Gear inspection device
The gear inspection device addresses phase alignment and luminance difference issues by acquiring and processing multiple images at predetermined angles, enhancing the accuracy of gear abnormality detection using machine learning.
Patent Information
- Authority / Receiving Office
- JP · JP
- Patent Type
- Patents
- Current Assignee / Owner
- TOYOTA JIDOSHA KK
- Filing Date
- 2023-05-11
- Publication Date
- 2026-06-02
AI Technical Summary
Existing gear inspection methods fail to accurately detect abnormalities such as scratches due to luminance differences not appearing in photographed images, and phase alignment issues lead to decreased determination accuracy, especially for large gears.
A gear inspection device comprising an image acquisition unit, an image extraction unit, and an abnormality determination unit, which acquires multiple images at predetermined angles, extracts inspection images by comparing with a master image, and determines abnormalities using machine learning models.
The device enables high-accuracy detection of gear abnormalities by aligning phase tolerance and using inspection images, effectively identifying scratches even if they occur on all teeth, thus improving detection precision.
Smart Images
Figure 0007868551000001 
Figure 0007868551000002 
Figure 0007868551000003
Abstract
Description
Technical Field
[0001] The present invention relates to a gear inspection device.
Background Art
[0002] Conventionally, in order to detect abnormalities such as scratches that can occur on a gear, it is known to inspect the gear by analyzing a photographed image of the gear without relying on human visual inspection.
[0003] In Patent Document 1, a gear is rotated by a predetermined angle and sequentially photographed, and in a difference image generated from the difference between the photographed image after rotation and the photographed image before rotation, a determination target region having a luminance exceeding a predetermined threshold is detected, and the surface state of the gear is determined by analyzing an inspection region surrounding the determination target region.
Prior Art Documents
Patent Documents
[0004]
Patent Document 1
Summary of the Invention
Problems to be Solved by the Invention
[0005] However, abnormalities in the gear do not necessarily appear as a luminance difference in the photographed images before and after rotation. For example, when the same scratches occur on all the teeth of the gear due to chipping of the gear tooth cutting cutter, the scratches of the gear to be detected do not appear as a luminance difference.
[0006] Also, in the method described in Patent Document 1, when the approach of the tooth of the gear is detected by a proximity sensor when the gear is rotated, the tooth of the gear is photographed according to the detection signal. However, when the gear to be inspected is large, it is not possible to align the gear to a desired phase (rotation angle) only by detecting the tooth of the gear, and there is a risk that the determination accuracy of the abnormality of the gear may decrease.
[0007] Furthermore, Patent Document 1 describes determining the presence or absence of damage to gears using a determination unit composed of a neurocomputer with learning capabilities. However, machine learning technology for image analysis has advanced further, and there is room for improvement in the gear image determination method.
[0008] Therefore, in view of the above problems, the object of the present invention is to accurately detect gear abnormalities from gear images. [Means for solving the problem]
[0009] To solve the above problems, the present invention provides a gear inspection device for inspecting gears, comprising: an image acquisition unit that acquires a predetermined number of images generated by photographing the teeth of the gear when the gear is rotated by a predetermined angle; an image extraction unit that extracts an inspection image from the images by comparing each of the predetermined number of images with a master image; and an abnormality determination unit that determines whether or not there is an abnormality in the gear using the inspection image, wherein the predetermined number is determined based on the predetermined angle and the number of teeth of the gear. [Effects of the Invention]
[0010] According to the present invention, abnormalities in gears can be detected with high accuracy from images of the gears. [Brief explanation of the drawing]
[0011] [Figure 1] Figure 1 is a schematic diagram showing the configuration of a gear inspection system according to an embodiment of the present invention. [Figure 2] Figure 2 is a functional block diagram of the edge PC processor. [Figure 3] Figure 3 is a flowchart showing the control routine for the abnormality detection process in this embodiment. [Modes for carrying out the invention]
[0012] Embodiments of the present invention will be described in detail below with reference to the drawings. In the following description, similar components will be given the same reference numerals.
[0013] Figure 1 is a schematic diagram showing the configuration of a gear inspection system 10 according to an embodiment of the present invention. The gear inspection system 10 is installed, for example, in a gear manufacturing plant or a factory that produces products having gears (such as an automobile assembly plant) to inspect gears. As shown in Figure 1, the gear inspection system 10 comprises an edge personal computer (edge PC) 1, an object detection sensor 2, a PLC (Programmable Logic Controller) 3, a camera 4, and a rotating device 5.
[0014] Edge PC1 is an example of a gear inspection device for inspecting gears. In this embodiment, Edge PC1 inspects the appearance of gears and detects gear abnormalities such as scratches. As shown in Figure 1, Edge PC1 comprises an input / output interface 11, memory 12, and processor 13. The input / output interface 11 and memory 12 are connected to the processor 13 via signal lines.
[0015] The input / output interface 11 has an interface circuit for connecting the edge PC 1 and peripheral devices. The outputs of peripheral devices such as the PLC 3 and camera 4 are input to the edge PC 1 via the input / output interface 11, and command signals generated by the processor 13 are output to peripheral devices such as the camera 4 and rotary device 5 via the input / output interface 11.
[0016] Memory 12 includes, for example, volatile semiconductor memory and non-volatile semiconductor memory. Memory 12 stores computer programs, data, etc., that are used when various processes are executed by the processor 13.
[0017] The processor 13 has one or more CPUs (Central Processing Units) and its peripheral circuits. Note that the processor 13 may further have an arithmetic circuit such as a logical arithmetic unit or a numerical arithmetic unit.
[0018] The object detection sensor 2 detects the presence or absence of the gear to be inspected. The object detection sensor 2 is configured as, for example, a photoelectric sensor, a proximity sensor, or a microphoto sensor. The object detection sensor 2 is connected to the PLC 3 by wire or wirelessly, and the output of the object detection sensor 2 is transmitted to the PLC 3.
[0019] The PLC 3 is communicably connected to the object detection sensor 2 and the edge PC 1. The PLC 3 determines the shooting timing of the gear based on the output of the object detection sensor 2 and transmits a shooting command for the gear to the edge PC 1. The PLC 3 is also communicably connected to other production facilities provided in the factory and transmits various command signals to each of the production facilities.
[0020] The camera 4 is provided at a predetermined position such as a factory so as to photograph the gear. The camera 4 includes a lens and an imaging element and is, for example, a CMOS (Complementary Metal Oxide Semiconductor) camera or a CCD (Charge Coupled Device) camera. The camera 4 is connected to the edge PC 1 by wire or wirelessly. The camera 4 photographs the teeth of the gear based on a command from the edge PC 1 and generates a photographed image (an image of the teeth of the gear). The output of the camera 4, that is, the photographed image generated by the camera 4 is transmitted to the edge PC 1. Note that a lighting device for irradiating the gear may be provided in order to obtain a clear photographed image.
[0021] The rotating device 5 has, for example, a holding table that holds a gear to be inspected and a drive unit (e.g., a stepping motor) that rotates the gear on the holding table. The rotating device 5 rotates the gear about its central axis. The rotating device 5 is connected to the edge PC1 by wire or wirelessly. The rotating device 5 rotates the gear by a predetermined angle based on a command from the edge PC1. The camera 4 photographs the teeth of the gear in synchronization with the rotation of the gear by the rotating device 5. That is, the camera 4 generates a still image of the teeth of the gear each time the gear rotates by a predetermined angle.
[0022] FIG. 2 is a functional block diagram of the processor 13 of the edge PC1. As shown in FIG. 2, the processor 13 has a photographing control unit 14, an image acquisition unit 15, an image extraction unit 16, and an abnormality determination unit 17. The photographing control unit 14, the image acquisition unit 15, the image extraction unit 16, and the abnormality determination unit 17 are functional modules realized by the processor 13 of the edge PC1 executing a computer program stored in the memory 12 of the edge PC1. Note that these functional modules may be realized by dedicated arithmetic circuits provided in the processor 13, respectively.
[0023] The photographing control unit 14 controls the photographing of the gear by the camera 4. Specifically, the photographing control unit 14 controls the phase (rotation angle) of the gear by the rotating device 5 and photographs the gear by the camera 4. When a photographing command is transmitted from the PLC 3 to the edge PC1, the photographing control unit 14 rotates the gear by the rotating device 5 and photographs the teeth of the gear by the camera 4.
[0024] The image acquisition unit 15 acquires the photographed image generated by the camera 4. In the present embodiment, the image acquisition unit 15 acquires a predetermined number of photographed images generated by photographing the teeth of the gear when the gear is rotated by a predetermined angle. The predetermined angle is the amount of phase shift of the gear allowed when determining the abnormality of the gear and is determined in advance.
[0025] The predetermined number is the number of images captured by camera 4 to determine gear abnormalities, and is determined based on a predetermined angle and the number of teeth on the gear. The predetermined number can be calculated, for example, by the following formula (1). The specified number = 360 / (number of teeth on the gear × specified angle) ... (1)
[0026] As can be seen from equation (1) above, the predetermined number increases as the predetermined angle decreases and as the number of teeth on the gear decreases. For example, if the predetermined angle is 1° and the number of teeth on the gear is 40, the predetermined number is 9. In this case, the shooting control unit 14 rotates the gear 1° eight times and takes a total of nine images of the gear: before it is rotated and after it has been rotated 1° at a time.
[0027] The image extraction unit 16 extracts inspection images suitable for gear abnormality detection from a predetermined number of captured images acquired by the image acquisition unit 15. In this embodiment, the image extraction unit 16 extracts inspection images from captured images by comparing a predetermined number of captured images with a master image. The master image is an image of a normal gear adjusted to the phase most suitable for gear abnormality detection and is prepared in advance. The image extraction unit 16 extracts the captured image with the smallest positional deviation (for example, deviation in the x and y directions) from the master image as the inspection image. As a specific example, the image extraction unit 16 extracts inspection images from a predetermined number of captured images using an image matching method such as the phase-limited correlation method. In this case, the image extraction unit 16 extracts the captured image with the highest similarity to the master image from the predetermined number of captured images as the inspection image.
[0028] The abnormality detection unit 17 determines whether or not there is an abnormality in the gear using the inspection image extracted by the image extraction unit 16. For example, the abnormality detection unit 17 determines whether or not there is an abnormality in the gear using a classifier that has been pre-trained to output whether or not there is an abnormality in the gear from the inspection image data. Examples of such classifiers include machine learning models such as neural networks, support vector machines, and random forests.
[0029] As described above, in this embodiment, the necessary number of images are acquired considering the tolerance for gear phase misalignment, and gear abnormalities are determined using inspection images extracted from these images. Therefore, since it is not necessary to photograph the gear after precisely adjusting its phase, it is possible to suppress a decrease in the accuracy of gear abnormality detection due to a decrease in the accuracy of gear phase adjustment. Furthermore, in this embodiment, gear abnormalities are determined using inspection images extracted from a predetermined number of images, instead of difference images generated from the difference between the image taken after rotation and the image taken before rotation. Therefore, even if similar scratches occur on all teeth of the gear, scratches shown in the inspection image can be detected as abnormalities. Accordingly, according to this embodiment, gear abnormalities can be detected with high accuracy from images of the gear.
[0030] Figure 3 is a flowchart showing the control routine for the abnormality detection process in this embodiment. This control routine is repeatedly executed by the processor 13 of the edge PC1.
[0031] First, in step S101, the image capture control unit 14 of the processor 13 determines whether or not it has received an image capture command from the PLC 3. The PLC 3 sends an image capture command to the edge PC 1 when the gear to be inspected is detected by the object detection sensor 2. If it is determined in step S101 that no image capture command has been received, this control routine terminates. On the other hand, if it is determined in step S101 that an image capture command has been received, this control routine proceeds to step S102.
[0032] In step S102, the shooting control unit 14 uses the camera 4 to photograph the gear teeth. The captured image generated by the camera 4 is transmitted to the edge PC1, and the image acquisition unit 15 of the processor 13 stores the captured image in, for example, the memory 12 of the edge PC1.
[0033] Next, in step S103, the imaging control unit 14 rotates the gear by a predetermined angle using the rotating device 5. The predetermined angle is the amount of phase shift of the gear that is permissible when determining a gear abnormality, and is predetermined.
[0034] Next, in step S104, the shooting control unit 14 determines whether the number of images generated by the camera 4, i.e., the number of times the camera 4 has taken pictures, has reached a predetermined number. The predetermined number is determined based on a predetermined angle and the number of teeth on the gear, and is calculated, for example, by formula (1) above. If it is determined in step S104 that the number of images taken has not reached the predetermined number, the control routine returns to step S102, and steps S102 and S103 are executed again.
[0035] On the other hand, if it is determined in step S104 that the number of captured images has reached a predetermined number, the control routine proceeds to step S105. In step S105, the image extraction unit 16 of the processor 13 extracts inspection images from the captured images by comparing a predetermined number of captured images with a master image using the method described above.
[0036] Next, in step S106, the abnormality detection unit 17 of the processor 13 performs preprocessing on the inspection image. For example, the abnormality detection unit 17 performs color correction, resizing, normalization, noise reduction, filtering, etc. as preprocessing. If color correction is performed as preprocessing, for example, a master image with optimized brightness is prepared, and the abnormality detection unit 17 brings the brightness of the inspection image closer to the brightness of the master image.
[0037] Next, in step S107, the abnormality detection unit 17 uses the inspection image to determine whether or not there is an abnormality in the gear. An abnormality in the gear is a scratch, dent, or the like on the surface of the gear. The result of the gear abnormality detection is displayed, for example, on the display of edge PC1. Alternatively, the result of the gear abnormality detection may be sent to PLC3. After step S107, this control routine terminates.
[0038] Note that the preprocessing in step S106 may be omitted. Also, although the above control routine determines an abnormality in one tooth of the gear, an abnormality may be determined in all teeth of the gear. In this case, the processing in steps S102 to S107 will be repeated a number of times corresponding to the number of teeth on the gear.
[0039] Although preferred embodiments of the present invention have been described above, the present invention is not limited to these embodiments, and various modifications and changes can be made within the scope of the claims. For example, communication between the edge PC1 and the peripheral device may be performed wirelessly, and the edge PC1 may be located outside the factory. [Explanation of Symbols]
[0040] 1 Edge PC 13 processors 14. Image capture control unit 15 Image acquisition unit 16 Image Extraction Unit 17 Abnormality determination section
Claims
[Claim 1] A gear inspection device for inspecting gears, An image acquisition unit that acquires a predetermined number of images generated by photographing the teeth of the gear when the gear is rotated by a predetermined angle, An image extraction unit that compares each of the predetermined number of captured images with a master image and extracts the captured image with the smallest positional deviation from the master image as the inspection image. An abnormality determination unit determines whether or not there is an abnormality in the gear using a classifier that has been pre-trained to output whether or not there is an abnormality in the gear from the data of the inspection image. Equipped with, A gear inspection device in which the predetermined number is determined based on the predetermined angle and the number of teeth of the gear.