A 3D vision measurement method for double-line crossed structured light

Abstract

A dual-line crossed structured light three-dimensional vision measurement method belongs to the field of structured light vision measurement systems. The present invention solves the problem of real-time pause calibration of the scanned light plane in the traditional single-line structured light measurement method based on a slide rail. Due to the characteristics of the stepper motor, the scanning process can be described as "step → pause → step", which cannot be called a continuous "scanning" process. The dual-line crossed structured light three-dimensional vision measurement method only requires a fixed slide rail, where the moving step size no longer needs to be controlled by a stepper motor and is not affected by the hysteresis of the stepper motor. It can make the scanned light strip move smoothly and continuously, while ensuring the accuracy. By calculating the real-time spatial attitude equation of the scanned light plane, the three-dimensional point cloud information of the measured object is obtained item by item, realizing the overall reconstruction of the three-dimensional shape of the object. Especially for the measurement of the three-dimensional surface of a spatial object, it has high efficiency and accuracy and can be applied to the field of three-dimensional reconstruction.

Description

Technical Field

[0001] The invention belongs to the field of structured light vision measurement systems, and in particular relates to a double-line cross structured light three-dimensional vision measurement method. Background Art

[0002] Visual measurement technology is a comprehensive cutting-edge measurement technology widely used in modern industry. Based on computer vision, it combines pattern recognition, image processing, optical communication and other related cutting-edge fields. It irradiates visible laser onto the object to be measured in a non-contact manner, and uses the feature points projected onto the surface of the object by laser, the position relationship between the camera and the laser to obtain the three-dimensional information of the object to be measured. It has the characteristics of fast speed, high precision and good stability, and can be divided into point structured light, line structured light and grating measurement technology.

[0003] Among them, line structured light measurement technology is the most representative. The basic principle of its measurement system is to project a laser plane with a known spatial posture emitted by a single-line laser to a certain object to be measured in space, forming a beam of projection light strip on the surface of the object. The laser plane is moved along a known direction, and the projection light strip on the corresponding object surface also moves accordingly. A high-precision camera is used to take pictures of the position of the projection light strip during its movement, and the light strip center detection technology is used to extract the pixel coordinates of the feature points of the projection light strip. Finally, the parameters of the structured light vision system obtained through calibration are used to project the pixel coordinates of the above feature points into the camera coordinate system, thereby obtaining the spatial morphological characteristics of the object to be measured.

[0004] However, the traditional single-line structured light measurement method first requires a linear slide to control the moving direction of the laser plane and the corresponding projected light strip, and secondly requires a stepper motor to control the moving step length of the laser plane and the corresponding projected light strip. The fixed-position laser is moved along the slide by a certain step length under the control of the stepper motor. The displacement of the step length must be a known fixed length, and the accuracy of the step length is affected by the moving accuracy of the stepper motor, so as to achieve line-by-line measurement of the three-dimensional morphology of the object being measured, and then restore the overall morphology by stitching multiple pictures. However, the accuracy of the stepper motor is limited, and the movement controlled by the stepper motor has hysteresis, which leads to a significant decrease in the accuracy and efficiency of the single-line structured light vision measurement system, affecting the measurement effect, and then affecting the restoration of the three-dimensional morphology of the object being measured. Summary of the invention

[0005] The purpose of the present invention is to solve the problem that the existing single-line structured light vision measurement system has poor accuracy and low efficiency resulting in poor measurement effect, and to provide a double-line cross-structured light three-dimensional vision measurement method, which can effectively solve the problems existing in the traditional single-line structured light measurement method based on a slide rail.

[0006] The present invention solves the problem of real-time pause calibration of the scanned light plane in the traditional single-line structured light measurement method based on a slide rail. Due to the characteristics of the stepper motor, the scanning process can be described as "step → pause → step", and strictly speaking, it cannot be called a continuous "scanning" process. The dual-line crossed structured light three-dimensional vision measurement method only requires a fixed slide rail. Among them, the moving step size no longer needs to be controlled by a stepper motor and is not affected by the hysteresis of the stepper motor. It can make the scanned light strip move smoothly and continuously. At the same time, the accuracy can be guaranteed. By calculating the real-time spatial attitude equation of the scanned light plane, the three-dimensional point cloud information of the measured object is obtained item by item, and the overall reconstruction of the three-dimensional shape of the object is realized. Especially for the measurement of the three-dimensional surface of a spatial object, it has high efficiency. The point cloud data obtained by this method has high accuracy and can be applied to the field of three-dimensional reconstruction.

[0007] To achieve the above object, the technical solution adopted by the present invention is as follows:

[0008] A dual-line crossed structured light three-dimensional vision measurement method, the method specifically being:

[0009] Step 1: Arrange a slide rail, two lasers, a camera, a target plane and an object to be measured at predetermined positions. The laser planes emitted by the two lasers intersect at a certain angle, and the intersecting angle can be determined according to the detection situation;

[0010] Step 2: Keep the laser arranged on the slide rail in the vertical direction and cover the entire area of the object to be measured. The light strip emitted by the other laser remains in the horizontal direction and its pose remains unchanged throughout the measurement, and it should be avoided from being projected onto the surface of the object to be measured;

[0011] Step 3: The intersection of the two light strips generates a cross-like intersection area. Determine the sub-pixel coordinates of the geometric center point within this area. At the same time, project the pixel coordinates of this intersection point into the camera coordinate system to calculate its spatial coordinates, and then use this spatial coordinate to calculate the equation of the moving laser plane to achieve real-time calibration. Then, use a high-precision camera to respectively take pictures of the real-time positions of the moving light strips. Finally, project the pixel coordinates of the light strip feature points on the measured object into the camera coordinate system by using the parameters of the dual-line crossed structured light three-dimensional vision measurement system obtained by calibration to obtain the spatial shape features of the measured object.

[0012] Further, Step 1 further includes calibrating the camera and two initial light planes to obtain the internal parameter matrix and distortion coefficient of the camera, and the initial light plane equation.

[0013] Further, in Step 3, calculate the scanned light plane VLP nThe real-time spatial posture equation represented by the dwell position in each picture; by the point normal equation theorem of the plane: In the spatial rectangular coordinate system, given a point M(x0,y0,z0) and a normal vector on the plane Then we can determine that this plane is: A(X-x0)+B(Y-y0)+C(z-z0)=0. It can be known that each spatial posture equation of the light plane VLPn in the scanning process when n≥2 can be calculated. The specific method is as follows:

[0014] First, the spatial coordinates of the intersection point P are obtained by calculation

[0015] Second, let the plane equation of the light plane VLP1 at the scanning starting point be A v1 x+B v1 y+C v1 z+1=0, where A v1 ,B v1 ,C v1 are the coefficients of x, y, and z, respectively, obtained by light plane calibration, vector is the representative normal vector of the light plane VLP1; the light planes VLP2, VLP3, ... VLP obtained by translation of the slide rail n are parallel to the light plane VLP1, so VLP n The normal vectors of (n≥2) are

[0016] Third, summarizing the above two points, we can write about the new light plane VLP n The following formula of the equation:

[0017]

[0018] After finishing, we can get: That is, the light plane VLP of the nth image is obtained n The spatial attitude equation of the image is used to realize the real-time calibration of the scanning light plane during the scanning process.

[0019] Compared with the traditional single-line structured light measurement method, the present invention has the following advantages:

[0020] 1. The accuracy can reach 1μm;

[0021] 2. It avoids the hysteresis limitation of the stepper motor, making the "scanning" process smooth and smooth, greatly improving the measurement efficiency;

[0022] 3. Easy to operate, with a certain level of semi-automation. The camera can continuously shoot images of light strip scanning through one operation; the light strip processing process in multiple images can also be automatically detected at the program level. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 It is a diagram of a slide rail, a laser, a laser plane, and their intersection poses;

[0024] Figure 2 It is a diagram of a target plane, a laser stripe, and an intersection area;

[0025] Figure 3 It is a diagram of the intersection area and its geometric center;

[0026] Figure 4 It is a schematic diagram of the initial light plane and the position of the object to be measured;

[0027] Figure 5 It is a diagram of the scanning process;

[0028] Figure 6 It is a diagram of a square pattern target and a light stripe;

[0029] Figure 7 It is a complete measurement flow chart. Specific implementation manners

[0030] The technical solutions of the present invention will be further described below in conjunction with the accompanying drawings and embodiments, but are not limited thereto. Any modification or equivalent replacement of the technical solutions of the present invention without departing from the spirit scope of the technical solutions of the present invention shall be covered within the protection scope of the present invention.

[0031] Based on the traditional single-line structured light measurement system, the present invention abandons the stepping motor, retains the slide rail, and at the same time introduces another laser plane, which intersects with the laser plane of the traditional method at a certain angle, and the intersection angle can be determined according to the detection situation. Thus, two projected light stripes are obtained: the traditional moving light stripe is kept in the vertical direction and can cover the entire area of the object to be measured; the other introduced light stripe is kept in the horizontal direction and remains unchanged in pose throughout the measurement process, and should be avoided from being projected onto the surface of the object to be measured. The two light stripes intersect to generate a cross-like intersection area. The sub-pixel coordinates of the geometric center point within this area are determined, and at the same time, the pixel coordinates of this intersection are projected into the camera coordinate system to calculate its spatial coordinates. Then, the equation of the moving laser plane is calculated using this spatial coordinate to achieve real-time calibration. Finally, pictures of the real-time positions of the moving light stripes are taken by a high-precision camera, and the pixel coordinates of the light stripe feature points on the object to be measured are projected into the camera coordinate system using the parameters of the double-line cross structured light three-dimensional vision measurement system obtained by calibration, so as to obtain the spatial morphology characteristics of the object to be measured.

[0032] To implement the above process, before applying the binocular crossed structured light three-dimensional vision measurement method to actual measurement, the parameters of the measurement system should first be obtained through calibration, that is, the internal parameters of the camera are obtained through camera calibration; the light plane equations of the initial poses of the two laser planes are obtained through light plane calibration.

[0033] Currently, in the field of vision measurement technology, a camera calibration method proposed by Zhang Zhengyou in his published literature Zhang Z. (2000). A flexible new technique for camera calibration. IEEE Trans. on Pattern Analysis and Machine Intelligence, 22(11): 1330-1334. is usually adopted to complete camera calibration.

[0034] In the camera calibration stage of the present invention, an improved calibration model based on Zhang Zhengyou's calibration method proposed in the publicly available doctoral dissertation "Axial Diameter Measurement Based on Machine Vision" published by Sun Qiucheng in 2010 is further adopted to adapt to the light plane calibration technology, improving the calculation efficiency and accuracy. In the light plane calibration stage, a calibration method for a line structured light vision system using a planar template with a square pattern disclosed in "A Flexible Calibration Method Using the Planar Target with a Square Pattern for Line Structured Light Vision System. Sun, Qiucheng, et al. A Flexible Calibration Method Using the Planar Target with a Square Pattern for Line Structured Light Vision System. PloS one 9.9: e106911. (SCI) 2014." is adopted. This method uses the coordinates of the four corner points of the square pattern and the projected light strips with different poses of the laser plane to obtain the spatial equation of the single line structured light plane in the camera coordinate system. The detection of the feature points at the center of the light strip is a key link in the line structured light vision measurement system. The present invention adopts a light strip center point detection algorithm published by Steger C. in his publicly available literature "An unbiased detector of curvilinear structures [J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 1988, 20(2): 113 - 125." to achieve sub-pixel level detection of the feature points, and this method has high accuracy and good robustness.

[0035] The following is the specific implementation process of the present invention

[0036] First, based on the traditional single line structured light measurement system, the stepping motor is discarded, the slide rail is retained, and at the same time, another single line laser is introduced, that is, another laser plane is introduced. Denote the slide rail as SR, the original laser as L1, and the newly introduced laser as L2. The laser planes they emit are denoted as LP1 and LP2 respectively. The two laser planes intersect at a certain angle, as Figure 1As shown, the intersecting angle can be determined according to the detection situation. The target plane is denoted as TP. Place the object to be measured on TP. The laser strips projected by the two laser planes onto TP are denoted as LS1 and LS2 respectively. Let LS1 be kept in the vertical direction as the scanning laser strip, and re-denoted as VLS1 (Vertical Light Strip), and the corresponding light plane is denoted as VLP1 (Vertical Light Plane). It is required that VLS1 can cover the entire surface of the object to be measured during the scanning process; let LS2 be kept in the horizontal direction as the fixed laser strip, re-denoted as HLS (Horizontal Light Strip), and the corresponding light plane is denoted as HLP (Horizontal Light Plane). It is required that HLS avoids projecting onto the surface area of the object to be measured. VLS1 and HLS always intersect during the entire binocular crossed structured light three-dimensional vision detection process, and a cross-like intersection area is generated at all times, as Figure 2 shown. Find the sub-pixel-level geometric center coordinates of the intersection point of the center lines of the two laser strips in this area by the intersection line method, denoted as P1 (Point), as Figure 3 shown.

[0037] Secondly, fix the positions of the camera and the laser L2, and rotate L2 to find a suitable angle on the target plane and fix it. Adjust the position of L1 so that VLS1 is kept at the starting position of the slide rail and adjusted to a suitable angle according to the position and shape of the object to be measured, as Figure 4 shown. Record the light plane VLP1 at this time. HLP is the light plane in the initial state. At this time, camera calibration and calibration of the two initial light planes are carried out successively to obtain the pre-parameters required by the present invention, that is, the internal parameter matrix and distortion coefficient of the camera, and the light plane equations of the light planes VLP1 and HLP.

[0038] Subsequently is the core part of the present invention: Translate the laser L1 along the slide rail once. The laser plane VLP1 emitted by it and the laser strip VLS1 irradiated on the surface of the object to be measured also move accordingly, generating a new laser plane, denoted as VLP2, and the new laser strip is denoted as VLS2. Among them, VLP1 and VLP2, VLS1 and VLS2 are all parallel relationships. At this time, the geometric center of the new intersection area generated by the intersection of the laser strip VLS2 and HLS, its coordinates are denoted as P2. Then, take a picture with the information of the two laser strips by the camera. Thus, one scanning action is completed, as Figure 5 shown. Repeat the above actions, continue to translate the laser L1 once, and a new laser plane is generated, denoted as VLP3, the new laser strip is denoted as VLS3, the new intersection coordinates P3, and then take a picture by the camera. And so on, slide in turn until the light plane VLP1 sweeps across the entire surface of the object to be measured, obtaining VLSn , P n and n pictures, the scanning can be considered completed.

[0039] Example 1:

[0040] The complete measurement process is as Figure 7 shown. The following are the specific implementation steps:

[0041] Step 1: Establish Figure 4 the measurement device of the binocular structured light vision measurement system composed of a slide rail, a laser, a camera, a target plane, and the object to be measured as shown. Among them, once the positions of the slide rail, the camera, the target plane, and the laser L2 are selected, their poses will not change during the entire scanning measurement process; the camera adjusts the internal parameters of the camera according to the fixed position to make the imaging effect of the camera on the target plane TP as clear as possible; on the premise of ensuring the measurement requirements, the laser L1 finds its initial position through the slide rail SR.

[0042] Step 2: Calibrate the camera parameters to obtain the internal parameters and external parameters of the camera. The specific method is as follows:

[0043] First, establish a conversion coordinate system, that is: world coordinate system, camera coordinate system, image physical coordinate system, and image pixel coordinate system; the world coordinate system can be freely placed according to the convenience of calculation. Generally, the origin of the world coordinate system is the upper left corner point on the plane calibration board, its XOY plane coincides with the calibration board plane, and let the points on the calibration board have Z = 0; the origin of the camera coordinate system is at the optical center of the camera, and the z-axis is parallel to the camera optical axis; the origin of the image physical coordinate system is at the intersection of the camera optical axis and the imaging plane; the origin of the image pixel coordinate system is at the upper left corner point of the picture; among them, the three-dimensional coordinates of a certain point in space under the world coordinate system are expressed as (X w , Y w , Z w ), and after the rigid body transformation captured by the camera, it is converted into the coordinates (X c , Y c , Z c ) under the camera coordinate system, and then through perspective projection, it is converted into the ideal image coordinates (x u , y u ) under the image physical coordinate system. Due to the distortion of the camera lens, (x u , y u ) is not the true position of a certain point. Through the conversion relationship, the actual image coordinates (x d , y d ) under the image physical coordinate system and the actual pixel coordinates (x p , y p ) under the image pixel coordinate system are obtained;

[0044] Second, use a checkerboard calibration board to make inclination changes in multiple different poses within the shooting range of the camera and take more than three pictures to calibrate the internal and external parameters of the camera and obtain the internal parameter matrix of the camera. And the distortion coefficient K=(k1, k2, p1, p2). Where α and β respectively represent the scale factors of the U-axis and V-axis in the image pixel coordinate system; γ represents the non-perpendicular factor of the two coordinate axes in the pixel plane, and (u0, v0) represents the coordinates of the intersection point of the camera optical axis and the image plane in the pixel coordinate system; k1 and k2 represent the coefficients of the radial distortion function in the image physical coordinate system; p1 and p2 represent the coefficients of the tangential distortion function in the image physical coordinate system.

[0045] Third, the specific calibration model adopts an improved calibration model based on Zhang Zhengyou's calibration method proposed by Sun Qiucheng. The formula for the coordinate system conversion relationship is as follows:

[0046]

[0047] where, t = [t1 t2 t3] T is the translation vector; is the external parameter rotation matrix (which can be represented by Euler angles);

[0048] Supplement to the calibration model formula:

[0049]

[0050] (X c , Y c , Z c ) is the coordinate of a point in the camera coordinate system; (X w , Y w , Z w ) is the coordinate of a point in the world coordinate system; t = [t1 t2 t3] T is the translation vector; is the external parameter rotation matrix. The meaning of this formula is that the coordinate of a point in the world coordinate system is converted into the corresponding coordinate in the camera coordinate system through rigid body transformations such as rotation and translation.

[0051]

[0052] (x u , y u ) is the ideal image coordinate of a point in the image physical coordinate system. The meaning of this formula is that the coordinate of a point in the camera coordinate system is converted into the corresponding ideal image coordinate in the image physical coordinate system through perspective projection.

[0053] where

[0054] (x d , y d ) is the actual image coordinate of a point in the image physical coordinate system; (x p , y p ) is the pixel coordinate of a point in the image pixel coordinate system; (k1, k2, p1, p2) are the camera lens distortion coefficients. The meaning of this formula is that the ideal coordinate of a point in the image physical coordinate system is converted to the corresponding actual coordinate after distortion correction.

[0055]

[0056] is the internal parameter matrix of the camera, and the specific meanings of the parameters have been mentioned in the second point of step two. The meaning of this formula is the conversion from the actual coordinate of a point in the image physical coordinate system to the corresponding pixel coordinate in the image pixel coordinate system.

[0057] Step 3: Calibrate the scanned light plane at the initial position and the light plane at the fixed position all the time, corresponding to VLP1 and HLP as shown in Figure 4 respectively, and obtain the plane equation of the laser plane in the camera coordinate system. The specific method is as follows:

[0058] First, the light plane calibration method uses a single-line structured light calibration method using a square pattern plane template disclosed by Sun Qiucheng. The plane equation of the light plane is fitted by the four corner points of the black square pattern and the feature points detected at the center of the light strip in multiple pictures. Use the single-line structured light calibration method twice to calibrate the two light planes VLP1 and HLP at the initial position respectively.

[0059] Second, taking the calibration of the light plane VLP1 as an example: Let the light strip VLS1 always stay on the square target, and the square target makes j (4 ≤ j ≤ 9) pose changes near the target plane. Each time it changes, use the camera to take a calibration picture including the square target and the projected light strip, as shown in Figure 6 until j pictures are obtained.

[0060] Third, perform corner detection on each calibration picture taken by the camera. Use the corner detection algorithm to detect the pixel coordinates of the 4 corner points of the square target in the calibration picture where m = 1, 2, 3……, representing the serial number of the picture; n = 1, 2, 3......, representing the serial number of the 4 corner points in the current single picture in this picture. Using the internal and external parameters of the camera, the following mathematical model about the coordinate conversion relationship can be established:

[0061]

[0062] Among them, represents the ideal image coordinate system coordinates of the corner points, represents the actual image coordinate system coordinates of the corner points. Using the pixel coordinates of the corner points and their corresponding two-dimensional coordinates in the world coordinate system , the following equation can be obtained:

[0063]

[0064] Among them, S represents the scale factor, and r i (i = 1, 2) represents the i-th column vector of the rotation matrix R, and t represents the translation vector. Let H = [r1 r2 t]. Then the above equation can be rewritten as: Canceling s, it can be written as: Among them, h i (i = 1, 2, 3) is the i-th row of the matrix H. Continuing the derivation, we get:

[0065]

[0066] Written in matrix form, it is:

[0067] Let When the pixel coordinates of 4 corner points are given, that is, when i takes values from 1 to 4 in sequence, 1 image can correspondingly obtain 8 equations and can be written as a homogeneous equation system in the form of Cx = 0. Among them, C is an 8×9 coefficient matrix. When x is defined as a scale factor, the solution of Cx = 0 is the right singular vector of L associated with the smallest singular value. Using the rotation matrix R and the translation vector T, the equation of the square target plane can be solved in the camera coordinate system.

[0068] Fourth, perform center detection on the laser projection light strips on each calibration image captured by the camera. Here, the Steger light strip center detection algorithm is used to obtain the pixel coordinates of the light strip center feature points in each calibration image. Similarly, using the external parameter H of the square target in each image, the pixel coordinates of the light strip center are projected into the camera coordinate system, that is, the corresponding three-dimensional camera coordinate system coordinate points are obtained. Repeating the processing of each image can obtain a set of laser projection light strip center points characterized by the light strip center points.

[0069] Fifth, project all the obtained point sets of the laser projection light strip centers into a unified camera coordinate system, and use these points to fit to obtain a unique spatial plane, thereby solving the spatial equation of the structured light plane and realizing the calibration of the light plane.

[0070] Step Four: Collect scanned images. As Figure 5As shown, the light plane VLP1 is continuously slid according to the slider on the slide rail, so that the light strip VLS1 smoothly transitions to the light strip VLS n , and use the camera to collect these n pictures at the same time. During the scanning operation, the camera can be set to a continuous shooting state through the driver, for example, it is set to shoot one picture every 50ms. Then, from the moment the "shoot" button is pressed, the slider on the rail that fixes the laser L1 is controlled so that the light strip VLS1 completely scans the object to be measured until the "end" button is pressed. Assuming that this process takes 10s, the camera will capture 200 pictures of the constantly changing scanning light strip, which can achieve efficient image acquisition.

[0071] Step 5: Find the pixel coordinates of the light bar intersection point P in the pixel coordinate system And find the corresponding spatial coordinates of point P Where p represents the intersection point P, q = 1, 2, 3 ... represents the order of the intersection point P, corresponding to the number of scanned images n. During the scanning process of the light plane VLP1, a scanning action (such as from VLP1 to VLP2) generates a new light plane and light strip intersection point, such as Figure 5 The specific location of the intersection of the light strips is shown in Figure 3 The specific method of finding the two coordinates of point P is as follows:

[0072] First, the pixel coordinates of the intersection point P It is obtained by the intersection method: the light strip center point detection algorithm is used on both sides of the intersection point P to obtain several sub-pixel coordinates falling on the light strip HLS, and a straight line equation y1 is fitted using these coordinates; similarly, several sub-pixel coordinates falling on the light strip VLS are found. n The sub-pixel coordinates on the light strip are obtained by fitting a straight line equation y2; the two straight lines intersect, that is, the simultaneous equations y1 and y2, and the pixel coordinates of the intersection point P of the light strip can be calculated. And the accuracy reaches sub-pixel level.

[0073] Second, the plane equation of the light plane HLP obtained in step 3 is known, and the internal and external parameters of the camera obtained in step 2 are known. At the same time, using the characteristic that the intersection of the light strips exists on both light strips, we can calculate Space coordinates in the camera coordinate system Let the plane equation of the light plane HLP be A H x+B H y+C H z+1=0, where A H ,B H ,C H are the coefficients of x, y, and z respectively, obtained from the light plane calibration in step 3. Substitute the camera calibration model, that is, project it into the camera coordinate system, and use the plane equation A of the known light plane HLP H x+B H y+C H z+1=0 can determine the depth of field of point P and obtain the spatial coordinates corresponding to point P.

[0074] Step 6: Calculate the scanned light plane VLP n The real-time spatial posture equation represented by the dwell position in each picture. According to the point normal equation theorem of the plane: In the spatial rectangular coordinate system, given a point M(x0,y0,z0) and a normal vector on the plane Then we can determine that this plane is: A(X-x0)+B(Y-y0)+C(z-z0)=0, so we know that the light plane VLP of the scanning process n When n≥2, each spatial posture equation can be solved as follows:

[0075] First, from step 5, we know that the spatial coordinates of the intersection point P are has been obtained through calculation.

[0076] Second, let the plane equation of the light plane VLP1 at the scanning starting point be A v1 x+B v1 y+C v1 z+1=0, where A v1 ,B v1 ,C v1 are the coefficients of x, y, and z respectively, which are obtained by the light plane calibration in step 3. It can be known that the vector is the representative normal vector of the light plane VLP1. The light planes VLP2, VLP3, ... VLP obtained by translation of the slide rail n are parallel to the light plane VLP1, so VLP n The normal vectors of (n≥2) are

[0077] Third, summarizing the above two points, we can write about the new light plane VLP n The following formula of the equation:

[0078]

[0079] After finishing, we can get:

[0080] That is, the light plane VLP of the nth image is obtained n The spatial attitude equation of the image is used to realize the real-time calibration of the scanning light plane during the scanning process.

[0081] Step 7: Use the light strip center point detection algorithm to find the scanning light strip VLS in the pixel coordinate system n The sub-pixel coordinates of the feature points Project these coordinates into the camera coordinate system to calculate their spatial coordinate set A set of spatial coordinates in the camera coordinate system can be obtained to complete the three-dimensional measurement of the surface feature points collected by a light strip corresponding to a picture of the measured object. The specific method is as follows:

[0082] First, the real-time spatial posture equation of the scanned light plane of each image has been calculated in step six.

[0083] Second, the pixel point set obtained by the light bar center point detection algorithm Projected to a unified camera coordinate system, after the camera calibration model, only an ideal two-dimensional image coordinate set without depth of field can be obtained. The process is as follows:

[0084]

[0085] in According to the camera imaging principle, the light reflected by the feature point in space is captured by the optical center of the camera. According to the reversibility of the optical path, it can also be considered that a ray is emitted from the optical center of the camera to the feature point, which falls on the image to form a sub-pixel point. The equation of the ray can be described as:

[0086] Third, the depth of field z is determined by the intersection of the ray and the plane. This can be written as follows:

[0087]

[0088] It can be introduced:

[0089] Once the depth of field z is determined, the spatial coordinates of a complete point can be calculated through the camera calibration model:

[0090] Finally, a set of spatial coordinates of the surface feature points of the measured object is obtained. That is, the three-dimensional measurement of the surface feature points collected from a picture of the object being measured is completed.

[0091] Step 8: Repeat steps 5, 6 and 7, calculate each picture in turn, and stitch all calculated spatial coordinate sets to obtain the point cloud data of the surface feature points of the measured object, and then obtain the three-dimensional picture of the measured object.

Claims

1. A three-dimensional visual measurement method for structured light with double-line intersection, characterized in that: The specific method is as follows: Step 1: Arrange the slide rail, two lasers, a camera, a target plane and the object to be measured at predetermined positions. The laser planes emitted by the two lasers intersect at a certain angle, and the intersection angle can be determined according to the detection situation; Step 2: Keep the laser arranged on the slide rail in the vertical direction and cover the entire area of the object to be measured. The light strip emitted by the other laser remains in the horizontal direction and keeps its pose unchanged throughout the measurement process, and it should be avoided from being projected onto the surface of the object to be measured; Step 3: The intersection of the two light strips generates a cross-like intersection area. Determine the sub-pixel coordinates of the geometric center point within this area. At the same time, project the pixel coordinates of this intersection point into the camera coordinate system to calculate its spatial coordinates, and then use these spatial coordinates to calculate the equation of the moving laser plane to achieve real-time calibration. Then, use a high-precision camera to respectively capture pictures of the real-time positions of the moving light strips. Finally, use the parameters of the double-line cross-structured light three-dimensional vision measurement system obtained by calibration to project the pixel coordinates of the light strip feature points on the object to be measured into the camera coordinate system to obtain the spatial morphology characteristics of the object to be measured.

2. The three-dimensional visual measurement method of double-line crossed structured light according to claim 1, characterized in that: Step 1 further includes calibrating the camera and the two initial light planes to obtain the internal parameter matrix and distortion coefficient of the camera, and the initial light plane equation.

3. A three-dimensional visual measurement method for double-line crossed structured light according to claim 1, characterized in that: In step 3, the scanning light plane VLP is calculated n The real-time spatial posture equation represented by the dwell position in each picture; by the point normal equation theorem of the plane: In the spatial rectangular coordinate system, given a point M(x0,y0,z0) and a normal vector on the plane Then we can determine that this plane is: A(X-x0)+B(Y-y0)+C(z-z0)=0, so we know that the light plane VLP of the scanning process n When n≥2, each spatial posture equation can be solved as follows: First, obtain the spatial coordinates of the intersection point P through calculation Second, the plane equation of the optical plane VLP1 at the scanning starting point is A v1 x + B v1 y + C v1 z + 1 = 0, where A v1 , B v1 , C v1 are the coefficients of x, y, and z respectively, obtained by calibrating the optical plane. The vector is a representative normal vector of the optical plane VLP1; the optical planes VLP2, VLP3... VLP n obtained by translating the slide rail are all parallel to the optical plane VLP1. It can be seen that the normal vectors of VLP n (n ≥ 2) are all Thirdly, summarizing the above two points, the following formula for the equation regarding the new optical plane VLP can be written: n ​ It can be sorted out as follows: That is, the optical plane VLP of the nth picture is obtained n The spatial attitude equation is realized, and the real-time calibration of the scanning optical plane during the scanning process is achieved.

Images