0.5e\\ 0.5d& 0.5e& f\end{array}\right]\·\left[\begin{array}{c}{u}_i\\ v_i\\ 1\end{array}\right]=0$

By above formula solve 5 variablees (b, c, d, e, f) after, uniquely can determine Metzler matrix.If equation quantity is more than 5, then least square method is utilized to ask for the optimum solution of Metzler matrix.

According to the method described above, solve concentric circles identification image equation (in the present embodiment, being namely obtain cylindrical identification image equation and inner circle identification image equation) successively, be designated as:

G _i(u,v)＝A _iu ²+B _iv ²+C _iuv+D _iu+E _iv+F _i＝0

Wherein, A _i=1.

(2-2) center of circle image coordinate is calculated.

If the image coordinate in the center of circle is (u _ov _o), the image coordinate in circular image equation and its center of circle meets following relation:

$\left[\begin{array}{c}{a}^{\′}\\ b^{\′}\\ c^{\′}\end{array}\right]=\left[\begin{array}{ccc}A& 0.5C& 0.5D\\ 0.5C& B& 0.5E\\ 0.5D& 0.5E& F\end{array}\right]\·\left[\begin{array}{c}{u}_0\\ v_0\\ 1\end{array}\right],$

Wherein, a', b', c' be respectively circle the coefficient of the line equation that goes out in the plane.Owing to having solved cylindrical identification image equation and inner circle identification image equation above, therefore lower equation can be able to:

$\left[\begin{array}{c}{a}^{\′}\\ b^{\′}\\ c^{\′}\end{array}\right]=\left[\begin{array}{ccc}{A}_1& 0.5C_1& 0.5D_1\\ 0.5C_1& B_1& 0.5E_1\\ 0.5D_1& 0.5E_1& F_1\end{array}\right]\·\left[\begin{array}{c}{u}_0\\ v_0\\ 1\end{array}\right]=\left[\begin{array}{ccc}{A}_2& 0.5C_2& 0.5D_2\\ 0.5C_2& B_2& 0.5E_2\\ 0.5D_2& 0.5E_2& F_2\end{array}\right]\·\left[\begin{array}{c}{u}_0\\ v_0\\ 1\end{array}\right]$

Can uniquely solve central coordinate of circle (u thus _ov _o).When rotating disk being provided with the concentric circles mark more than two, after solving all concentric circles identification image equations, then the optimum solution of central coordinate of circle can be solved by least square method.

(3) circular image equation or inner circle mark circular image equation is identified, according to equation by center of circle image coordinate, cylindrical $\left[\begin{array}{c}{a}^{\′}\\ b^{\′}\\ c^{\′}\end{array}\right]=\left[\begin{array}{ccc}A& 0.5C& 0.5D\\ 0.5C& B& 0.5E\\ 0.5D& 0.5E& F\end{array}\right]\·\left[\begin{array}{c}{u}_0\\ v_0\\ 1\end{array}\right]$ (line that goes out is defined as: in projective mapping to solve the line that goes out, in plane, all infinity points are positioned on straight line, this straight line is " go out line " or " blanking line ", English is: vanishline) value of coefficient a', b', c' of equation, then the equation of the line L that goes out of plane of rotor disc (i.e. the plane at cylindrical mark and inner circle mark place) is asked for, the line equation that goes out is: a'u+b'v+c'=0, and is saved in the supporting data processing unit of digital camera.

It should be noted that, if for changing digital camera and/or rotating disk in measuring process, then need to re-execute step (1) to (3), again ask for center of circle image coordinate, go out line equation etc.

(4) when time t1, digital camera gathers rotating disk image (comprising at least one complete mark of encoding strip clearly in rotating disk image), identifies, is partitioned into clear, the most complete encoding strip logo image.The implication of the encoding strip logo image of most complete display is herein, adopts the sharpness of out of focus decision algorithm to all rotating disk images obtained to calculate, chooses the encoding strip logo image that sharpness is maximum.After digital camera installs, its most blur-free imaging district determines thereupon, and therefore, the encoding strip logo image of most complete display preferentially obtains from most blur-free imaging district.

(5) decoding and coding bar mark C1, namely on discrimination encoding strip mark C1, the encoded radio of each bits of coded is 1 or is 0, and composition coding d1, is saved in data processing unit.

(6) adopt Algorithm of fitting a straight line, the image equation on a matching partition encoding bar mark C1 left side (or right) border, and solve the image equation on this left side (or right) border and the intersection point P1 (u of the line L that goes out ₁v ₁).It should be noted that, intersection point P1 (u herein ₁v ₁) actual be an encoding strip mark C1 left side on the line that goes out (or right) border institute vanishing point in the straight direction, in actual treatment, the acquisition of vanishing point coordinate is exactly ask for absorbing boundary equation and the intersecting point coordinate of the line that goes out.

(7) at time t2, repeat step (4) and (5), extract the coding d2 of encoding strip mark C2, and obtain the image equation on a left encoding strip mark C2 left side (or right) border and the intersection point P2 (u of the line L that goes out ₂v ₂).Intersection point P2 (u herein ₂v ₂) actual be an encoding strip mark C2 left side on the line that goes out (or right) border vanishing point in the straight direction.

It should be noted that, if the angle that rotating disk rotates within t1 to the t2 time is, the angle between the encoding strip mark C1 in t1 moment and the encoding strip mark C2 position in t2 moment and the integral multiple sum of 360 °, so just cannot detect the velocity of rotation of rotating disk.Therefore, by test of many times verification mode, reduce the mistiming between t1 and t2 gradually, two different coding bar target images that final acquisition is taken in a rotation period, that is, the angle that rotating disk rotates within t1 to the t2 time is the angle between the encoding strip mark C1 in t1 moment and the encoding strip mark C2 position in t2 moment.If encoding strip mark C1 and C2 photographed in t1 and the t2 moment is one and same coding bar mark, so the angle of encoding strip mark C1 (or C2) respectively between t1 moment and t2 moment position is the angle that rotating disk rotates within t1 to the t2 time.In order to measure the rotational angle with the rotating disk of higher rotation speed more accurately, get the image of the encoding strip mark C2 different from C1 in the t2 moment, the encoding strip mark namely on rotating disk should be set at least two.

(8) ask for the angle between t1 moment encoding strip mark C1 and t2 moment encoding strip mark C2, and ask for velocity of rotation according to this angle.This step comprises:

(8-1) by intersection point P1, P2, camera internal parameter K and twice encoding strip target image information, the direction vector that encoding strip mark C1 and C2 points to respectively on t1, t2 time point is solved.

If the unit direction vector of place, encoding strip mark border straight line is then it can by the vanishing point coordinate (u that this side up _j, v _j) substitute into following formula acquisition:

${\mathrm{\ρ}}_j\·\left[\begin{array}{c}{v}_{xj}\\ v_{yj}\\ v_{zj}\end{array}\right]=K^{-1}\left[\begin{array}{c}{u}_j\\ v_j\\ 1\end{array}\right]$

Wherein, when asking for the unit direction vector v1 on a left side (or right) border of encoding strip mark C1, j=1, when asking for the unit direction vector v2 on a left side (or right) border of encoding strip mark C2, j=2, ρ _ibe any non-zero constant, use vectorial method for normalizing, may separate for two groups that unit direction vector can be obtained.And the position of digital camera is once be provided with, the corresponding relation of image-region and direction vector can be determined, that is, according to the appearance position of encoding strip logo image on general image, can determine that it is truly separated.Unit direction vector v1, the v2 on a left side (or right) border of encoding strip mark C1, C2 can be obtained thus respectively, unit direction vector v1, the v2 on a left side of encoding strip mark C1, C2 (or right) border are the direction vector that encoding strip mark C1 and C2 points to respectively on t1, t2 time point, the direction vector that encoding strip mark C1 and C2 points to respectively on t1, t2 time point gets final product the current location of presentation code bar mark C1 and C2 respectively on t1, t2 time point, and the angle namely between v1, v2 is the angle between the encoding strip mark C1 in t1 moment and the encoding strip mark C2 in t2 moment.

(8-2) according to the angle between the t1 moment encoding strip mark C1 asked for and t2 moment encoding strip mark C2, and the angle that rotating disk is fixed between encoding strip mark C1 and C2, ask for the angle that rotating disk rotates within t1 to the t2 time: α=β ± θ, wherein, α is the angle of rotating in t1 to the t2 time period, β is the angle between t1 moment first encoding strip mark and t2 moment second encoding strip target unit direction vector, and θ is the angle between the first encoding strip mark and the second encoding strip mark.

If definition rotate counterclockwise as positive dirction, and reality for rotating counterclockwise time, so within t1 to the t2 time, the angle of dial rotation is α=β+θ.

If definition rotates counterclockwise as positive dirction, when reality is for rotating clockwise, so within t1 to the t2 time, the angle of dial rotation is α=β-θ.

Introduce time interval parameter, velocity of rotation, rotation acceleration parameter can be obtained.

All features disclosed in this instructions, or the step in disclosed all methods or process, except mutually exclusive feature and/or step, all can combine by any way.

Arbitrary feature disclosed in this instructions (comprising any accessory claim, summary and accompanying drawing), unless specifically stated otherwise, all can be replaced by other equivalences or the alternative features with similar object.That is, unless specifically stated otherwise, each feature is an example in a series of equivalence or similar characteristics.

Claims (4)

1. the angle detection device based on machine vision, it is characterized in that, comprise housing and the rotating disk being arranged at enclosure interior, the center of described rotating disk is provided with rotating shaft, rotating disk is provided with identical but at least two that the vary in size marks of shape, and described at least two are designated circle that center coincides, parallelogram, equilateral triangle or positive even numbers limit shape; Be provided with at least two encoding strip marks between mark, and every bar encoding strip target center line crosses center, the corresponding coding of every bar encoding strip mark, and the coding difference that every two encoding strip marks are corresponding; Also comprise the image collecting device for absorbing rotating disk image and lighting source, described lighting source and digital image acquisition device are fixed by housing;

Angle between adjacent described two coding bar codes is known, the installation site of described lighting source and described digital image acquisition device must meet can obtain image clearly, described image clearly comprise mark in circle, the outer mark of circle part or all of, at least two complete encoding strip marks.

2. the angle detection device based on machine vision according to claim 1, is characterized in that, described in be designated two, be respectively the first mark and second mark.

3. the angle detection device based on machine vision according to claim 2, is characterized in that, described first mark and the second mark are circle.

4. application rights requires that the angle detection device based on machine vision described in 1 carries out the method for angle detection, it is characterized in that, comprises the following steps:

(1) uncalibrated image harvester inner parameter matrix K;

(2) utilize image collecting device to absorb rotating disk image, from rotating disk image, be partitioned into the identification images of at least two marks, ask at least two identification image equations, and the image coordinate at the center of knowing according at least two identification image Solving Equations label takings;

(3) at moment t1, picked-up encoding strip logo image, resolves the first encoding strip target coding, and extracts the first encoding strip target first boundary image equation;

(4) at moment t2, picked-up encoding strip logo image, resolves the second encoding strip target coding, and extracts the second encoding strip target the second boundary image equation;

(5) ask for the line equation that goes out of mark place plane, and ask for the vanishing point coordinate of the first place, border rectilinear direction on the line that goes out and the vanishing point coordinate of the second boundary place rectilinear direction on the line that goes out;

(6) according to two intersecting point coordinates and parameter matrix K, ask for t1 moment first encoding strip mark and t2 moment second encoding strip target unit direction vector respectively, ask for the angle of rotating in t1 to the t2 time period according to the first encoding strip mark and the second encoding strip target unit direction vector; Described in step (2), at least two marks are circle;

Utilize image collecting device to absorb rotating disk image described in step (2), from rotating disk image, be partitioned into the identification image of at least two marks, the method asking at least two identification image equations is:

If g1 is the image that a circular curve obtains through image acquisition device, extract the border of this circular curve, and on border, choose arbitrarily at least 5 points, seek out at least 5 corresponding image coordinate of point, be expressed as then the equation general formula of g1 can be described as: $g(u,v)={\left[\begin{array}{c}u\\ v\\ 1\end{array}\right]}^T\·M\·\left[\begin{array}{c}u\\ v\\ 1\end{array}\right]=0,M=\left[\begin{array}{ccc}A& 0.5C& 0.5D\\ 0.5C& B& 0.5E\\ 0.5D& 0.5E& F\end{array}\right],$ Get A=1, by least 5 corresponding image coordinate (u of point _tv _t) substitute in above formula, have

${\left[\begin{array}{c}{u}_i\\ v_i\\ 1\end{array}\right]}^T\·\left[\begin{array}{ccc}1& 0.5c& 0.5d\\ 0.5c& b& 0.5e\\ 0.5d& 0.5e& f\end{array}\right]\·\left[\begin{array}{c}{u}_i\\ v_i\\ 1\end{array}\right]=0,$ Wherein, b=B/A, c=C/A, d=D/A, e=E/A, f=F/A; Solve b, c, d, e, f, then identification image equation is:

G _i(u, v)=A _iu ²+ B _iv ²+ C _iuv+D _iu+E _iv+F _i=0 wherein, A _i=1;

Described in step (2) according to the method for the image coordinate at the center of at least two identification image Solving Equations label takings knowledges be:

If the image coordinate at center is (u _ov _o), the image coordinate at circular image equation and its center meets relation $\left[\begin{array}{c}{a}^{\′}\\ b^{\′}\\ c^{\′}\end{array}\right]=\left[\begin{array}{ccc}A& 0.5C& 0.5D\\ 0.5C& B& 0.5E\\ 0.5D& 0.5E& F\end{array}\right]\·\left[\begin{array}{c}{u}_0\\ v_0\\ 1\end{array}\right],$ Wherein, a', b', c' be respectively circle the coefficient of the line equation that goes out in the plane; According to $\left[\begin{array}{c}{a}^{\′}\\ b^{\′}\\ c^{\′}\end{array}\right]=\left[\begin{array}{ccc}{A}_1& 0.5C_1& 0.5D_1\\ 0.5C_1& B_1& 0.5E_1\\ 0.5D_1& 0.5E_1& F_1\end{array}\right]\·\left[\begin{array}{c}{u}_0\\ v_0\\ 1\end{array}\right]=\left[\begin{array}{ccc}{A}_2& 0.5C_2& 0.5D_2\\ 0.5C_2& B_2& 0.5E_2\\ 0.5D_2& 0.5E_2& F_2\end{array}\right]\·\left[\begin{array}{c}{u}_0\\ v_0\\ 1\end{array}\right]$ Image coordinate (the u at center can be drawn _ov _o); Wherein, A ₁, B ₁, C ₁, D ₁, E ₁, F ₁for the coefficient of one of them identification image equation, A ₂, B ₂, C ₂, D ₂, E ₂, F ₂for another knows the coefficient of image equation;

The method asking for the line equation that goes out of mark place plane of described step (5) is:

By center image coordinate (u _ov _o), at least two identification image equations, according to equation $\left[\begin{array}{c}{a}^{\′}\\ b^{\′}\\ c^{\′}\end{array}\right]=\left[\begin{array}{ccc}A& 0.5C& 0.5D\\ 0.5C& B& 0.5E\\ 0.5D& 0.5E& F\end{array}\right]\·\left[\begin{array}{c}{u}_0\\ v_0\\ 1\end{array}\right]$ Solve the value of coefficient a', b', the c' of the line equation that goes out, then the line equation that goes out identifying place plane is: a ' u+b ' v+c '=0;

According to two intersecting point coordinates and parameter matrix K described in step (6), ask for t1 moment first encoding strip mark and t2 moment second encoding strip target unit direction vector respectively, its method is:

If the unit direction vector of place, encoding strip mark border straight line is then it can by the vanishing point coordinate (u that this side up _j, v _j) substitute into ${\mathrm{\ρ}}_j\·\left[\begin{array}{c}{v}_{xj}\\ v_{yj}\\ v_{zj}\end{array}\right]=K^{-1}\left[\begin{array}{c}{u}_j\\ v_j\\ 1\end{array}\right]$ Draw, wherein, ρ _iit is any non-zero constant; Use vectorial method for normalizing, and be marked on the appearance position in binary encoding bar logo image according to encoding strip, the coordinate of unit direction vector can be determined; Obtain the first encoding strip mark and the second encoding strip target unit direction vector thus respectively;

The angle of rotating in t1 to the t2 time period is asked for according to the first encoding strip mark and the second encoding strip target unit direction vector described in step (6), its method is: α=β ± θ, wherein, α is the angle of rotating in t1 to the t2 time period, β is the angle between t1 moment first encoding strip mark and t2 moment second encoding strip target unit direction vector, and θ is the angle between the first encoding strip mark and the second encoding strip mark.