CN115345924A - A Bend Pipe Measurement Method Based on Multi-camera Line Laser Scanning - Google Patents

Abstract

The invention provides a bent pipe measuring method based on multi-camera line laser scanning, which is used for automatically carrying out digital three-dimensional geometric parameter interpretation on a bent pipe. The method comprises the following steps: projecting line laser by using line laser projection equipment, and moving the projection equipment through a sliding rail to enable the line laser to traverse the surface of the bent pipe to be detected in a push-sweeping manner; meanwhile, the multi-view camera synchronously photographs the bent pipe from different angles, line laser center line pixels are extracted from multi-view images and three-dimensional reconstruction is carried out, high-precision three-dimensional point cloud of the bent pipe is obtained, and measuring parameters of the bent pipe are extracted from three-dimensional point cloud coordinates.

Description

A Bend Pipe Measurement Method Based on Multi-camera Line Laser Scanning

technical field

The invention relates to the field of machine vision industrial measurement, and specifically provides a multi-eye camera-based line laser scanning bending pipe measurement method.

Background technique

In the field of modern aerospace, high-precision bent pipes are key components of important equipment such as aerospace engines. The reliability of their assembly is very important in the product life cycle and will be affected by the quality of processing. To produce high-precision pipe bends, first of all, high-precision pipe-bend inspection equipment is required. After obtaining the high-precision parameters of the pipe bend, they are used for adjustment and calibration of the pipe bender and registration of pipe types.

Methods for pipe bend measurement can be divided into two categories: contact methods and non-contact methods. A common method is the tactile measurement method, which uses a mechanical mold or a coordinate measuring machine (CMM) to measure the bend. The method based on the mechanical mold requires that the mold be manufactured according to the shape of the pipe, and the machining accuracy of the mold should be high. Therefore, this method has low versatility, and can only classify the pipes with "high" or "low" accuracy according to the degree of fit between the pipe and the mold, and cannot provide the geometric error of the low-precision bend. In the CMM-based approach, the operator controls a probe at each sampling point on the surface of the line and arc. Then, the model fitting of the pipe bend is performed according to the sampling points, but this method is inefficient and consumes a lot of manpower.

The non-contact method is a vision-based method, and the vision-based method uses computer vision technology to take pictures of the surface of the elbow parts and obtain a digital image. Then, analyze the pixel value of the digital image on the computer to obtain the three-dimensional point information of the elbow, and further reconstruct the three-dimensional point cloud. However, the existing vision-based measurement methods have the problems of low efficiency and low precision of the reconstructed 3D model.

In summary, the accuracy of the contact measurement method is unstable, and the measurement results of different personnel are different for the same bend, and this method requires high manual experience. Among non-contact measurement methods, scanners and existing visual measurement methods have the problems of low precision and low efficiency. Therefore, it is necessary to design a method for quickly and accurately generating the bend pipe model and then measuring the bend pipe parameters.

Contents of the invention

The invention provides a method based on multi-eye camera line laser scanning to reconstruct the measurement parameters of the bent pipe, which solves the problems of low measurement accuracy and low efficiency in the related art.

This method can quickly extract the 3D point cloud of the bent pipe from the images collected by the line laser scanner, and obtain the geometric parameters of the bent pipe from the 3D point cloud.

Compared with the prior art, the present invention designs a line laser three-dimensional scanning step to measure the three-dimensional structure of the object to be measured, obtain a digital three-dimensional point cloud, and simultaneously obtain the geometric parameters of the bent pipe.

The technical scheme that the present invention adopts is as follows:

A method based on multi-camera line laser scanning, comprising:

(1) Use a special calibration board to calibrate the multi-eye camera, and the multi-eye camera collects the image information of the elbow;

(2) Use the line laser scanning method to extract the pixel coordinates of the line laser centerline from the curved pipe image, and perform centerline pixel matching between any two images;

(3) According to the pixel coordinates of the center line of the laser line, using the principle of triangulation, calculate the three-dimensional point coordinates;

(4) Generate a 3D point cloud based on the 3D point information, and calculate the geometric parameters of the elbow from the coordinate values of the 3D point cloud.

Further, a three-dimensional measurement device composed of a line laser projector and an industrial camera is constructed, the device performs three-dimensional surface model reconstruction, and the hardware of the device includes a projector device and an imaging camera;

Further, use the calibration board to calibrate the relative position and posture of the camera and the internal parameters of the image to obtain the calibration parameters; the internal parameters include the principal point coordinates, the principal distance length and the distortion parameters of the image of each camera, and the relative position and posture parameters include each The baseline distance value between two pairs of cameras, the relative position and relative rotation angle parameters of the camera's space coordinate system;

The calibration board is a regular calibration board in the field of machine vision. The geometric model of the camera can be obtained by taking pictures of the flat plate with a fixed-pitch pattern array by the camera and calculating by the calibration algorithm, so as to obtain high-precision measurement and reconstruction results. By processing the image of the calibration board through a nonlinear calibration algorithm, the internal parameters and relative position and attitude parameters of the camera can be calculated. In this case, the camera is used to take images of the calibration plate from different angles and positions for calibration, and the calibration parameters are calculated using a nonlinear calibration algorithm;

Further, the projector projects a high-brightness line laser, which irradiates the surface of the bent pipe under test. Due to the irradiation of the line laser, the surface of the bent pipe under test will form a bright laser line image. The change of the laser line on the image is consistent with the measured The surface shape change of the bending pipe is consistent in the spatial shape;

Further, moving the projector equipment through the slide rail makes the line laser projector move at a constant speed, so that the line laser push-broom scans the entire object, and the camera takes a picture every time the line laser projector moves a fixed distance. The slide rail moves the projector equipment at a constant speed, and the camera takes pictures at intervals of time. The multi-eye camera takes pictures synchronously at equal time intervals, records the image of the modulated line laser on the surface of each bend, and finally obtains a set of modulated line laser images;

Further, according to the line laser images obtained by different cameras, the center line of the line laser is extracted from the line laser image, and the Gaussian fitting curve method is used to extract the line laser center line, and the pixel value of the line laser on a row of pixels obeys the Gaussian distribution. Fit the pixel brightness value to a Gaussian curve, find the position of the maximum value and record its coordinates, which are the center positions of the line laser stripes, and then perform center line pixel matching between the two images;

Further, according to the calibration parameters of the camera and the pixel coordinate values of the line laser centerline, the triangulation spatial relationship is established, and the three-dimensional point cloud coordinates of the elbow in the physical world corresponding to the image pixels are calculated according to the principle of triangulation. From the coordinate value of the three-dimensional point cloud of the elbow, the measurement of the parameters of the elbow can be realized.

Among them, the specific steps for calculating the three-dimensional point cloud coordinates of the elbow are:

Using the relative position and attitude of the cameras obtained through calibration, the epipolar line correction is performed on the multi-eye images. After the epipolar line correction is done for the two images, the coordinates of the pixels of the line laser centerline on the two images on the two images are marked as (x _l , y _l ) and (x _r , y _r ), corresponding to the same physical point The pixels of the matched laser centerlines of the two lines only have pixel coordinate differences in the row direction, and the pixel coordinates of the matched pixels in the column direction are the same. On the epipolar-corrected image, the formula for calculating the distance value from the camera corresponding to each matching pixel point is:

Among them, f is the main distance length value of the camera obtained from calibration, d=x _l -x _r represents the parallax, x _l and x _r are the column coordinates of the pixels of the two images respectively, and B is the optical connection of the two images Baseline length value, B is the relative position and attitude parameter of the calibration result, and Z is the distance value of the target point of the matching pixel from the z-axis direction of the camera. Then the X coordinate and Y coordinate of the observed three-dimensional point are:

是图像的主点坐标，是标定结果的内参数。in,

is the principal point coordinate of the image, and is the internal parameter of the calibration result.

From a pair of line laser center lines of a pair of images captured by two cameras, a pair of pixel coordinates obtained by matching can calculate a three-dimensional point coordinate (X, Y, Z). The point-by-point matching calculation along the center line of the line laser can obtain the three-dimensional points on the line laser. By moving the line laser projector, multiple line lasers can be obtained, and the three-dimensional point cloud covering the surface of the curved pipe can be calculated successively.

The 3D point cloud accurately reflects the 3D geometry of the elbow, and the geometric measurement parameters of the elbow are directly interpreted from the coordinates of the 3D point cloud.

In summary, due to the adoption of the above technical solution, the beneficial effects of the present invention are: the present invention uses a method based on multi-eye camera line laser scanning to quickly and accurately calculate the three-dimensional model and geometric dimensions of the bend, and improve the parameters of the bend. The accuracy and efficiency of the calculation provide auxiliary technical means for the adjustment and calibration of the pipe bender, the registration and reverse of the pipe shape.

Description of drawings

Fig. 1 method data acquisition and processing flowchart of the present invention;

Fig. 2 Composition of line laser scanning equipment;

Figure 3 Line laser scanning bend pipe image;

Figure 4 Gaussian fitting curve method to extract the centerline of the line laser.

Detailed ways

In this embodiment, a method for measuring bent pipes based on multi-eye camera line laser scanning is provided. Fig. 1 is a flow chart of data acquisition and processing according to an embodiment of the present invention. As shown in Fig. 1, the process includes the following steps:

Step 1, build a curved pipe model reconstruction device composed of a line laser projector and an industrial camera, the hardware equipment of the device includes a line laser projection device and an imaging camera, as shown in Figure 2;

Calibrate the internal parameters and relative position and attitude parameters of the multi-camera, take pictures of the calibration board from different angles and distances, and obtain 12 calibration board images. Through the nonlinear calibration algorithm, the image of the calibration board is processed, and the relative position and attitude of the camera and the internal parameters of the camera are obtained for calibration. The internal parameters include the principal point coordinates, the principal distance length and the distortion parameters of the image of each camera, and the relative position and posture parameters include the baseline distance value between every two pairs of cameras, the relative position and relative rotation angle parameters of the camera's spatial coordinate system;

Step 2, the line laser projector projects a bright line laser, which irradiates the surface of the measured object. Due to the irradiation of the line laser, the surface of the measured object will form a bright laser line image. The change of the laser line on the image is consistent with the The surface shape change of the measured elbow is consistent in spatial shape, as shown in Figure 3;

Step 3. Move the projector equipment through the slide rail to make the line laser projector move at a constant speed, so that the line laser push-broom scans the entire object. Every time the line laser projector moves a fixed distance, the camera takes a photo. The slide rail moves the projector equipment at a constant speed, and the camera takes pictures at intervals of time. The above realizes that every time the projector moves a fixed distance, the camera takes a picture, records the image of the modulated line laser on the surface of each object, and finally obtains a set of modulated line laser images;

Step 4, according to the line laser centerline extraction algorithm, extract the centerline of each line laser image to obtain the pixel coordinates of the line laser centerline; in the process of online laser scanning image processing, the most crucial step is to align the line The centerline of the laser is extracted and its coordinates are recorded. Taking advantage of the fact that the pixel values of the line laser on a row of pixels obey the Gaussian distribution, the Gaussian fitting curve method is used to extract the center line of the line laser, as shown in Figure 4, and then the center line pixel matching is performed between the two images.

Step 5, according to the calibration parameters of the camera and the pixel value of the center line, the triangulation spatial relationship is established, and the three-dimensional point cloud coordinates of the elbow in the physical world corresponding to the image pixels are calculated according to the triangulation principle.

Using the relative position and attitude of the cameras obtained through calibration, the epipolar line correction is performed on the multi-eye images. After the epipolar line correction is done for the two images, the coordinates of the pixels of the line laser centerline on the two images on the two images are marked as (x _l , y _l ) and (x _r , y _r ), corresponding to the same physical point The pixels of the matched laser centerlines of the two lines only have pixel coordinate differences in the row direction, and the pixel coordinates of the matched pixels in the column direction are the same. On the epipolar-corrected image, the formula for calculating the distance value from the camera corresponding to each matching pixel point is:

Among them, f is the main distance length value of the camera obtained from calibration, d=x _l -x _r represents the parallax, x _l and x _r are the column coordinates of the pixels of the two images respectively, and B is the optical connection of the two images Baseline length value, B is the relative position and attitude parameter of the calibration result, and Z is the distance value of the target point of the matched pixel from the z-axis direction of the camera. Then the X coordinate and Y coordinate of the observed three-dimensional point are:

是图像的主点坐标，是标定结果的内参数。in,

is the principal point coordinate of the image, and is the internal parameter of the calibration result.

From a pair of line laser center lines of a pair of images captured by two cameras, a pair of pixel coordinates obtained by matching can calculate a three-dimensional point coordinate (X, Y, Z). The point-by-point matching calculation along the center line of the line laser can obtain the three-dimensional points on the line laser. By moving the line laser projector, multiple line lasers can be obtained, and the three-dimensional point cloud covering the surface of the curved pipe can be calculated successively.

From the coordinates of the 3D point cloud, the geometric parameters of the bent pipe can be measured.

Through the above steps, the three-dimensional point cloud of the curved pipe is generated according to the laser scanning technology of the line multi-eye camera, thereby avoiding the damage of the curved pipe in the prior art in the contact measurement, and the efficiency is low; the problem of the low efficiency of the non-contact measurement is proposed. The new bend measurement method provides a new solution for the bend detection industry.

Claims (4)

1. A bent pipe measuring method based on multi-view camera line laser scanning is characterized in that the bent pipe measuring method combining line laser scanning and multi-view camera shooting comprises the following steps:

(1) Constructing a bent pipe measuring device consisting of a line laser projector and an industrial camera, wherein hardware equipment of the device comprises line laser projection equipment and a multi-purpose camera;

(2) Calibrating internal parameters and relative position posture parameters of a camera, photographing a calibration plate at multiple angles and multiple positions, and calculating the internal parameters and the relative position posture of the camera by using a nonlinear calibration algorithm to obtain calibration parameters;

(3) The line laser projector projects a piece of high-brightness line laser to irradiate the surface of a measured object, the surface of the measured object can form a high-brightness laser line image due to the irradiation of the line laser, and the change of the laser line on the image is consistent with the change of the surface shape of the measured elbow on the spatial shape;

(4) Moving the projector equipment through a sliding rail to enable the linear laser projector to move at a constant speed, so that the linear laser pushes and scans the whole bent pipe, simultaneously, when the projector moves for a fixed distance, a camera takes a picture once, an image of modulating linear laser on the surface of an object every time is recorded, and finally a group of modulated linear laser images are obtained;

(5) Extracting pixels of a line laser central line of each line laser image according to a line laser center extraction algorithm to obtain pixel coordinates of the line laser central line, and performing central line pixel matching between every two images;

(6) According to the calibration parameters of the camera and the pixel coordinates of the line laser central line, three-dimensional reconstruction is carried out, a triangulation spatial relationship is established, according to a triangulation algorithm, the three-dimensional point cloud coordinates of the bent pipe in the physical world corresponding to the pixels are calculated, and the measurement parameters of the bent pipe are extracted from the three-dimensional point cloud.

2. The method according to claim 1, wherein the calibration board in step (2) is a regular calibration board in the field of machine vision, a checkerboard calibration board and a solid origin calibration board can be used, the internal parameters obtained by calibration include principal point coordinates, principal distance length and distortion parameters, the relative position and attitude parameters include a baseline distance value between each two pairs of cameras, and the relative position and relative rotation angle parameters of the spatial coordinate system of the cameras.

3. The elbow pipe measuring method based on the line laser scanning of the multi-view camera as claimed in claim 1, wherein the line laser in step (3) is projected onto the surface of the object by using a line laser projection device, the line laser is modulated by the surface of the object, and the modulated line laser image is formed by taking pictures from different angles by the multi-view camera.

4. The elbow measuring method based on line laser scanning of the multi-camera according to claim 1, wherein the camera in step (4) takes a picture, the camera takes a picture each time the line laser projector moves a fixed distance, the fixed distance is controlled by means of the uniform motion of the sliding rail and the taking of pictures by the camera at equal time intervals, and the projector and the camera are moved synchronously.

Images