Workpiece shape quality determination method
By comparing workpiece images at multiple angles with a reference, the method efficiently determines shape quality, eliminating blind spots and reducing computational burden, thus ensuring accurate judgments.
Patent Information
- Application Number
- JP2023085658
- Authority / Receiving Office
- JP · JP
- Patent Type
- Applications
- Current Assignee / Owner
- Filing Date
- 2023-05-24
- Publication Date
- 2024-12-05
AI Technical Summary
Existing methods for determining the shape quality of a workpiece require significant effort in synthesizing three-dimensional data, leading to computational burdens and inefficiencies.
A method that compares images of the workpiece taken at multiple angles with a reference value, determining the workpiece's quality by matching differences at all angles, eliminating the need for comprehensive three-dimensional data synthesis and using a laser scanner to obtain point cloud data.
Enables quick and reliable shape quality determination of workpieces by eliminating blind spots and reducing computational effort, ensuring accurate pass/fail judgments without synthesizing three-dimensional data.
Smart Images

Figure 2024168741000002 
Figure 2024168741000003 
Figure 2024168741000004
Abstract
Description
[Technical field]
[0001] The present invention relates to a method for determining the quality of a shape of a workpiece by optical measurement and, more particularly, to a method for detecting the precise shape of a workpiece to be inspected and determining whether the shape is good or bad. [Background technology]
[0002] Laser scanners that can obtain the precise shape of the workpiece to be measured as point cloud data that can be directly processed in a computer are becoming more widely used, but in this case, blind spots from the laser light occur, making it impossible to obtain data. Patent Document 1 proposes applying dimensional data from standard information to the blind spots to generate an overall model of the object to be measured, including the blind spots. [Prior art documents] [Patent documents]
[0003] [Patent Document 1] Patent No. 6545065 Summary of the Invention [Problem to be solved by the invention]
[0004] Conventionally, the workpiece is placed on a tiltable measuring stage, tilted to multiple angles, and the three-dimensional data obtained at each tilt angle is synthesized to obtain an overall image of the workpiece without blind spots, and the shape is then judged to be good or bad. However, there is a problem in that synthesis of three-dimensional data requires a great deal of effort.
[0005] Therefore, the present invention is intended to solve such problems, and aims to provide a method for determining whether the shape of a workpiece is good or bad, which can easily and reliably determine whether the shape of the workpiece being inspected is good or bad, without the need for the time-consuming synthesis of three-dimensional data. [Means for solving the problem]
[0006] In order to achieve the above-mentioned objective, in the first invention, each image of the test work obtained by optical measurement at a number of different tilt angles is matched with each image of a reference work acquired in advance at each tilt angle identical to the tilt angles mentioned above, the difference between the two matched images is calculated, and the shape of the test work is judged to be good when the difference between the two images at all tilt angles is smaller than a predetermined value.
[0007] In the first invention, the blind spots of the test workpiece caused by the optical measurement at each tilt angle are different for each tilt angle, so that even if there is a blind spot at one tilt angle, it is possible to prevent it from becoming a blind spot at another tilt angle, and by matching the images of the test workpiece and the reference workpiece for each tilt angle and calculating the difference between the two images, and judging the shape of the test workpiece as good only when the difference between the two images at all tilt angles is smaller than a predetermined value, the quality of the test workpiece can be reliably judged without being affected by the blind spots. In addition, according to the first invention, there is no need to synthesize three-dimensional data obtained at multiple tilt angles to obtain an entire image of the test workpiece as in the conventional method, so there is no need to synthesize three-dimensional data, and the burden of calculation is reduced.
[0008] In the second invention, the optical measurement uses a laser scanner to obtain point cloud data.
[0009] In the third invention, a bounding box is obtained for each of the matched images, and the shapes of the bounding boxes are compared to determine whether the shape of the test workpiece is good or bad.
[0010] In the third invention, the quality of the shape of the test work can be determined quickly and easily by comparing the bounding boxes of the images of the test work and the reference work. Effect of the Invention
[0011] As described above, according to the method for determining whether the shape of a workpiece of the present invention is acceptable, it is possible to easily and reliably determine whether the shape of a test workpiece is acceptable. [Brief description of the drawings]
[0012] [Figure 1] FIG. 1 is a perspective view showing the appearance of an apparatus for carrying out the method of the present invention. [Diagram 2] 1 is a flowchart showing a processing procedure performed in a computer. [Diagram 3] 1 is a conceptual three-dimensional image of the workpiece. [Figure 4] These are two-dimensional images of the test workpiece and the reference workpiece viewed from three axial directions before matching. [Diagram 5] These are two-dimensional images of a reference arc and a test workpiece with a good shape viewed from three axial directions in a matching state. [Figure 6] These are two-dimensional images of a reference workpiece and a test workpiece with a defective shape viewed from three axial directions in a matching state. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS
[0013] It should be noted that the embodiment described below is merely an example, and various design improvements made by those skilled in the art without departing from the gist of the present invention are also included in the scope of the present invention.
[0014] The external appearance of an apparatus for carrying out the method of the present invention is shown in Fig. 1. In Fig. 1, the tilting device is equipped with a motor-driven frame 1, which is capable of rotating around a rotation axis 11 within a predetermined angular range, and is also capable of rotating around a rotation axis 12 perpendicular to the rotation axis 11 at a rotation position around the rotation axis 11 within a predetermined angular range. A test workpiece is placed on a stage (not shown) supported by the frame 1, and the posture of the test workpiece can be changed three-dimensionally to a plurality of tilt positions.
[0015] A laser scanner 2 is disposed above the frame 1 (i.e., above the test work), and for the test work, whose posture has been changed to a plurality of tilt positions to eliminate blind spots, three-dimensional data (point cloud data of the three-dimensional shape) is obtained by the laser scanner 2 at each tilt position. The obtained three-dimensional data is sent to a computer 3 connected to the laser scanner 2, and the three-dimensional images of the reference work and the test work (three-dimensional images obtained from the point cloud data) are matched by the process described below to judge whether the shape of the test work is good or bad.
[0016] The reference work is a work that satisfies the dimensional specifications, with the same shape as the test work and with the dimensions of each part accurately finished. In this embodiment, prior to the process of determining whether the shape of the test work is good or bad, which will be described later, the reference work is placed on the stage and tilted to the above-mentioned multiple predetermined tilt positions, and the three-dimensional data (point cloud data of the three-dimensional shape) obtained by the laser scanner 2 at each tilt position is stored in advance in the computer 3.
[0017] The shape pass / fail judgment process carried out in the computer 3 is as shown in the flowchart of Fig. 2. That is, in step 101 of Fig. 2, the frame 1 (i.e., the workpiece to be inspected) is tilted to a predetermined angle position in response to a command from the computer 3, and in step 102, three-dimensional data (point cloud data of the three-dimensional shape) of the workpiece to be inspected as viewed from this direction is obtained by the laser scanner 2, and this three-dimensional data is converted into voxel data, and the normal vector of each voxel is calculated. Fig. 3 shows an example of a three-dimensional image in which the normal vector Nv has been calculated for each voxel.
[0018] In step 103 in FIG. 2, the three-dimensional data obtained by tilting the reference workpiece by the same angle is read out, and the normal vector of each voxel is calculated in the same manner. Then, the test workpiece and the reference workpiece are matched by, for example, the ICP (Iterative Closest Point) method so that the normal vectors of the three-dimensional images of the test workpiece and the reference workpiece are most closely matched (step 104 in FIG. 2). This allows the three-dimensional image to be well matched with the three-dimensional image of the reference workpiece regardless of whether the test workpiece slides on the attachment due to its own weight due to tilting and its position or angle is shifted. In this embodiment, since there is no need to provide an anti-slip structure when the test workpiece is tilted, it is possible to prevent blind spots caused by the structure.
[0019] In this state, in this embodiment, the 3D bounding boxes of the matched stereoscopic images of the reference workpiece and the test workpiece are calculated (step 105). Then, in step 106, if the difference in the external dimensions of both bounding boxes is within the allowable range, the overall shape is deemed to be good and the process proceeds to the next step 107. On the other hand, if the difference in the external dimensions of both bounding boxes is outside the allowable range, the shape is determined to be defective (step 108).
[0020] To further explain the above process using two-dimensional images of the three-dimensional images of the test work and reference work viewed from three directions, the X-axis, Y-axis, and Z-axis directions, respectively, Figure 4 shows a two-dimensional image Ft of the test work and a two-dimensional image Fs of the reference work before matching, with (1), (2), and (3) being views from the X-axis, Y-axis, and Z-axis directions, respectively.
[0021] Figure 5 (1), (2), and (3) show the two-dimensional images Fs, Ft seen from the X-axis, Y-axis, and Z-axis directions, respectively, after matching when the shape of the test workpiece matches the shape of the reference workpiece, and the size of the simple rectangular bounding box (white line in the figure) also matches.
[0022] On the other hand, (1), (2), and (3) in Figure 6 are the two-dimensional images Ft, Fs viewed from the X-axis, Y-axis, and Z-axis directions, respectively, after matching when the shape of the test workpiece does not match the shape of the reference workpiece, and the outlines of the simple rectangular bounding boxes (white lines in the figures) do not match in size.
[0023] 2, the presence or absence of local shape defects such as flatness and the angle between the faces of the three-dimensional image of the test workpiece matched with the three-dimensional image of the reference workpiece is judged, and if there is a local shape defect, it is judged to be a shape defect (step 108). If there is no local shape defect, the test workpiece at the angle position is judged to be a good shape (step 109).
[0024] As described above, the above steps 101 to 109 are repeated by tilting the inspected workpiece to a plurality of tilt positions to eliminate blind spots.
[0025] Table 1 shows an example of the external dimensional difference that occurs in the bounding boxes of the two-dimensional images Fs, Ft of the reference workpiece and the test workpiece viewed from the X-axis, Y-axis, and Z-axis directions when the test workpiece is tilted to each tilt position a to e. In Table 1, a dimensional difference of 0.05 mm or more occurs at tilt positions c and d, so the test workpiece is ultimately determined to have a defective shape.
[0026] [Table 1]
[0027] Ultimately, only when the shape is judged to be good in step 109 at all of the plurality of tilt positions, the shape of the inspected workpiece is finally judged to be good.
[0028] (Other embodiments) In the above embodiment, the reference workpiece is placed on the stage and tilted to a number of predetermined tilt positions, and the three-dimensional data obtained by the laser scanner 2 at each tilt position is stored in advance in the computer 3, but it may also be stored in advance in the computer as three-dimensional data created by CAD. In the above embodiment, the optical measurement is performed by a laser scanner, but the present invention is not limited to this. [Explanation of symbols]
[0029] 1...frame, 11, 12...rotating axis, 2...laser scanner, 3...computer.
Claims
1. A method for determining the quality of a shape of a workpiece, in which each image of a test workpiece obtained by optical measurement at a number of different tilt angles is matched with each image of a reference workpiece previously acquired at each tilt angle identical to the tilt angles, the difference between the two matched images is calculated, and the shape of the test workpiece is determined to be good when the difference between the two images at all tilt angles is smaller than a predetermined value.
2. 2. The method for determining the quality of a shape of a workpiece according to claim 1, wherein the optical measurement is performed by using a laser scanner to obtain point cloud data.
3. A method for determining whether the shape of a workpiece is good or bad as described in claim 1 or 2, in which a bounding box is obtained for each of the matched images, and the shape of the test workpiece is determined to be good or bad by comparing the shapes of the bounding boxes.
Citation Information
Patent Citations
3D point cloud data modeling device, modeling program, and elevator guide rail modeling method
JP6545065B2
