CN103528568B - A kind of object pose image measuring method based on wireless channel - Google Patents

Abstract

A kind of object pose image measuring method based on wireless channel data transmission system of the present invention, relates to image procossing and pose measurement technology.Including arranging transmitting terminal and receiving terminal, transmitting terminal includes video camera and graphics processing unit, and graphics processing unit extracts at least 3 characteristic points of shooting image, determines its coordinate, set up the matching relationship of described coordinate and the priori features of target, generate image characteristic information data；Image characteristic information data is transferred to receiving terminal by wireless channel from transmitting terminal；The pose solving unit of receiving terminal, uses the image characteristic information data received to carry out pose parameter resolving.The image measurement of pose is divided into feature extraction and pose parameter to resolve two steps by the present invention, uses transmission image characteristic information data to replace entire image, meets the wireless channel requirement that capacity is relatively low.Feature extraction completes under conditions of high-resolution and high frame rate image, thus ensure that the certainty of measurement of pose parameter.

Description

A kind of object pose image measuring method based on wireless channel

Technical field

The present invention relates to image procossing and pose measurement technology, particularly relate to a kind of based on wireless channel data transmission system Object pose image measuring method.

Background technology

Pose image measurement technology has the advantage not contacting testee, at field tools such as scientific research, military affairs, space developments There is highly important using value.In the scene of some pose measurement, need that observation station is arranged on aircraft and (such as fly Machine, airship, rocket etc.) on, measure the phase contraposition of the airbound target (such as satellite, rocket etc.) with aircraft with relative motion Appearance parameter.It is arranged on carry-on observation station to be communicated with grounded receiving station by wireless channel.

The precision of image measurement is highly dependent on image resolution ratio and frame per second, when image resolution ratio and frame per second are sufficiently high High measurement accuracy can be obtained.In the field such as space flight, military affairs, when system of measuring uses wireless channel transmission data, channel capacity is relatively The low image resolution ratio making measurement system and frame per second are relatively low, need, by rational conceptual design, to optimize pose image measurement skill Each step of art, reaches higher pose measurement precision.

The object pose remote image observation system of Beijing Institute of Technology's space flight measurement and control laboratory development, uses data compression Technology achieves being wirelessly transferred at relatively low channel capacity hypograph, and image can be used for seeing object pose qualitatively Survey.But being limited by radio channel capacity, image resolution ratio and code check that this system is transmitted are relatively low, it is impossible to object pose Carry out quantitative measurment.

The extraterrestrial target pose measuring method of view-based access control model is studied by the National University of Defense Technology, at paper " base Extraterrestrial target position attitude measurement method research in vision " in, it is proposed that accurately survey under conditions of relatively high image resolution The method of amount object pose, but the method does not accounts for the problem that high-definition picture cannot transmit under low channel capacity.

Typical wireless channels capacity is not higher than 2Mbps, is unable to realize under the channel capacity that said method is relatively low at this The high precision image of object pose is measured, is required for the lower data transmission conditions design object position of this channel capacity for this Appearance image measuring method.

Summary of the invention

In order to solve drawbacks described above present in prior art, the technical problem to be solved in the present invention is for wireless channel The transmission conditions of low channel capacity a kind of object pose image measuring method is proposed, the method is capable of object pose High precision image is measured, it is adaptable to have various types of cooperative targets of prior information.

The present invention solves its technical problem and is adopted the technical scheme that:

A kind of object pose image measuring method based on wireless channel, position-pose measurement is decomposed into feature extraction and Pose parameter resolves two links, and the transmission code rate of the image characteristic information data obtained in feature extraction step meets wireless communication The low capacity in road limits.

The method of the present invention includes, arranges transmitting terminal and receiving terminal；

Described transmitting terminal includes video camera and graphics processing unit；The high resolution graphics of described video camera shooting airbound target Picture；Wherein the image resolution ratio of video camera shooting is higher than 512 × 512, and frame per second is higher than 50bps；

Described graphics processing unit extracts at least 3 characteristic points of described video camera shooting image, determines that characteristic point is at figure Coordinate in Xiang；And set up the matching relationship of the priori features of described coordinate and target, generate image characteristic information data.

In the image shooting video camera carries out the step of feature extraction, the characteristic point extracted can be described image Angle point, the characteristics of image such as edge.

In described step, the image characteristic information data of described generation need to be through Image semantic classification, target detection, feature The links such as detection, characteristic matching, image characteristic information data output, above link is optional to be become with what measurement scene adapted Ripe known algorithm realizes.

Owing to target to be measured in the method for the present invention is cooperative target, therefore there is priori features information, with image procossing Relevant algorithm can take known corresponding Feature Correspondence Algorithm, the image feature information of generation according to the difference of target signature Data can calculate object pose parameter.

The image characteristic information data of described generation is transferred to receiving terminal by wireless channel from transmitting terminal.

Described receiving terminal includes pose solving unit, and pose solving unit uses the described image feature information number received According to carrying out pose parameter resolving.

The radio channel capacity of data transmission system is relatively low, it is impossible to transmit high-resolution and high frame rate image in real time, but The transmission code rate of image characteristic information data can meet the capacity limit of wireless channel.

The data stream of transmission is used image characteristic information data to replace entire image by the present invention, thus it is relatively low to meet capacity Wireless channel requirement；And the image measurement of pose is divided into feature extraction and pose parameter resolve two steps, feature extraction Complete under conditions of high-resolution and high frame rate image, it is ensured that the precision of image characteristic information data, thus ensure that position The certainty of measurement of appearance parameter.

After by way of example embodiments of the present invention being described in detail below in conjunction with accompanying drawing, the present invention's Other characteristics, features and advantages will be the most obvious.

Accompanying drawing explanation

Fig. 1 is the system of a kind of based on wireless channel data transmission system the object pose image measuring method of the present invention Composition frame chart.

Fig. 2 is the measurement of a kind of based on wireless channel data transmission system the object pose image measuring method of the present invention Flow chart.

Fig. 3 is the flight of a kind of based on wireless channel data transmission system the object pose image measuring method of the present invention Target appearance figure.

Fig. 4 is the flight of a kind of based on wireless channel data transmission system the object pose image measuring method of the present invention Clarification of objective point schematic diagram.

Fig. 5 is the signal of a kind of based on wireless channel data transmission system the object pose image measuring method of the present invention Figure.

Fig. 6 is the center of a kind of based on wireless channel data transmission system the object pose image measuring method of the present invention Projection model figure.

Fig. 7 is the image of a kind of based on wireless channel data transmission system the object pose image measuring method of the present invention Characteristic information data extracts flow chart.

Fig. 8 be the present invention a kind of based on wireless channel data transmission system object pose image measuring method in fly Target image gradient direction schematic diagram.

Fig. 9 is the pixel of a kind of based on wireless channel data transmission system the object pose image measuring method of the present invention Coordinate system and image coordinate system schematic diagram.

Detailed description of the invention

The present invention is elaborated by below in conjunction with the accompanying drawings with one typical detailed description of the invention.

Seeing shown in accompanying drawing, Fig. 1 is that a kind of object pose image based on wireless channel data transmission system of the present invention is surveyed The composition frame chart of the system of metering method.The system of the present invention is made up of transmitting terminal and receiving terminal two parts.Wherein, transmitting terminal is by taking the photograph Camera, graphics processing unit, modulation and amplifirer, emitter composition；Receiving terminal is by receiver, detection and demodulator, pose solution Calculation unit forms.Wherein, the system transmitting terminal of the present invention completes IMAQ, image characteristics extraction, signal modulation and amplifies, believes The functions such as number transmission；Receiving terminal completes the functions such as signal receives, signal demodulates, pose resolves, parameter shows.

Shown in Figure 2, the measurement procedure of the present invention is as follows: use high-resolution camera photographic subjects image, shooting The image resolution ratio of machine is higher than 512 × 512, and frame per second is higher than 50bps；Captured image passes to image processor, image procossing Device carries out feature extraction to image, and by image characteristic information data, is modulated and is sent by emitter after power amplifier.Can be through The wireless channel of the low bandwidth crossing channel capacity not higher than 2Mbps sends receiving terminal to.It is special that the receiver of receiving terminal receives image After levying information data, after testing and demodulator is passed to pose solving unit and carried out pose parameter resolving.

The core procedure of the present invention is, is divided into image characteristics extraction and pose parameter to ask the position-pose measurement of target Solving two parts, transmitting terminal and receiving terminal in system complete respectively, so that system completes the pose parameter to target and measures.

Below as a example by the conical target that target to be measured is with indicia patterns, the object pose illustrating the present invention is surveyed The method of amount.

Fig. 3 is the outside drawing of airbound target to be measured, and Fig. 4 is the characteristic point schematic diagram of airbound target, with the class ladder shown in Fig. 4 4 summits of shape are as clarification of objective point.

(1) pose parameter defines with coordinate system

Fig. 5 is the pose measurement system sketch of the present invention, and Fig. 6 is the central projection illustraton of model of the measuring method of the present invention. As shown in Fig. 5, Fig. 6 and Fig. 9, to defining 5 coordinate systems during the pose measurement of airbound target to be measured altogether, wherein, image is sat Mark system o-xy is parallel to plane X_cO_cY_c, and with plane X_cO_cY_cThe focal distance f that distance is video camera, image coordinate system initial point o exists The Z of camera coordinate system_cOn axle.5 coordinate systems are respectively as follows:

(1) target-based coordinate system O-XYZ

(2) coordinate system O is measured_w-X_wY_wZ_w

(3) camera coordinate system O_c-X_cY_cZ_c

(4) image coordinate system o-xy

(5) pixel coordinate system o'-uv

Meanwhile, if characteristic point P_iCoordinate in above-mentioned 5 coordinate systems is respectively as follows: W_i=(X_i,Y_i,Z_i)^T,W=(x_i,y_i)^T, w'=(u_i,v_i)^T.(this patent acquiescence i=1, 2,3,4)

W_iWith W_i ^wBetween there is following relation:

W_i ^w=RW_i+T (1)

T is translation vector, is a three-dimensional vector T=[T_X,T_Y,T_Z]^T.Represent the relative position between Two coordinate system, i.e. The initial point of target-based coordinate system is measuring the coordinate of coordinate system.

R is spin matrix, is the trigonometric function combination of three angles (α, beta, gamma).Around X-axis rotation angle α, around the Y-axis anglec of rotation β, about the z axis anglec of rotation γ.Target-based coordinate system rotates rotating around three reference axis in order, just reach three reference axis respectively with survey Measure the attitude that three change in coordinate axis direction corresponding to coordinate system is consistent.R is used for describing target-based coordinate system relative to measuring coordinate system Attitude.R with the relation of (α, β, γ) is:

$\begin{array}{c}R=R_Z{R}_Y{R}_X\\ =\left[\begin{array}{ccc}\cos \mathrm{\γ}& \sin \mathrm{\γ}& 0\\ -\sin \mathrm{\γ}& \cos \mathrm{\γ}& 0\\ 0& 0& 1\end{array}\right]\left[\begin{array}{ccc}\cos \mathrm{\β}& 0& -\sin \mathrm{\β}\\ 0& 1& 0\\ \sin \mathrm{\β}& 0& \cos \mathrm{\β}\end{array}\right]\left[\begin{array}{ccc}1& 0& 0\\ 0& \cos \mathrm{\α}& \sin \mathrm{\α}\\ 0& -\sin \mathrm{\α}& \cos \mathrm{\α}\end{array}\right]\\ =\left[\begin{array}{ccc}\cos \mathrm{\γ}\cos \mathrm{\β}& \sin \mathrm{\γ}\cos \mathrm{\α}+\cos \mathrm{\γ}\sin \mathrm{\β}\sin \mathrm{\α}& \sin \mathrm{\γ}\sin \mathrm{\α}-\cos \mathrm{\γ}\sin \mathrm{\β}\cos \mathrm{\α}\\ -\sin \mathrm{\γ}\cos \mathrm{\β}& \cos \mathrm{\γ}\cos \mathrm{\α}-\sin \mathrm{\γ}\sin \mathrm{\β}\sin \mathrm{\α}& \cos \mathrm{\γ}\sin \mathrm{\α}+\sin \mathrm{\γ}\sin \mathrm{\β}\cos \mathrm{\α}\\ \cos \mathrm{\β}& -\cos \mathrm{\β}\sin \mathrm{\α}& \cos \mathrm{\β}\cos \mathrm{\α}\end{array}\right]\end{array}---\left(2\right)$

Pose measurement seeks to solve translation vector T and attitude angle (α, β, γ).

(2) feature point extraction

Feature point extraction seeks to the seat extracting target signature in the pixel coordinate system of the image of shot by camera Mark.The outward appearance conceptual design to be marked of target to be measured produces prior image characteristic information data, for different targets Signature design is different.Prior image characteristic information data can be characteristic point or characteristic curve, and its quantity determines The wireless signal-path band width that image characteristic information data is to be taken.Image solves object pose at least needs the elder generation of 3 characteristic points Test information.The bandwidth that the upper limit of feature point number distributes to measurement system to channel is relevant, but the feature point number used is more Many computation complexities are the biggest, and the load causing graphics processing unit is excessive, and Optimal units is 3 or 4.

As a example by the target shown in Fig. 3, the process of feature point extraction is described.Target signature is the 4 of each class trapezoid area Individual summit P1, P2, P3, P4, according to the high-precision method for extracting for this feature point of the present invention, extract flow process such as Fig. 7, specifically Step is as follows:

(1) the detection trapezoidal whole pixel edge of class, the detailed description of the invention of the present invention uses Canny operator detection class trapezoidal side Edge, step is:

The first step: use Gaussian filter smoothed image.Gaussian smoothing function is:

$\begin{array}{c}H(x,y)=e^{-\frac{{\mathrm{\α}}^2+b^2}{2{\mathrm{\σ}}^2}}\\ G(x,y)=f(x,y)*H(x,y)\end{array}---\left(3\right)$

Second step: calculate amplitude and the direction of gradient by the finite difference of single order local derviation.First-order difference convolution mask is:

$\begin{array}{cc}{H}_1=\left|\begin{array}{cc}-1& -1\\ 1& 1\end{array}\right|& H_2=\left|\begin{array}{cc}1& -1\\ 1& -1\end{array}\right|\end{array}---\left(4\right)$

3rd step: gradient magnitude is carried out non-maxima suppression.The gradient only obtaining the overall situation is not sufficient to determine edge, Therefore for determining edge, it is necessary to retaining the point that partial gradient is maximum, and suppress non-maximum, the present invention utilizes the direction of gradient to ask Gradient.

ξ [i, j]=Sector (θ [i, j]) (6)

Airbound target image gradient direction to be measured schematic diagram as shown in Figure 8, four sectors be numbered 0 to 3, corresponding 3 Four kinds of × 3 neighborhoods may combination.To a point, the center pixel M of neighborhood with compared with two pixels of gradient line.As Really the Grad of M is big unlike two neighbor Grad along gradient line, then make M=0.I.e.

N [i, j]=NMS (M [i, j], ξ [i, j]) (7)

4th step: with the detection of dual threshold algorithm and connection edge.Reduce the typical method of false edge section quantity be to N [i, J] use a threshold value, all values that will be less than threshold value composes null value.The present invention uses dual threshold algorithm.Choose dual threshold: τ₁With τ₂, and τ₁≈2τ₂, obtain two threshold skirt image T₁[i, j] and T₂[i, j], at T₁In collect edge, by T₂In all gaps Couple together.

Canny algorithm is utilized to obtain the whole pixel edge point P that class is trapezoidal_i(m,n)。

(2) separate waist, the upper end in the trapezoidal edge of class, go to the bottom.Owing to black class is trapezoidal and white group is trapezoidal adjacent, class ladder Shape waist both sides color have black and white two kinds；And the color on both sides, the class trapezoidal upper and lower end, it is grey (background) and another While being white or black (signature).According to this feature, trapezoidal for class waist is separated from marginal point.Class is trapezoidal Waist is straight line, with least square fitting out.In the trapezoidal marginal point of class, remove the marginal point on waist, obtain class trapezoidal upper, Go to the bottom marginal point.The class trapezoidal upper and lower end is circular arc, projects to become elliptic arc (or circular arc) in image, intends respectively with these points Close the elliptic curve at place, two upper and lower ends.

(3) the trapezoidal sub-pixel edge of class is extracted.The trapezoidal whole pixel edge point P of class is being sought based on Canny operator_i(m, n) it After, by known whole pixel edge point P_i(m, the gradient side of the sub-pixel edge point of gradient direction its unknown of approximate substitution n) To, and at the enterprising row interpolation of gradient direction of whole pixel edge point, it is thus achieved that (x, y), at image border for difference functions φ Gray scale derivative value is maximum, therefore, φ (x, the coordinate of maximum of points y) is the coordinate of sub-pixel edge point, then, then pass through (x, maximum y) just can obtain the coordinate P of sub-pixel edge point to seek difference functions φ_i′(m′,n′)。

Let R be the marginal point P that class is trapezoidal_i(m, gradient magnitude n), R₀For marginal point P_i(m, the mould of shade of gray n), R_-1、R₁Be respectively on gradient direction with P_iThe two pixel P that point is adjacent_i-1、P_i+1Gradient magnitude, R₀、R_-1、R₁By eight templates Sobel operator obtains, then marginal point P_i(m, subpixel coordinates P n)_i' (m ', n ') is:

$\begin{array}{c}{m}^{\′}=m+\frac{R_{-1}-R_1}{R_{-1}-2R_0+R_1}.\frac{W}{2}\cos \left(\mathrm{\θ}\right)\\ n^{\′}=n+\frac{R_{-1}-R_1}{R_{-1}-2R_0+R_1}.\frac{W}{2}\sin \left(\mathrm{\θ}\right)\end{array}---\left(8\right)$

In formula, W is the neighbor pixel distance to marginal point, W=1 orθ is the folder of gradient direction and the image longitudinal axis Angle.

According to said method, obtain waist, the upper end, the subpixel coordinates P of each marginal point of going to the bottom_i′(m′,n′)。

(4) characteristic point subpixel coordinates is solved

Owing to the waist that class is trapezoidal is the straightway along cone bus, therefore, the waist that class is trapezoidal can use least square method Fit to straight line model y=kx+b.If P_L1(m_L1,n_L1) and P_L2(m_L2,n_L2) it is that the whole pixel on trapezoidal two waists of class is sat respectively Mark, P '_L1(m′_L1,n′_L1) and P '_L2(m′_L2,n′_L2) be the sub-pixel edge point coordinates of respective point, then the waist L that class is trapezoidal₁、L₂For:

$\begin{array}{c}{L}_1:n_{L1,i}^{\′}=k_1{m}_{L1,i}^{\′}+b_1\\ L_2:n_{L2,j}^{\′}=k_2{m}_{L2,j}^{\′}+b_2\end{array}---\left(9\right)$

In formula, k₁、b₁、k₂、b₂It is respectively as follows:

$\begin{array}{c}{b}_1=\stackrel{\‾}{n_{L1}^{\′}}-k_1\stackrel{\‾}{m_{L1}^{\′}}\\ k_1=\frac{\Σ(\stackrel{\‾}{m_{L1}^{\′}}-m_{L1}^{\′})(\stackrel{\‾}{n_{L1}^{\′}}-n_{L1}^{\′})}{\Σ{(\stackrel{\‾}{m_{L1}^{\′}}-m_{L1}^{\′})}^2}\\ b_2=\stackrel{\‾}{n_{L2}^{\′}}-k_2\stackrel{\‾}{m_{L2}^{\′}}\\ k_2=\frac{\Σ(\stackrel{\‾}{m_{L2}^{\′}}-m_{L2}^{\′})(\stackrel{\‾}{n_{L2}^{\′}}-n_{L2}^{\′})}{\Σ{(\stackrel{\‾}{m_{L2}^{\′}}-m_{L2}^{\′})}^2}\end{array}---\left(10\right)$

In formula,Represent pixel average.

Because the class trapezoidal upper end and go to the bottom as circular arc, project to image becomes elliptic arc, therefore, the class trapezoidal upper end and under The end, fits to quadratic curve equation model ax²+bxy+cy²+ dx+ey+f=0.Make P_C1(m_C1,n_C1)、P_C2(m_C2,n_C2) it is that class is trapezoidal The upper end and the whole pixel coordinate gone to the bottom, P '_C1(m′_C1,n′_C1)、P′_C2(m′_C2,n′_C2) it is the subpixel coordinates of respective point, class is trapezoidal The upper end and go to the bottom respectively L₃、L₄。

Due to cone distance video camera farther out time, the upper end and the marginal point negligible amounts on going to the bottom, the secondary simulated J curve effectJ is undesirable, bigger with real curve error.For this problem, the present invention utilizes the end in Lagrange differential technique matching Conic section with place of going to the bottom.

Seek the trapezoidal L of class₁,L₃Intersecting point coordinate P_L1L3(m, n), method as follows:

A () finds out L₃On away from L₁Two sub-pixel edge point P for d ≈ 1, d ≈ 3₁′(m′₁,n′₁)、P₂′(m′₂,n′₂)。

B () extracts L₁Away from L at opposite side class the trapezoidal upper end₁Sub-pixel edge point P for d ≈ 1₃′(m′₃,n′₃)。

C () is based on formula, P₁′(m′₁,n′₁)、P₂′(m′₂,n′₂)、P₃′(m′₃,n′₃) carry out Lagrange difference

$\begin{array}{c}{c}_1\left(x\right)=\frac{(x-m_2^{\′})(x-m_3^{\′})}{(m_1^{\′}-m_2^{\′})(m_1^{\′}-m_3^{\′})}\\ c_2\left(x\right)=\frac{(x-m_1^{\′})(x-m_3^{\′})}{(m_2^{\′}-m_1^{\′})(m_2^{\′}-m_3^{\′})}\\ c_3\left(x\right)=\frac{(x-m_1^{\′})(x-m_2^{\′})}{(m_3^{\′}-m_1^{\′})(m_3^{\′}-m_2^{\′})}\end{array}---\left(11\right)$

(4) based on formula, L is obtained₁,L₃Intersecting point coordinate

$\begin{array}{c}y=n_1{c}_1\left(x\right)+n_2{c}_2\left(x\right)+n_3{c}_3\left(x\right)\\ n_{L1,i}^{\′}=k_1{m}_{L1,i}^{\′}+b_1\end{array}---\left(12\right)$

It is similar to, the trapezoidal waist of class and the upper end and other 3 the intersecting point coordinate P gone to the bottom can be obtained_L2L3(m,n),P_L1L4(m, N), P_L2L4(m,n)。

According to the detailed description of the invention of the present invention, features described above point extracts, sets up image characteristic point and target priori features Matching relationship, generate image characteristic information data whole links completed by the graphics processing unit with dsp chip as core.

(3) transmission code rate calculates

After generating image characteristic information data, image characteristic information data be passed to for equipped with position by wireless channel The pose solving unit of appearance Survey Software.

The airbound target to be measured image in the image captured by high-resolution camera is obtained special through feature point extraction Levy information data, the present invention uses image characteristic information data replace the image transferring content as wireless channel itself, from And meet the capacity limit of wireless channel.As a example by target shown in Fig. 3, its characteristic point is 4, calculates transmission code rate.

The image characteristic information data of every two field picture includes coordinate and the numbering of 4 characteristic points, and each characteristic point coordinate accounts for With the memory space of 8Bytes, characteristic point numbering takies the memory space of 2Bytes, the therefore image feature information of single-frame images Data Data amount is 40Bytes.

Calculate and use 4 characteristic points, the asynchronous code check of image frame per second.As shown in table 1:

Table 1 code check calculates (1)

Image frame per second	Bit rate output
		50bps	15.63Kbps
100bps	31.25Kbps
		200bps	62.50Kbps

Result of calculation as shown in table 1, when using higher frame per second 200bps, it is special that wireless channel is used for transmitting image graph picture The a width of 62.5Kbps of band of reference breath Data Data, far below the channel capacity of typical wireless channels 2Mbps, can be wireless communication Massive band width is saved in road.

Calculating image frame per second is 100bps, the asynchronous code check of feature point number of employing.As shown in table 2:

Table 2 code check calculates (2)

Feature point number	Bit rate output
		4	31.25Kbps
6	46.88Kbps
		8	62.50Kbps

Result of calculation as shown in table 2, when using 8 characteristic points, wireless channel is used for transmitting image characteristic information data The a width of 62.5Kbps of band, far below the channel capacity of typical wireless channels 2Mbps.

(4) pose parameter resolves

Pose parameter resolves and is completed by graphics processing unit, can be completed by known position-pose measurement.Can in practice Pose parameter by the computer-solution target equipped with pose measurement software.In the method for the present invention, target to be measured is cooperation Target, therefore has prior image characteristic information data.In preceding step, characteristic point seat in target-based coordinate system O-XYZ Mark W_i=(X_i,Y_i,Z_i) it is prior information, for known；Characteristic point coordinate w'=(u in pixel coordinate system_i,v_i)^TBy upper The feature point extracting step stated obtains.Pose parameter resolves the high-resolution shooting utilizing above known quantity to combine the present invention exactly The parametric solution of machine goes out the pose parameter of target.

Fig. 5 is the pose measurement system sketch of the present invention, and the high-resolution camera coordinate system of the present invention is O-X_cY_cZ_c, Video camera principal point is coordinate origin, and video camera transverse and longitudinal direction is respectively X_cAxle and Y_cAxle, camera optical axis is Z_cAxle.Measure and sit Mark system is O_w-X_wY_wZ_w。

(1) characteristic point image coordinate system coordinate is solved

If the distance in x-axis and y-axis direction, i.e. pixel actual physical size between each pixel in pixel coordinate system, Being dx and dy respectively, the pixel coordinate of picture centre is (u₀,v₀), such as Fig. 9, by pixel coordinate system and the conversion of image coordinate system Relation can be by arbitrfary point in image at pixel coordinate system coordinate w'=(u_i,v_i)^TSolve image coordinate system coordinate w=(x_i,y_i )^T, i.e.

$\left\{\begin{array}{c}{x}_i=(u_i-u_0)dx\\ y_i=(v_i-v_0)dy\end{array}\right.---\left(13\right)$

(2) characteristic point camera coordinate system coordinate is solved

F represents the focal length of video camera, then 4 characteristic point image coordinate system coordinate (x_i,y_i)^TWith camera coordinate system coordinateThere is following relation:

$\left\{\begin{array}{c}\frac{x_i}{f}=\frac{X_i^c}{Z_i^c}\\ \frac{y_i}{f}=\frac{Y_i^c}{Z_i^c}\end{array}\right.---\left(14\right)$

Three dimensions distance between in these 4 points any 2 is known (prior information) simultaneously, it may be assumed that

$d(P_i,P_j)=\sqrt{{(X_i^c-X_j^c)}^2+{(Y_i^c-Y_j^c)}^2+{(Z_i^c-Z_j^c)}^2}=d_{ij}---\left(15\right)$

Can be obtained 8 equatioies by formula (14), formula (15) can obtain 6 equatioies, this can obtain 1 over-determined systems, by minimum Square law can solve W_i ^c=(X_i ^c, Y_i ^c, Z_i ^c)^T。

(3) characteristic point world coordinate system coordinate is solved

If the Relative Transformation relation between camera coordinate system and world coordinate system is by spin matrix Rc and translation vector Tc Represent.Rc Yu Tc video camera installation process before measuring is known (camera calibration).That is:

W_i ^w=RcW_i ^c+Tc (16)

(4) pose parameter of target is solved

The pose parameter of target is represented by spin matrix R and translation vector T, world coordinate system coordinate and object coordinates system There is following relation in coordinate:

W_i ^w=RW_i+T (17)

If the coordinate that the barycenter of 4 characteristic points is in world coordinate system and target-based coordinate system is respectively as follows:

$\stackrel{\‾}{P}=\frac{1}{4}\underset{i=1}{\overset{4}{\Σ}}{W_i}^w,\stackrel{\‾}{Q}=\frac{1}{4}\underset{i=1}{\overset{4}{\Σ}}{W}_i$

Then 4 characteristic points new coordinate under the coordinate system with barycenter as initial point is:

${W_i^w}^{\′}=W_i^w-\stackrel{\‾}{P},{W_i}^{\′}=W_i-\stackrel{\‾}{Q}$

Wherein:

${W_i^w}^{\′}={({x_i^w}^{\′},{y_i^w}^{\′},{x_i^w}^{\′})}^T,{W_i}^{\′}={({x_i}^{\′},{y_i}^{\′},{x_i}^{\′})}^T$

Thus obtain:

$T=\stackrel{\‾}{P}+R\stackrel{\‾}{Q}---\left(18\right)$

If:

$\left\{\begin{array}{c}{S}_{xx}=\underset{i=1}{\overset{4}{\Σ}}{x_i^w}^{\′}{x_i}^{\′}\\ S_{xy}=\underset{i=1}{\overset{4}{\Σ}}{x_i^w}^{\′}{y_i}^{\′}\\ S_{xz}=\underset{i=1}{\overset{4}{\Σ}}{x_i^w}^{\′}{z_i}^{\′}\\ .\\ .\\ .\end{array}\right.$

$N=\left[\begin{array}{cccc}{S}_{xx}+S_{yy}+S_{zz}& S_{yz}-S_{yz}& S_{zx}-S_{xz}& S_{xy}-S_{yx}\\ S_{yz}-S_{yz}& S_{xx}-S_{yy}-S_{zz}& S_{xy}+S_{yx}& S_{zx}+S_{xz}\\ S_{zx}-S_{xz}& S_{xy}+S_{yx}& -S_{xx}+S_{yy}-S_{zz}& S_{yz}+S_{zy}\\ S_{xy}-S_{yx}& S_{zx}+S_{xz}& S_{yz}+S_{zy}& -S_{xx}-S_{yy}+S_{zz}\end{array}\right]$

Quaternary number corresponding to spin matrix RIt it is the characteristic vector corresponding to eigenvalue of maximum of N.OrderThen its spin matrix represented is:

$R=\left[\begin{array}{ccc}{r_0}^2+{r_1}^2-{r_2}^2-{r_3}^2& 2(r_1{r}_2-r_0{r}_3)& 2(r_1{r}_3+r_0{r}_2)\\ 2(r_1{r}_2+r_0{r}_3)& {r_0}^2-{r_1}^2+{r_2}^2-{r_3}^2& 2(r_2{r}_3-r_0{r}_1)\\ 2(r_1{r}_3-r_0{r}_2)& 2(r_2{r}_3+r_0{r}_1)& {r_0}^2-{r_1}^2-{r_2}^2+{r_3}^2\end{array}\right]$

Thus try to achieve R, and solved (α, beta, gamma) by formula (2), formula (18) solve T=[T_X,T_Y,T_Z]^T, obtain target Pose parameter.

It should be appreciated that above description is one particular embodiment of the present invention, the present invention be not limited only to Upper diagram or the specific structure of description, all changes side that claim will cover in the connotation of the present invention and scope Case.

Claims (3)

1. an object pose image measuring method based on wireless channel, it is characterised in that

Transmitting terminal and receiving terminal are set；

Described transmitting terminal includes video camera and graphics processing unit；

The resolution ratio of the described video camera shooting cooperation airbound target image higher than 512 × 512；

Described graphics processing unit extracts at least 3 characteristic points of described video camera shooting high-definition picture, determines characteristic point Coordinate in the picture；And set up the matching relationship of the priori features of described coordinate and described cooperation airbound target, generate image Characteristic information data；

The image characteristic information data of every two field picture includes coordinate and the numbering of characteristic point；

The image characteristic information data of described generation is transferred to receiving terminal by wireless channel from transmitting terminal；

Described receiving terminal includes pose solving unit；

Pose solving unit uses the described image characteristic information data received to carry out pose parameter resolving.

Object pose image measuring method based on wireless channel the most according to claim 1, it is characterised in that video camera The image resolution ratio of shooting is higher than 512 × 512, and frame per second is higher than 50bps.

Object pose image measuring method based on wireless channel the most according to claim 1 and 2, it is characterised in that institute State feature point extraction, set up the image characteristic point matching relationship with cooperative target priori features, generate image characteristic information data Whole links completed by the graphics processing unit with dsp chip as core.