A slurry and slag identification method based on stereovision

Abstract

The application discloses a stereovision-based slurry and slag identification method, and belongs to the technical field of tunnel construction, and specifically comprises the following steps: two same cameras are placed above the slurry and slag, fixed after being adjusted to the same height and the imaging half-frames are overlapped; a calibration plate is shot to obtain the baseline distance of the optical centers of the two cameras; the exposure time is determined in combination with the camera accuracy, the slurry and slag movement speed and the imaging magnification; two images of the slurry and slag are synchronously shot, the coordinate system is established, the sight distance difference and the imaging point spacing are calculated, the object distance is solved based on the geometric relationship constructed by the Gaussian imaging formula and the slurry and slag point position and the imaging plane and the camera optical center; finally, the three-dimensional point cloud model is established according to the object distance, and the slurry and slag shape, volume and surface area are output. The application solves the problems of motion blur, image distortion and the inability to accurately identify the size of the slurry and slag in the identification of the slurry and slag, can improve the identification accuracy, and meets the intelligent and refined construction requirements of large shield projects.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of tunnel construction, in particular to a slurry and slag identification method based on stereovision. BACKGROUND

[0002] A large amount of rock debris is generated in the interaction between a shield machine and a rock-soil body during tunneling, and the rock debris mixes with the slurry to form slurry and slag. The size, shape and composition information of the slurry and slag directly reflect the geological conditions of the tunneling face and the working state of the shield machine. The traditional manual slurry and slag analysis method has low efficiency and strong subjectivity, and it is difficult to meet the intelligent and refined construction standards of large-scale shield projects. The existing technology has gradually replaced the traditional manual slurry and slag analysis with intelligent algorithms, but the following key problems have not been solved:

[0003] (1) Motion blur problem. The slurry and slag on the prescreening machine moves at a speed of 0.5-1.0 m / s and is accompanied by strong vibration. If the camera exposure time is not set properly, it is easy to cause image smearing, affecting the recognition accuracy.

[0004] (2) Image distortion problem. The inherent optical properties of the industrial camera lens will cause distortion in the captured image. If not corrected, it will cause calculation errors in the physical parameters such as size and shape of the slurry and slag.

[0005] (3) Pixel and physical size conversion problem. When accurately calculating the actual physical size of the slurry and slag from the two-dimensional image pixel information, there is a lack of pixel equivalent conversion model. SUMMARY

[0006] The present application provides a slurry and slag identification method based on stereovision, which solves the problems of motion blur, image distortion, and inaccurate pixel and size conversion in the slurry and slag identification process.

[0007] The technical solution of the present application is as follows:

[0008] A slurry and slag identification method based on stereovision, comprising the following steps:

[0009] S1, stably placing two identical cameras above the slurry and slag to be identified, with the left camera being the first camera and the right camera being the second camera;

[0010] S2, synchronously adjusting the first camera and the second camera to make the two cameras at the same horizontal height, and making the right half of the imaging picture of the first camera coincide with the left half of the imaging picture of the second camera, and fixing the positions of the first camera and the second camera after coincidence;

[0011] S3, operating the first camera and the second camera to capture the calibration board, processing the captured image through the calibration software, outputting the calibration parameters of the first camera and the second camera, and obtaining the baseline distance x of the optical centers of the first camera and the second camera based on the calibration parameters;

[0012] S4, determining the exposure time based on the camera lens accuracy, the slurry residue movement speed and the imaging magnification:

[0013] ;

[0014] In the formula, t represents the exposure time, p represents the blur circle diameter, v represents the slurry residue movement speed, and K represents the imaging magnification;

[0015] S5, operating the first camera and the second camera to simultaneously shoot the slurry residue, and obtaining a first image and a second image on the imaging plane respectively;

[0016] A first rectangular coordinate system is established with the first image center point as the origin O1, and the optical center of the first camera is located below the origin of the coordinate system, denoted as P; a second rectangular coordinate system is established with the second image center point as the origin O2, and the optical center of the second camera is located below the origin of the coordinate system, denoted as Q; for any point M of the slurry residue in the superimposed image, the following steps are performed:

[0017] S51, assuming that the point M corresponds to L and R in the first image and the second image respectively, the coordinates of the points L and R are (X L ,0) and (X R ,0) respectively, the distance difference d=(X L -X R )×i of the point M in the two images is calculated based on (X L ,0) and (X R ,0), and i represents the unit pixel distance;

[0018] S52, calculating the imaging point distance l LR of the point M in the first image and the second image based on the baseline distance x and the distance difference d;

[0019] S53, drawing perpendicular lines from the point M to the two origin connecting lines and the two optical center connecting lines to intersect the two origin connecting lines and the two optical center connecting lines at B and N respectively, and based on the similarity relationship of each triangle contained in the triangle MPQ and l LR , solving the distances of the point M to the points P and Q respectively, denoted as l MP and l MQ , and the l MP and l MQ respectively represent the object distance of the point M to the first camera and the second camera;

[0020] S6, establishing a three-dimensional point cloud model of the slurry residue based on l MP and l MQ , and outputting the shape, volume and surface area of the slurry residue based on the three-dimensional point cloud model.

[0021] Furthermore, in S53, based on △MLR and △MPQ contained in △MPQ, we derive Equation 1 using △MLR∽△MPQ:

[0022] ;

[0023] , Let M represent the distance from point M to point B, and B represent the distance from point B to point N, respectively.

[0024] Furthermore, in S53, based on △MLB and △MPN contained in △MPQ, equation two is derived using △MLB∽△MPN:

[0025] ;

[0026] In the formula, , Let P represent the distance from point P to point N, and L represent the distance from point B to point B, respectively.

[0027] Based on △MBR and △MNQ contained in △MPQ, and using the fact that △MBR∽△MNQ, we can derive equation three:

[0028] ;

[0029] In the formula, , Let N represent the distance from point N to point Q, and B represent the distance from point B to point R, respectively.

[0030] Furthermore, based on △MBR, △MNQ, △MLB, and △MPN contained in △MPQ, equations four and five are derived:

[0031] ;

[0032] .

[0033] Furthermore, according to the Gaussian imaging formula, assuming the focal length is f, equation six is ​​derived from △MPN contained within △MPQ:

[0034] ;

[0035] In the formula, This represents the distance from point M to point P.

[0036] Furthermore, based on the right triangles △MPN and △MQN contained in △MPQ, equations seven and eight are obtained according to the Pythagorean theorem:

[0037] ;

[0038] .

[0039] Due to the adoption of the above technical solutions, the application has the following beneficial effects:

[0040] 1. The application provides an exposure time determination formula based on the motion speed of the slurry and the camera accuracy. Through the exposure time calculation formula, the exposure time can be adjusted according to the change of the on-site conveying belt speed, thereby solving the motion blur problem in the photographing process.

[0041] 2. The application obtains the real object distance through the camera parameters and the image parameters of the camera, and establishes a "binocular vision" through the 50% overlapping area of the two cameras. The "binocular vision" can effectively avoid the image distortion problem caused by a single image. Two images are mutually constrained through two groups of image parameters to obtain a real object distance with higher accuracy. Finally, the object distance of each point is calculated to perform three-dimensional point cloud modeling, and the real slurry shape is fitted.

[0042] 3. More specifically, the application realizes the inverse calculation of the actual physical size from two-dimensional image information through various transformation forms of triangle similarity, basic space parameters and camera parameters, and obtains the real object distance through the photographed photos. BRIEF DESCRIPTION OF DRAWINGS

[0043] The accompanying drawings, which are included to provide a further understanding of the application, form a part of the application and illustrate the illustrative embodiments of the application and together with the description serve to explain the application. The accompanying drawings are included to provide a further understanding of the application and constitute a part of the application.

[0044] Figure 1 The flow chart for the setup process of the first camera and the second camera of the application is shown in Figure 1.

[0045] Figure 2 The imaging principle diagram of △MPQ of the application is shown in Figure 2. DETAILED DESCRIPTION

[0046] Based on the background art, the application provides a slurry recognition method based on stereo vision, which includes the following steps:

[0047] S1, two same cameras are stably placed above the slurry to be recognized. The left camera is referred to as the first camera, and the right camera is referred to as the second camera.

[0048] S2, the first camera and the second camera are synchronously adjusted to be at the same horizontal height, and the right half of the imaging picture of the first camera is overlapped with the left half of the imaging picture of the second camera. After the overlap, the positions of the first camera and the second camera are fixed.

[0049] In the specific implementation, the first camera and the second camera are fixed on a rigid platform or support to take pictures of the pulp dregs from a top-down angle. The height of the first camera and the second camera is adjusted by using a total station to make the two cameras be located on the same plane. A laser is installed on the hot shoe of the camera to emit a laser beam, and the two laser beams are made to coincide at the position of the far wall to finely adjust the angle of the camera, so as to represent that the two optical axes are parallel. Finally, the magnification is adjusted to make the right half of the picture of the first camera coincide with the left half of the picture of the second camera.

[0050] S3, operating the first camera and the second camera to take pictures of the calibration board, processing the taken pictures by using a calibration software, outputting the calibration parameters of the first camera and the second camera, and obtaining the baseline distance x of the optical centers of the first camera and the second camera based on the calibration parameters.

[0051] The camera calibration process can be realized based on the prior art. In the specific implementation, the checkerboard calibration board is aligned with the camera, and the position of the calibration board is constantly changed to take different pictures. The taken pictures are processed by using a calibration software to respectively output the intrinsic parameters and the extrinsic parameters of the two cameras. The intrinsic parameters include the focal length and the principal point coordinates (the center point of the image), and the extrinsic parameters include the relative position relationship of the two cameras, which are represented by a rotation matrix and a translation vector. The intrinsic parameters and the extrinsic parameters obtained by the calibration are used to perform geometric transformation by using the functions of stereoRectify and initUndistortRectifyMap, so as to obtain a correction transformation matrix. According to the transformation matrix, the original images are corrected, and finally the first image and the second image are made to be coplanar. After the correction, the distance between the optical centers of the two cameras, that is, the baseline distance, is output, which is denoted as x.

[0052] After the correction, an ideal stereo vision system that is virtual and has completely parallel optical axes is obtained. At this time, the pictures taken by the first camera and the second camera can be perfectly spliced.

[0053] S4, determining the exposure time based on the camera lens accuracy, the pulp dregs movement speed and the imaging magnification:

[0054] ;

[0055] In the formula, t represents the exposure time, p represents the diameter of the blur circle, v represents the movement speed of the pulp dregs, and K represents the imaging magnification. In the embodiment, p is the size of one pixel.

[0056] In the specific implementation, the movement speed of the pulp dregs is between 0.5 m / s and 1.0 m / s, the magnification has been determined in the camera fixing process, and when the speed of the pulp dregs conveying belt changes, the exposure time of the camera is adjusted.

[0057] S5, operating the first camera and the second camera to simultaneously take pictures of the pulp dregs to respectively obtain a first image and a second image on the imaging plane.

[0058] As shown in the accompanying drawingsFigure 2 As shown in the figure, a first rectangular coordinate system is established with the first image center point as the origin O1, and then the first camera optical center is located above the origin of the coordinate system, denoted as P; a second rectangular coordinate system is established with the second image center point as the origin O2, and then the second camera optical center is located above the origin of the coordinate system, denoted as Q. The optical center is located above the image center. For any point M of the slag in the superimposed image, the following steps are performed:

[0059] S51, assuming that the imaging points of M in the first image and the second image are L and R respectively, the coordinates of L and R are (X L ,0) and (X R ,0) respectively, the parallax of M in the two images is calculated based on (X L ,0) and (X R ,0), d=(X L -X R )×i, i represents the unit pixel distance. When establishing the coordinate system, right is positive and left is negative, and the parallax is the horizontal coordinate difference of the same object in the images of the left and right cameras.

[0060] S52, based on the baseline distance x and the parallax d, the distance l LR between the imaging points of M in the first image and the second image is calculated, l LR =x-d;

[0061] S53, the perpendiculars of M to the two origin connecting lines and the two optical center connecting lines intersect the two origin connecting lines and the two optical center connecting lines at B and N respectively, based on the similarity of each triangle contained in △MPQ and l LR , the distances of M to P and Q are solved respectively, denoted as l MP and l MQ , l MP and l MQ represent the object distance of M to the first camera and the second camera respectively, which can be expressed as , ; the image distance of the first camera and the second camera is the distance from the optical center to the imaging plane, denoted as l PO1 and l PO2 , which can be expressed as , .

[0062] In S53, based on △MLR and △MPQ contained in △MPQ, the three angles of △MLR and △MPQ are equal, and equation one is obtained by using △MLR≈△MPQ:

[0063] △MLR≈△MPQ then has:

[0064] ;

[0065] ;

[0066] Then: ;

[0067] , respectively represent the distance from point M to point B, the distance from point B to point N.

[0068] In S53, based on the fact that the three angles of ΔMLB and ΔMPN contained in ΔMPQ are all equal, equation two is obtained by using ΔMLB ΔMPN:

[0069] Then ΔMLB ΔMPN has:

[0070] ;

[0071] ;

[0072] ;

[0073] In the formula, , respectively represent the distance from point P to point N, the distance from point L to point B;

[0074] Based on the fact that the three angles of ΔMBR and ΔMNQ contained in ΔMPQ are all equal

[0075] , equation three is obtained by using ΔMBR ΔMNQ:

[0076] Then ΔMBR ΔMNQ has:

[0077] ;

[0078] ;

[0079] Then ;

[0080] In the formula, , respectively represent the distance from point N to point Q, the distance from point B to point R.

[0081] Based on the fact that ΔMBR, ΔMNQ, ΔMLB, ΔMPN contained in ΔMPQ, equation four and equation five are obtained:

[0082] ;

[0083] .

[0084] According to the Gaussian imaging formula, let the focal length be f:

[0085] ;

[0086] In the formula, and respectively represent the object distance and image distance obtained by the first image, and after deformation, the following is obtained:

[0087] ;

[0088] Take the reciprocal: ;

[0089] Similarly, the following can be obtained: , , which respectively represent the object distance and image distance obtained by the second image;

[0090] In △MPQ, contained in △MPN, equation six is obtained:

[0091] ;

[0092] In the formula, represents the distance from point M to point P.

[0093] Based on the right triangles △MPN and △MQN contained in △MPQ, equations seven and eight are obtained according to the Pythagorean theorem:

[0094] ;

[0095] ;

[0096] Integrate the above equations one to eight:

[0097] ;

[0098] Through the eight equations, eight unknowns can be solved. Since M has arbitrary nature, the object distance of all points in the image can be calculated, and the next step is performed according to the object distance of each point.

[0099] S6, based on l MP and l MQ , a three-dimensional point cloud model of the slurry is established, and the shape, volume, and surface area of the slurry are output based on the three-dimensional point cloud model, and the slurry recognition is completed.

[0100] A three-dimensional rectangular coordinate system is established with the imaging plane of the camera as the bottom surface. The distance from any point in the slurry to the imaging plane (object distance) has been calculated, and the projection (L or R) of the point to the camera imaging plane is known. The computer automatically inputs the object distance and projection coordinates, and calculates the three-dimensional coordinates of the point. Repeat this process to obtain the three-dimensional coordinates of all points, construct a three-dimensional point cloud model, and output the shape, volume, and surface area of the slurry after the slurry model is constructed.

[0101] The places not mentioned in the application can be realized by using or referring to the existing technology.

[0102] The above only describes the embodiments of the application and is not used to limit the application. The application can have various changes and variations for those skilled in the art. Any modification, equivalent replacement, improvement, etc. within the spirit and principle of the application shall be included in the scope of claims of the application.

Claims

1. A slurry and slag recognition method based on stereovision, characterized in that, The method comprises the following steps: S1, placing two same cameras stably above the pulp sludge to be identified, taking the left camera as a first camera and the right camera as a second camera; S2, synchronously adjusting the first camera and the second camera to make the two cameras at the same horizontal height, and making the right half of the imaging picture of the first camera coincide with the left half of the imaging picture of the second camera, and fixing the positions of the first camera and the second camera after the coincidence; S3, operating the first camera and the second camera to shoot a calibration board, processing the shooting image by a calibration software, outputting calibration parameters of the first camera and the second camera, and obtaining a baseline distance x of optical centers of the first camera and the second camera based on the calibration parameters; S4, determining an exposure time based on camera accuracy of a camera lens, pulp sludge movement speed and imaging magnification: ； wherein t represents the exposure time, p represents a blur circle diameter, v represents the pulp sludge movement speed, and K represents the imaging magnification; S5, operating the first camera and the second camera to shoot the pulp sludge at the same time, and obtaining a first image and a second image on imaging planes respectively; establishing a first rectangular coordinate system with a center point of the first image as an origin point O1, and then the optical center of the first camera is below the origin point of the coordinate system and is recorded as P; establishing a second rectangular coordinate system with a center point of the second image as an origin point O2, and then the optical center of the second camera is below the origin point of the coordinate system and is recorded as Q; for a point M of the pulp sludge in the coincident image, the following steps are executed: S51, set M in the first image and the second image point respectively corresponding to L, R, L point and R point coordinates are (X L ,0) and (X R ,0), based on (X L ,0) and (X R ,0) calculation M point in two images of the distance difference d=(X L -X R )×i, i indicates unit pixel distance; S52, calculate the imaging point spacing l between the first image and the second image of M based on the baseline distance x and the difference d between the sight distances LR = x - d; S53, the perpendicular lines from the M point to the two origin connecting lines and the two optical center connecting lines intersect the two origin connecting lines and the two optical center connecting lines at B and N, based on the similarity relationship of each triangle contained in the △MPQ and l LR , the distance from the M point to the P point and the Q point is solved, denoted as l MP and l MQ , the l MP and l MQ respectively represent the object distance from the M point to the first camera and the second camera; S6、based on l MP and l MQ A three-dimensional point cloud model of the slurry and the slag is established, and a shape, a volume and a surface area of the slurry and the slag are output based on the three-dimensional point cloud model.

2. The pulp sludge identification method based on stereovision according to claim 1, characterized in that, in S53, based on the fact that △MLR and △MPQ are contained in △MPQ, equation one is obtained by using △MLR △MPQ: ； , respectively denote the distance from point M to point B, the distance from point B to point N.

3. The pulp sludge identification method based on stereovision according to claim 2, characterized in that, in S53, based on the fact that △MLB and △MPN are contained in △MPQ, equation two is obtained by using △MLB △MPN: ； wherein , respectively represent the distance of point P to point N, the distance of point L to point B. based on the fact that △MBR and △MNQ are contained in △MPQ, equation three is obtained by using △MBR △MNQ: ； wherein , respectively represent the distance from point N to point Q, and the distance from point B to point R.

4. The pulp sludge identification method based on stereovision according to claim 3, characterized in that, based on the fact that △MBR, △MNQ, △MLB and △MPN are contained in △MPQ, equation four and equation five are obtained: ； 。 5. The pulp sludge identification method based on stereovision according to claim 4, characterized in that, according to a Gaussian imaging formula, assuming that a focal length is f, in △MPN contained in △MPQ, equation six is obtained: ； wherein represents the distance from point M to point P.

6. The pulp sludge identification method based on stereovision according to claim 5, characterized in that, based on the fact that right-angled triangles △MPN and △MQN are contained in △MPQ, equation seven and equation eight are obtained according to the Pythagorean theorem: ； 。

Images