Wide-field-of-view high-precision vision measurement system and method imitating human eye perception mode

By using a wide field-of-view high-precision visual measurement system that mimics human eye perception, combined with spherical and plane mirror binocular sensors and servo devices, the contradiction between field of view and accuracy in visual perception is resolved. This enables rapid capture of large-scale scenes and high-precision 3D measurement, improving measurement efficiency and reducing system complexity.

CN121540083APending Publication Date: 2026-02-17BEIHANG UNIV
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202511756473.9
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2025-11-27
Publication Date
2026-02-17

AI Technical Summary

Technical Problem

In existing visual perception technologies, it is difficult to achieve both a wide field of view and high precision. Traditional binocular systems have high computational complexity and limited field of view. Multi-view stitching has a large field of view, but its precision is affected by the stitching quality. Single-camera rotation efficiency is low, and single-camera mirror distortion is large and lacks an adaptive adjustment mechanism.

Method used

The wide-field-of-view high-precision visual measurement system adopts a human-eye-like perception mode. It achieves wide-field-of-view scanning and rapid target positioning through a spherical mirror binocular sensor, performs high-precision three-dimensional measurement using a plane mirror binocular sensor, and automatically adjusts the sensor attitude with a servo device to achieve automatic target acquisition and high-precision measurement.

Benefits of technology

It achieves a balance between field of view and accuracy, significantly improves measurement efficiency, reduces system size and cost, and enables fully automated high-precision 3D measurement.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN121540083A_ABST
    Figure CN121540083A_ABST
Patent Text Reader

Abstract

The invention belongs to the technical field of vision measurement, and particularly provides a wide-view-field high-precision vision measurement system and a wide-view-field high-precision vision measurement method simulating a human eye perception mode. The system comprises a single-camera spherical mirror binocular sensor, a single-camera plane mirror binocular sensor, a servo device, a data transmission line, a computer and a calibration device. The measurement method comprises a calibration stage and a measurement stage. In the calibration stage, imaging model parameters of the virtual camera, an essential matrix and structural parameters of the binocular sensor, structural parameters of the measurement system and structural parameters of the calibration device are calibrated. In the measuring stage, the single-camera spherical mirror binocular sensor is used for carrying out wide-view-field rough scanning, the servo device is guided to adjust the single-camera plane mirror binocular sensor so that a target object can be located in the view field of the single-camera plane mirror binocular sensor, an image is captured and transmitted to a computer for processing, and high-precision three-dimensional measurement is achieved. According to the invention, by constructing a measurement system in a human eye perception simulation mode, the contradiction between the field of view and the precision of a traditional vision measurement system is solved, a new intelligent self-adaptive measurement mode is established for wide-field-of-view high-precision three-dimensional vision measurement, and the method can be widely applied to the fields of automatic driving, robot navigation, industrial detection and the like.
Need to check novelty before this filing date? Find Prior Art

Description

TECHNICAL FIELD

[0001] The present application relates to the field of visual measurement technology, in particular to a wide field of view high precision visual measurement system and measurement method imitating human eye perception mode. BACKGROUND

[0002] In 2024, the "Central Enterprise Artificial Intelligence Special Promotion Conference" proposed the concept of AI empowering industry. Intelligent environmental perception, as the core technology in the fields of AI robots, unmanned aerial vehicles, and autonomous driving, is the key prerequisite for realizing high-precision positioning and navigation, collaborative perception, and intelligent operation. The mainstream environmental perception solutions such as laser radar have limitations such as high cost and large size, while pure visual perception methods have become an important development direction due to their compact structure and low cost. With the expansion of complex application scenarios such as on-orbit spacecraft monitoring and industrial precision detection, the requirements for visual measurement range and precision of the system are increasing, and wide field of view high precision real-time perception technology has become a key research direction.

[0003] Existing visual perception technology mainly faces the following limitations: binocular system has high computational complexity and limited field of view; multi-eye stitching has a large field of view but the precision is affected by the stitching quality; fisheye lens has large distortion and complex correction; single camera rotation is low in efficiency; single camera mirror has the potential of wide field of view and high precision, but lacks an adaptive adjustment mechanism. The human eye's "wide first, then precise" perception mode inspires us to first scan and locate the target with a wide field of view, and then focus on high-precision measurement, which can break through the inherent contradiction between field of view and precision. Based on this, the present application proposes a visual measurement system imitating human eye perception mode, which realizes the unification of fast capture and high-precision three-dimensional measurement in a large range of scenes, and provides a new idea for intelligent environmental perception. SUMMARY

[0004] The present application aims to solve the problem of wide field of view and high precision in existing technology, and provides a wide field of view high precision visual measurement system and measurement method imitating human eye perception mode. The system simulates the human eye visual perception mechanism, first locates the target quickly through wide field of view scanning, and then obtains the accurate three-dimensional information of the target through high precision measurement.

[0005] The technical solution of the present application is: a wide field of view high precision visual measurement system and measurement method imitating human eye perception mode, characterized by:

[0006] 1. The wide field of view high precision visual measurement system and measurement method imitating human eye perception mode provided by the present application comprises a spherical mirror binocular sensor (1), a planar mirror binocular sensor (2), a servo device (3), a data transmission line (4), a computer (5), and a calibration device (6).

[0007] The spherical mirror binocular sensor (1) is composed of an industrial camera A (7), a spherical mirror group (8) and a sensor shell (9), which is used for shooting the image of the target object (16) located in the field of view A (17), realizing wide field of view rough scanning, and transmitting the image into the computer (5) through the data transmission line (4) for processing to obtain the position of the target object (16).

[0008] The servo device (3) is composed of a deflection shaft (13), a pitch shaft (14) and a servo bracket (15), which adjusts the position of the plane mirror binocular sensor (2) according to the processing result of the computer (5), so that the target object (16) is located in the field of view B (18) of the plane mirror binocular sensor (2).

[0009] The plane mirror binocular sensor (2) is composed of an industrial camera B (10), a plane mirror group (11) and a sensor bracket (12), which shoots the image of the target object (16) and transmits the image into the computer (5) through the data transmission line (4), extracts and matches feature points, and realizes high-precision three-dimensional measurement of the target object (16) according to the binocular vision measurement principle.

[0010] 2. The present application also provides a measurement method for three-dimensional measurement using the above-mentioned system, which comprises a calibration stage and a measurement stage. The calibration stage is used for obtaining the imaging model parameters of each virtual camera, the essential matrix and the structure parameters of the binocular sensor, and calibrating the structure parameters of the measurement system and the structure parameters of the calibration device (6); the measurement stage uses the spherical mirror binocular sensor (1) to perform wide field of view rough scanning to obtain the position of the target object (16), adjusts the servo device (3) according to the position information, so that the target object (16) is located in the field of view B (18) of the plane mirror binocular sensor (2), and then uses the plane mirror binocular sensor (2) to perform high-precision three-dimensional measurement.

[0011] Compared with the prior art, the present application has the following beneficial effects:

[0012] 1. The unity of field of view and precision: the wide field of view scanning is realized by the spherical mirror binocular sensor to quickly position, and then the high-precision measurement is performed by the plane mirror binocular sensor, which solves the inherent contradiction between the field of view and the precision in the traditional vision system.

[0013] 2. The measurement efficiency is improved: the human eye perception mode is adopted, the wide field of view scanning is performed first, and then the high-precision measurement is performed, which significantly reduces the search time and improves the measurement efficiency.

[0014] 3. Compact structure and light weight: the binocular vision is realized by using a single camera combined with a mirror group, which avoids the complex structure of the traditional binocular system which needs two cameras, and reduces the system volume and cost.

[0015] 4. Full automation measurement: through servo device, the sensor posture is automatically adjusted, the target is automatically captured and high-precision measurement is realized, and manual intervention is reduced.

[0016] 5. High-precision three-dimensional measurement: through accurate calibration, the imaging model parameters of the virtual camera, the essential matrix and structure parameters of the binocular sensor, the structure parameters of the measurement system and the structure parameters of the calibration device are obtained, and high-precision three-dimensional measurement is realized in combination with a stereo matching algorithm. BRIEF DESCRIPTION OF DRAWINGS

[0017] Figure 1 It is a schematic diagram of a wide-field high-precision vision measurement system and a measurement method imitating a human eye perception mode.

[0018] Figure 2 It is a structural schematic diagram of a wide-field high-precision vision measurement system imitating a human eye perception mode in an embodiment of the present application.

[0019] Figure 3 It is a structural schematic diagram of a spherical mirror binocular sensor in an embodiment of the present application.

[0020] Figure 4 It is a structural schematic diagram of a planar mirror binocular sensor in an embodiment of the present application.

[0021] Figure 5 It is a structural schematic diagram of a servo device in an embodiment of the present application.

[0022] Figure 6 It is a structural schematic diagram of a calibration device in an embodiment of the present application.

[0023] Figure 7 It is a measurement model schematic diagram of a spherical mirror binocular sensor in an embodiment of the present application.

[0024] Figure 8 It is a measurement model schematic diagram of a planar mirror binocular sensor in an embodiment of the present application.

[0025] Figure 9 It is an initial state calibration model schematic diagram of a wide-field high-precision vision measurement system imitating a human eye perception mode in an embodiment of the present application.

[0026] Figure 10 It is a global calibration model and a vision servo principle schematic diagram of a wide-field high-precision vision measurement system imitating a human eye perception mode in an embodiment of the present application.

[0027] In the drawings,

[0028] 1: spherical mirror binocular sensor 2: plane mirror binocular sensor 3: servo device 4: data transmission line 5: computer 6: calibration device 7: industrial camera A 8: spherical mirror group 9: sensor shell 10: industrial camera B 11: plane mirror group 12: sensor support 13: deflection shaft 14: pitch shaft 15: servo support 16: target object 17: field of view A 18: field of view B 19: plane target A 20: plane target B 21: connecting piece 22: small spherical mirror 23: large spherical mirror 24: large plane mirror 25: small plane mirror 26: virtual camera A 27: virtual camera B 28: virtual camera C 29: virtual camera D DETAILED DESCRIPTION

[0029] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only some of the embodiments of the present application, rather than all the embodiments. The components of the embodiments of the present application described and shown in the drawings herein can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present application provided in the drawings is not intended to limit the scope of the claimed present application, but only represents selected embodiments of the present application.

[0030] The present application will be further described below. The present application provides a wide field of view high precision vision measurement system simulating human eye perception mode, calibrating imaging model parameters of virtual cameras, essential matrix and structure parameters of binocular sensors, structure parameters of the measurement system and structure parameters of the calibration device, and then realizing three-dimensional measurement. The schematic of the measurement system and the measurement method is shown in Figure 1 .

[0031] 1. The wide field of view high precision vision measurement system simulating human eye perception mode provided by the present application is composed of a spherical mirror binocular sensor (1), a plane mirror binocular sensor (2), a servo device (3), a data transmission line (4), a computer (5) and a calibration device (6), as shown in Figure 2 .

[0032] The spherical mirror binocular sensor (1) is used to shoot images of the target object (16) in the field of view A (17), to realize wide field of view rough scanning, and to transmit the images into the computer (5) through the data transmission line (4) for processing, to obtain the position of the target object (16).

[0033] The servo device (3) adjusts the position of the plane mirror binocular sensor (2) according to the processing result of the computer (5), so that the target object (16) is located in the field of view B (18) of the plane mirror binocular sensor (2).

[0034] The planar mirror binocular sensor (2) takes the image of the target object (16) and transmits the image into the computer (5) through the data transmission line (4). Through feature point extraction and matching, high-precision three-dimensional measurement of the target object (16) is realized according to the binocular vision measurement principle.

[0035] S1: The spherical mirror binocular sensor (1) is composed of an industrial camera A (7), a spherical mirror group (8), and a sensor shell (9), as shown in Figure 3 .

[0036] The spherical mirror group (8) is composed of two spherical mirrors with a common optical axis. The industrial camera A (7) is placed coaxially with the spherical mirror group (8), and the field of view faces the spherical mirror group (8). The spherical mirror inside the field of view of the industrial camera A (7) is defined as a small spherical mirror (22), and the spherical mirror outside the field of view is defined as a large spherical mirror (23). The optical center of the industrial camera A (7) is defined as C1, and the optical axis intersects the small spherical mirror (22) at point B. The industrial camera A (7) images through the small spherical mirror (22) and the large spherical mirror (23) as virtual camera A (26) and virtual camera B (27), respectively.

[0037] The sensor shell (9) is used to fix the industrial camera A (7) and the spherical mirror group (8), and is connected with the deflection shaft (13) of the servo device (3).

[0038] The geometric parameters of the spherical mirror binocular sensor (1) include: the mirror radius of the small spherical mirror (22) is 15-35 mm, and the base radius is 10-30 mm; the mirror radius of the large spherical mirror (23) is 65-105 mm, and the base radius is 30-50 mm; The length of the sensor shell (9) is 40-80 mm, the diameter is 80-120 mm, and the height is 90-130 mm. The optical parameters include: the lens focal length of the industrial camera A (7) is 6-10 mm, the field of view angle is 40-80 ° , and the working distance perpendicular to the optical axis direction is 40-200 mm.

[0039] S2: The planar mirror binocular sensor (2) is composed of an industrial camera B (10), a planar mirror group (11), and a sensor support (12), as shown in Figure 4 .

[0040] The planar mirror group (11) is composed of two planar mirrors with a common prism line. The planar mirror on the left side of the field of view of the industrial camera B (10) is defined as a large planar mirror (24), and the planar mirror on the right side of the field of view is defined as a small planar mirror (25). The optical center of the industrial camera B (10) is defined as C2, and the optical axis intersects the small planar mirror (25) at point D. The industrial camera B (10) images through the large planar mirror (24) and the small planar mirror (25) as virtual camera C (28) and virtual camera D (29), respectively. The industrial camera B (10) is imaged into virtual camera C (28) and virtual camera D (29) through the large plane mirror (24) and the small plane mirror (25) respectively.

[0041] The sensor support (12) is used for fixing the industrial camera B (10) and the plane mirror group (11) and is connected with the pitch axis (14) of the servo device (3).

[0042] The geometric parameters of the plane mirror binocular sensor (2) include that the thicknesses of the large plane mirror (24) and the small plane mirror (25) are 1-3 mm, the lengths in the direction parallel to the direction are 60-100 mm, the lengths in the direction perpendicular to the direction are 40-60 mm and 20-40 mm respectively, and the length of C2M is 10-50 mm, the angles between the large plane mirror (24) and the small plane mirror (25) and the plane where the large plane mirror (24) and the small plane mirror (25) are located are 100-130 ° and 30-50 ° respectively, and the diameter and the height of the sensor support (12) are 90-120 mm and 60-100 mm respectively. The optical parameters include that the focal length of the industrial camera B (10) is 2-5 mm, the field of view angle is 40-60°, and the working distance in the direction perpendicular to the measuring window is 25-150 mm.

[0043] S3: the servo device (3) is composed of a yaw axis (13), a pitch axis (14) and a servo support (15), as shown in Figure 5 The diameter and the height of the yaw axis (13) are 70-110 mm and 5-15 mm respectively, the diameter of the pitch axis (14) is 20-40 mm, the heights on both sides are 10-30 mm, the diameter and the height of the servo support (15) are 80-120 mm and 90-120 mm respectively.

[0044] The yaw axis (13) is used for connecting the spherical mirror binocular sensor (1) and the servo support (15) to realize the rotation of the sensor in the horizontal direction, and the pitch axis (14) is used for connecting the plane mirror binocular sensor (2) and the servo support (15) to realize the rotation of the sensor in the vertical direction.

[0045] S4: the calibration device (6) is a global calibration target composed of a plane target A (19) and a plane target B (20) connected through a connecting piece (21), as shown in Figure 6 There are 16-64 black solid circles arranged in a matrix on the two plane targets, the diameter is 2-8 mm, the center distance between adjacent circles is 4-10 mm, the accuracy is 0.01 mm, and the feature point is defined as the center of the black solid circle.

[0046] 2. This invention provides a measurement method for three-dimensional measurement of a target object (16) using a wide field-of-view high-precision visual measurement system that mimics human eye perception, comprising the following steps:

[0047] Calibration stage: The spherical mirror binocular sensor (1) and the plane mirror binocular sensor (2) are calibrated to obtain the imaging model parameters of each virtual camera, the essential matrix and structural parameters of the binocular sensor, and the structural parameters of the measurement system and the calibration device (6) are calibrated.

[0048] Measurement stage: Use the spherical binocular sensor (1) to perform a wide field of view coarse scan to obtain the position of the target object (16); adjust the servo device (3) according to the position information so that the target object (16) is located within the field of view B (18) of the plane mirror binocular sensor (2); use the plane mirror binocular sensor (2) to perform high-precision three-dimensional measurement.

[0049] The three-dimensional coordinate system of the wide field-of-view high-precision visual measurement system that mimics human eye perception modes includes: defining the global coordinate system O-xyz as the three-dimensional coordinate system of the servo device (3); and defining the three-dimensional coordinate system of the planar target A (19) as O. b1 -x b1 y b1 z b1 The three-dimensional coordinate system of the planar target B(20) is O. b2 -x b2 y b2 z b2 The three-dimensional coordinate system of industrial camera A(7) is O. c1 -x c1 y c1 z c1 The three-dimensional coordinate system of industrial camera B(10) is O. c2 -x c2 y c2 z c2 The three-dimensional coordinate system of virtual camera A(26) is O c3 -x c3 y c3 z c3 The three-dimensional coordinate system of virtual camera B(27) is O. c4 -x c4 y c4 z c4 The three-dimensional coordinate system of the virtual camera C(28) is O c5 -x c5 y c5 z c5 The three-dimensional coordinate system of the virtual camera D(29) is O. c6 -x c6 y c6 z c6 .

[0050] The structural parameters of the spherical mirror binocular sensor (1) include: O c4 -x c4 y c4 z c4 To O c3 -x c3 y c3 z c3 The transformation is represented by a 3×3 rotation matrix R3 and a 3×1 translation vector T3.

[0051] The structural parameters of the plane mirror binocular sensor (2) include: O c6 -x c6 y c6 z c6 To O c5 -x c5 y c5 z c5 The transformation is represented by a 3×3 rotation matrix R4 and a 3×1 translation vector T4.

[0052] The structural parameters of the calibration device (6) include: O b2 -x b2 y b2 z b2 To O b1 -x b1 y b1 z b1 The transformation is performed using a 3×3 rotation matrix R. b and a 3×1 translation vector T b express.

[0053] The structural parameters of the wide-field-of-view, high-precision visual measurement system that mimics human eye perception include:

[0054] Initial position O c2 -x c2 y c2 z c2 To O c1 -x c1 y c1 z c1 The transformation is represented by the rotation matrix R0 and the translation vector T0;

[0055] After the spatial position of the plane mirror binocular sensor (2) changes, O c2 -x c2 y c2 z c2 To O c1 -x c1 y c1 z c1 The transformation is represented by the rotation matrix R′0 and the translation vector T′0;

[0056] O c1 -x c1 y c1 z c1 The transformation to O-xyz is represented by the rotation matrix R1 and the translation vector T1;

[0057] O c2 -x c2 y c2 z c2 The transformation to O-xyz is represented by the rotation matrix R2 and the translation vector T2;

[0058] After the spatial position of the plane mirror binocular sensor (2) changes, O c2 -x c2 y c2 z c2 The transformation to O-xyz is represented by the rotation matrix R′2 and the translation vector T′2;

[0059] O c3 -x c3 y c3 z c3 To O c1 -x c1 y c1 z c1 The transformation is represented by rotation matrix R5 and translation vector T5;

[0060] O c5 -x c5 y c5 z c5 To O c2 -x c2 y c2 z c2 The transformation is represented by the rotation matrix R6 and the translation vector T6;

[0061] O b1 -x b1 y b1 z b1 To O c3 -x c3 y c3 z c3 The transformation is performed using the rotation matrix R. b1 Translation vector T b1 express;

[0062] O b2 -x b2 y b2 z b2 To O c5 -x c5 y c5 z c5 The transformation is performed using the rotation matrix R.b2 Translation vector T b2 It represents; where the rotation matrix is ​​a 3×3 identity orthogonal matrix and the translation vector is a 3×1 vector.

[0063] The calibration phase includes the following steps:

[0064] Step S101: Place the planar target A (19) and the planar target B (20) within the field of view A (17) of the spherical binocular sensor (1) and the field of view B (18) of the planar binocular sensor (2), respectively. Move the calibration device (6) freely to at least three positions. Each time it moves to a position, each sensor takes an image. The image taken by the spherical binocular sensor (1) is called the spherical mirror group calibration image, and the image taken by the planar binocular sensor (2) is called the planar mirror group calibration image.

[0065] like Figure 7 and Figure 8 As shown, measurement models of a spherical mirror binocular sensor (1) and a plane mirror binocular sensor (2) are presented respectively; with the optical center of the camera as the origin O. ci , z ci The axis coincides with the optical axis and points towards the subject being photographed, x ci axis, y ci The axes are parallel to the horizontal and vertical axes of the image plane, respectively. A virtual camera coordinate system and a camera coordinate system O are established according to the right-hand rule. ci -x ci y ci z ci .

[0066] Step S102: Using the spherical mirror group to calibrate the image, the imaging model parameters of virtual camera A (26) and virtual camera B (27) in the spherical mirror binocular sensor (1) are calibrated respectively, as well as the essential matrix and structural parameters of the spherical mirror binocular sensor (1). The specific steps are as follows:

[0067] The first step is to split the inner and outer parts of the spherical mirror group calibration image into calibration image A and calibration image B, which correspond to virtual camera A (26) and virtual camera B (27) respectively; extract the image coordinates of the feature points of calibration images A and B respectively, and calibrate the imaging model parameters of virtual camera A (26) and virtual camera B (27) in the spherical mirror binocular sensor (1), including equivalent focal length, principal point coordinates, lens and spherical mirror distortion coefficients;

[0068] The second step is to use the lens and spherical mirror distortion coefficients to correct and calibrate images A and B, and obtain distortion-free calibration images A and B. The image coordinates of the feature points of the distortion-free calibration images A and B are extracted respectively. The feature points of the two images are matched and then matched with the world coordinates of the feature points of the planar target A (19) to calibrate the essential matrix E1 = R3S1 and the structural parameters (R3|T3) of the spherical mirror binocular sensor (1), where S1 is an antisymmetric matrix composed of elements in T3.

[0069] Step S103: Using the plane mirror group to calibrate the image, calibrate the imaging model parameters of the virtual camera C (28) and the virtual camera D (29) in the plane mirror binocular sensor (2), as well as the essential matrix and structural parameters of the plane mirror binocular sensor (2). The specific steps are as follows:

[0070] The first step is to split the left and right parts of the calibration image of the plane mirror group into calibration image C and calibration image D, which correspond to virtual camera C (28) and virtual camera D (29) respectively; extract the image coordinates of the feature points of calibration images C and D respectively, and calibrate the imaging model parameters of virtual camera C (28) and virtual camera D (29) in the plane mirror binocular sensor (2), including equivalent focal length, principal point coordinates and lens distortion coefficient;

[0071] The second step is to use the lens distortion coefficient to correct and calibrate images C and D, and obtain distortion-free calibration images C and D. The image coordinates of the feature points of the distortion-free calibration images C and D are extracted respectively. The feature points of the two images are matched and then matched with the world coordinates of the feature points of the planar target B (20) to calibrate the essential matrix E2 = R4S2 and the structural parameters (R4|T4) of the planar mirror binocular sensor (2), where S2 is an antisymmetric matrix composed of elements in T4.

[0072] The description method for the equivalent focal length, principal point coordinates, and radial and tangential distortion coefficients of the lens obtained by calibration of virtual cameras A(26), B(27), C(28), and D(29) can be found in the paper Z. Zhang, “A flexible new technique for camera calibration”, in IEEE Transactions on Pattern Analysis and Machine Intelligence, vol.22, no.11, pp.1330-1334, Nov.2000;

[0073] Similar to lens distortion, the distortion of spherical mirrors can also be considered as consisting of radial and tangential distortion. The description method for calibrating the radial and tangential distortion coefficients of small spherical mirrors (22) and large spherical mirrors (23) can be found in the paper Y. Chen, F. Zhou, et al, “A Novel Mirrored Binocular Vision Sensor Based on Spherical Catadioptric Mirrors”, in IEEE Sensors Journal, vol.21, no.17, pp.18670-18681, Sept.2021;

[0074] The description method of the essential matrix and structural parameters of the spherical mirror binocular sensor (1) and the plane mirror binocular sensor (2) obtained by calibration can be found in the paper Y.Cui, F.Zhou, et al, “Precise calibration of binocular vision system used for vision measurement”, in Optics express, vol.22, no.18, pp.9134-49, Apr.2014.

[0075] Step S104: Using the distortion-free calibration images A, B and C, D, and the fixed positional relationship of the two planar targets in the calibration device (6), calibrate the structural parameters of the measurement system and the structural parameters of the calibration device (6). The specific steps are as follows:

[0076] The first step is to use the image coordinates of the feature points of the distortion-free calibration image A extracted in step S102 to calibrate the structural parameters (R) of the measurement system. b1 |T b1 (R5|T5);

[0077] The second step is to use the image coordinates of the feature points of the distortion-free calibration image C extracted in step S103 to calibrate the structural parameters (R) of the measurement system. b2 |T b2 ) and (R6|T6);

[0078] The third step is to calibrate the structural parameters (R) of the calibration device (6) by utilizing the fixed positional relationship between the two planar targets in the calibration device (6). b |T b );

[0079] The fourth step is to calibrate the structural parameters (R0|T0) of the measurement system using the structural parameters of the measurement system and the calibration device (6) that have been calibrated in the first three steps.

[0080] Fifth step: Based on the geometric positional relationship between the spherical binocular sensor (1) and the servo device (3), calibrate the structural parameters (R1|T1) of the measurement system.

[0081] like Figure 9 As shown, establish the initial position O. c2 -x c2 y c2 z c2 To O c1 -x c1 y c1 z c1 Coordinate transformation model:

[0082]

[0083] Where the coordinate transformation matrix The form of the coordinate transformation matrix in the following steps is the same as that described above; These are spatial feature points P in O. c1 -x c1 y c1 z c1 O c2 -x c2 y c2 z c2 Homogeneous coordinates.

[0084] like Figure 10 As shown, establish O c1 -x c1 y c1 z c1 Coordinate transformation model to O-xyz:

[0085]

[0086] in Let P be the homogeneous coordinates of the spatial feature point P in the O-xyz coordinate system.

[0087] The measurement phase includes the following steps:

[0088] Step S201: The spherical mirror binocular sensor (1) performs a wide field-of-view coarse scan of the object in the field of view A (17) and captures an image of the target object (16), which is called the spherical mirror group measurement image; the lens and spherical mirror distortion coefficients obtained in step S102 are used to correct the spherical mirror group measurement image to obtain distortion-free measurement images A and B.

[0089] Step S202: Extract the image coordinates of feature points in distortion-free measurement images A and B respectively. Using the essential matrix of the stereo sensor obtained in step S102, establish epipolar constraints to perform stereo matching of corresponding image feature points. Using the imaging model parameters of the virtual camera and the structural parameters of the stereo sensor obtained in step S102, calculate the feature points in O. c3 -x c3 y c3 z c3 The three-dimensional coordinates below.

[0090] Step S203: Using the structural parameters of the measurement system obtained in step S104, calculate the three-dimensional coordinates of the feature points in O-xyz, and guide the servo device (3) to adjust the pitch and yaw angles so that the target object (16) is located within the field of view B (18) of the plane mirror binocular sensor (2). The specific steps are as follows:

[0091] First, based on the structural parameters (R5|T5) of the measurement system obtained in step S104, calculate the feature points at O. c1 -x c1 y c1 z c1 The three-dimensional coordinates below;

[0092] The second step involves using the structural parameters (R1|T1) of the measurement system obtained in step S104 to calculate the three-dimensional coordinates of the feature point in the O-xyz region, thereby obtaining the pitch angle θ1 and yaw angle of the feature point in the O-xyz region.

[0093] The third step is to define the center point of the measurement window of the sensor bracket (12) of the plane mirror binocular sensor (2) as Q, and obtain the position of point Q from O based on the geometric position relationship. c2 -x c2 y c2 z c2 The three-dimensional coordinates below;

[0094] Fourth, using the structural parameters (R0|T0) and (R1|T1) of the measurement system obtained in step S104, the structural parameter (R2|T2) of the measurement system is obtained. The three-dimensional coordinates of point Q in O-xyz are calculated, and the pitch angle θ2 and yaw angle of point Q in O-xyz are obtained.

[0095] Fifth step, based on pitch angle θ1 and yaw angle Pitch angle θ2, yaw angle The servo device (3) is instructed to adjust the pitch and yaw angles so that... Pointing to the spatial feature point, the target object (16) is located within the field of view B (18) of the plane mirror binocular sensor (2).

[0096] like Figure 10 As shown, the visual servo principle of the measurement system is illustrated, and the O-axis is established. c2 -x c2 y c2 z c2 Coordinate transformation model to O-xyz:

[0097]

[0098] Where N2 = N1N0.

[0099] Define pitch angles θ1 and θ2 as positive in space where y>0, and as negative in space where y<0. The pitch angle θ of the servo device (3) rotates in the positive z-axis direction and the yaw angle in the counterclockwise direction in the O-xy plane. for:

[0100]

[0101] Step S204: The plane mirror binocular sensor (2) after the spatial position change captures an image of the target object (16), which is called the plane mirror group measurement image; the plane mirror group measurement image is corrected using the lens distortion coefficient obtained in step S103 to obtain distortion-free measurement images C and D, and the image coordinates of the feature points of the distortion-free measurement images C and D are extracted respectively. Referring to step S202, the coordinates of the feature points at O ​​are calculated. c5 -x c5 y c5 z c5 The three-dimensional coordinates below.

[0102] Step S205: Using the structural parameters of the measurement system obtained in step S104, and the pitch and yaw angles obtained in step S203, calculate the three-dimensional coordinates of the feature points in O-xyz after the spatial position of the plane mirror binocular sensor (2) changes, and realize the high-precision three-dimensional measurement of the target object (16). The specific steps are as follows:

[0103] First, based on the structural parameters (R6|T6) of the measurement system obtained in step S104, calculate the feature points at O. c2 -x c2 y c2 z c2 The three-dimensional coordinates below;

[0104] In the second step, based on the structural parameters (R0|T0) and (R1|T1) of the measurement system obtained in step S104, and the pitch and yaw angles of the servo device (3) obtained in step S203, the structural parameters (R′0|T′0) and (R′2|T′2) of the measurement system after the spatial position of the plane mirror binocular sensor (2) changes are calculated, and the three-dimensional coordinates of the feature points in O-xyz are calculated to achieve high-precision three-dimensional measurement of the target object (16).

[0105] After the spatial position of the plane mirror binocular sensor (2) changes, O c2 -x c2 y c2 z c2 To O c1 -x c1 y c1 z c1 The rotation matrix and translation vector are transformed from (R0|T0) to (R′0|T′0):

[0106] N′0=KN0 (7)

[0107] Where K is the value derived from θ, The transformation matrix formed.

[0108] O c2 -x c2 y c2 z c2 The coordinate transformation matrix N2 to O-xyz becomes N′2=N1N′0, and the global measurement model is:

[0109]

[0110] The above are merely embodiments of the present invention and are not intended to limit the scope of protection of the present invention. For those skilled in the art, the present invention can have various modifications and variations. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A wide field-of-view high-precision visual measurement system with a human eye perception mode, characterized in that it includes a spherical mirror binocular sensor (1), a plane mirror binocular sensor (2), a servo device (3), a data transmission line (4), a computer (5), and a calibration device (6); The spherical binocular sensor (1) is used to capture images of the target object (16) located within the field of view A (17), realize a wide field of view coarse scanning, and transmit the images to the computer (5) for processing through the data transmission line (4) to obtain the position of the target object (16); the servo device (3) adjusts the position of the plane mirror binocular sensor (2) according to the processing result of the computer (5) so that the target object (16) is located within the field of view B (18) of the plane mirror binocular sensor (2); the plane mirror binocular sensor (2) captures images of the target object (16) and transmits the images to the computer (5) through the data transmission line (4), and realizes high-precision three-dimensional measurement of the target object (16) through feature point extraction and matching based on the binocular vision measurement principle; 1.

1. The wide field-of-view high-precision visual measurement system based on the human eye perception mode is characterized in that the spherical binocular sensor (1) is composed of an industrial camera A (7), a spherical lens group (8), and a sensor housing (9); the spherical lens group (8) is composed of two stacked spherical mirrors with a common optical axis, the industrial camera A (7) and the spherical lens group (8) are placed on the same optical axis, and the field of view faces the spherical lens group (8); the sensor housing (9) is used to fix the industrial camera A (7) and the spherical lens group (8), and is connected to the servo device (3); 1.

2. According to the wide field of view high-precision visual measurement system based on the human eye perception mode, the spherical mirror group (8) is characterized in that the spherical mirror located inside the field of view of the industrial camera A (7) is a small spherical mirror (22), and the spherical mirror outside the field of view is a large spherical mirror (23); the optical center of the industrial camera A (7) is C1, and the optical axis intersects the small spherical mirror (22) at point B; the industrial camera A (7) is imaged by the small spherical mirror (22) and the large spherical mirror (23) to form virtual camera A (26) and virtual camera B (27) respectively; 1.

3. The wide field-of-view high-precision visual measurement system based on the human eye perception mode is characterized in that the small spherical mirror (22) has a mirror radius of 15-35 mm and a bottom radius of 10-30 mm; the large spherical mirror (23) has a mirror radius of 65-105 mm and a bottom radius of 30-50 mm. The length is 40-80mm; the diameter of the sensor housing (9) is 80-120mm, and the height is 90-130mm; the focal length of the lens of the industrial camera A (7) is 6-10mm, and the field of view is 40-80°. ° The working distance perpendicular to the optical axis is 40–200 mm; 1.

4. The wide field-of-view high-precision visual measurement system based on the human eye perception mode is characterized in that the planar mirror binocular sensor (2) consists of an industrial camera B (10), a planar mirror group (11), and a sensor bracket (12); the planar mirror group (11) consists of a common ridge line. The sensor bracket (12) is used to fix the industrial camera B (10) and the plane mirror group (11) and is connected to the servo device (3); 1.

5. The wide field-of-view high-precision visual measurement system based on the human eye perception mode is characterized in that, in the plane mirror group (11), the plane mirror located to the left of the field of view of the industrial camera B (10) is a large plane mirror (24), and the plane mirror to the right of the field of view is a small plane mirror (25); the optical center of the industrial camera B (10) is C2, and the optical axis is perpendicular to the optical axis of the industrial camera B (10). They intersect at point M; the industrial camera B (10) is imaged by the large plane mirror (24) and the small plane mirror (25) to form virtual camera C (28) and virtual camera D (29) respectively; 1.

6. The wide field-of-view high-precision visual measurement system based on the human eye perception mode is characterized in that the thickness of the large plane mirror (24) and the small plane mirror (25) is 1-3 mm, and they are parallel to the... The lengths in both directions are 60–100 mm, perpendicular to The lengths in the directions are 40–60 mm and 20–40 mm, respectively; The length is 10-50mm, and the angles between it and the planes containing the large plane mirror (24) and the small plane mirror (25) are 100-130° respectively. ° and 30-50 ° The sensor bracket (12) has a diameter of 90-120mm and a height of 60-100mm; the industrial camera B (10) has a lens focal length of 2-5mm and a field of view of 40-60 degrees. ° The working distance perpendicular to the measuring window direction is 25–150 mm; 1.

7. The wide field-of-view high-precision visual measurement system based on the human eye perception mode is characterized in that the servo device (3) is composed of a deflection shaft (13), a pitch shaft (14), and a servo bracket (15); the deflection shaft (13) has a diameter of 70-110 mm and a height of 5-15 mm; the pitch shaft (14) has a diameter of 20-40 mm and a height of 10-30 mm on both sides; the servo bracket (15) has a diameter of 80-120 mm and a height of 90-120 mm. 1.

8. The wide-field-of-view, high-precision visual measurement system based on the human eye-like perception mode, characterized in that, The calibration device (6) is a global calibration target consisting of planar target A (19) and planar target B (20) connected by connector (21); there are 16 to 64 black solid circles arranged in a matrix on the two planar targets, with a diameter of 2 to 8 mm, a center-to-center distance of 4 to 10 mm between adjacent circles, and an accuracy of 0.01 mm; the feature point is defined as the center of the black solid circle.

2. A measurement method for performing three-dimensional measurement of a target object (16) using a wide field-of-view high-precision visual measurement system with a human-eye-like perception mode as described in claim 1, characterized in that, Includes the following steps: Calibration stage: The spherical mirror binocular sensor (1) and the plane mirror binocular sensor (2) are calibrated to obtain the imaging model parameters of each virtual camera, the essential matrix and structural parameters of the binocular sensor, and the structural parameters of the measurement system and the calibration device (6) are calibrated. Measurement stage: Use the spherical mirror binocular sensor (1) to perform a wide field of view coarse scan to obtain the position of the target object (16); adjust the servo device (3) according to the position information so that the target object (16) is located within the field of view B (18) of the plane mirror binocular sensor (2); use the plane mirror binocular sensor (2) to perform high-precision three-dimensional measurement; 2.

1. According to the aforementioned measurement method, the characteristic is that, The three-dimensional coordinate system of the wide field-of-view high-precision visual measurement system that mimics the human eye's perception mode includes: The global coordinate system O-xyz is defined as the three-dimensional coordinate system of the servo device (3); the three-dimensional coordinate system of the planar target A (19) is O b1 -x b1 y b1 z b1 The three-dimensional coordinate system of the planar target B(20) is O. b2 -x b2 y b2 z b2 The three-dimensional coordinate system of industrial camera A(7) is O. c1 -x c1 y c1 z c1 The three-dimensional coordinate system of industrial camera B(10) is O. c2 -x c2 y c2 z c2 The three-dimensional coordinate system of virtual camera A(26) is O c3 -x c3 y c3 z c3 The three-dimensional coordinate system of virtual camera B(27) is O. c4 -x c4 y c4 z c4 The three-dimensional coordinate system of the virtual camera C(28) is O c5 -x c5 y c5 z c5 The three-dimensional coordinate system of the virtual camera D(29) is O. c6 -x c6 y c6 z c6 ; 2.

2. According to the measurement method described above, the characteristic is that, The structural parameters of the spherical mirror binocular sensor (1) include: O c4 -x c4 y c4 z c4 To O c3 -x c3 y c3 z c3 The transformation is represented by a 3×3 rotation matrix R3 and a 3×1 translation vector T3; 2.

3. According to the aforementioned measurement method, the characteristic is that, The structural parameters of the plane mirror binocular sensor (2) include: O c6 -x c6 y c6 z c6 To O c5 -x c5 y c5 z c5 The transformation is represented by a 3×3 rotation matrix R4 and a 3×1 translation vector T4; 2.

4. According to the aforementioned measurement method, the characteristic is that, The structural parameters of the calibration device (6) include: O b2 -x b2 y b2 z b2 To O b1 -x b1 y b1 z b1 The transformation is performed using a 3×3 rotation matrix R. b and a 3×1 translation vector T b express; 2.

5. The measurement method described above, characterized in that, The structural parameters of the wide field-of-view high-precision visual measurement system that mimics the human eye's perception mode include: Initial position O c2 -x c2 y c2 z c2 To O c1 -x c1 y c1 z c1 The transformation is represented by the rotation matrix R0 and the translation vector T0; After the spatial position of the plane mirror binocular sensor (2) changes, O c2 -x c2 y c2 z c2 To O c1 -x c1 y c1 z c1 The transformation is performed using the rotation matrix R. ′ 0 and translation vector T ′ 0 represents; O c1 -x c1 y c1 z c1 The transformation to O-xyz is represented by the rotation matrix R1 and the translation vector T1; O c2 -x c2 y c2 z c2 The transformation to O-xyz is represented by the rotation matrix R2 and the translation vector T2; After the spatial position of the plane mirror binocular sensor (2) changes, O c2 -x c2 y c2 z c2 The transformation to O-xyz is achieved using the rotation matrix R. ′ 2. Translation vector T ′ 2 represents; O c3 -x c3 y c3 z c3 To O c1 -x c1 y c1 z c1 The transformation is represented by rotation matrix R5 and translation vector T5; O c5 -x c5 y c5 z c5 To O c2 -x c2 y c2 z c2 The transformation is represented by the rotation matrix R6 and the translation vector T6; O b1 -x b1 y b1 z b1 To O c3 -x c3 y c3 z c3 The transformation is performed using the rotation matrix R. b1 Translation vector T b1 express; O b2 -x b2 y b2 z b2 To O c5 -x c5 y c5 z c5 The transformation is performed using the rotation matrix R. b2 Translation vector T b2 It is represented as follows: where the rotation matrix is ​​a 3×3 identity orthogonal matrix, and the translation vector is a 3×1 vector. 2.

6. The measurement method described above, characterized in that, The calibration phase includes the following steps: Step S101: Place planar target A (19) and planar target B (20) in the field of view A (17) of the spherical binocular sensor (1) and the field of view B (18) of the planar binocular sensor (2), respectively. Move the calibration device (6) freely to at least three positions. Each time it moves to a position, each sensor takes an image. The image taken by the spherical binocular sensor (1) is called the spherical mirror group calibration image, and the image taken by the planar binocular sensor (2) is called the planar mirror group calibration image. Step S102: Use the spherical mirror group to calibrate the image, and calibrate the imaging model parameters of the virtual camera A (26) and virtual camera B (27) in the spherical mirror binocular sensor (1), including the equivalent focal length, principal point coordinates, lens and spherical mirror distortion coefficients, as well as the essential matrix and structural parameters of the spherical mirror binocular sensor (1); Step S103: Use the plane mirror group to calibrate the image, and calibrate the imaging model parameters of the virtual camera C (28) and virtual camera D (29) in the plane mirror binocular sensor (2), including the equivalent focal length, principal point coordinates, lens distortion coefficient, and the essential matrix and structural parameters of the plane mirror binocular sensor (2); Step S104: Using the spherical mirror group calibration image and the plane mirror group calibration image, as well as the fixed positional relationship of the two planar targets in the calibration device (6), calibrate the structural parameters of the measurement system and the structural parameters of the calibration device (6); 2.

7. The measurement method described above is characterized in that, The measurement phase includes the following steps: Step S201: The spherical binocular sensor (1) performs a wide field-of-view coarse scan of the object within the field of view A (17) and captures an image of the target object (16), which is called the spherical lens group measurement image. Step S202: Extract the image coordinates of the feature points in the spherical mirror group measurement image. Using the imaging model parameters of the virtual camera, the essential matrix and structural parameters of the spherical mirror binocular sensor (1) obtained in step S102, calculate the feature points in O c3 -x c3 y c3 z c3 The three-dimensional coordinates below; Step S203: Using the structural parameters of the measurement system obtained in step S104, calculate the three-dimensional coordinates of the feature points in O-xyz, and guide the servo device (3) to adjust the pitch angle and yaw angle so that the target object (16) is located within the field of view B (18) of the plane mirror binocular sensor (2); Step S204: The plane mirror binocular sensor (2) after the spatial position change captures an image of the target object (16), which is called the plane mirror group measurement image; the image coordinates of the feature points in the plane mirror group measurement image are extracted, and the imaging model parameters of the virtual camera, the essential matrix and structural parameters of the plane mirror binocular sensor (2) obtained in step S103 are used to calculate the feature points in O c5 -x c5 y c5 z c5 The three-dimensional coordinates below; Step S205: Using the structural parameters of the measurement system obtained in step S104 and the pitch and yaw angles obtained in step S203, calculate the three-dimensional coordinates of the feature points in O-xyz after the spatial position of the plane mirror binocular sensor (2) changes, and realize the high-precision three-dimensional measurement of the target object (16).