FI3926318T3 - Method and device for adjusting a tyre on a rim to a precise angle - Google Patents

Description

METHOD AND DEVICE FOR ADJUSTING A TYRE ON A RIM TO A PRECISE

ANGLE

Description

The invention relates to a method and a device for aligning a tyre on a rim at a precise angle of rotation in a preassembled complete wheel, wherein the actual angular position of a match point of the tyre and/or a match point of the rim is detected based on an image recorded by a camera relative to a respective reference alignment and/or relative to one — other, and a desired angular position is established by rotating the tyre and/or the rim.

A method of this kind is known from EP 1 564 028 B1. The image evaluation is not specified in greater detail there. Further methods for aligning a wheel on a rim at a precise angle of rotation according to the preamble of Claim 1 can be found in documents

EP1738937A2 and EP1746401A2.

Matching is a method used in tyre assembly which helps to achieve optimal true running of the complete wheel consisting of the rim, tyre, and valve. This involves the radial run- out deviations of the rim and tyre being combined with one another in such a way that they largely offset one other. For example, it may be provided that the geometric low point of the rim is assigned to the force high point of the tyre. The tyres are measured by the manufacturer and marked with a match point on the sidewall. The same is done with the rim. These two match points are moved on top of one other or into another predetermined angular position relative to one other in a matching machine by rotating the tyre and/or the rim.

With the rim, it is customary for the angular position of a match point to relate to a valve.

The valve itself can therefore be regarded as a match point. Therefore, automated detection of the valve or a similar structural feature on the rim or tyre in the image evaluation is desirable.

The object of the invention is therefore to expand the method known from the prior art with areliable, practical and rapid image evaluation that allows automated detection and position determination of structural features of the rim and/or the tyre, in particular of a valve on the rim, in computer-implemented form.

The object is achieved according to the invention by a method having the features of Claim 1 and an associated device according to Claim 12.

Consequently, the detection of the actual angular position comprises the following steps: e recording at least three, preferably four individual images of the complete wheel by the camera, each with different (independent) illumination directions; e creating an output image containing information on the (particularly three- dimensional) surface structure of the complete wheel by fusion from the individual images; e software-based (algorithmic) object search in the output image for a predetermined — structural feature of the tyre and/or the rim and determination of the object position.

The device for carrying out the method comprises a support for receiving the complete wheel and is eguipped with a camera directed at the support, with at least three, preferably four, light sources for illuminating the complete wheel on the support in each case from different illumination directions, with a controller for the coordinated control of the camera and the light sources, and with an evaluation unit for creating the output image and for the object search and position determination.

Surprisingly, it has been found that the method known per se as photometric stereo measurement, which involves a reconstruction of three-dimensional surface properties of the object recorded with different illumination directions, easily provides substantially more reliable results in the present context with the detection of match points during wheel assembly than image evaluation of the original camera image. In particular, it is therefore possible for different types and/or sizes of tyres and/or rims to be flexibly processed, — without the underlying algorithm having to be learned using reference images.

Advantageous embodiments of the method are disclosed in the dependent claims in conjunction with the description of the figures.

An exemplary embodiment of the invention is explained in greater detail below with reference to the accompanying drawings. The drawings show in a simplified schematic representation:

FIG. 1 a perspective view of a matching machine for aligning (matching) a tyre on a rim at a precise angle of rotation,

FIG. 2 a side view of a camera and illumination system integrated into the matching machine according to FIG. 1,

FIGS. 3 to9 various images captured with the camera and illumination system according to FIG. 2 at different stages of processing during a matching process.

The term rim is used here in the colloquial sense and refers to the wheel of a vehicle without the tyre. The unit made up of the rim and tyre is also referred to as a complete wheel or simply as a wheel. — The device schematically shown in FIG. 1, a so-called matching machine 2, is used for aligning (matching) a tyre 4 on a rim 6 (see also FIG. 2) at a precise angle of rotation. In this case, a predetermined reference or match point (marking) on the circumference of the tyre 4 is brought into a defined angular position relative to a predetermined reference or match point on the circumference of the rim 6 by rotating the tyre 4 preassembled on the rim 6 relative to said rim 6. For example, this can be achieved by axially pressing together the tyre 4, which is fixed in its rotational position by a suitable gripping tool in the matching machine 2, using a compressing or clamping tool in the region of the tyre sidewall, so that the firm fit on the rim 6 loosens, and the rim 6 is turned into the desired rotational position with the help of a turning tool. The compression is subsequently released and due to its elasticity, the tyre 4 resumes its firm fit on the rim 6 - now in the desired angular position.

The mechanical components required in order to convey, centre, hold, rotate and discharge the wheels or their components in the matching machine 2 can be realized as described in

EP 1 564 028 BI, for example, or in a different manner. The detail of this is not crucial for the present invention. In particular, the matching machine 2 can also be designed for the preassembly of the complete wheel 8, in other words, for mounting the tyre 4 on the rim 6 before the matching process.

To determine the required angle of rotation, the current state, in other words the current angular position of the match points of the tyre 4 and the rim 6 relative to a spatially fixed reference coordinate system or a reference alignment, has to be detected. For this purpose, an automated computer-based image evaluation of camera images is provided in the present case, which images are recorded by a camera 10 in a top view of the complete wheel 8.

Specifically, during the matching process, the preassembled complete wheel 8 rests horizontally on a clamping support or support 20 of the matching machine 2, which is realized by rails, for example, in such a manner that its wheel axis A is vertically aligned.

The camera 10 mounted on a frame or housing part of the matching machine 2 looks down from above onto the complete wheel 8, wherein the optical axis of the camera 10 preferably coincides with the wheel axis A (central arrangement). This means that the camera 10 is substantially aligned perpendicular to the scene to be recorded. The field of view of the camera 10 illustrated by a circular cone in FIG. 1 preferably covers the entire complete wheel 8 up to the outer circumference of the tyre 4.

The match point of the tyre 4 may, for example, be marked by a coloured dot on the typically dark tyre surface and is generally easily recognizable through automatic image evaluation. On the other hand, the match point of the rim 6 is realized by a structural feature of the rim 6, namely a valve 12 or a valve hole which is more difficult to determine. This is especially true when not only a single, but multiple types of wheels or rims are handled in the matching machine 2.

Even if a conceptually different match point is to be provided on the rim 6, by knowing the angular position of said match point relative to the valve 12, through a simple conversion the valve 12 can be used as the match point. 5 For image evaluation, at least three, preferably four images (photos) of the complete wheel 8 are taken with the camera 10, wherein the arrangement and alignment of the camera 10 relative to the complete wheel 8 remains the same as described above, but the illumination is varied. This is illustrated in the schematic side view according to FIG. 2. Two light sources 14 are shown by way of example there, which, when turned on, illuminate the — complete wheel 8 obliquely from above, namely at a specific angle of inclination a relative to the optical axis. In reality, at least three, preferably four, light sources 14 are distributed around the camera 10, namely preferably evenly distributed on an imaginary horizontal circular line around the optical axis. In other words, with three light sources 14, the corresponding main illumination directions in the top view or projection from above are preferably each offset or rotated by 120° from each other, and with four light sources, preferably by 90°. The preferably identical light sources 14 are preferably each aligned with the centre of the complete wheel 8 and are preferably at the same distance from this centre.

The angle of inclination a relative to the optical axis is preferably the same for all light sources 14.

The light sources 14 are preferably emitters, in particular flash emitters, that produce as parallel and uniform light beams as possible (telecentric light sources). Where necessary, they have suitable lenses for this purpose. Each of the light sources 14 should be able to illuminate the wheel uniformly over the entire visible area from above, in order to avoid local overexposure or underexposure. Suitable enclosures, or the like, can be attached to the matching machine 2 around the wheel, in order to block out ambient light.

For a simple and robust mechanical structure, the light sources 14 are preferably fixedly arranged and aligned on a frame part or housing part of the matching machine 2. The angle of inclination or the incident angle a is therefore normally fixed during operation but can optionally be adjustable.

For creating the images, the light sources 14 are switched on and off in rapid succession, similar to a flash cascade, wherein a corresponding image is taken by the camera 14 during the respective activation period. To accommodate different types of wheels or rims, one or more of the following parameters are preferably adjustable in a corresponding controller 16: illumination duration (flash duration), illumination intensity, camera exposure time, aperture, or camera sensitivity. This allows the controller 16 to flexibly react to different rim designs (colouring, reflection, etc.), for example, and perform suitable adjustments.

In an alternative scenario, there is only one light source 14 which can be moved to predetermined illumination positions along rails using an adjusting drive, for example.

The result of the operation is exemplified in FIG. 3, showing by way of example four images of the complete wheel 8 taken with a total of four independent illumination directions.

The images of the complete wheel 8 taken by the camera 10 under different illumination conditions in each case are then fused or overlaid and evaluated in a corresponding computer using image evaluation software based on a photometric stereo measurement technique.

Photometric stereo measurement enables the surface of a three-dimensional object to be reconstructed based on the two-dimensional images recorded.

The evaluation, based on Woodham's algorithm, for example, requires a minimum of three images of the same object taken with different known illumination directions. The alignment of the camera with the object must be the same for all images.

The three-dimensional shape of the object is primarily calculated in the form of local gradients of the three-dimensional surface. These gradients can be further integrated to create a height model, in other words an image in which each pixel value corresponds to a relative height. The two-dimensional texture is referred to as albedo and represents the local light absorption and reflection characteristics of the surface without the influence of shadows.

The evaluation software advantageously outputs at least one of the following images: e A gradient image, in which the pixel values correspond to the value of a (possibly normalized) surface gradient of the recorded object. e A height image, in which the pixel values correspond to a relative height of the recorded object. e An albedo image, in which the pixel values describe the ratio of reflected light to incident light. White surfaces, for example, are assigned a value of one, while black surfaces are assigned a value of zero.

The gradient image has proved to be particularly preferred for subseguent evaluation steps.

However, they are also possible in principle with the other two types of output images.

As mentioned earlier, the photometric stereo method reguires a minimum of three images with different illumination directions. Redundancy is advantageous for more robust results.

Therefore, more than three light sources 14 with different illumination directions should normally be chosen. However, an increasing number of illumination directions also leads to a higher number of images to be processed and therefore to a longer processing time. In the present application, the number four has proved to be a reasonable compromise.

However, more than four light sources 14 and a corresponding number of recordings are possible in principle.

The output image of the photometric stereo evaluation, which serves as the basis for the subseguent automated object or structure search, is preferably a monochromatic image, in particular a grey-scale image. This simplifies the search process. Otherwise, for a colour image, the operations described below would need to be performed for each colour channel, which would be correspondingly more time-consuming. However, this is also possible in principle.

FIG. 4 shows by way of example an overlay image generated from the superposition of the four individual images according to FIG. 3, wherein the photometric evaluation described above produces a gradient image, represented here as a grey-scale image, from the four individual images.

In an optional intermediate step, an identification of the circular outer circumference of the tyre 4, and consequently a determination of the wheel or tyre outer diameter, can be performed through a pattern recognition algorithm based on the superposed image or the gradient image. This is illustrated in FIG. 5.

Similarly, in an optional intermediate step, an identification of the rim circumference and determination of the rim diameter (referring to the outer diameter in this case) can be performed. This is illustrated in FIG. 6.

Based on the data obtained in these two steps, the tyre and/or rim type can be determined and/or checked, for example.

For the automated search for the valve 12, or valve hole, acting as the match point of the rim 6, it is advantageous to restrict the search area in the gradient image to the relevant circular ring-shaped area in the outer rim ring region (which is typically connected to the wheel flange by the wheel disc or the wheel rim). This can be done in a simple variant of the method by manually inputting the corresponding radii or diameters, or by retrieving the values from a database of wheel type data based on the current wheel type. In particular, the wheel type can be automatically identified in this case based on the previous intermediate steps. In a preferred variant of the method, the identification of the relevant circular ring-shaped area is performed fully automatically based on image-based pattern recognition software in the gradient image, similar to the subsequent valve search. To a certain extent, the determination of the search area is an advantageous precursor to the valve search within the search area described further on.

The circular ring-shaped search area determined or detected in this way is illustrated by way of example in FIG. 7.

The actual software-based search of the valve 12, or valve hole, is now carried out in the search area which is restricted in this manner. The circular ring-shaped area is advantageously unwound by the evaluation software for this purpose and transformed into a corresponding rectangular area. The processed search area is illustrated in FIG. 8.

The background for these measures is the subsequent identification of, and differentiation between, relevant and non-relevant objects or areas. For example, when looking at the radially extending spokes and their shading in the original, as yet unwound, images, they always run in different directions. This makes it difficult to identify these linear objects.

Therefore, the unwinding process is advantageously used to transform the circular structure into substantially horizontal lines and radial structure into vertical lines. These objects mainly present in the output image transformed in this way of the photometric stereo method, in other words, vertical and horizontal lines, can be found far more easily by object recognition algorithms and discarded as non-relevant.

In general and slightly simplified terms, finding the valve 12 is not based on specifically searching for the valve 12. Instead, it is preferably approached conceptually in the opposite way, searching for areas that definitely cannot be the valve 12, with the algorithm. These areas are excluded or discarded, and what remains at the end is the valve 12.

According to this exclusion principle, the horizontal and vertical line objects found in the unwound representation are excluded from the search area. Then, coherent areas of the same grey-scale range are identified. The grey-scale range is predefined and empirically determined. All areas with a size above a given upper size limit are excluded as too large and all areas below a lower size limit are also excluded. Afterwards, the remaining objects are classified by shape. This means that an area of a grey-scale range which was not eliminated by the above filters is now checked for its shape. Since a circular valve 12 or valve hole is sought, the algorithm excludes all elongated, non-circular objects. Interfering —embossments and the like can also be detected and discarded in this way. Ultimately, only one area, the valve 12, remains.

At the end of the identification process, the circular ring-shaped area is unwound again, metaphorically speaking, and the angular position of the valve 12 relative to a pre-defined reference orientation is calculated. This therefore involves a reverse transformation of the valve coordinates determined based on the unwound representation into the original coordinate system, followed by the angle calculation. This is illustrated in FIG. 9.

Once the current actual angular position of the rim 4 and tyre 6 on the support 20 of the matching machine 2 is known, the angle of rotation required in order to achieve the desired target angular position is calculated and, as described earlier, this can be achieved, for example, by rotating the rim 6 relative to the fixed tyre 4.

The described algorithm may, in principle, be applied to any type and any size of rim 6.

There is no need for any teaching of the algorithm, such as by using reference images and/or initial manual identification of the valve 12, as is required in certain types of object recognition methods based on artificial intelligence.

All the described method steps of image evaluation, object recognition, and angle calculation can be implemented through corresponding software 18 on a standard commercial or specialized computer. The camera 10 is also connected to this computer via suitable interfaces and provides the images used as the basis for evaluation. Furthermore, the light sources 14 are controlled directly by the computer or the camera 10, preferably in a manner synchronized with the function of the camera 10, so that the light flashes are synchronized with the camera images. This is schematically indicated in FIG. 2. The software 18 is advantageously modular and integrated into the general system control or controller 16 of the matching machine 2, ensuring that the evaluations underlying the matching process are aligned with the mechanical workflow and seamlessly integrated without significant delay. Alternatively, the image evaluation may also run on a separate computer. The electronic camera 10, preferably a digital camera with an image sensor, for example based on CCD or CMOS technology, can be a still image camera or a video camera with a still image function, advantageously designed for rapid image sequence recording (typically 50 ms) in combination with flash-like light sources 14. The evaluations which have been described practically run in real-time, meaning that a high cycle rate for the matching processes can be achieved.

The method described with the help of a valve 12 can also be similarly applied to other unique structural features of the rim 6 or tyre 4.

List of reference numbers 2 matching machine 4 tyre 6 rim 8 complete wheel camera 12 valve 14 light source 16 controller 10 18 software support

A angle of inclination

A wheel axis (= optical axis)

Claims (12)

Patent Claims 1. A method for aligning the tire (4) to the rim (6) with a precise angle of rotation into a pre-installed ready-made wheel (8), whereby the actual angular position of the fitting point of the tire (4) and/or rim (6) is recorded based on the image recorded by the camera (10) and the target angular position is achieved by rotating the ring (4) and/or rim (6), characterized by the fact that detecting the actual angular position includes the following steps: * — at least three, preferably four, individual images, each of which has different lighting directions , recording with a camera (10) of the whole bike (8); * creation of a result image containing information about the surface structure of the entire wheel (8) by combining the individual images; + software-based object search from the output image according to a specific structural feature of the tire (4) and/or rim (6) and determining the location of the object nose 2. The method according to claim 1, wherein the print image is reconstructed according to the principle of photometric stereo measurement. 3. The method according to claim 2, wherein the print image is a gradient image of the entire surface of the wheel (8). 4. The method according to one of the previous patent claims, wherein the print image is a monochrome image, especially a grayscale image. 5. — A method according to one of the preceding patent claims, wherein a light source (14) is arranged for each direction of illumination, which is indicated. 6. The method according to claim 5, wherein the light sources (14) are — configured as flash light sources and the individual images are recorded in rapid succession, preferably between 10 and a few 100 milliseconds. 7. — A method according to one of the preceding patent claims, wherein the structural feature is a valve (12) or a valve hole in the rim/rim (6). 8. The method according to claim 7, wherein the object search is limited to a circular search area that represents the outer rim of the rim (6). 9. The method according to claim 8, the circular area of the object search is converted into a rectangular area in the manner of development. 10. The method according to claim 8 or 9, wherein the valve (12) or the hole in the valve is distinguished from other structures in the search area based on its characteristic shape and size. 11. A method according to one of the preceding patent claims, whereby the evaluation of the image — is not learned on the basis of a comparison image. 12. A device for carrying out a method according to one of the previous patent claims, with a support (20) for receiving the entire wheel (8), a camera (10) directed at the support (20), at least three, preferably four, light sources (14) of the support ( 20) for lighting from different lighting directions, a controller (16) for controlling the camera (10) and light sources (14) in a coordinated manner, and an evaluation unit for creating a print image and searching for objects and determining the location.