CN104296695B - A kind of method for obtaining video camera spatial attitude - Google Patents

Abstract

The invention discloses a method for obtaining the spatial attitude of a camera. Firstly, a coordinate system is established, and the camera target surface and the ground are represented by a geometric relationship. Using the internal parameters of the camera and the corresponding function between the picture pixels and the ground position, according to the plane geometry and the two The meta-equations determine the installation pose data of the camera. The invention utilizes the conversion relationship between the picture pixel position and the ground coordinate position to directly calculate the installation attitude of the camera without going to the site for measurement and calibration, which greatly reduces the workload of the project; it is easy to operate and easy to grasp, avoiding the inability to The problem with sampling.

Description

A Method of Obtaining Camera Space Attitude

technical field

The invention belongs to the technical field of measurement, and in particular relates to a method for acquiring the spatial attitude of a camera.

Background technique

In the application of computer vision technology, it is necessary to obtain the installation attitude data of the camera, also known as the external parameters of the camera. The installation attitude data includes: height, X coordinate of the vertical projection position on the ground, Y coordinate of the vertical projection position on the ground, direction angle, tilt angle, vertical angle etc.

It is difficult to guarantee the measurement accuracy by using physical methods for manual measurement. The previous camera space attitude algorithms need to use complex calculations of three-dimensional analytic geometry and ternary quadratic, especially when performing actual calibration and collecting calibration data. There are many strict requirements, which are difficult to achieve for many occasions. Some methods need to measure points in some special positions, such as the four corner points of the camera screen, and often because of the relationship between the shooting content, the shooting position of the corner points is not on the ground, such as on buildings, trees, etc., or in dangerous areas such as water Can not be measured.

Contents of the invention

The purpose of the present invention is to provide a simple, fast and easy-to-grasp method for acquiring the spatial attitude of a camera.

The technical scheme that realizes the object of the present invention is: a kind of method for obtaining camera space attitude, comprises the following steps:

Step 1. Build a coordinate system: the coordinates of the image screen are in pixels, the upper left corner of the image screen is the origin of the coordinates, the horizontal direction of the screen is the x-axis, and the vertical direction of the screen is the y-axis; specifically:

Use [x,y] to represent the screen coordinate point, and [X,Y] to represent the ground coordinate point; it is determined by the corresponding conversion function F([X,Y]) between the screen coordinate [x,y] and the ground coordinate [X,Y] Ground position coordinates [X, Y]; the projective point of the screen center point o of the optical axis of the camera lens on the ground is recorded as the ground center point O, and the coordinates of the screen center point o are [ox, oy]. The coordinates are [OX,OY];

Step 2. Determine the intersection line between the camera target surface and the ground, which is recorded as the ground horizon Lh;

Step 3. Draw a vertical line from the ground center point O to the ground horizon Lh, which is recorded as the ground vertical center line Lv, and the ground position of the camera is on the ground vertical center line;

Step 4. Take the intersection of the ground vertical centerline Lv and the ground horizon Lh, and record it as the ground perspective point F; the ground coordinates of F are [FX, FY];

Determine the distance from the ground center point O to the ground perspective point F, which is recorded as the perspective distance D:

Step 5, take the midpoint of the line connecting the ground center point O and the ground perspective point F, record it as the ground middle point P, and the ground coordinates of P are [PX, PY]; determine the projective point of the ground middle point P on the image screen, It is recorded as the middle point p of the screen, and the screen coordinates of p are [px,py];

Draw a ground line LP parallel to the ground horizon from the middle point of the ground;

Randomly select a ground reference point T on LP, and the ground coordinates of T are [TX,TY];

Determine the projective point of the ground reference point T on the screen, which is recorded as the screen reference point t, and the image frame coordinates of t are [tx, ty];

Determine the distance D1 between the middle point P on the ground and the reference point T on the ground:

Determine the distance d1 between the middle point p of the picture and the reference point t of the picture:

Determine the distance d2 between the middle point p of the picture and the center point o of the picture:

Determine the distance D2 from the middle point P on the ground to the straight line Oo:

D2=d2*D1/d1;

Step 6, take the line connecting the center point o of the screen and the middle point p of the screen, and record it as the vertical center line lv of the screen;

Step 7. Determine the space attitude parameters: the space attitude parameters include camera direction angle α, camera rotation angle β, camera vertical angle γ, camera height CH and camera position [CX, CY];

The camera direction angle α is the angle between the vertical centerline Lv of the ground and the Y-axis of the ground;

Camera rotation angle β is the angle between the vertical center line lv of the screen and the y-axis of the screen;

Camera vertical angle γ, that is, the angle between the camera shooting axis and the ground, that is, the supplementary angle between the camera target surface and the ground:

γ=arcsin(D2*2/D);

0<γ<π/2;

The camera height CH is the vertical height of the camera relative to the ground, namely

CH=D2*2*cos(γ);

The camera position [CX, CY] is the position of the vertical point of the camera on the ground; determine the distance D3 from the camera position [CX, CY] to the perspective point F on the ground:

D3＝D2*2*D2*2/D;

CX＝FX+D3/D*(OX‐FX)

CY=FY+D3/D*(OY-FY).

Compared with the prior art, the remarkable advantages of the present invention are: (1) The present invention utilizes the conversion relationship between the picture pixel position and the ground coordinate position to directly calculate the installation attitude of the camera, without going to the site for measurement and calibration, which greatly reduces the engineering workload; (2) the present invention is easy to operate, easy to master, and avoids the problem that sampling cannot be performed due to scene restrictions.

The present invention will be described in further detail below in conjunction with the accompanying drawings.

Description of drawings

FIG. 1 is a schematic geometric diagram of the method for obtaining the spatial attitude of a camera in the present invention.

FIG. 2 is a schematic diagram of the spatial relationship of the method for obtaining the spatial attitude of the camera according to the present invention.

detailed description

The internal parameters of the camera and the corresponding function between the picture pixels and the ground position are known;

Combining Fig. 1 and Fig. 2, a method for obtaining the spatial attitude of a camera includes the following steps:

Step 1. Build a coordinate system: the coordinates of the image screen are in pixels, the upper left corner of the image screen is the origin of the coordinates, the horizontal direction of the screen is the x-axis, and the vertical direction of the screen is the y-axis; specifically:

Use [x,y] to represent the screen coordinate point, and [X,Y] to represent the ground coordinate point; it is determined by the corresponding conversion function F([X,Y]) between the screen coordinate [x,y] and the ground coordinate [X,Y] Ground position coordinates [X, Y]; the projective point of the screen center point o of the camera lens optical axis on the ground is recorded as the ground center point O, and the coordinates of the screen center point o are [ox, oy], and the corresponding ground coordinates of O is [OX,OY];

The corresponding conversion function F([x,y]) of the picture pixel coordinates [x, y] and the ground position coordinates [X, Y] (see glFrustum(), gluPerspective() and glOrtho() of "OpenGL") , and the inverse transformation function F ^T ([X,Y]) is:

[X,Y]=F([x,y])

[x,y]=F ^T ([X,Y])

Then [OX,OY]=F([ox,oy]);

Step 2. Determine the intersection line between the camera target surface and the ground, which is recorded as the ground horizon Lh; specifically includes the following steps:

Step 2-1. Take four vertices v1, v2, v3 and v4 of any rectangle on the screen;

Step 2-2. Use the corresponding conversion function F([x,y] of the screen pixel coordinates [x,y] and the ground position coordinates [X,Y] to determine the projective points V1 of v1, v2, v3, and v4 on the ground , V2, V3, V4; V1, V2, V3, V4 form a quadrilateral Q on the ground;

Step 2-3, determine respectively by 2 groups of opposite sides L1, L2 and L3, L4 of quadrilateral Q as two intersections F1 and F2 of straight line, the connecting line of F1 and F2 is ground horizon Lh;

The linear equation of Lh is: Y=Lh_a*X+Lh_b, where Lh_a is the coefficient term of X, and Lh_b is the constant term;

Step 3, draw a vertical line from the ground center point O to the ground horizon Lh, which is recorded as the ground vertical center line Lv, and the ground position of the camera is on the ground vertical center line; the equation expression of the ground vertical center line Lv is:

Y=Lv_a*X+Lv_b, wherein Lv_a is a coefficient term of X, and Lv_b is a constant term;

Step 4, take the intersection of the vertical centerline Lv of the ground and the horizon Lh of the ground, and record it as the ground perspective point F; the coordinates of F are [FX, FY];

Determine the distance from the ground center point O to the ground perspective point F, which is recorded as the perspective distance D:

Step 5, take the midpoint of the line connecting the ground center point O and the ground perspective point F, record it as the ground middle point P, and the ground coordinates of P are [PX, PY]; determine the projective point of the ground middle point P on the image screen, It is recorded as the middle point p of the screen, and the screen coordinates of p are [px,py];

Draw a ground line LP parallel to the ground horizon from the middle point of the ground;

Randomly select a ground reference point T on LP, and the ground coordinates of T are [TX,TY];

Determine the projective point of the ground reference point T on the screen, which is recorded as the screen reference point t, and the image frame coordinates of t are [tx, ty];

Determine the distance D1 between the middle point P on the ground and the reference point T on the ground:

Determine the distance d1 between the middle point p of the picture and the reference point t of the picture:

Determine the distance d2 between the middle point p of the picture and the center point o of the picture:

Determine the distance D2 from the middle point P on the ground to the straight line Oo:

D2=d2*D1/d1;

Step 6, get the line connecting the picture center point o and the picture middle point p, and record it as the picture vertical center line lv; the equation expression of the picture vertical center line lv is: y=lv_a*x+lv_b, wherein lv_a is x The coefficient term of , lv_b is a constant term;

Step 7. Determine the space attitude parameters: the space attitude parameters include camera direction angle α, camera rotation angle β, camera vertical angle γ, camera height CH and camera position [CX, CY];

The camera direction angle α is the angle between the vertical centerline Lv of the ground and the Y-axis of the ground:

If OX-PX>0, then α=arcctg(Lv_a);

If OX-PX≤0, then α=arcctg(Lv_a)+π;

The camera rotation angle β is the angle between the vertical center line lv of the screen and the y-axis of the screen:

If ox-px>0, then β=arcctg(lv_a);

If ox-px≤0, then β=arcctg(lv_a)+π;

Camera vertical angle γ, that is, the angle between the camera shooting axis and the ground, that is, the supplementary angle between the camera target surface and the ground:

γ=arcsin(D2*2/D);

0<γ<π/2;

The camera height CH is the vertical height of the camera relative to the ground, namely

CH=D2*2*cos(γ);

The camera position [CX, CY] is the vertical point of the camera on the ground; first determine the distance D3 from the camera position [CX, CY] to the perspective point F on the ground:

D3＝D2*2*D2*2/D;

CX＝FX+D3/D*(OX‐FX)

CY=FY+D3/D*(OY-FY).

The present invention will be further described below in conjunction with specific embodiments.

Example 1

Combined with Figure 1 and Figure 2, take the actual data of a camera as an example.

The picture pixel position of the center point o of the picture optical axis is [367,322];

After the calculation of the plane conversion relationship, the projective point O of the center point o of the picture on the ground is obtained, and the ground coordinates of O are

[1511.72,2930.61];

Take four vertices v1, v2, v3, and v4 of any parallelogram on the screen, and the pixel positions are [267,322], [367,222], [467,322], [367,422];

Determine the projective points V1, V2, V3, and V4 of v1, v2, v3, and v4 on the ground, and the coordinate positions are [1792.93, 2716.66], [1813.26, 3368.80], [1220.26, 3152.34], [1299.68, 2622.49];

V1, V2, V3, and V4 form a quadrilateral Q on the ground, and the straight lines L1, L2, L3, and L4 of the four sides of the quadrilateral Q are calculated as follows:

Y=32.08*X–54803.16;

Y＝-6.67*X+11294.03;

Y=0.37*X+2706.93;

Y=0.19*X+2374.33;

Determine the intersection F1 of L1 and L2 and the intersection F2 of L3 and L4, which are [1705.58,-85.73] and [-1910.55,2009.54] respectively;

The connection line between F1 and F2 is called the ground horizon Lh, and the equation of Lh is: Y=-0.58*X+902.53;

Draw a vertical line from the ground center point O to the ground horizon Lh, which is recorded as the ground vertical center line Lv, and the equation of Lv is: Y=1.73*X+321.60;

The ground position of the camera is on the vertical centerline of the ground;

Determine the intersection of the ground vertical centerline Lv and the ground horizon Lh, and record it as the ground perspective point F, and the coordinates of F are [252.00,756.51];

Determine the distance from the ground center point O to the ground perspective point F, which is recorded as the perspective distance D, D=2512.68;

Determine the midpoint of the line connecting the ground center point O and the ground perspective point F, which is recorded as the ground midpoint P, and the ground coordinates of P are [881.86, 1843.56];

Determine the projection point of the middle point P on the screen, which is recorded as the middle point p of the screen, and the screen coordinates of p are [313.12,890.87];

Draw a ground line LP parallel to the ground horizon from the middle point of the ground, Y=-0.58+2354.53;

Select a ground reference point T on LP, and the ground coordinates of T are [981.86,1785.62];

Determine the projective point of the ground reference point T on the screen, record it as the screen reference point t, and the screen coordinates of t are [248.97,884.29];

Determine the distance D1 between the middle point P on the ground and the reference point T on the ground, D1=115.57;

Determine the distance d1 between the middle point p of the picture and the reference point t of the picture, d1=64.48;

Determine the distance d2 between the middle point p of the picture and the center point o of the picture, d2=570.83;

Determine D2=d2*D1/d1=1023.09;

The line connecting the center point o of the screen and the middle point p of the screen is the vertical center line lv of the screen, and the equation of lv is:

y=-10.48*x+4173.08;

Finally calculated:

Camera direction angle α (unit degree), α=30.09;

Camera rotation angle β (unit degree), β=-5.45;

Camera vertical angle γ (unit degree), γ=35.48;

Camera height CH (in centimeters), CH=1187.57;

Camera position [CX, CY] (unit cm), [CX, CY] = [1087.39, 2198.27]

After verification, the parameters are consistent with the actual, and the accuracy is higher than the accuracy calculated by corner point measurement.

The present invention is not limited by the above-mentioned embodiments, and what described in the above-mentioned embodiments and the description only illustrates the principle of the present invention, and without departing from the spirit and scope of the present invention, the present invention also has various changes and improvements, these changes All modifications and improvements are within the scope of the claimed invention.

Claims (7)

1. A method for obtaining camera space attitude, characterized in that, comprising the following steps: Step 1. Build a coordinate system: the coordinates of the image screen are in pixels, the upper left corner of the image screen is the coordinate origin, the horizontal direction of the image screen is the x-axis, and the vertical direction of the image screen is the y-axis; specifically: Use [x, y] to represent the coordinate points of the image screen, and [X, Y] to represent the ground coordinate points; through the corresponding conversion function F([X, Y] of the image screen coordinates [x, y] and the ground coordinates [X, Y] ) to determine the ground position coordinates [X, Y]; the projective point o of the image screen center point o of the optical axis of the camera lens on the ground is recorded as the ground center point O, and the coordinates of the image screen center point o are [ox, oy], and the ground center point The ground coordinates of point O are [OX,OY]; Step 2. Determine the intersection line between the camera target surface and the ground, which is recorded as the ground horizon Lh; Step 3. Draw a vertical line from the ground center point O to the ground horizon Lh, which is recorded as the ground vertical center line Lv, and the ground position of the camera is on the ground vertical center line; Step 4. Take the intersection of the ground vertical centerline Lv and the ground horizon Lh, and record it as the ground perspective point F; the ground coordinates of F are [FX, FY]; Determine the distance from the ground center point O to the ground perspective point F, which is recorded as the perspective distance D: $\mathrm{<span\; class="notranslate">\; D.</span>}<span\; class="notranslate">\; =</span>\sqrt{{<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; o</span>}\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; f</span>}\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; )</span>}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; +</span>{<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; o</span>}\mathrm{<span\; class="notranslate">\; Y</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; f</span>}\mathrm{<span\; class="notranslate">\; Y</span>}<span\; class="notranslate">\; )</span>}^{\mathrm{<span\; class="notranslate">\; 2</span>}}}$ Step 5, take the midpoint of the line connecting the ground center point O and the ground perspective point F, record it as the ground middle point P, and the ground coordinates of P are [PX, PY]; determine the projective point of the ground middle point P on the image screen, It is recorded as the middle point p of the image screen, and the image screen coordinates of p are [px, py]; Draw a ground line LP parallel to the ground horizon from the middle point of the ground; Randomly select a ground reference point T on LP, and the ground coordinates of T are [TX,TY]; Determine the projective point of the ground reference point T on the image frame, which is recorded as the image frame reference point t, and the image frame coordinates of t are [tx, ty]; Determine the distance D1 between the middle point P on the ground and the reference point T on the ground: $\mathrm{<span\; class="notranslate">\; D.</span>}\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; =</span>\sqrt{{<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; P</span>}\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; T</span>}\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; )</span>}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; +</span>{<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; P</span>}\mathrm{<span\; class="notranslate">\; Y</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; T</span>}\mathrm{<span\; class="notranslate">\; Y</span>}<span\; class="notranslate">\; )</span>}^{\mathrm{<span\; class="notranslate">\; 2</span>}}}$ Determine the distance d1 between the middle point p of the image frame and the reference point t of the image frame: $\mathrm{<span\; class="notranslate">\; d</span>}\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; =</span>\sqrt{{<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; p</span>}\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; t</span>}\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; )</span>}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; +</span>{<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; p</span>}\mathrm{<span\; class="notranslate">\; the\; y</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; t</span>}\mathrm{<span\; class="notranslate">\; the\; y</span>}<span\; class="notranslate">\; )</span>}^{\mathrm{<span\; class="notranslate">\; 2</span>}}}$ Determine the distance d2 between the middle point p of the image frame and the center point o of the image frame: $\mathrm{<span\; class="notranslate">\; d</span>}\mathrm{<span\; class="notranslate">\; 2</span>}<span\; class="notranslate">\; =</span>\sqrt{{<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; p</span>}\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; o</span>}\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; )</span>}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; +</span>{<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; p</span>}\mathrm{<span\; class="notranslate">\; the\; y</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; o</span>}\mathrm{<span\; class="notranslate">\; the\; y</span>}<span\; class="notranslate">\; )</span>}^{\mathrm{<span\; class="notranslate">\; 2</span>}}}$ Determine the distance D2 from the middle point P on the ground to the straight line Oo: D2=d2*D1/d1; Step 6, get the connection line between the center point o of the image frame and the middle point p of the image frame, and record it as the vertical center line lv of the image frame; Step 7. Determine the space attitude parameters: the space attitude parameters include camera direction angle α, camera rotation angle β, camera vertical angle γ, camera height CH and camera position [CX, CY]; The camera direction angle α is the angle between the vertical centerline Lv of the ground and the Y-axis of the ground; The camera rotation angle β is the angle between the vertical center line lv of the image frame and the y-axis of the image frame; Camera vertical angle γ, that is, the angle between the camera shooting axis and the ground, that is, the supplementary angle between the camera target surface and the ground: γ=arcsin(D2*2/D); 0<γ<π/2; The camera height CH is the vertical height of the camera relative to the ground, namely CH=D2*2*cos(γ); The camera position [CX, CY] is the position of the vertical point of the camera on the ground; determine the distance D3 from the camera position [CX, CY] to the perspective point F on the ground: D3＝D2*2*D2*2/D; CX＝FX+D3/D*(OX-FX) CY=FY+D3/D*(OY-FY). 2. the method for obtaining camera space attitude according to claim 1, is characterized in that, the corresponding conversion function F([x ,y]) and the inverse transformation function F ^T ([X,Y]) is: [X,Y]=F([x,y]) [x,y]=F ^T ([X,Y]) Then [OX,OY]=F([ox,oy]). 3. the method for obtaining camera space attitude according to claim 1, is characterized in that, the intersection line Lh of determining camera target surface and ground described in step 2 specifically comprises the following steps: Step 2-1. Take four vertices v1, v2, v3 and v4 of any rectangle on the screen; Step 2-2. Use the corresponding conversion function F([x,y] of the screen pixel coordinates [x,y] and the ground position coordinates [X,Y] to determine the projective points V1 of v1, v2, v3, and v4 on the ground , V2, V3, V4; V1, V2, V3, V4 form a quadrilateral Q on the ground; Step 2-3, determine respectively by 2 groups of opposite sides L1, L2 and L3, L4 of quadrilateral Q as two intersections F1 and F2 of straight line, the connecting line of F1 and F2 is ground horizon Lh; The linear equation of Lh is: Y=Lh_a*X+Lh_b, where Lh_a is the coefficient term of X, and Lh_b is the constant term. 4. the method for obtaining camera space attitude according to claim 1, is characterized in that, the equation expression of ground vertical center line Lv described in step 3 is: Y=Lv_a*X+Lv_b, wherein Lv_a is a coefficient term of X, and Lv_b is a constant term. 5. the method for obtaining camera space posture according to claim 1, is characterized in that, the equation expression of picture vertical center line lv among the step 6 is: y=lv_a*x+lv_b, wherein lv_a is the coefficient item of x, lv_b is a constant item. 6. The method for obtaining camera space attitude according to claim 1, characterized in that, determining the camera direction angle α in step 7 is specifically: If OX-PX>0, then α=arcctg(Lv_a); If OX−PX≦0, then α=arcctg(Lv_a)+π. 7. The method for obtaining the spatial attitude of the camera according to claim 1, wherein determining the camera rotation angle β in step 7 is specifically: If ox-px>0, then β=arcctg(lv_a); If ox-px≤0, then β=arcctg(lv_a)+π.