{"id":10976,"date":"2022-11-01T01:33:17","date_gmt":"2022-11-01T01:33:17","guid":{"rendered":"https:\/\/www.szlaser.com\/?page_id=10976"},"modified":"2022-11-01T04:00:01","modified_gmt":"2022-11-01T04:00:01","slug":"optical-height-measurement-system","status":"publish","type":"page","link":"https:\/\/www.szlaser.com\/index.php\/wiki\/optical-height-measurement-system\/","title":{"rendered":"Optical Height Measurement System Based on Binocular Camera"},"content":{"rendered":"

Optical Height Measurement System Based on Binocular Camera<\/h1><\/h1><\/div>

According to the binocular stereo vision model, this paper builds a laboratory-specific height measurement system. By using the Matlab toolbox to calibrate the binocular camera to obtain many parameters required by the experiment, the paper establishes the relationship equation between the corrected world coordinates and the pixel coordinates through image correction technology. Under the programming environment of VS2015 combined with OpenCV, the stereo matching algorithm and the parallax theory are used to obtain the three-dimensional spatial information of the target surface. Finally, the space coordinates of the same target are obtained by selecting two characteristic points to achieve the purpose of height measurement. The experimental results show that the system can achieve 3D information measurement of real scenes well.<\/p>\n<\/div>

Key words<\/h3>\n

binocular stereo vision; camera calibration; disparity map; height measurement<\/p>\n<\/div>

Introduction<\/h3><\/h3><\/div>

Since ancient times, human beings have been trying to use tools to describe everything they see. With the continuous development of modern technology, new height measurement technologies are constantly being proposed. For colleges and universities, obtaining the size of surrounding objects more accurately and quickly in the laboratory is helpful to verify the experimental results in stages. Compared with ordinary measurement methods that are time-consuming, laborious and inefficient, binocular vision height measurement technology has Multi-target simultaneous detection, non-contact, high-precision, high-efficiency advantages.<\/p>\n

In the field of binocular vision measurement, many scholars at home and abroad have done a lot of research and practice. In 1966, the least squares method was first used by Hallert in the processing of calibrating observation data, and the application in the field stereo coordinate measuring instrument achieved breakthrough and accurate results. Zhou Kejie et al. introduced a 3D measurement system using a binocular camera combined with encoded structured light. The system can obtain a 3D point cloud matrix by moving the measuring instrument, and then complete the measurement of large curved surfaces. Although the current theoretical research on stereo vision technology has gradually improved, the requirements in different scenarios are also different, which puts forward higher requirements for the robustness and stability of the entire stereo vision system.<\/p>\n

In order to improve the accuracy of the obtained camera-related parameters, the calibration method of binocular vision is mainly improved, and the method of pre-calibration + single-target + dual-target calibration is pioneered for stereo calibration, and the experimental results are finally demonstrated. This method works.<\/p>\n<\/div>

1. The principle of binocular height measurement<\/h3><\/h3><\/div>

As an important branch of machine vision, binocular vision measurement technology uses a pair of binocular modules to simulate the visual mechanism of the human eye, so as to obtain two images of the same scene from two different perspectives, and then establish a corresponding mathematical model based on the principle of parallax. To restore the depth information in the scene, and finally use the similarity principle to calculate the spatial three-dimensional coordinates of the highest and lowest points on the same surface of the target object.<\/p>\n

The binocular stereo vision model is shown in Figure 1. In Figure 1, PL<\/sub> and PR<\/sub> are the projections of the point P (x, y, z) on the imaging planes of the left and right cameras respectively, OR<\/sub> and OL<\/sub> are the optical centers of the left and right cameras, n is the principal optical axis vector of the camera, and \u0192 is the focal length of the camera. . If the projection coordinates PL<\/sub> and PR<\/sub> of the spatial point P on the left and right image planes and the camera calibration parameters are known, the three-dimensional spatial coordinates of the point P can be calculated.<\/p>\n

\"\"

Fig.1 Binocular stereo vision model<\/p><\/div>\n<\/div>

1.1 Dual goal setting<\/h4><\/h4><\/div>

Binocular camera calibration requires the determination of the position of each feature point on the reference object relative to the world coordinate system (XW<\/sub>, YW<\/sub>, ZW<\/sub>), and the corresponding relationship between the image plane coordinates and the three-dimensional space coordinates. This mapping relationship is mainly determined by the binocular camera. , the external parameters are determined. In other words, the main purpose of binocular camera calibration is to obtain the internal and external parameters of the camera and the corresponding distortion coefficients, because this will directly affect the results of binocular measurement.<\/p>\n

The calibration method adopts the single-plane checkerboard method proposed by Professor Zhang Zhengyou, hereinafter referred to as Zhang’s calibration method, and the calibration board is shown in Figure 2. Zhang’s calibration method directly fuses the world coordinate system with the calibration plate plane, that is, ZW<\/sub> = 0, the homogeneous coordinates of any point P on the calibration plate plane are (XW<\/sub>, YW<\/sub>, 0, 1), which is expressed in the image pixel coordinate system as (u, \u03bd, 1 ), image physical coordinate system (x, y, 1 ), camera coordinate system (XC<\/sub>, YC<\/sub>, ZC<\/sub>), expressed in matrix form as:<\/p>\n

\"\"<\/p>\n<\/div>

Among them, M1<\/sub> represents the built-in parameter matrix of the camera, which is determined by fx<\/sub>, fy<\/sub>, u0<\/sub>, \u03bd0<\/sub>, and \u03b3, but \u03b3 is generally regarded as 0 and used. These five parameters are only related to the internal structure of the camera, and M2<\/sub> contains the rotation matrix MR<\/sub> and translation matrix MT<\/sub>. .<\/p>\n

\"\"

Fig. 2 Calibration board<\/p><\/div>\n

Since many parameters are obtained in the calibration stage, in order to make the calibration result more accurate, the method of pre-calibration + single-target calibration + dual-target calibration is creatively used.<\/p>\n

The pre-calibration is shown in Figure 3. In the pre-calibration stage, Matlab’s automatic calibration toolbox-Stereo Camera Calibrator was used to screen and calibrate the collected 24 sets of checkerboard images, and then the image pairs with large errors in the calibration process were eliminated according to the Mean Error Pixels index.<\/p>\n

\"\"

Fig. 3 Pre-calibration<\/p><\/div>\n

The single target setting is shown in Figure 4. The corner selection is shown in Figure 4(a). In the second stage, use the calib toolbox to manually calibrate the filtered left and right views respectively. This process mainly includes manually setting the corner points and automatically extracting the corner points, thereby obtaining two sets of parameters for single-object calibration, see Table 1.<\/p>\n

Table 1 Build-in parameters of left and right camera<\/strong><\/p>\n\n

\n \n \n\n \n \n \n \n\n\n wdt_ID<\/th> Attributes<\/th> Left Camera<\/th> Right Camera<\/th> <\/tr>\n<\/thead>\n \n\n \n \n\n \n \n \n
1<\/td>\n Focal Length<\/td>\n [1 230.788 42 1 244.18806] \u00b1 [2.853 18 2.978 42]<\/td>\n [1 247.161 89 1 234.775 46] \u00b1 [2.395 85 2.711 14]<\/td>\n <\/tr>\n
2<\/td>\n Main Point<\/td>\n [608.242 70 257.838 62] \u00b1 [4.342 02 4.447 98]<\/td>\n [589.279 35 272.085 24] \u00b1 [4.626 46 4.436 35]<\/td>\n <\/tr>\n
3<\/td>\n Distortion Coefficient<\/td>\n [0.253 87 -0.506 67 -0.052 99 -0.012 18 0.000 00] \u00b1 \n[0.075 60 0.582 88 0.009 54 0.003 38 0.000 00]\n<\/td>\n [0.239 34 -0.186 53 -0.059 75 -0.021 71 0.000 00]\u00b1\n[0.071 40 0.285 46 0.010 34 0.004 28 0.000 00] \n<\/td>\n <\/tr>\n <\/tbody> \n\n \n \n \n\n <\/table>\n\n<\/div>