Image processing method

Abstract

To provide an image processing method that can be used in checking characteristics of a tire on the basis of a contour line of a tire cross-sectional image.SOLUTION: An image processing method of an embodiment is a method for processing a rimmed tire cross-sectional image including the steps of: extracting an entire contour line that is a contour line of an entire rimmed tire cross section; extracting an outside contour line representing an outer face of the rimmed tire from the entire contour line; and extracting an inside contour line representing an inner face of the rimmed tire from the entire contour line.SELECTED DRAWING: Figure 5

Description

Technical Field

[0001] The present invention relates to an image processing method.

Background Art

[0002] As described in Patent Document 1, a tire cross-sectional image is taken with an imaging device such as a CT device, and image processing is performed on the acquired tire cross-sectional image to check the shape of the components of the tire. The tire cross-sectional image can be used not only to check the shape of the components of the tire but also to check various features of the tire.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] By the way, depending on the purpose of using the tire cross-sectional image, it is necessary to extract the contour line of the tire cross-sectional image. However, simply extracting the contour line of the tire cross-sectional image cannot recognize which part of the object each line segment in the contour line or each point forming the contour line belongs to. Therefore, it has been difficult to check the features of the tire based on the contour line of the tire cross-sectional image.

[0005] Therefore, an object of the present invention is to provide an image processing method that can be used when checking the features of a tire based on the contour line of a tire cross-sectional image.

Means for Solving the Problems

[0006] The present invention includes the following embodiments.

[0007] [1] A method for processing a cross-sectional image of a rimmed tire, comprising: a step of extracting an overall contour line which is the contour line of the entire cross-section of the rimmed tire; a step of extracting an outer contour line representing the outer surface of the rimmed tire from the overall contour line; and a step of extracting an inner contour line representing the inner surface of the rimmed tire from the overall contour line. Furthermore, the overall contour line consists of a point cloud, and the extraction of the outer contour line is performed by repeatedly selecting the outermost point in the tire diameter direction among the intersections of a straight line extending radially through the cavity inside the tire and the overall contour line, and the nearest neighbor point of the outer contour line as the next outer point. An image processing method characterized by the following features.

[0009] [ 2 ] A point located at a predetermined distance from the aforementioned outer point is not considered the aforementioned outer point, 1 The image processing method described in [ ].

[0010] [ 3 The overall contour line consists of a point cloud, and the extraction of the inner contour line is performed by taking a straight line that passes through the cavity inside the tire and extends in the radial direction of the tire, excluding the intersection point that is furthest out in the radial direction of the tire and the intersection points that are continuous with it, and designating the intersection point that is furthest out in the radial direction of the tire as an inner point that is part of the inner contour line, and the nearest neighbor point of the inner point as the next inner point, and repeating this process. [1] or [2] The image processing method described above.

[0011] [ 4 ] A point located at a predetermined distance from the aforementioned inner point is not considered the aforementioned inner point, 3 The image processing method described in [ ].

[0012] [ 5 [1] The line consisting of the point cloud remaining after the extraction of the outer contour line is considered to be the inner contour line. or [2] The image processing method described above. [Effects of the Invention]

[0013] The image processing method described above allows the overall contour line to be classified into an outer contour line and an inner contour line, which can then be used to identify the characteristics of the tire. [Brief explanation of the drawing]

[0014] [Figure 1] Figure showing the configuration of the image processing apparatus. [Figure 2] (a) is the overall image before brightness adjustment. (b) is the overall image after brightness adjustment. [Figure 3] (a) is the overall image before rotation. (b) is the overall image after rotation. [Figure 4] Enlarged view of the vicinity of the bead core of the cross-sectional image of the tire with rim. (a) is the image before binarization. (b) is the image after binarization and before dilation and contraction processing. (c) is the image after the first dilation processing on the image of (b). (d) is the image after the first contraction processing on the image of (c). [Figure 5] Figure showing the simplified contour line of the cross-sectional image of the tire with rim. [Figure 6] Figure showing the contour lines before and after dilation and contraction processing. (a) is the cross-sectional image of the tire with rim before dilation and contraction processing. (b) is the figure showing the contour line of the tread after dilation and contraction processing. [Figure 7] Partial enlarged view of the tire cross-sectional image. Figure for explaining the method of determining the inner reference point. [Figure 8] Flowchart of image processing. [Figure 9] Flowchart of brightness adjustment. [Figure 10] Flowchart of the process of rotating the image. [Figure 11] Flowchart of noise processing. [Figure 12] Flowchart of the classification of the outer contour line and the inner contour line. [Figure 13] Flowchart of creating a model for length measurement. [Figure 14] Flowchart of calculating the thickness of the tire. [Figure 15] Cross-sectional image of the tire with rim in which the tire cross-sectional image is on top and the rim cross-sectional image is at the bottom.

Embodiments for Carrying Out the Invention

[0015] Embodiments will be described based on the drawings. Note that the embodiments described below are merely examples, and any modifications made as appropriate without departing from the spirit of the present invention are included within the scope of the present invention.

[0016] Figure 1 shows the image processing device 10 of this embodiment. The image processing device 10 of this embodiment is a device that automatically measures the dimensions of a predetermined part in a cross-sectional image of a pneumatic tire (hereinafter referred to as "tire cross-sectional image").

[0017] The tire cross-sectional image in this embodiment is a cross-sectional image of a pneumatic tire (hereinafter referred to as "tire"), and is an image taken when the cross-section is a plane passing through the perpendicular to the outer surface of the tire and the tire rotation axis. Unless otherwise specified, the tire axis direction in the description of the tire cross-sectional image refers to the direction of extension of the tire rotation axis. In addition, in the tire cross-sectional image, the tire radial direction coincides with the direction perpendicular to the tire axis direction.

[0018] Furthermore, the tire cross-sectional image in this embodiment is an image taken with a CT (Computed Tomography) device that utilizes X-rays.

[0019] The image processing device 10 is implemented by a computer including a processing unit, a storage unit, an input unit, and a display unit. The storage unit includes RAM (Random Access Memory), ROM (Read Only Memory), HDD (Hard Disk Drive), etc. The storage unit stores a program for executing this embodiment, data of tire cross-section images taken by a CT scanner, and image data after various processing steps described later.

[0020] The processing unit consists of a CPU (Central Processing Unit), etc. The processing unit performs image processing in this embodiment by reading a program stored in ROM, etc., onto RAM and executing it. The input unit is, for example, a mouse and keyboard, and receives input from the user of the image processing device 10 (hereinafter referred to as "user"). The display unit is, for example, a display, and displays various data such as tire cross-section images, and input screens from the input unit, etc.

[0021] The program may be distributed by storing it on computer-readable storage media such as CD-ROMs, CD-Rs, or USB drives, or by storing it on an internet-connected computer and allowing it to be downloaded via the internet.

[0022] The image processing device 10 is realized by the processing unit executing a program as described above. As shown in Figure 1, the image processing device 10 functionally includes a grayscale conversion unit 11, a brightness adjustment unit 12, an image rotation unit 13, a binarization processing unit 14, a noise processing unit 15, a contour line detection unit 16, a contour line type unit 17, a length measurement model creation unit 18, and a length calculation unit 19.

[0023] The CT scanner captures cross-sectional images of the tire while it is mounted on the rim. The captured images are cross-sectional images of the tire mounted on the rim at any position in the circumferential direction of the tire. As a result of the capture, a cross-sectional image of the rim portion on the same cross-section (hereinafter referred to as the "rim cross-sectional image") is obtained along with the tire cross-sectional image. The entire acquired image, including the tire cross-sectional image and the rim cross-sectional image, is referred to as the "overall image." Furthermore, the combined image of the tire cross-sectional image and the rim cross-sectional image is referred to as the "tire cross-sectional image with rim." The proportion (area ratio) of the tire cross-sectional image with rim within the overall image is preferably 5% to 15%, and more preferably 8% to 12%.

[0024] In a CT scan, areas that absorb X-rays more easily appear brighter, while areas that transmit X-rays more easily appear darker. Therefore, metal parts such as the bead core and rim appear brighter, while rubber parts (hereinafter referred to as "rubber parts") appear darker. The brightness of the rubber parts varies depending on the compound, resulting in different brightness levels for each type of tire. Similarly, the brightness of the metal parts varies depending on the type of metal. Areas with no visible material will appear black or dark.

[0025] The grayscale conversion unit 11 shown in Figure 1 converts the overall image captured by the CT scanner and taken into the image processing device 10 into a grayscale image. This conversion results in an image with 256 gradations, ranging from 0 (black) to 255 (white). In the following explanation, the numerical value representing the gradation is referred to as luminance. High luminance means closer to white, and low luminance means closer to black.

[0026] The grayscale overall image consists of point cloud data. The point cloud data includes, at a minimum, the coordinates and brightness data of each of the numerous points (pixels) that make up the image. The overall image size is at least 1000 pixels in both height and width. A specific example is 1200 pixels high and 1800 pixels wide. In this embodiment, pixels that make up the image are sometimes referred to as points.

[0027] The brightness adjustment unit 12 shown in Figure 1 adjusts the brightness of the grayscale tire cross-section image to a reference brightness (hereinafter referred to as "reference brightness"). As shown in Figure 1, the brightness adjustment unit 12 includes a tire image extraction unit 12a and a brightness changing unit 12b.

[0028] The tire image extraction unit 12a extracts the rubber portion of the tire cross-section image from the overall image. The rubber portion includes tread rubber, sidewall rubber, bead filler, belt topping rubber, etc. From the results of investigating cross-section images of various tires, it is known in advance what level of brightness the rubber portion of the tire cross-section image will have within the overall image. Therefore, the tire image extraction unit 12a extracts the rubber portion of the tire cross-section image by extracting the portion of the overall image whose brightness falls within a predetermined range. The predetermined range in this case is, for example, a brightness value (the numerical value of the gradation when the overall image is represented by 256 gradations as described above) of 30 or more and 150 or less.

[0029] When the rubber portion of the tire cross-section image is extracted, the brightness adjustment unit 12b first calculates the average brightness of the rubber portion. The average brightness is the value obtained by dividing the sum of the brightness of all pixels that make up the calculation range by the number of pixels that make up the calculation range. Therefore, the average brightness of the rubber portion is the value obtained by dividing the sum of the brightness of all pixels that make up the rubber portion by the number of pixels that make up the rubber portion. Note that if image processing is to be performed on multiple tire cross-section images, the average brightness is calculated for each tire cross-section image.

[0030] Next, the brightness adjustment unit 12b uses the average brightness of the rubber portion to adjust the brightness of the tire cross-section image to approximate the reference brightness. To do this, the brightness adjustment unit 12b sets the scaling factor to the value obtained by dividing the reference brightness by the average brightness of the rubber portion (i.e., (reference brightness) / (average brightness of the rubber portion)). Then, for each pixel of the rubber portion in the overall image, the brightness adjustment unit 12b performs a calculation (scaling calculation) by multiplying the brightness of the pixel by the scaling factor. Through the scaling calculation, the brightness of the rubber portion approaches the reference brightness.

[0031] Furthermore, there is almost no variation in brightness between tires or between images taken in the rim cross-section image and the background, so there is little need to perform scaling calculations on the rim cross-section image and the background. Therefore, in this embodiment, scaling calculations are performed only on the rubber portion of the overall image.

[0032] As the brightness of the tire cross-section image approaches the reference brightness, if the tire cross-section image was originally relatively dark, it will become brighter overall, and if it was originally relatively bright, it will become darker overall. For example, the overall image, which was originally dark as shown in Figure 2(a), will become a brighter image as shown in Figure 2(b). In this way, as the brightness of the tire cross-section image approaches the reference brightness, all tire cross-section images will become approximately the same brightness.

[0033] The image rotation unit 13 shown in Figure 1 rotates the overall image on its plane so that the tire axis direction (hereinafter referred to as the "tire axis direction," and shown by arrow D in Figure 3) on the tire cross-section image is parallel to the reference line. At this time, the image rotation unit 13 is positioned so that the tire cross-section image is at the top and the rim cross-section image is at the bottom. Here, the reference line is the X-coordinate axis.

[0034] The cross-sectional image of a rimmed tire is represented in a Cartesian coordinate system consisting of mutually orthogonal X and Y coordinate axes (shown in Figure 3(a)). However, as shown in Figure 3(a), the cross-sectional image of a rimmed tire immediately after being captured by the image processing device 10 has its axis direction tilted with respect to the X coordinate axis, which serves as the reference line. Therefore, as described above, the image rotation unit 13 rotates the overall image so that the axis direction of the tire cross-sectional image becomes parallel to the X coordinate axis, which serves as the reference line.

[0035] As shown in Figure 1, the image rotation unit 13 includes a bead core detection unit 13a and a rotation unit 13b.

[0036] The bead core detection unit 13a detects bead cores from the overall image. In the overall image, the bead core is the part with the highest brightness. Therefore, the bead core detection unit 13a detects the part with a brightness above a predetermined threshold as a bead core. As the threshold, for example, 250 is used, which is the number of grayscale levels when the overall image is represented by 256 grayscale levels.

[0037] Since the bead core is located on both the serial side and the non-serial side of the tire, the bead core detection unit 13a detects two bead cores from the overall image. The serial side refers to the side of the tire axial direction where the serial lines are displayed on the surface. The non-serial side refers to the side of the tire axial direction opposite to the serial side.

[0038] Furthermore, the bead core detection unit 13a calculates the coordinates of the centroids of the bead cores on both the serial and anti-serial sides. The X-coordinate of the centroid of the bead core is obtained by dividing the sum of the X-coordinates of each pixel forming the bead core by the number of pixels forming the bead core. The Y-coordinate of the centroid of the bead core is obtained by dividing the sum of the Y-coordinates of each pixel forming the bead core by the number of pixels forming the bead core. The bead core detection unit 13a calculates the coordinates of the centroids of the bead cores on both the serial and anti-serial sides using this calculation method.

[0039] The rotation unit 13b of the image rotation unit 13 first finds a straight line connecting the centroids of the bead cores on the serial side and the non-serial side. If the centroid coordinates of the bead core on the serial side are (x1, y1) and the centroid coordinates of the bead core on the non-serial side are (x2, y2), then the straight line can be expressed by the following equation.

[0040]

number

[0041]

number

[0042] Once the rotation angle θ' is determined in this way, the rotating unit 13b rotates the overall image, including the cross-sectional image of the tire with a rim. The rotation angle is the rotation angle θ' that was determined earlier. Since the rotation is performed on the plane of the overall image, the image rotation axis is an axis that extends in a direction perpendicular to the planar overall image.

[0043] Rotation causes the tire axis direction in the tire cross-section image to become the X-coordinate direction, and the tire diameter direction in the tire cross-section image to become the Y-coordinate direction. In other words, even if the tire axis direction was tilted relative to the X-coordinate direction before rotation, as shown in Figure 3(a), after rotation the tire axis direction becomes the X-coordinate direction, as shown in Figure 3(b). After rotating the overall image, cropping is performed as needed.

[0044] The binarization processing unit 14 shown in Figure 1 performs binarization of the entire image. In the binarization process, a luminance value greater than 0 and less than 255 is set as the threshold. The method for determining the threshold can be selected as appropriate. Pixels with a luminance greater than the threshold are converted to white pixels with a signal value of 1. Pixels with a luminance less than the threshold are converted to black pixels with a signal value of 0. As a result, the tire cross-section image and rim cross-section image, which originally had high luminance, become white, and the background, which originally had low luminance, becomes black.

[0045] Figure 4(a) is a magnified view of the area near the bead core in a cross-sectional image of a rimmed tire before binarization. In Figure 4(a), it can be seen that the brightness varies from place to place (coordinate). For example, the bead core and rim have high brightness, the background has low brightness, and the rubber part has intermediate brightness. This image is converted into an image consisting only of black and white, as shown in Figure 4(b), through binarization.

[0046] The noise processing unit 15 shown in Figure 1 removes noise from the binarized cross-sectional image of the rimmed tire. First, the noise processing unit 15 saves the coordinates and signal values ​​of a predetermined range within the binarized cross-sectional image of the rimmed tire as image data before expansion and contraction processing. This predetermined range is referred to as the "save range".

[0047] The saved range is the area where the original contour lines of the rimmed tire cross-section image may be distorted by the subsequent expansion and contraction process. The saved range is a predetermined shape, such as a rectangle. The saved range may be selected manually by the user or automatically by recognizing the shape. In this embodiment, the area enclosed by the square frame in Figure 4(b), that is, the area including the gap between the flange and the tire, is selected as the saved range.

[0048] The noise processing unit 15 performs dilation and deflation on the entire image after binarization, following the saving of the image data within the specified range. During the dilation process, a kernel with a predetermined shape and size is scanned within the image. At each position during scanning, if the kernel centered on the pixel of interest contains at least one pixel with a signal value of 1 (white), the signal value of the pixel of interest is set to 1 (i.e., if the signal value of the pixel of interest was originally 1, it remains 1, and if it was 0, it is converted to 1). As a result, the white areas are expanded by the dilation process. In addition, small black noises (such as missing pixels) are eliminated by the dilation process.

[0049] During the scaling process, the kernel described above is scanned within the image. At each scanning position, if the kernel centered on the pixel of interest contains even one pixel with a signal value of 0 (black), the signal value of the pixel of interest is set to 0 (i.e., if the signal value of the pixel of interest was originally 0, it remains 0, and if it was 1, it is converted to 0). As a result, the white areas are scaled by the scaling process. Additionally, small white noises are eliminated by the scaling process.

[0050] The noise processing unit 15 first performs an expansion process, followed by a contraction process. The image after the expansion process is shown in Figure 4(c), and the image after the subsequent contraction process is shown in Figure 4(c). Although not shown in the illustration, the contraction process after the expansion process fills in any gaps in the cross-sectional image of the tire with a rim.

[0051] Next, the noise processing unit 15 performs a shrinking process, followed by an expansion process. By performing the expansion process after the shrinking process, small white noises and other noises that were present around the cross-sectional image of the tire with a rim are removed. In this way, the noise processing unit 15 performs the processes in the order of expansion, shrinking, shrinking, and expansion.

[0052] In this embodiment, the shape of the kernel for the expansion and contraction process is, for example, a circle or a rhombus, and the size of the kernel (radius if the kernel is a circle, or the length from the center to the vertex (the vertex furthest from the center) if it is a rhombus) is, for example, 10 pixels.

[0053] The above expansion and contraction process removes noise present in the overall image. However, in the saved area enclosed by the square frame in Figure 4(b), the contour of the rimmed tire cross-section image after expansion and contraction is deformed from the contour before expansion and contraction.

[0054] Therefore, the noise processing unit 15 replaces the image data of the saved range within the overall image after expansion and contraction processing with the saved image data before expansion and contraction processing. As a result, the signal values ​​of each coordinate within the saved range are returned to the signal values ​​before expansion and contraction processing. Consequently, the image within the saved range becomes the image before expansion and contraction processing, and the image in other areas becomes the image after expansion and contraction processing. In the saved range, the contour of the tire cross-section image with a rim appears correctly, and in other areas, the noise is removed and the contour of the tire cross-section image with a rim appears correctly. Note that since the saved range is originally a range with little noise, there is no problem even if the image data of the saved range is replaced with the image data before expansion and contraction processing in this way.

[0055] The contour detection unit 16 shown in Figure 1 detects contour lines of a rimmed tire cross-section image from the overall image after it has been binarized and noise-processed. Detection is performed using well-known methods such as the use of an edge function. The detected contour lines consist of a point cloud. The entire set of detected contour lines is referred to as the "overall contour line".

[0056] Figure 5 shows a simplified representation of the detected overall contour (the actual detected overall contour is equivalent to what a human can recognize from images such as Figure 2, and is smoother than the overall contour in Figure 5). Each square in the figure represents one point (pixel). The gray squares in the figure are the points that form the overall contour. The thickness of the overall contour is basically 1 pixel, but there are places where it is 2 pixels, as seen in the figure where B and N101 are consecutive vertically.

[0057] The contour line classification section 17 shown in Figure 1 classifies the overall contour lines into outer contour lines 30 and inner contour lines 31, as shown in Figures 5 and 6. The outer contour line 30 represents the outer surface of the rimmed tire. The inner contour line 31 represents the inner surface of the rimmed tire. Both the outer contour line 30 and the inner contour line 31 are annular lines. The inner contour line 31 is located inside the outer contour line 30.

[0058] The contour line classification unit 17 extracts the outer contour line 30 from the point cloud representing the overall contour line. First, the contour line classification unit 17 draws a tire axis-axis centerline on the cross-sectional image of the rimmed tire. The tire axis-axis centerline is a straight line that passes through the midpoint of the maximum and minimum coordinates in the X-coordinate direction (tire axis direction) of the overall contour line (however, the overall contour line of the rimmed tire image where the tire axis direction is parallel to the X-coordinate axis) and is parallel to the Y-coordinate direction (tire diameter direction). In Figure 5, the tire axis-axis centerline is shown as a dashed line.

[0059] Next, the contour line type section 17 designates the point furthest outward in the tire radial direction among the intersections of the tire axial center line and the overall contour line (i.e., the point with the largest Y coordinate) as the first outer point. An outer point is a point that forms the outer contour line 30. In Figure 5, point A is the point furthest outward in the tire radial direction among the intersections of the tire axial center line and the overall contour line.

[0060] Next, the contour line classification unit 17 searches for the point closest to the first outer point (nearest neighbor) from among the points included in the point cloud that form the overall contour line but have not yet been identified as an outer point, and designates the found nearest neighbor as the second outer point. In Figure 5, such a second outer point is point N1.

[0061] Next, the contour line classification unit 17 searches for the point closest to the second outer point (nearest neighbor) from among the points included in the point cloud that form the overall contour line and have not yet been identified as an outer point, and sets the found nearest neighbor as the third outer point. In Figure 5, such a third outer point is point N3. In this way, the contour line classification unit 17 repeatedly sets the nearest neighbor of the previously identified outer point (however, a point that has not been identified as an outer point up to that point) as the next outer point.

[0062] If we let the coordinates of the previously identified outer point be (x0, y0) and the coordinates of the other points that make up the overall contour be (xn, yn) (n=1, 2, 3, ...), then the distance Ln between the previously identified outer point and the other points can be calculated using the following formula. The point with the smallest distance Ln is the nearest neighbor of the previously identified outer point.

[0063]

number

[0064] In Figure 5, the nearest neighbors N4, N5, etc. are identified one after another, until finally point N100 is identified as the nearest neighbor of point N99. Among the points that have not yet been identified as outer points by the time point N100 is identified, the closest point to point N100 is point B, but point B is more than 2 pixels away from point N100 and cannot be selected. Therefore, the search for outer points ends when the last nearest neighbor point N100 is identified. The outer points identified by the end of the search are defined as the points that form the outer contour line 30.

[0065] The contour line classification unit 17 also extracts the inner contour line 31. The contour line classification unit 17 designates the second point from the outermost side in the tire radial direction among the intersections of the tire axial center line and the overall contour line as the first inner point. An inner point is a point that forms the inner contour line 31. In Figure 5, the second point from the outermost side in the tire radial direction among the intersections of the tire axial center line and the overall contour line is point B.

[0066] Next, the contour line classification unit 17 searches for the point closest to the first inner point (nearest neighbor) from among the points included in the point cloud that form the overall contour line but have not yet been identified as an inner point, and sets the found nearest neighbor as the second inner point. In Figure 5, such a second inner point is point N101.

[0067] Next, the contour line classification unit 17 searches for the point closest to the second inner point (nearest neighbor) from among the points included in the point cloud that form the overall contour line and that have not yet been identified as an inner point, and sets the found nearest neighbor as the third inner point. In Figure 5, such a third inner point is point N102. In this way, the contour line classification unit 17 repeatedly sets the nearest neighbor of the previously identified inner point (however, a point that has not been identified as an inner point up to that point) as the next inner point. The method for calculating the distance between the previously identified inner point and other points is the same as the formula in equation 3 above.

[0068] In the process of identifying inner points, points that are more than a predetermined distance away from previously identified inner points are not selected as inner points. Therefore, even if a point is the nearest neighbor to a previously identified inner point, if it is more than a predetermined distance away from that point, it will not be selected as the next inner point. The predetermined distance is, for example, 2 pixels. This measure prevents points that are far from the actual outer contour line 30 (for example, points that form the inner contour line 31) from being selected as inner points. Furthermore, this measure causes the contour line classification unit 17 to terminate the search for inner points when it becomes impossible to identify the next inner point, for example, because the nearest neighbor is more than a predetermined distance (for example, 2 pixels) away.

[0069] In Figure 5, the nearest neighbors N103, N104, etc. are identified one after another, until finally point N200 is identified as the nearest neighbor of point N199. Among the points not yet identified as interior points, the closest to point N200 is point N100, but point N100 is more than 2 pixels away from point N200 and cannot be selected. Therefore, the search for interior points ends when the last nearest neighbor point N200 is identified. The interior points identified by the end of the search are defined as the points that form the interior contour line 31.

[0070] The length measurement model creation unit 18 shown in Figure 1 creates a model (hereinafter referred to as the "length measurement model") for measuring the thickness of the tire at a predetermined position. As can be seen from Figure 6(a), in the original tire cross-section image, multiple main grooves 34 appear on the outer contour line 30 of the tread portion. When the original tire cross-section image is 1200 pixels vertically and 1800 pixels horizontally, the width of the main grooves 34 is, for example, 200 pixels, and the depth of the main grooves 34 is, for example, 50 pixels.

[0071] However, when measuring the thickness of a tire, the measurement position is generally set at a predetermined distance along the outer surface of the tire from a predetermined starting point on the outer surface of the tire in the tire cross-section. And, in that case, the "predetermined distance" is generally measured while ignoring the main groove 34 (i.e., as if the main groove 34 does not exist). Therefore, when attempting to automatically determine the measurement position for measuring the thickness of a tire, the presence of the main groove 34 in the tire cross-section image is an obstacle. To address this, the main groove 34 is filled in by the length measurement model creation unit 18, and the line representing the outer surface of the tire in the length measurement model is changed to a single smooth line 35 (see Figure 6(b)).

[0072] As shown in Figure 1, the length measurement model creation unit 18 includes a smooth line creation unit 18a and a length calculation unit 18b.

[0073] The smoothing line creation unit 18a performs expansion and contraction processing on a selected area within the overall image. Here, the selected area is a sufficiently large area that includes the part along which the length along the smoothing line 35 is to be measured, for example, an area that includes the entire tire tread. This area can be selected manually or automatically.

[0074] The shape of the kernel for the expansion and contraction process by the smoothing line creation unit 18a is, for example, a circle or a rhombus. The size of the kernel for the expansion and contraction process (radius if the kernel is a circle, or length from the center to the vertex (the vertex furthest from the center) if the kernel is a rhombus) is such that it can fill the main groove 34 in the tire cross-sectional image with one expansion and contraction process, respectively. Furthermore, since it is not necessary to expand or contract beyond the thickness of the tire, the size of the kernel is less than or equal to the thickness from the outer contour line 30 to the inner contour line 31 in the tread. Here, thickness refers to the length in the normal direction of the inner contour line 31. Also, the thickness referred to here is, for example, the thickness along the center line in the tire axial direction. The size of the kernel corresponds to the amount of expansion in the expansion process and the amount of contraction in the contraction process.

[0075] In this embodiment, the kernel size is, for example, 100 pixels. The smoothing line creation unit 18a performs an expansion process once, followed by a contraction process once.

[0076] Through this expansion and contraction process, the main grooves 34 that were visible in the outer contour line 30 of the tire cross-section image are filled in, and the outer contour line 30 of the selected region becomes a smooth line 35 as shown in Figure 6(b). Outside the selected region, the outer contour line 30 in the tire cross-section image is smooth even without the expansion and contraction process performed by the smooth line creation unit 18a. Hereafter, the entire outer contour line 30 in the tire cross-section image will be referred to as a smooth line 35.

[0077] The length calculation unit 18b determines the length along the smooth line 35, starting from the intersection point of the smooth line 35 created by the smooth line creation unit 18a and the tire axial center line, to each position (point) on the smooth line 35.

[0078] First, the length calculation unit 18b sets the intersection of the smooth line 35 and the tire axial center line (dotted line) as the starting point A, as shown in Figures 5 and 6 (Note that Figure 5 was used in the explanation of the classification of the outer contour line 30 and the inner contour line 31, but for convenience, Figure 5 will also be used here). Next, the length calculation unit 18b searches for the nearest point to the starting point A and identifies point N1 as the nearest point. Next, the length calculation unit 18b calculates the distance from the starting point A to point N1 (referred to as the "distance between A and N1").

[0079] Next, the length calculation unit 18b searches for the nearest neighbor of point N1 and identifies point N2 as the nearest neighbor. Note that the search range does not include points that have already been identified (points A and N1). Next, the length calculation unit 18b calculates the distance from point N1 to point N2 (referred to as the "distance between N1 and N2"). Next, the length calculation unit 18b adds the distance between N1 and N2 to the distance between A and N1, and takes this value as the length along the smoothed line 35 from the starting point A to point N2.

[0080] In this manner, the length calculation unit 18b repeatedly identifies the nearest point to the previously identified point and calculates the distance between those two points. The length calculation unit 18b also integrates the distances between the two points that have been calculated so far, and uses this integrated value as the length along the smoothed line 35 from the starting point A.

[0081] Note that the search range for the nearest neighbor does not include points that have already been identified. Also, the search for the nearest neighbor is performed in one direction along the smooth line 35 (i.e., in the clockwise or counterclockwise direction along the outer ring contour line 30). Furthermore, since the smooth line 35 is basically formed by consecutive points (pixels), the search range for the nearest neighbor is a range of 2 pixels or less. Therefore, the distance between the two points mentioned above is a distance of 2 pixels or less.

[0082] Table 1 shows the lengths along the smoothed line 35 from the starting point A to each point, which are determined in this way. N1, N2, etc. listed in Table 1 are the same as N1, N2, etc. in Figure 5.

[0083] [Table 1] Using such a table, it is possible to identify a point that is a predetermined distance away from the starting point A along the smoothing line 35. For example, if we want to identify a point (pixel) that is 4 pixels away from the starting point A along the smoothing line 35, the point whose combined distance from the starting point A is closest to 4 will be identified. Based on Table 1, that point is point N4 (see Figure 5).

[0084] Furthermore, if the actual length of one pixel on a tire is known in advance, it is possible to identify which point on the smooth line 35 corresponds to a predetermined distance from the starting point A on the actual tire. For example, if it is known in advance that one pixel corresponds to 1 mm, and we want to identify a position 8 mm away from the starting point A, the point closest to 8 mm (= 8 pixels) in the cumulative distance between two points from the starting point A will be identified. Based on Table 1, that point is point N7.

[0085] The length calculation unit 19 shown in Figure 1 includes a reference point determination unit 19a and a tire thickness calculation unit 19b. The reference point determination unit 19a determines the outer reference point on the outer contour line 30 and the inner reference point on the inner contour line 31. Then, the tire thickness calculation unit 19b calculates the length from the outer reference point to the inner reference point, and calculates the tire thickness based on that length.

[0086] The optimal method for determining the outer and inner reference points differs depending on where the thickness is measured. There are three methods for determining these points.

[0087] The first method is primarily used to calculate the thickness at the center position in the tire axial direction and at the maximum width position of the tire. In the first method, the reference point determination unit 19a uses the intersection point of a straight line drawn on the tire cross-sectional image and the outer contour line 30 as the outer reference point, and the intersection point of the same straight line and the inner contour line 31 as the inner reference point.

[0088] When calculating the thickness at the center position in the tire axial direction, the reference point determination unit 19a designates the intersection of the tire axial center line and the outer contour line 30 as the outer reference point (indicated by reference numeral 32 in Figure 5 for reference), and the intersection of the tire axial center line and the inner contour line 31 as the inner reference point (indicated by reference numeral 33 in Figure 5 for reference). Alternatively, the point with the largest Y-coordinate value among the intersections of the tire axial center line and the overall contour line may be identified as the outer reference point, and the point with the second largest Y-coordinate value may be identified as the inner reference point.

[0089] Furthermore, when calculating the tire thickness at the position of maximum tire width (the position where the length in the tire axis direction is longest), a virtual line parallel to the tire axis direction is first assumed. Here, it is assumed that the tire axis direction is parallel to the X coordinate axis, and that the virtual line is a straight line parallel to the X coordinate axis with a constant Y coordinate value. The reference point determination unit 19a changes the Y coordinate value of such a virtual line and finds the difference between the maximum and minimum X coordinate values ​​at the intersection points of the virtual line and the outer contour line 30. Then, the reference point determination unit 19a identifies the virtual line when this difference is maximum as a straight line passing through the position of maximum tire width.

[0090] Next, the reference point determination unit 19a determines the coordinates of the intersection point between the straight line passing through the tire's maximum width position, as determined in this way, and the outer contour line 30, and further determines the coordinates of the intersection point between the straight line passing through the same tire's maximum width position and the inner contour line 31. From these determined coordinates, the reference point determination unit 19a calculates the tire's thickness at the tire's maximum width position.

[0091] The second method is suitable for calculating the thickness of the tire's tread rubber and sidewall rubber. In the second method, first, a point on the smoothing line 35 is identified as the point corresponding to the outer reference point. Specifically, the user inputs the distance from the starting point A along the smoothing line 35 that will be used as the outer reference point.

[0092] The reference point determination unit 19a identifies the point (pixel) closest to the starting point A by the input distance from the point cloud forming the smooth line 35 as the point corresponding to the outer reference point (referred to as the "corresponding point"). At this time, the reference point determination unit 19a also identifies the coordinates of the corresponding point. Here, the point closest to the starting point A by the input distance is identified based on Table 1. For example, if the input distance is 4 pixels, point N4, which is the point whose combined distance from the starting point A is closest to 4, is identified as the point closest to the starting point A by the input distance (4 pixels).

[0093] Next, the reference point determination unit 19a identifies a point with the same coordinates as the corresponding point in the image before it is used as a length measurement model (the image before the outer contour line 30 is smoothed out to form a smooth line 35) as the outer reference point. Here, the outer reference point is often a point on the outer contour line 30, but it may also be a point in the location of the tire's main groove 34, where there is no outer contour line 30.

[0094] Note that the images before and after changing the outer contour line 30 to the smooth line 35 are the same size and have the same coordinates. The only difference between the images before and after changing the outer contour line 30 to the smooth line 35 is the portion where the smooth line 35 is located.

[0095] Next, the reference point determination unit 19a draws a straight line (perpendicular line) through the outer reference point to the inner contour line 31 in the image before it is used as a length measurement model, and selects the point closest to that perpendicular line among the point group forming the inner contour line 31 as the inner reference point. Specifically, the reference point determination unit 19a first extracts several points close to the outer reference point from the point group forming the inner contour line 31. In Figure 7, the outer reference point is indicated by reference numeral 32, and the several points close to the outer reference point are indicated by the other black dots. However, the number and arrangement of points in Figure 7 are simplified for illustrative purposes.

[0096] Here, "multiple points close to the outer reference point" refers to multiple points within a predetermined range from the outer reference point. When the outer reference point is on the tread contact surface, the predetermined range is, for example, within ±10 pixels in the X coordinate. When the outer reference point is on the sidewall surface, the predetermined range is, for example, within ±10 pixels in the Y coordinate. When the outer reference point is at or near the tread contact edge, the predetermined range is, for example, 30 to 50 pixels in the direction of the tire axial centerline in the X coordinate (that is, in this case, all "multiple points close to the outer reference point" are located further in the direction of the tire axial centerline than the outer reference point).

[0097] Next, the reference point determination unit 19a approximates several points close to the outer reference point, extracted from the point cloud forming the inner contour line 31, with a straight line (referred to as the "approximation line"). In Figure 7, the approximation line is the dashed line labeled L1. Next, the reference point determination unit 19a finds a straight line that passes through the outer reference point and intersects the approximation line perpendicularly. This line can be considered a perpendicular to the inner contour line 31. In Figure 7, this perpendicular is the dashed line labeled L2. Next, the reference point determination unit 19a calculates the distance between the perpendicular to the inner contour line 31 and each point forming the inner contour line 31, and identifies the point with the shortest distance as the inner reference point. In Figure 7, the inner reference point thus identified is indicated by the symbol 33.

[0098] The third method is suitable for calculating the thickness of the tire buttresses and rim strips. In the third method, the point closest to a specified position on the tire cross-sectional image is identified as the outer reference point from the point cloud forming the outer contour line 30.

[0099] Specifically, first, the user identifies the coordinates near the point they want to select as an outer reference point on the tire cross-section image. For example, on the tire cross-section image with the outer contour line 30, the user clicks with the mouse near the point they want to select as an outer reference point. The coordinates of the clicked position are then identified. Next, the reference point determination unit 19a calculates the distance between the identified coordinates and each point on the outer contour line 30. Then, the reference point determination unit 19a identifies the point on the outer contour line 30 with the shortest distance as the outer reference point.

[0100] The method for identifying the inner reference point in the third method is basically the same as in the second method. Specifically, the reference point determination unit 19a first extracts multiple points close to the outer reference point from the point cloud that forms the inner contour line 31. Multiple points close to the outer reference point are multiple points that are within a predetermined range based on the outer reference point. When the outer reference point is on the buttress, the predetermined range is, for example, an area 20 to 60 pixels away from the outer reference point in the direction of the tire axis center in the X coordinate. When the outer reference point is on the rim strip, the predetermined range is, for example, an area 20 to 50 pixels away from the outer reference point in the direction of the tire rotation axis (direction where the Y coordinate value is 0) in the Y coordinate.

[0101] Next, the reference point determination unit 19a approximates several points close to the outer reference point, extracted from the point cloud forming the inner contour line 31, with a straight line (approximation line). Next, the reference point determination unit 19a finds a straight line (perpendicular line) that passes through the outer reference point and intersects the approximation line perpendicularly. Next, the reference point determination unit 19a calculates the distance between the perpendicular line to the inner contour line 31 and each point forming the inner contour line 31, and defines the point with the shortest distance as the inner reference point.

[0102] After the outer and inner reference points are determined by one of the methods described above, the tire thickness calculation unit 19b calculates the distance from the outer reference point to the inner reference point (this distance is referred to as the "distance between reference points"). The unit of the distance between reference points obtained in this calculation is pixels.

[0103] Next, the tire thickness calculation unit 19b converts the distance between reference points into the actual thickness of the tire. For this conversion, a coefficient is used that shows the relationship between the number of pixels in the overall image and the length of the actual tire. Such a coefficient is predetermined by known methods, such as determining it from the relationship between the length of a predetermined part of a reference object and the number of pixels in that predetermined part in an image of that object captured by the image processing device 10.

[0104] Based on the flowcharts in Figures 8 to 14, the tire thickness is calculated using the image processing device 10 described above.

[0105] First, the cross-sectional image of the tire with the rim, captured by the CT scanner, is taken into the image processing device 10 and converted into a 256-level grayscale image by the grayscale conversion unit 11 (S1 in Figure 8). The proportion of the cross-sectional image of the tire with the rim within the overall image is preferably 5% to 15%, as described above.

[0106] Next, the brightness adjustment unit 12 adjusts the brightness of the overall image (S2 in Figure 8). Specifically, first, as a tire image extraction step, the rubber portion of the tire cross-section image is extracted from the overall image (S2-1 in Figure 9). The extraction of the rubber portion is performed by extracting a portion of the overall image whose brightness is within a predetermined range (for example, 30 or more and 150 or less). Next, as an average brightness calculation step, the average brightness of the rubber portion is calculated (S2-2 in Figure 9).

[0107] Next, as a brightness adjustment step, the brightness of the tire cross-section image is adjusted to approximate the reference brightness using the average brightness of the rubber portion of the tire cross-section image. To do this, first, a scaling factor is calculated by dividing the reference brightness by the average brightness of the rubber portion (S2-3 in Figure 9). Next, a scaling calculation is performed for each pixel of the rubber portion of the tire cross-section image, multiplying the pixel's brightness by the scaling factor (S2-4 in Figure 9). Through this scaling calculation, the brightness of the rubber portion of the tire cross-section image approaches the reference brightness. As a result, all tire cross-section images become roughly the same brightness.

[0108] Next, the image rotation unit 13 automatically rotates the entire image so that the axis direction of the tire cross-section image is parallel to the X coordinate axis (S3 in Figure 8). In detail, first, the portion of the entire image whose brightness is above a predetermined threshold is detected as the bead core of the tire cross-section image (S3-1 in Figure 10). Two bead cores are detected. Next, the centroid coordinates of the two bead cores are calculated and identified (S3-2 in Figure 10). Next, a straight line connecting the centroids of the two bead cores is identified (S3-3 in Figure 10), and the slope of that line is determined (S3-4 in Figure 10).

[0109] Next, based on the determined inclination, the angle between the line connecting the centers of gravity of the bead cores and the X-coordinate axis is calculated, and the rotation angle θ' is determined based on this angle (S3-5 in Figure 10). Next, the overall image is rotated by the determined rotation angle θ' (S3-6 in Figure 10). As a result of this rotation, the tire axis direction of the tire cross-section image becomes the X-coordinate direction, and the tire diameter direction of the tire cross-section image becomes the Y-coordinate direction. Also, the tire cross-section image is at the top and the rim cross-section image is at the bottom.

[0110] Next, the binarization processing unit 14 performs binarization on the overall image after brightness adjustment and rotation (S4 in Figure 8). Through binarization, the pixels in the tire cross-section image and rim cross-section image become white pixels with a signal value of 1, while the background pixels become black pixels with a signal value of 0.

[0111] Next, the noise processing unit 15 removes noise from the binarized cross-sectional image of the rimmed tire (S5 in Figure 8). Specifically, the user selects the range within the binarized overall image that includes the gap between the flange and the tire as the save range (S5-1 in Figure 11). Then, the coordinates and signal values ​​of each pixel in the selected save range are saved as image data before the dilation and deflation process (S5-2 in Figure 11).

[0112] Next, the entire image after binarization is subjected to dilation and condensation (S5-3 in Figure 11). The dilation and condensation process removes noise present in the entire image. However, the dilation and condensation process deforms the contours of the portion corresponding to the saved area.

[0113] Next, the image data corresponding to the saved area in the overall image after expansion and contraction is replaced with the image data of the saved area before expansion and contraction (S5-4 in Figure 11). As a result, the image in the saved area becomes the image before expansion and contraction, and the image in the other areas becomes the image after expansion and contraction. Consequently, the entire contour of the rimmed tire cross-section image is correctly displayed. Furthermore, since there is little noise in the saved area even before expansion and contraction, and noise is removed in the areas outside the saved area by expansion and contraction, the overall image becomes less noisy.

[0114] Next, the contour line detection unit 16 detects the contour lines of the rimmed tire cross-section image from the binarized and noise-processed overall image (S6 in Figure 8). The detected overall contour lines consist of a point cloud.

[0115] Next, the contour line classification unit 17 classifies the overall contour line into an annular outer contour line 30 representing the outer surface of the rimmed tire and an annular inner contour line 31 representing the inner surface of the rimmed tire (S7 in Figure 8).

[0116] In detail, first, the outer contour line 30 is extracted from the point cloud representing the overall contour line (S7-1 to S7-4 in Figure 12). In extracting the outer contour line 30, the point that is furthest outward in the tire radial direction among the intersections of the overall contour line and the tire axial center line (for example, point A in Figure 5) is identified as the first outer point on the outer contour line 30 (S7-1 in Figure 12).

[0117] Next, it is checked whether there are any other points not identified as outer points within a predetermined distance (2 pixels) from the first outer point (S7-2 in Figure 12). If there are any points not identified as outer points within the predetermined distance (No. in S7-2 in Figure 12), the point closest to the first outer point (nearest neighbor) is identified as the second outer point (S7-3 in Figure 12).

[0118] Next, it is checked whether there are any other points not identified as outer points within a predetermined distance from the second outer point (S7-2 in Figure 12). If there are any points not identified as outer points within the predetermined distance (No. in S7-2 in Figure 12), the point closest to the second outer point (nearest neighbor) is identified as the third outer point (S7-3 in Figure 12).

[0119] Thus, as long as there is a point within a predetermined distance that has not been identified as an outer point (No. in S7-2 of Figure 12), the point closest to the previously identified outer point (nearest neighbor) is identified as the next outer point (S7-3 of Figure 12), and this process is repeated.

[0120] Then, when there are no more points within a predetermined distance that have not been identified as outer points (Yes in S7-2 of Figure 12), the search for outer points is completed, and the line consisting of the points identified as outer points up to that point is extracted as the outer contour line 30 (S7-4 of Figure 12).

[0121] Next, the inner contour line 31 is extracted from the point cloud representing the overall contour line (S7-5 to S7-8 in Figure 12). In extracting the inner contour line 31, first, the second point from the outermost side in the tire radial direction among the intersection points of the overall contour line and the tire axial center line (for example, point B in Figure 5) is identified as the first inner point (S7-5 in Figure 12).

[0122] Next, it is checked whether there are any other points not identified as interior points within a predetermined distance (2 pixels) from the first interior point (S7-6 in Figure 12). If there are any points not identified as interior points within the predetermined distance (No. in S7-6 in Figure 12), the point closest to the first interior point (nearest neighbor) is identified as the second interior point (S7-7 in Figure 12).

[0123] Next, it is checked whether there are any other points not identified as interior points within a predetermined distance from the second interior point (S7-6 in Figure 12). If there are any points not identified as interior points within the predetermined distance (No. in S7-6 in Figure 12), the point closest to the second interior point (nearest neighbor) is identified as the third interior point (S7-7 in Figure 12).

[0124] Thus, as long as there is a point within a predetermined distance that has not been identified as an interior point (No. S7-6 in Figure 12), the point closest to the previously identified interior point (nearest neighbor) is identified as the next interior point (S7-7 in Figure 12), and this process is repeated.

[0125] Then, when there are no more points within a predetermined distance that have not been identified as interior points (Yes in S7-6 of Figure 12), the search for interior points is completed, and the line consisting of the points identified as interior points up to that point is extracted as the interior contour line 31 (S7-8 of Figure 12).

[0126] As described above, the outer contour line 30 and the inner contour line 31 are extracted, and the overall contour line is classified into the outer contour line 30 and the inner contour line 31.

[0127] Next, the length measurement model creation unit 18 fills in the main grooves 34 of the tread in the tire cross-sectional image, and a length measurement model is created in which the outer contour line 30 becomes a smooth line 35. Then, the distances along the smooth line 35 from the starting point A to each point on the smooth line 35 are determined, as shown in Table 1 (S8 in Figure 8).

[0128] In detail, first, a sufficiently wide area (for example, an area including the entire tread) containing the portion along which the length along the smooth line 35 is to be measured is selected from the overall image, and expansion and contraction processing is performed on the selected area (S8-1 in Figure 13). Due to the expansion and contraction processing, the main grooves 34 of the tread are filled, and the outer contour line 30 in the tire cross-sectional image becomes the smooth line 35. The amount of expansion and contraction in the expansion and contraction processing is set to be less than or equal to the thickness from the outer contour line 30 to the inner contour line 31 of the tire tread.

[0129] Next, the intersection of the smoothing line 35 and the tire axial center line is set as the starting point A (S8-2 in Figure 13). Next, the nearest neighbor point of the starting point A is identified from the point cloud forming the smoothing line 35 (S8-3 in Figure 13). Next, the distance from the starting point A to its nearest neighbor point is calculated (S8-4 in Figure 13).

[0130] Next, from the point cloud forming the smoothing line 35, the point closest to the previously identified nearest neighbor is identified as the next nearest neighbor (S8-5 in Figure 13). Next, the distance between the previously identified nearest neighbor and the next nearest neighbor identified (distance between two points) is calculated (S8-6 in Figure 13). Note that the distance between two points is less than or equal to 2 pixels. Next, the distances between two points found up to that point are added together (S8-7 in Figure 13). This added value corresponds to the distance along the smoothing line 35 from the starting point A to the most recently identified nearest neighbor.

[0131] The process of identifying the next nearest neighbor, calculating the distance between two points, and calculating the cumulative distance between two points is repeated until the calculation within the predetermined range is complete (No. S8-8 in Figure 13). Once the calculation within the predetermined range is complete (Yes. S8-8 in Figure 13), the creation of the length measurement model is finished. This allows us to determine the distance from the starting point A to each point forming the smooth line 35, as shown in Table 1.

[0132] Next, the length calculation unit 19 calculates the tire thickness at an arbitrary position (S9 in Figure 8). Specifically, the user determines the location for which they want to find the thickness and selects a method for determining the outer and inner reference points suitable for that location from the first to third methods described above. Alternatively, the system may be configured so that one of the first to third methods is automatically selected when the user inputs the location for which they want to find the thickness through the input unit. Here, we assume that the second method, which utilizes a length measurement model, is selected.

[0133] First, the user inputs the distance from the starting point A along the smoothing line 35 to be designated as the outer reference point (S9-1 in Figure 14). Here, in step S8, the distance from the starting point A to each point forming the smoothing line 35 is determined. Then, on the smoothing line 35, the point whose distance from the starting point A is closest to the distance input in step S9-1 is identified as the point corresponding to the outer reference point (corresponding point) (S9-2 in Figure 14). At this time, the coordinates of the corresponding point are also identified.

[0134] Next, in the image before the outer contour line 30 is smoothed out to form the smooth line 35, a point with the same coordinates as the corresponding point mentioned above is identified as the outer reference point (S9-3 in Figure 14).

[0135] Next, a straight line (perpendicular line, for example, L2 in Figure 7) passing through the outer reference point and perpendicular to the inner contour line 31 is identified. Then, from the point group forming the inner contour line 31, the point closest to the perpendicular line (for example, the point labeled 31 in Figure 7) is identified as the inner reference point (S9-4 in Figure 14).

[0136] If the first method is selected, straight lines such as the tire axis centerline are drawn within the overall image. The intersection points of these lines with the outer contour line 30 are identified as outer reference points, and the intersection points of these lines with the inner contour line 31 are identified as inner reference points. The identification of outer and inner reference points is performed simultaneously.

[0137] Furthermore, if the third method is selected, the user identifies the coordinates near the point they wish to select as the outer reference point on the tire cross-section image. Then, the point on the outer contour line 30 closest to the identified coordinates is identified as the outer reference point. Next, from the point group forming the inner contour line 31, the point closest to a straight line (perpendicular line) passing through the outer reference point and perpendicular to the inner contour line 31 is identified as the inner reference point.

[0138] After the outer and inner reference points are determined, the distance from the outer reference point to the inner reference point (distance between reference points) is calculated (S9-5 in Figure 14). Next, the distance between reference points is converted into the actual thickness of the tire (S9-6 in Figure 14). Based on the thickness of each part of the tire calculated using the above method, the tire is evaluated.

[0139] If evaluation is performed on multiple tires, the above steps (S1 to S9 in Figure 8) are performed for each of those tires.

[0140] As described above, in this embodiment, after extracting the overall contour line of the rimmed tire cross-sectional image (S6 in Figure 8), the steps of extracting the outer contour line 30 and the inner contour line 31 from the overall contour line are performed, so that the overall contour line can be classified into the outer contour line 30 and the inner contour line 31 (S7 in Figure 8). Therefore, it becomes clear whether the points and line segments within the overall contour line belong to the outer contour line 30 or the inner contour line 31, and based on this, the characteristics of the tire can be confirmed. Specifically, as in this embodiment, it becomes possible to confirm the thickness of the tire at any position.

[0141] Furthermore, by detecting the outermost point in the tire's radial direction among the intersections of a straight line extending radially through the tire's internal cavity and the overall contour line, the outermost point can be reliably detected, and this detected point can be designated as the first outermost point. Subsequently, by repeatedly designating the nearest neighbor of the previously identified outermost point as the next outermost point, the outer contour line 30 can be reliably extracted.

[0142] However, points located at a predetermined distance from previously identified outer points are not considered outer points. This prevents points that should be included in the inner contour line 31 from being recognized as outer points.

[0143] Furthermore, by detecting the second point from the outermost side in the tire's radial direction among the intersections of a straight line extending radially through the tire's internal cavity and the overall contour line, an inner point can be reliably detected, and this detected point can be designated as the first inner point. Subsequently, by repeatedly designating the nearest neighbor of the previously identified inner point as the next inner point, the inner contour line 31 can be reliably extracted.

[0144] Furthermore, it is possible that multiple points on the outer contour line are aligned in the tire radial direction, and that multiple points on the outer contour line are included at the intersection of the straight line extending radially through the internal cavity of the tire and the overall contour line. To prepare for such cases, it is preferable to exclude the intersection point that is furthest out in the tire radial direction and its adjacent intersection points (i.e., exclude multiple points on the outer contour line that are aligned in the tire radial direction) from the intersection points of the straight line extending radially through the internal cavity of the tire and the overall contour line, and to designate the intersection point that is furthest out in the tire radial direction as the inner point. By performing the process of excluding not only the intersection point that is furthest out in the tire radial direction but also its adjacent intersection points, multiple points on the outer contour line can be reliably excluded, and the inner point can be accurately identified. Of course, even if there is only one point on the outer contour line that is an intersection point of the straight line extending radially through the internal cavity of the tire and the overall contour line, the inner point can be accurately identified by performing this process.

[0145] However, points located at a predetermined distance from previously identified interior points are not considered interior points. This prevents points that should be included in the outer contour line 30 from being recognized as interior points.

[0146] Various modifications can be made to the above embodiments. One of the modification examples described below may be applied to the above embodiments, or two or more may be selected and combined and applied to the above embodiments.

[0147] <Example of change 1> As tire cross-sectional images, not only those taken with X-ray CT scanners but also those taken with various non-destructive testing devices such as MRI (Magnetic Resonance Imaging) scanners can be used.

[0148] Additionally, images obtained by actually cutting a tire with a cutter and photographing the cross-section with a camera can also be used as tire cross-section images.

[0149] <Example of change 2> In adjusting the brightness using the brightness adjustment unit 12, the brightness may be adjusted not only based on the rubber portion as in the above embodiment, but also based on the average brightness of the entire tire cross-section image with rim (i.e., the entire tire cross-section image and the entire rim cross-section image).

[0150] In this case, the cross-sectional image of the tire with the rim is first extracted from the overall image. Within the overall image, the rubber portion of the tire cross-section has high brightness, and the metal portion (bead core, etc.) and the rim cross-section within the tire cross-section have even higher brightness. Therefore, the entire cross-sectional image of the tire with the rim is extracted by extracting the portion of the overall image whose brightness exceeds a threshold. As a threshold, for example, a value of 30 is used when the overall image is represented by 256 grayscale levels.

[0151] Next, the average brightness of the entire tire cross-section image with the rim is calculated. Then, the value obtained by dividing the reference brightness by the average brightness is set as the scaling factor. Next, for each pixel of the entire image or a part thereof (for example, the tire cross-section image with the rim), a calculation is performed in which the brightness of the pixel is multiplied by the scaling factor. Through this calculation, the brightness of the tire cross-section image with the rim approaches the reference brightness.

[0152] <Example of change 3> In adjusting the brightness using the brightness adjustment unit 12, the overall average brightness of the entire image may be calculated, and a scaling coefficient may be determined from that average brightness and the reference brightness.

[0153] In this case, the proportion (area ratio) of the rimmed tire cross-section image within the overall image affects the average brightness. Therefore, the image is captured so that the area ratio falls within a predetermined range. The predetermined range is preferably 5% to 15%, and more preferably 8% to 12%.

[0154] Furthermore, in this case, it is preferable that the number of pixels in the white areas of the overall image (for example, areas with a grayscale level between white and black or higher) is 5% to 10% of the total number of pixels in the overall image. If this condition is met, the average brightness will be appropriate.

[0155] <Example of change 4> When using a bead core to rotate a cross-sectional image of a rimmed tire, the method for selecting the bead core and the method for determining the center of gravity of the bead core are not limited to the method of the above embodiment.

[0156] For example, the user may manually select the bead core from the cross-sectional images of the rimmed tire. Alternatively, the user may manually determine the center of gravity of the bead core.

[0157] <Example of change 5> When rotating a cross-sectional image of a rimmed tire so that it is parallel to the reference line (X-axis), a straight line indicating the inclination of the rim may be used instead of a straight line connecting the bead cores on both sides of the tire axis.

[0158] Specifically, first, the flanges on both sides of the rim in the tire axial direction are identified, and then the outermost vertex in the tire radial direction of each flange is identified. After these two vertices are identified, a straight line connecting them is determined. Here, the method for determining the straight line from two points is the same as the method used to determine the straight line from the centroids of the two bead cores in the above embodiment.

[0159] The straight line obtained in this way represents the inclination of the rim. The rotation angle of the tire cross-sectional image is then determined based on the angle between the straight line representing the rim's inclination and the reference line. Here, the method for determining the rotation angle from the straight line is the same as the method used in the above embodiment.

[0160] In this example of modification, the method for identifying the flange and its radially outer vertex is not limited. For example, the user may manually identify the flange and its radially outer vertex. Alternatively, the rim may be automatically detected based on brightness, or the radially outer vertex of the flange may be automatically identified from the rim shape.

[0161] Alternatively, by taking advantage of the high brightness of the rim cross-section image in the rim-attached tire cross-section image, a direction in which many pixels with high brightness (for example, a brightness within a predetermined range of 180 to 250) are aligned can be identified, and the straight line extending in that direction can be used as a straight line indicating the inclination of the rim.

[0162] The basic shape of the rim on which the tire is mounted is the same regardless of the type of tire. Therefore, by rotating the cross-sectional image of the tire with the rim attached using a straight line that indicates the inclination of the rim, as in this example, all tires can be oriented in the same direction.

[0163] <Example of change 6> This example of a change illustrates how the rotation angle is determined when rotating a cross-sectional image of a rimmed tire so that it is parallel to the reference line (X-axis).

[0164] In this modification example, multiple small rotations are performed on the rim section image (or tire section image with rim) by an arbitrary small angle. The center of rotation for each small rotation is one of the points on the rim section image, and the rotation occurs on the plane of the overall image. The axis of rotation for each small rotation is an axis that extends perpendicular to the planar overall image. An arbitrary angle is, for example, 1°.

[0165] For each small rotation by the aforementioned arbitrary angle, the maximum coordinates in the direction perpendicular to the reference line (Y-coordinate direction) can be determined for one portion of the rim in the tire axial direction and the other portion. In many cases, these maximum coordinates are the coordinates of the rim flange ends on one and the other portions in the tire axial direction. Furthermore, for each small rotation by the aforementioned arbitrary angle, the absolute value of the difference between the maximum coordinates of the one portion in the tire axial direction and the other portion can be determined.

[0166] When small rotations are repeated, there is a point when the absolute value of the difference between the maximum coordinates on one side of the tire axis and the other side reaches its minimum. This minimum value occurs when the axis direction of the rim cross-section image (which coincides with the axis direction of the tire cross-section image) becomes parallel to the reference line, or as close to parallel as possible. At that point, the number of small rotations performed until the absolute value reaches its minimum becomes clear. Furthermore, the cumulative rotation angle (cumulative angle of small rotations) from the start of the small rotations until the absolute value reaches its minimum becomes clear.

[0167] The cumulative rotation angle revealed in this way is set as the rotation angle used to rotate the cross-sectional image of the tire with the rim. Then, the image of the tire with the rim is rotated by that rotation angle.

[0168] This process makes the axis direction of the rim cross-section image parallel or approximately parallel to the reference line, and further, the axis direction of the tire cross-section image parallel or approximately parallel to the reference line.

[0169] Note that the above method can be performed when, in the pre-rotation image of the tire cross-section with the rim, the tire cross-section image is at the top (i.e., the one with the larger Y-coordinate value) and the rim cross-section image is at the bottom. If, in the pre-rotation image of the tire cross-section with the rim, the tire cross-section image is at the bottom and the rim cross-section image is at the top, then it is necessary to rotate the rim cross-section image by 180° before performing the above method.

[0170] <Example of change 7> This example demonstrates a modification to the method of rotating the cross-sectional image of a rimmed tire so that it is parallel to the reference line (X-axis).

[0171] In this modification example, multiple small rotations are performed, rotating the rim section image (or tire section image with rim) by an arbitrary small angle. The center of rotation for each small rotation is one of the points in the tire section image with rim, and the rotation occurs on the plane of the overall image. The axis of rotation for each small rotation is an axis extending perpendicular to the planar overall image. An arbitrary angle is, for example, 1°.

[0172] For each small rotation by the aforementioned arbitrary angle, the maximum coordinates of one and the other portions of the rim in the tire axial direction can be determined in the direction perpendicular to the reference line (Y-coordinate direction). In many cases, these maximum coordinates are the coordinates of the ends of the rim flanges on the one and the other portions in the tire axial direction. Furthermore, for each small rotation by the aforementioned arbitrary angle, the absolute value of the difference between the maximum coordinates of the one and the other portions in the tire axial direction can be determined.

[0173] As small rotations are repeated, there will come a point when the absolute value of the difference between the maximum coordinates on one side of the tire axis and the other side reaches its minimum. This minimum absolute value occurs when the axis direction of the rim cross-section image becomes parallel to the reference line, or as close to parallel as possible. At this time, the axis direction of the tire cross-section image is also parallel to the reference line, or as close to parallel as possible. Therefore, the small rotation is terminated when the absolute value reaches its minimum.

[0174] This process makes the axis direction of the rim cross-section image parallel or approximately parallel to the reference line, and further, the axis direction of the tire cross-section image parallel or approximately parallel to the reference line.

[0175] <Example of change 8> As shown in Figure 15, in the overall image captured by the image processing device 10, the rim-attached tire cross-section image may not be upside down (i.e., the tire cross-section image is at the top and the rim cross-section image is at the bottom). In such cases, in the step of rotating the image (S3 in Figure 8), the rotation angle of the rim-attached tire cross-section image is set to the angle θ between the straight line connecting the centers of gravity of the bead cores (or the straight line indicating the inclination of the rim as explained in the above modification example) and the reference line.

[0176] Whether or not a cross-sectional image of a tire with a rim is upside down is automatically determined by the following method. First, the overall image is divided into two regions, upper and lower. At this time, the areas of the two regions are equal. Next, the average brightness within each region is calculated. Then, it is determined that the rim cross-sectional image exists in the region with the higher average brightness (because the rim and the bead core near it appear particularly bright). If the region where the rim cross-sectional image exists is the upper region, it is determined that the cross-sectional image of the tire with a rim is upside down.

[0177] <Example of change 9> In the noise processing step (S5 in Figure 8), the storage range for restoring the image data to its state before expansion and contraction after expansion and contraction is not limited to the range including the gap between the flange and the tire (within the frame in Figure 4(b)) as in the above embodiment.

[0178] The storage range preferably includes areas where the original contour lines are likely to be distorted by the expansion and contraction process, or areas in the original image where there is little noise such as missing parts. Examples of such areas include any part of the rim.

[0179] <Example of change 10> In the noise processing step of the above embodiment (S5 in Figure 8), the processing was performed in the order of expansion, contraction, contraction, and expansion, but the order of processing is not limited to this. For example, if there are few defects in the cross-sectional image of the tire with a rim, and there is no risk of the thin parts of the cross-sectional image of the tire with a rim being cut off even if the contraction processing is performed first, the processing may be performed in the order of contraction, expansion, expansion, and contraction.

[0180] <Example of change 11> As a method for classifying the overall contour line into an outer contour line 30 and an inner contour line 31, the line consisting of the point cloud remaining after extracting the outer contour line 30 may be extracted as the inner contour line 31.

[0181] Specifically, the outer contour line 30 is first extracted using the method described in the above embodiment. Then, among the point cloud that forms the overall contour line, those not included in the point cloud that forms the outer contour line 30 are identified as the point cloud that forms the inner contour line 31.

[0182] <Example of change 12> In the method of classifying the overall contour line into an outer contour line 30 and an inner contour line 31, in the above embodiment, the intersection points of the tire axial center line and the overall contour line were identified as the first outer point and the first inner point. However, the straight line used to identify the intersection points is not limited to the tire axial center line.

[0183] If a straight line is parallel to the Y-axis (i.e., a straight line extending in the radial direction of the tire) and can intersect both the outer contour line 30 and the inner contour line 31, then the intersection points of that line with the overall contour line can be identified as the first outer point and the first inner point. A straight line parallel to the Y-axis and that can intersect both the outer contour line 30 and the inner contour line 31 is a straight line that passes through the cavity inside the tire and is parallel to the Y-axis. The center line in the axial direction of the tire is also a straight line that passes through the cavity inside the tire and is parallel to the Y-axis.

[0184] <Example of change 13> In the above embodiment, Table 1 was created by determining the length from the starting point A to each point, using the intersection point of the smooth line 35 created by the smooth line creation unit 18a and the tire axial center line as the starting point A. However, the position of the starting point is not limited. A table similar to Table 1 can be created by determining the length from any point on the smooth line 35 to each point.

[0185] Furthermore, in the above embodiment, in the second method for identifying the outer and inner reference points, the intersection point of the smooth line 35 and the tire axial center line was set as the starting point A, and the outer reference point was identified based on the distance along the smooth line 35 from the starting point A. However, the starting point for identifying the outer reference point is not limited to the starting point A, which is the intersection point of the smooth line 35 and the tire axial center line. The starting point can be any point on the smooth line 35.

[0186] <Example of change 14> In the above embodiment, the tire thickness was the length in the direction of extension of the perpendicular to the inner contour line 31. However, the tire thickness may also be the length in the direction of extension of the perpendicular to the outer contour line 30.

[0187] <Example of change 15> The order of the steps in the above embodiment can be changed. For example, the order of the step of adjusting the brightness of the overall image (S2 in Figure 8) and the step of rotating the cross-sectional image of the rimmed tire (S3 in Figure 8) can be swapped. Also, the order of the step of classifying the overall contour line into an outer contour line 30 and an inner contour line 31 (S7 in Figure 8) and the step of filling in the main groove 34 and creating a length measurement model by making a single smooth line 35 representing the outer surface of the tire (S8 in Figure 8) can be swapped.

[0188] Furthermore, in the step of classifying the overall contour into outer contour lines 30 and inner contour lines 31 (S7 in Figure 8), the outer contour lines 30 may be extracted after the inner contour lines 31 have been extracted. Alternatively, the extraction of the outer contour lines 30 and the inner contour lines 31 may be performed simultaneously.

[0189] <Example of change 16> The scaling calculation may be performed only on the entire tire cross-section image (i.e., including not only the rubber part but also metal parts such as the bead core, etc., but excluding the rim cross-section image and background). Alternatively, the scaling calculation may be performed on the entire image. [Explanation of Symbols]

[0190] 10…Image processing device, 11…Grayscale conversion unit, 12…Brightness adjustment unit, 12a…Tire image extraction unit, 12b…Brightness change unit, 13…Image rotation unit, 13a…Bead core detection unit, 13b…Rotation unit, 14…Binarization processing unit, 15…Noise processing unit, 16…Contour line detection unit, 17…Contour line type unit, 18…Length measurement model creation unit, 18a…Smooth line creation unit, 18b…Length calculation unit, 19…Length calculation unit, 19a…Reference point determination unit, 19b…Tire thickness calculation unit, 30…Outer contour line, 31…Inner contour line, 32…Outer reference point, 33…Inner reference point, 34…Main groove, 35…Smooth line

Claims

1. In a method for processing images of cross-sections of tires with rims, Steps include: extracting the overall contour line, which is the contour line of the entire cross-section of the tire with a rim; The steps include extracting an outer contour line representing the outer surface of a rimmed tire from the overall contour line, The steps include extracting an inner contour line representing the inner surface of a rimmed tire from the overall contour line, Includes, The aforementioned overall contour line consists of a point cloud. The extraction of the outer contour line is performed by repeatedly identifying the outermost point in the tire's radial direction, which is part of the outer contour line, as the intersection point of a straight line extending radially through the cavity inside the tire and the overall contour line, and the nearest neighbor point of the outer contour line as the next outer point. An image processing method characterized by the following features.

2. The image processing method according to claim 1, wherein a point located at a predetermined distance from the aforementioned outer point is not considered to be the aforementioned outer point.

3. The aforementioned overall contour line consists of a point cloud. The image processing method according to claim 1, wherein the extraction of the inner contour line is performed by excluding the intersection point that is furthest outward in the tire diameter direction from the intersection points of a straight line extending in the tire diameter direction through the cavity inside the tire and the overall contour line, and the intersection point that is consecutive thereto, and the intersection point that is furthest outward in the tire diameter direction is designated as an inner point that is part of the inner contour line, and the nearest neighbor point of the said inner point is designated as the next inner point, and so on, and this process is repeated.

4. The image processing method according to claim 3, wherein a point located at a predetermined distance from the aforementioned inner point is not considered to be the aforementioned inner point.

5. The image processing method according to claim 1, wherein the line consisting of the point cloud remaining after the extraction of the outer contour line is considered to be the inner contour line.

Images