JP5628596B2 - Tire appearance inspection method and appearance inspection apparatus - Google Patents

Description

  The present invention relates to a tire appearance inspection method and an appearance inspection apparatus, and in particular, detects a region periodically formed on a tire surface at the time of tire molding, and accurately removes the region from a captured image of the tire surface. The present invention relates to a tire appearance inspection method and an appearance inspection apparatus that perform surface appearance inspection.

Conventionally, among tire surfaces, for example, as one of the methods for inspecting the tire inner peripheral surface, the tire inner peripheral surface is imaged along the circumferential direction by a light cutting method, and tire molding defects are determined based on the captured image. A method of inspection has been proposed. The method irradiates slit light in the width direction of the tire by a plurality of irradiation means on the side part and the crown part of the inner peripheral surface of the tire while rotating the tire by the holding means for holding the tire rotatably in the circumferential direction. The slit light irradiating part is imaged by a plurality of imaging means arranged corresponding to the irradiating means to obtain a captured image for each part of the tire, and after the synthesized images are synthesized by the image processing means, the tire inner periphery Get an inspection image of the surface. And the quality of a tire inner peripheral surface is determined based on the brightness | luminance information and height information which are contained in a test | inspection image.
However, when a tire is molded, a convex part called a ridge is periodically formed on the tire inner peripheral surface by a bladder that presses the tire against the molding die from the inside. Even Ridge is subject to inspection, which hinders improvement in inspection efficiency. Further, dirt and scratches in the vicinity of the ridge cannot be distinguished from the ridge and misjudged, which hinders improvement in inspection efficiency.
One of the methods for removing a specific region from the inspection image is, for example, in order to remove a knurled portion formed with a constant width along the side surface of the tire surface from an image obtained by imaging the tire side surface. The image is transformed into a frequency image by Fourier transform, and the Fourier transform image is removed from the frequency image on the tire side surface as a knurled portion as a region appearing on one straight line at a predetermined interval. There is a method of removing a knurled portion from a captured image of a tire side surface by performing inverse Fourier transform.

  However, the above method is based on the premise that the region to be removed appears on a straight line at a predetermined interval in the Fourier-transformed image, in order to improve the inspection efficiency and inspection accuracy of the tire inner peripheral surface. When removing the ridge from the inspection image, the area corresponding to the ridge is not necessarily represented on a single straight line at a predetermined interval in the image obtained by Fourier transforming the captured image. Therefore, it is difficult to remove the ridge from the captured image.

JP 2009-139297 A

  In order to solve the above-described problems, an object of the present invention is to provide a tire appearance inspection method and an appearance inspection apparatus that can reliably remove a region periodically formed on a tire surface from a captured image.

As a method for solving the above problem, a pixel length that appears periodically in the captured image of the tire surface imaged along the tire circumferential direction, is equal to or larger than the pixel width set by the threshold value, and is set to the threshold value. Detecting a region having the above shape as a specific shape from a captured image, and calculating an inclination angle with respect to a tire circumferential direction in a direction in which the pixel lengths are continuous for all the detected specific shapes; Detecting an inclination angle of the image, a step of obtaining an average pixel width and an average pixel length from the pixel widths and pixel lengths of all the detected specific shapes, and a direction of the average pixel length is a tire circumference. Creating a mask image having the shape of the average pixel width and the average pixel length so as to be orthogonal to the direction, and the mask image so that the mask image matches the specific shape. A step of rotating the click image so as to be parallel to the large inclination angles of the most frequent number of times, and the mask image and the captured image is Fourier transform, converting the said mask image and the captured image to the power spectrum image A step of binarizing and bit-inverting the power spectrum image of the mask image; a step of multiplying the power spectrum image of the mask image by bit-inverting the power spectrum image of the captured image; and the mask image And a step of performing an inverse Fourier transform on the power spectrum image of the captured image multiplied by the bit-inverted power spectrum image .
According to this method, a mask image having a specific shape that periodically appears in the captured image of the tire surface is created, the mask image is rotated so as to coincide with the removal of the specific shape detected from the captured image, and the captured image is By superimposing the mask images, the specific shape can be efficiently and accurately removed from the captured image.
Further , according to this method, since the power spectrum image of the mask image is the mask image itself, noise is removed from the mask image by binarizing the mask image, and the binarized mask image is bit-inverted. By multiplying the power spectrum image of the captured image by the image, a component corresponding to the specific shape can be removed from the power spectrum image of the captured image. That is, since the power spectrum image of the bit-reversed mask image is represented by the brightness of the pixel corresponding to the specific shape being zero and the brightness of the pixel that is not to be removed is represented by 1, the power spectrum image of the captured image is represented by the bit-reversed mask image. By multiplying with, the region corresponding to the specific shape can be easily removed from the power spectrum image of the captured image without requiring a complicated calculation. Furthermore, the power spectrum image of the captured image obtained by multiplying the power spectrum image of the captured image multiplied by the power spectrum image obtained by bit-inversion of the mask image is subjected to inverse Fourier transform, thereby obtaining an image of the tire surface from which the specific shape has been removed from the captured image. can do. In addition, since the specific shape that appears periodically in the captured image is represented in the same part in the power spectrum image obtained by Fourier transforming the captured image, a mask image for one specific shape is created and converted into a power spectrum image. By multiplying the power spectrum image obtained by digitizing and then bit-inverting the power spectrum image of the captured image and then performing inverse Fourier transform, specific shapes that appear periodically can be collectively removed. Therefore, since the obtained image does not include a portion unnecessary for the appearance inspection, inspection efficiency and inspection accuracy can be improved by inspecting the image.
As a configuration for solving the above-mentioned problem, a pixel length that appears periodically in a captured image of the tire surface imaged along the tire circumferential direction, is equal to or larger than the pixel width set by the threshold, and is set to the threshold Specific shape detection means for detecting a region having the above shape as a specific shape from a captured image, and for all the detected specific shapes, an inclination angle with respect to a tire circumferential direction in a direction in which the pixel lengths are continuous is calculated. , Detecting the most frequent inclination angle, obtaining the average pixel width and average pixel length from the pixel widths and pixel lengths of all the detected specific shapes, and the direction of the average pixel length is the tire circumferential direction A mask image creating means for creating a mask image having a shape with the average pixel width and the average pixel length so as to be orthogonal to the mask image so that the mask image matches the specific shape. A mask image adapting means for rotating and adapting the image so as to be parallel to the inclination angle having the highest frequency, and transforming the mask image and the captured image into a power spectrum image by Fourier transforming the mask image and the captured image. FFT processing means, binarization processing means for binarizing the bit spectrum of the power spectrum image of the mask image, and specific shape removal for multiplying the power spectrum image of the captured image by the power spectrum image of the bit inverted mask image And an inverse FFT processing means for performing an inverse Fourier transform on the power spectrum image of the captured image multiplied by the power spectrum image of the bit-inverted mask image.
According to this configuration, a specific shape that periodically appears in the captured image is detected from the captured image by the specific shape detection unit, a mask image is generated by using the specific shape as a shape to be removed by the mask image generation unit, and the mask image is specified. The rotated mask image and the picked-up image after being rotated so as to coincide with the shape are converted into the power spectrum image of the mask image and the power spectrum image of the picked-up image by the FFT processing means, and the mask is obtained by the binarization processing means. Only the power spectrum image of the image is binarized and then bit-reversed, and the power spectrum of the captured image is multiplied by the power spectrum image of the mask image that has been bit-reversed with the power spectrum image of the captured image by the specific shape removing means. A region corresponding to a specific shape included in the image can be removed. Then, by performing inverse Fourier transform on the power spectrum image of the captured image processed by the specific shape removing unit by the inverse FFT processing unit, all of the specific shapes that appear periodically in the captured image of the tire surface are removed. A captured image can be obtained. Therefore, since the obtained image does not include a portion unnecessary for the appearance inspection, the inspection efficiency and the inspection accuracy can be improved by inspecting the tire surface based on the image.

1 is a schematic configuration diagram of a tire appearance inspection apparatus according to the present invention. The layout of a camera set with respect to the tire which concerns on this invention. The figure which shows the captured image for every site | part of the tire internal peripheral surface which concerns on this invention. The block diagram of the image processing means which concerns on this invention. The partial enlarged view when the captured image concerning this invention is image-processed. The enlarged view when image processing is performed on the mask image according to the present invention. The figure which shows the power spectrum image, image processing, and test | inspection image of the captured image and mask image which concern on this invention. The flowchart of the image processing by an image processing means.

  Hereinafter, the present invention will be described in detail through embodiments of the invention. However, the following embodiments do not limit the invention according to the claims, and all combinations of features described in the embodiments are included in the invention. It is not necessarily essential to the solution, but includes a configuration that is selectively adopted.

FIG. 1 shows an overall schematic view of the appearance inspection apparatus 1, and FIG. 2 shows an arrangement of camera sets 11 to 13 with respect to a tire T. Hereinafter, the appearance inspection apparatus 1 will be described with reference to FIGS. 1 and 2. As shown in FIG. 1, the appearance inspection apparatus 1 is roughly composed of an imaging device 10 that images a tire surface and an image processing device 30.
The imaging device 10 includes a rotary table 5 on which a tire T that is a subject is placed, and camera sets 11 to 13 that image the tire inner peripheral surface Ts for each part.
The turntable 5 is rotated by driving a motor 4 built in a pedestal portion of a turntable (not shown), and rotates a tire T placed sideways. The motor 4 is connected to an image processing device 30 described later, and rotates based on a signal output from the imaging control means 31 of the image processing device 30.

  The camera sets 11 to 13 are configured by a projector 11A and a camera 11B, a projector 12A and a camera 12B, a projector 13A and a camera 13B, and are arranged on the tire inner diameter portion. The camera sets 11 to 13 are arranged so as to be shifted from each other in the tire circumferential direction Cr by an angle α and an angle β. The projectors 11 </ b> A to 13 </ b> A are arranged so that, for example, the belt-shaped slit light made of red laser light is irradiated, and the extending direction of the slit light is the tire width direction W. The cameras 11B to 13B are constituted by, for example, CCD area cameras, and images the slit light irradiation units 11a, 12a, and 13a irradiated by the projectors 11A to 13A with the corresponding cameras. The camera set 11 images the lower tire side T1, the camera set 12 images the tire crown T2, and the camera set 13 images the upper tire side T3. In other words, the tire inner peripheral surface Ts to be inspected is divided into three regions and imaged. The camera sets 11 to 13 capture images of different portions of the tire inner peripheral surface Ts so as to overlap each other (see FIG. 2).

Specifically, the projector 11A of the camera set 11 irradiates slit light in the tire radial direction R and the tire width direction W from the tire side T1 to a range including a part of the tire crown T2, and the projector 12A of the camera set 12 The slit light is irradiated in the width direction of the crown T2, and the projector 13A of the camera set 13 irradiates the slit light in the tire radial direction R and the tire width direction W from the tire side T3 to a range including a part of the tire crown T2. That is, the camera set 11 images one tire side T1 on the turntable 5 side of the tire inner peripheral surface Ts, the camera set 12 images the tire crown T2, and the camera set 13 images the other tire side T3. .
Note that the angle α and the angle β at which the camera sets 11 to 13 are displaced may be determined as appropriate, and the camera sets 11 to 13 may be disposed on the inner diameter portion of the tire T so as to be capable of imaging the tire inner circumferential surface Ts.
In addition, the camera sets 11 to 13 are connected to the image processing device 30, and the imaging control unit 31 controls slit light irradiation by the projectors 11 </ b> A to 13 </ b> A and imaging by the cameras 11 </ b> B to 13 </ b> B.

The images of the tire inner circumferential surface Ts imaged along the tire radial direction by the cameras 11B to 13B include the height information and luminance information of the cross section at the imaged position of the tire inner circumferential surface Ts, and are provided for each camera 11B to 13B. The images are sequentially output to the connected image processing device 30. The images of the tire inner peripheral surface Ts imaged by the cameras 11B to 13B are acquired by dividing the tire inner peripheral surface Ts in the tire circumferential direction Cr, for example, by 10,000 divisions.
The captured image P1 is an image of the tire inner surface on the mounting surface side of the tire inner peripheral surface Ts, and the captured image P2 is an image of the tire crown portion of the tire inner peripheral surface Ts. The captured image P3 is an image of the tire inner surface on the opposite side to the placement surface of the tire inner peripheral surface Ts (see FIG. 3).

The image processing apparatus 30 includes an imaging control unit 31, a captured image preprocessing unit 32, an image processing unit 33, and an appearance inspection unit 34. The image processing apparatus 30 is a so-called computer including a CPU as a calculation means, a ROM and a RAM as storage means, an interface as a communication means, and the like, and operates based on a program stored in the ROM.
The imaging control means 31 controls the imaging of the tire inner peripheral surface Ts by outputting an imaging start signal and an imaging end signal to the camera sets 11 to 13 and the motor 4.

The captured image preprocessing unit 32 performs preprocessing so that each part of the tire inner peripheral surface Ts captured by the camera sets 11 to 13 becomes a belt-shaped image having the same length.
Specifically, since the images captured by the camera sets 11 to 13 are obtained by acquiring the shape curved in the width direction from the coordinates of each camera, the images are represented on one common coordinate. In this way, coordinate conversion is performed, and the image is converted into three strip-shaped captured images P1 to P3 as shown in FIGS. In the captured images P1 to P3, the height information and color tone (luminance) information of the unevenness of the tire inner peripheral surface Ts is stored for each pixel constituting the captured images P1 to P3. A in FIGS. 3A to 3C is a convex portion called a ridge formed by the bladder surface at the time of molding the tire, and is formed periodically on the tire inner circumferential surface Ts.

FIG. 4 shows a block diagram of the image processing means 33 that processes the captured images P1 to P3. 5A and 5B are partially enlarged views when the captured image P2 is subjected to image processing. 6A and 6B are enlarged views when the mask image M is subjected to image processing. FIG. 7A shows the power spectrum image of the captured image P2, FIG. 7B shows the image processing of the mask image M, and FIG. 7C shows the inspection image Q. Hereinafter, image processing will be described with reference to FIG. 4, but will be described with reference to FIGS. Note that X shown in FIGS. 5A and 5B is scratches or dirt on the tire inner peripheral surface Ts.
The image processing unit 33 includes a low frequency component removing unit 51, a binarization processing unit 52, a specific shape detecting unit 53, an angle mode value detecting unit 54, a mask image creating unit 55, and a mask image fitting unit 56. And FFT processing means 57, specific shape removing means 58, and inverse FFT processing means 59.
The low-frequency component removing unit 51 is, for example, a high-pass filter that emphasizes edges in which the brightness of the image changes abruptly from the captured images P1 to P3, and emphasizes ridges A and scratches included in the captured images P1 to P3. And the undulation of the image is removed (see FIG. 5A).
The binarization processing unit 52 is a filter that binarizes an image and performs bit inversion of the binarized image, and the captured images P1 to P3 from which low-frequency components are removed, and a mask image The power spectrum image of M is binarized (see FIGS. 5B and 7B).
The specific shape detection unit 53 detects, for example, a specific shape having a predetermined pixel width and a predetermined length from the binarized captured images P1 to P3. That is, the shape corresponding to the ridge A formed on the tire inner peripheral surface Ts is detected from the captured images P1 to P3. The ridge A is a region that is continuous in the tire width direction and formed in a convex shape with a predetermined width on the tire inner circumferential surface Ts. In the captured images P1 to P3, FIGS. As shown in FIG. 5, images are taken with a predetermined pixel width that is continuous in the tire width direction. Therefore, the detection of the ridge A from the captured images P1 to P3 is performed by setting a threshold value so as to detect a continuous region with a predetermined pixel width and a predetermined pixel length, thereby specifying a specific shape corresponding to the ridge A. Can be detected.
The angle mode detection means 54 calculates the inclination angle θ of the plurality of ridges A detected by the specific shape detection means 53 with respect to the tire circumferential direction, and the most frequent inclination among the calculated inclination angles θ of the ridge A. The angle θ is detected.
The mask image creation means 55 creates a removal target shape as an image from the captured images P1 to P3. For example, in the case of the captured image P2, the specific shape detecting unit 53 creates a straight line having a pixel width α and a continuous pixel length β used for detecting the ridge A as a mask image M. As shown in FIG. 5A, the mask image M is created, for example, in a single linear shape perpendicular to the tire width direction.
Further, when the ridge A includes a curve as in the captured images P1 and P3, the mask image M is created by a curve obtained by adding a curvature to the pixel width α and the continuous pixel length β used for detecting the ridge A. You should do it. The mask image M may be set as a mask image by extracting the shape to be removed from the captured images P1 to P3.
The mask image M may be created with any brightness of a color scale, a gray scale, and a black and white scale. Preferably, when the mask image M is created with a binary brightness of black and white, the processing speed of image processing is improved. In this example, description will be made with a gray scale of 256 gradations. Note that the background of the mask image M is set to, for example, white.
The mask image matching unit 56 rotates the mask image M so as to coincide with the most frequent inclination angle θ detected by the angle mode value detecting unit 54. Specifically, the mask image matching means 56 uses the mode inclination of the ridge A detected by the angle mode value detection means 54 in the mask image M created as shown in FIGS. 6 (a) and 6 (b). Rotate to match the angle θ.
The FFT processing means 57 performs Fourier transform on the captured images P1 to P3 and the mask image M, and outputs the power spectrum image of the mask image M and the power spectrum images of the captured images P1 to P3. The Fourier transform is processed by an algorithm of an FFT transform (Fast Fourier Transform). Since the mask image M consists only of images of the shapes to be removed in the captured images P1 to P3, the power spectrum image of the mask image M shows the mask image M itself. Further, the same periodic shape included in the captured images P1 to P3 is represented at the same position in the power spectrum image (see FIGS. 7A and 7B).
The specific shape removing unit 58 performs processing for removing the power spectrum image of the mask image M from the power spectrum images of the captured images P1 to P3. Specifically, by multiplying the power spectrum image of the captured images P1 to P3 by the bit-inverted image after binarizing the power spectrum image of the mask image M, the ridge A is obtained from the captured images P1 to P3. The mask image M corresponding to is removed.
The inverse FFT processing means 59 performs inspection images Q1 to Q3 in which a plurality of ridges are removed from the captured images P1 to P3 by performing inverse Fourier transform on the power spectrum images of the captured images P1 to P3 processed by the specific shape removing means 58. Is output (see FIG. 7C). The inverse Fourier transform is processed by an algorithm of IFFT transform (Inverse Fast Fourier Transform).

  The appearance inspection means 34 is, for example, a dirt detection means for detecting dirt on the basis of a change in the luminance of the pixels constituting the inspection images Q1 to Q3 or the size thereof, and an inspection image acquired based on the light cutting method. The presence or absence of dirt or scratches is inspected by defect detection means for inspecting scratches based on the height information of the peripheral surface.

Hereinafter, a method for inspecting the tire inner peripheral surface Ts by the tire appearance inspection apparatus 1 will be described.
First, the tire T, which is the subject, is placed on the rotary table 5 sideways by the inspector. At this time, the tire T is placed so that the center thereof coincides with the rotation center of the turntable 5.
Next, the inspector arranges the camera sets 11 to 13 on the inner diameter portion of the tire T, and inputs an inspection start signal to the image processing apparatus 30 from an input unit (not shown).
The imaging control means 31 of the image processing device 30 outputs an imaging start signal to the camera sets 11 to 13 and outputs a rotation start signal of the turntable 5 to the motor 4. Imaging of the tire inner peripheral surface Ts by the camera sets 11 to 13 is controlled so that the rotational speed of the motor 4 and the number of times of imaging of the camera sets 11 to 13 are synchronized so that 10000 times of imaging corresponds to one round of the tire. The The images of the tire inner peripheral surface output by the camera sets 11 to 13 are sequentially output to the image processing device 30.

FIG. 8 shows a flowchart of image processing by the image processing means 33.
When the imaging for one round of the tire is completed, the captured image preprocessing unit 32 performs the front so that each part of the tire inner circumferential surface Ts captured by the camera sets 11 to 13 becomes a belt-shaped image having the same length. Processing is performed, and the images are converted into captured images P1 to P3 as shown in FIGS.
Hereinafter, the image processing by the image processing means 33 will be described. Of the captured images P1 to P3, the captured image P2 shown in FIG. 3B will be described as an example.
First, filter processing for removing low-frequency components of the captured image P2 is performed, the edges of the image included in the captured image P2 are emphasized, and the scratches on the tire inner peripheral surface Ts and the contour of the ridge A are enhanced. (S101).
Next, the captured image P2 from which the low frequency component is removed is binarized (S102).
Next, the ridge A is detected from the binarized captured image P2 (S103). Specifically, as shown in FIG. 5B, since the ridge A is acquired as a periodic appearance of straight lines inclined with respect to the tire circumferential direction in the captured image P2, a threshold value is obtained from the captured image P2. By detecting a region having the shape having the above pixel width α and a pixel length β equal to or greater than the threshold value, a region considered to be ridge A can be detected. That is, in FIG. 5B, the region indicated by the scratch X is excluded from detection as not being a ridge A because the pixel width α and the pixel length β do not satisfy the threshold range.
Next, the inclination angles θ of all the ridges A detected by the angle mode detection means 54 are calculated, and the inclination angle θ having the highest frequency of occurrence among the inclination angles θ is detected (S104).
Next, a mask image M for deleting the ridge A is created from the captured image P2 (S105). Specifically, the average pixel width and average pixel length of the ridge A are obtained from the plurality of ridges A detected by the specific shape detection means 53, and the shape corresponding to the ridge A is orthogonal to the tire circumferential direction. In addition, a straight line having an average pixel width and an average pixel length is created as a mask image M. In the present embodiment, the straight line image is created by, for example, gray scale. It should be noted that the image is preferably created using a black and white binary image. The straight line in the mask image M may be created at any angle in the image. The method for creating the mask image M is not limited to the above, and the pixel length of the mask image M is determined by the width in the tire width direction of the captured image P2 and the most frequent inclination angle θ detected by the angle mode value detecting means 54. The pixel width may be set based on the average pixel width of the ridge A to the ridge A detected by the specific shape detecting means 53.
Next, the mask image M is rotated so that the mask image M created by the mask image creating means 55 is parallel to the most frequent inclination angle θ (S106).
Next, the captured image P2 and the rotated mask image M are Fourier transformed to create a power spectrum image of the captured image P2 and a power spectrum image of the mask image M (S107).
Next, the power spectrum image of the mask image M is binarized and then bit-inverted (S108).
Next, an area corresponding to ridge A is removed from the power spectrum image of the captured image P2 by multiplying the power spectrum image of the captured image P2 by the power spectrum image of the mask image M obtained by bit inversion (S109).
Next, by performing inverse Fourier transform on the power spectrum image of the captured image P2 multiplied by the power spectrum image of the mask image M, an inspection image Q2 from which all the ridges A have been removed from the original captured image P2 is generated ( S110). Therefore, as shown in FIG. 7C, the ridge A is removed from the captured image P2 and only the scratch X is clearly left in the obtained inspection image Q2.
Next, based on the unevenness information on the tire inner peripheral surface of the inspection image acquired based on the stain detection means or the light cutting method based on the change in the luminance of the pixels constituting the inspection image Q2 by the inspection means. Defects are detected by inspection.
By performing the above process on each of the captured images P1 and P3, it is possible to obtain inspection images Q1 and Q3 from which the plurality of ridges are removed from the captured images P1 and P3.

  As described above, after detecting the removal target shape that periodically appears from the captured image of the tire inner peripheral surface, and separately creating an image corresponding to the removal target shape as a mask image, the captured image and the mask image are Apply FFT processing to represent the frequency spectrum, remove the frequency spectrum of the mask image from the frequency spectrum of the captured image, and perform inverse Fourier transform to easily remove the shape to be removed from the captured image of the tire inner peripheral surface Can do. In particular, a mask image corresponding to one of the ridges periodically formed on the inner circumferential surface of the tire is created, and the captured image and the mask image are subjected to FFT processing to be expressed as a frequency spectrum. From the frequency spectrum of the captured image, By removing the frequency spectrum of the mask image and performing inverse Fourier transform, a plurality of ridges are removed at a time, which is efficient.

  As described above, the ridge A periodically formed from the captured image of the tire inner peripheral surface has been described as being removed from the image. Of the tread patterns formed on the tire outer peripheral surface, for example, the tire width formed periodically The lateral groove extending in the direction may be removed, and the appearance of the tire outer peripheral surface may be inspected. By removing the lateral grooves on the tire outer peripheral surface by the above-described method, the lateral grooves are removed from the inspection target in the appearance inspection on the tire outer peripheral surface, so that the inspection efficiency and inspection accuracy for molding defects on the tire outer peripheral surface are improved.

  As mentioned above, although this invention was demonstrated using embodiment, the technical scope of this invention is not limited to the range as described in the said embodiment. Various modifications or improvements can be added to the above embodiment.

1 visual inspection device, 4 motor, 5 rotary table, 10 imaging device,
11-13 camera set, 11A-13A projector, 11B-13B camera,
30 image processing apparatus, 31 imaging control means, 32 captured image preprocessing means,
33 Image processing means, 34 Appearance inspection means, 51 Low frequency component removal means,
52 binarization processing means, 53 specific shape detection means, 54 angle mode value detection means,
55 mask image creation means, 56 mask image adaptation means, 57 FFT processing means,
58 specific shape removing means, 59 inverse FFT processing means,
A ridge, M mask image, P1; P2; P3 captured image,
Q1; Q2; Q3; Q Inspection image, Ts tire inner peripheral surface, θ inclination angle.

Claims (2)

A region having a shape that appears periodically in the captured image of the tire surface imaged along the tire circumferential direction and has a shape that is equal to or larger than the pixel width set by the threshold and that is equal to or larger than the pixel length set to the threshold Detecting from the captured image;
Calculating a tilt angle with respect to a tire circumferential direction in a direction in which the pixel length is continuous for all the detected specific shapes, and detecting a most frequent tilt angle;
Obtaining an average pixel width and an average pixel length from the pixel widths and pixel lengths of all the detected specific shapes;
Creating a mask image of the shape of the average pixel width and the average pixel length so that the direction of the average pixel length is orthogonal to the tire circumferential direction;
Rotating the mask image so as to be parallel to the most frequently inclined angle so that the mask image matches the specific shape ;
Fourier transforming the captured image and the mask image, and converting the captured image and the mask image into a power spectrum image;
Binarizing and bit-inverting the power spectrum image of the mask image;
Multiplying the power spectrum image of the captured image by a power spectrum image obtained by bit-inversion of the mask image;
A method for inspecting a tire appearance, comprising: inverse Fourier transforming a power spectrum image of a captured image multiplied by a power spectrum image obtained by bit-inversion of the mask image . A region having a shape that appears periodically in the captured image of the tire surface imaged along the tire circumferential direction and has a shape that is equal to or larger than the pixel width set by the threshold and that is equal to or larger than the pixel length set to the threshold Specific shape detection means for detecting from a captured image;
For all the detected specific shapes, the inclination angle with respect to the tire circumferential direction in the direction in which the pixel length is continuous is calculated, the most frequent inclination angle is detected, and the pixel widths of all the detected specific shapes In addition, an average pixel width and an average pixel length are obtained from the pixel length, and a mask image having the shape of the average pixel width and the average pixel length so that the direction of the average pixel length is orthogonal to the tire circumferential direction Mask image creating means for creating
A mask image adaptation means for rotating and adapting the mask image so as to be parallel to the most frequently inclined angle so that the mask image matches the specific shape;
FFT processing means for Fourier transforming the mask image and the captured image to convert the mask image and the captured image into a power spectrum image;
Binarization processing means for binarizing the power spectrum image of the mask image and inverting the bit;
Specific shape removing means for multiplying the power spectrum image of the captured image by the power spectrum image of the bit-inverted mask image;
A tire appearance inspection apparatus comprising: an inverse FFT processing unit that performs inverse Fourier transform on a power spectrum image of a captured image multiplied by a power spectrum image of the mask image with the bit inverted.