Search

CN-122015693-A - Three-dimensional morphology measurement method for cross-shaped transparent mechanism imaging

CN122015693ACN 122015693 ACN122015693 ACN 122015693ACN-122015693-A

Abstract

A three-dimensional morphology measuring method for imaging a cross-shaped transparent mechanism comprises the steps of respectively arranging a binocular camera and a calibration plate on two sides of the shaped transparent mechanism in a calibration stage, collecting images of the calibration plate through the shaped transparent mechanism, thereby establishing a binocular refraction model containing characteristics of the shaped transparent mechanism, realizing calibration through a single image calibration method, and processing real-time collected images of an object to be measured through the calibrated binocular refraction model in an on-line stage to obtain the three-dimensional morphology. Aiming at the defect that the measurement accuracy is low because the calibration process is not suitable for on-machine measurement and the porthole refraction influence is not considered, the method can correct the influence of the refraction effect on the three-dimensional shape measurement method of the object, and the full-field shape measurement accuracy is 0.4%, so that the actual engineering measurement requirement is met.

Inventors

  • DING TIANHAO
  • SUN CHEN
  • CHEN JUBING

Assignees

  • 上海交通大学

Dates

Publication Date
20260512
Application Date
20260120

Claims (6)

  1. 1. A three-dimensional morphology measurement method crossing an abnormal transparent mechanism imaging is characterized in that a binocular camera and a calibration plate are respectively arranged at two sides of the abnormal transparent mechanism in a calibration stage, images of the calibration plate are acquired through the abnormal transparent mechanism, so that a binocular refraction model containing characteristics of the abnormal transparent mechanism is established, calibration is realized through a single image calibration method, and the three-dimensional morphology is obtained by processing an object image to be measured acquired in real time through the calibrated binocular refraction model in an online stage; the special-shaped transparent mechanism comprises a plane transparent mechanism and a circular tube type transparent mechanism.
  2. 2. The method for measuring three-dimensional morphology through imaging of special-shaped transparent mechanism according to claim 1, wherein the binocular refraction model is established by assuming a camera as a reverse light source and utilizing a refraction law Wherein: The index is a refraction matrix, k is a depth coefficient, [ X, Y, Z ] 'is a physical coordinate of any point in space, [ u, v ]' is a pixel coordinate of the point mapped to an imaging plane, A is a camera internal parameter matrix, I is a unit matrix, [ R, T ] is a rotation and translation relation between two cameras, and subscripts 1 and 2 respectively represent the two cameras; the refraction matrix Wherein, matrix , Rodrigues (m, n) is a rodgers rotation formula, which represents rotation with n as a rotation axis and m as an angle; , N 1 is the normal vector of the surface of the flat-plate transparent mechanism, N 2 、N 3 is the normal vector corresponding to the intersection point when imaging light passes through the front surface and the rear surface of the circular-tube transparent mechanism, d 1 is the distance from the camera to the flat-plate transparent mechanism, d 2 is the thickness of the transparent mechanism, d 3 is the distance from the flat-plate transparent mechanism to the cylindrical transparent mechanism, d 4 is the thickness of the cylindrical transparent mechanism, v 0 is the unit direction vector of emergent light obtained according to the camera internal reference A, pixel coordinates [ u, v ] and camera focal length f, and r= II [ u, v, f ] |, air refractive index N 1 and porthole refractive index N 2 .
  3. 3. The three-dimensional morphology measurement method for cross-shaped transparent mechanism imaging according to claim 1 or 2, wherein the calibration is to extract and collect image point coordinate information of characteristic points in an image of a calibration plate arranged outside a side window by using a Harris angular point extraction algorithm based on pose relation between a camera and the calibration plate and internal parameters of the camera, and substituting the information into a binocular refraction model.
  4. 4. The three-dimensional morphology measurement method for imaging across a special-shaped transparent mechanism according to claim 1 or 2, wherein the calibrating specifically comprises: Step 1) obtaining an objective function according to the porthole and the pose relation between the single calibration plate and the two cameras Wherein, a first error , 、 The space coordinates of each characteristic point on the calibration plate calculated by the binocular refraction model relative to the two cameras respectively, and a second error , Calculated for using a binocular refraction model The const is the actual distance between the adjacent feature points on the calibration plate, and n is the number of the feature points; step 2) setting initial parameters and range of a binocular refraction model; Step 3) substituting the initial parameters into the binocular refraction model and calculating an objective function, executing step 6 when the objective function is smaller than a first threshold value, otherwise, continuing step 4; step 4) changing the value of the parameter to be optimized based on a sequence quadratic programming algorithm (SQP), substituting the value into a binocular refraction model and calculating an updated objective function; Step 5) executing step 6 when the updated objective function is smaller than a first threshold value or the difference between the objective function before and after updating is larger than a second threshold value, otherwise returning to step 4; Step 6) substituting the obtained parameters into the calibrated binocular refraction model Wherein: 、 A calibrated homography matrix for each camera.
  5. 5. The method for measuring the three-dimensional morphology of the cross-shaped transparent mechanism imaging according to claim 4, wherein the initial parameters comprise parameters to be optimized and fixed parameters, wherein the parameters to be optimized comprise space positioning parameters of a flat-plate-type transparent mechanism, space positioning parameters of a cylindrical-type transparent mechanism and space positioning parameters of a calibration plate; All the space positioning parameters in the parameters to be optimized can be roughly measured by simple measuring tools such as calipers, and the optimization range is set to be +/-10%.
  6. 6. The method for measuring the three-dimensional morphology imaged by the cross-shaped transparent mechanism according to claim 1, wherein the three-dimensional morphology is obtained by determining the corresponding pixel coordinates of any feature point of the surface of the object to be measured in the two cameras through a feature point matching algorithm according to a binocular refraction model ) And% ) Traversing all characteristic points by a least square method to obtain the spatial coordinates of the full-field morphology of the object to be detected , , )。

Description

Three-dimensional morphology measurement method for cross-shaped transparent mechanism imaging Technical Field The invention relates to a technology in the field of binocular vision, in particular to a three-dimensional morphology measurement method for cross-shaped transparent mechanism imaging. Background When the existing optical measurement system arranged in the cabin is used for measuring the appearance of an object outside the cabin, the phenomenon that image points are distorted due to the refraction effect of the porthole multilayer glass structure is not considered, so that the accuracy of a measurement result is affected. In addition, in existing methods of correcting for cross-media imaging effects, only the refractive path of light through a parallel transparent mechanism is often described and is not applicable to multi-layer, differently shaped structures. Meanwhile, prior information of the position of the calibration plate is required in the prior art, which means that the transparent structure is required to be disassembled firstly to perform pre-calibration to determine the position of the plate and then the transparent structure is installed to perform calibration of the whole model in the operation process, and the disassembly and installation process is not suitable for on-line measurement on a machine and cannot describe the influence of the special-shaped transparent mechanism on the light path. Disclosure of Invention Aiming at the defects existing in the prior art, the invention provides a three-dimensional morphology measurement method for imaging across a special-shaped transparent mechanism, and aiming at the defect that the measurement precision is low because the calibration process is not suitable for on-machine measurement and the porthole refraction influence is not considered, the influence of the porthole refraction effect on the three-dimensional morphology measurement method of an object can be corrected, the whole-field morphology measurement precision is 0.4%, and the actual engineering measurement requirement is met. The invention is realized by the following technical scheme: the invention relates to a three-dimensional morphology measurement method for cross-shaped transparent mechanism imaging, which comprises the steps of respectively arranging a binocular camera and a calibration plate at two sides of a shaped transparent mechanism in a calibration stage, and acquiring images of the calibration plate through the shaped transparent mechanism, so as to establish a binocular refraction model containing the characteristics of the shaped transparent mechanism and realize calibration through a single image calibration method; and processing the real-time acquired image of the object to be measured through the calibrated binocular refraction model in the online stage to obtain the three-dimensional morphology. The binocular refraction model is established by assuming a camera as a reverse light source and utilizing a refraction lawWherein: The index is a refraction matrix, k is a depth coefficient, [ X, Y, Z ] 'is a physical coordinate of any point in space, [ u, v ]' is a pixel coordinate of the point mapped to an imaging plane, A is a camera internal parameter matrix, I is a unit matrix, [ R, T ] is a rotation and translation relation between two cameras, and subscripts 1 and 2 respectively represent the two cameras. The special-shaped transparent mechanism comprises a plane transparent mechanism and a circular tube type transparent mechanism. The refraction matrixWherein, matrix,Rodrigues (m, n) is a rodgers rotation formula, which represents rotation with n as a rotation axis and m as an angle;, N 1 is the normal vector of the surface of the flat-plate transparent mechanism, N 2、N3 is the normal vector corresponding to the intersection point when imaging light passes through the front surface and the rear surface of the circular-tube transparent mechanism, d 1 is the distance from the camera to the flat-plate transparent mechanism, d 2 is the thickness of the transparent mechanism, d 3 is the distance from the flat-plate transparent mechanism to the cylindrical transparent mechanism, d 4 is the thickness of the cylindrical transparent mechanism, v 0 is the unit direction vector of emergent light obtained according to the camera internal reference A, pixel coordinates [ u, v ] and camera focal length f, and r= II [ u, v, f ] |, air refractive index N 1 and porthole refractive index N 2. The calibration, namely, based on pose relation between a camera and a calibration plate and internal parameters of the camera, extracts and collects image point coordinate information of characteristic points in an image of the calibration plate arranged outside the side window by using a Harris angular point extraction algorithm, and substitutes the information into a binocular refraction model, and the method specifically comprises the following steps: Step 1) obtaining an objecti