Model-based spring shape measurement

Abstract

Images of the wire spring (10) are acquired from a plurality of cameras (150) mounted at different locations, and then a parameterized three-dimensional shape model is fitted to the acquired images. The parameterized three-dimensional shape model is defined by a plurality of reference points which are aligned with the trajectory of the line spring (10) in the image via the fitting.

Description

TECHNICAL FIELD

[0001] The present invention relates to a method and a system for measuring wire springs. BACKGROUND

[0002] Wire springs can be used in various types of products, for example in spring mattresses or other types of bedding and seating appliances. Depending on the product, wire springs can be manufactured with various different properties and geometries, such as helical coil springs with a cylindrical, hourglass or barrel-like outer shape. The desired geometry is generally achieved by winding the wire in a corresponding set winding process, which can be performed by an automatic or semi-automatic spring winding machine.

[0003] However, during the manufacturing of wire springs, it can happen that the properties of the manufactured springs do not correspond to the desired properties. The reason for this can be, for example, that the geometry of the wire spring does not correspond to the target geometry and / or that the design of the spring geometry is not perfect.

[0004] WO 2022 / 017785 A1 describes a method for manufacturing coil springs by spring winding, in which a parametric three-dimensional (3D) helix model of the coil spring is adjusted using coil spring images acquired during the winding process, and then, based on the three-dimensional helix model, measurement values representing the geometry of the coil spring, such as the pitch, diameter or extension length, are obtained. However, such a modeling of wire springs can not be applicable to certain types of wire springs, for example, wire springs with a more complex geometry or wire springs that deviate significantly from the target helical spring geometry due to severe defects. Furthermore, in some cases, it can be necessary to perform a measurement of the wire spring independently of the manufacturing process. For example, at an assembly site where the wire spring end product is assembled in a product.

[0005] Therefore, there is a need for a technology that can flexibly and accurately measure the geometry of a wire spring. SUMMARY

[0006] The invention provides a machine according to claim 1 and a method according to claim 10. The dependent claims constitute further embodiments.

[0007] Accordingly, according to an embodiment, the invention provides a method for measuring the shape of a wire spring. According to the method, images of the wire spring are acquired from a plurality of cameras mounted at different positions, and a parametric three-dimensional shape model is fitted to the acquired images. The parametric three-dimensional shape model is defined by a plurality of reference points, which are fitted to align with the wire spring trajectory in the images.

[0008] By using multiple cameras installed at different positions, such as two, three or even more cameras, images of the wire spring can be obtained at different viewing angles, thus enabling an accurate evaluation of the shape of the spring. Moreover, the taking of the images can be made independent of the specific position of the spring. In addition, multiple images can be taken at the same time at different viewing angles, not only enabling a measurement at a specific moment in time, but even in a time-resolved manner. In addition thereto, a measurement in the free state of the wire spring is possible. By defining the parametric three-dimensional shape model with multiple reference points, a modeling of the wire spring in an efficient and accurate manner is possible without imposing too many restrictions on the nominal geometry of the wire spring to be modeled. For example, even for wire springs which are not helical or which have a geometry which deviates significantly from the ideal helical geometry, a modeling in an efficient and accurate manner is possible.

[0009] According to an embodiment, each reference point defines a reference sphere which is fitted to align with the trajectory of the wire spring in the image. In this way, regardless of the viewing angle of the image, the projection of the reference sphere in the image remains circular due to the image being two-dimensional, thus enabling an efficient alignment of the reference points with the trajectory of the wire spring in the image. In this way, the requirements on the processing required for the alignment can be reduced. However, it should be noted that depending on the cross-sectional geometry of the wire material used for the wire spring, other types of geometrical reference objects can be used in addition to spheres, such as reference ellipsoids, reference cubes, etc.

[0010] As mentioned above, the number of cameras can be more than three. However, in some cases, even with only two cameras, satisfactory results can be achieved. In other cases, by using four or even more cameras, a further improvement of the modeling accuracy can be achieved.

[0011] According to an embodiment, the images are taken in the free state of the wire spring. In this way, changes of the shape of the spring due to moving the spring while taking the images can be avoided. As an alternative, more than one of the images can also be taken during the movement of the spring. Moreover, the movement of the spring can be performed in the gap between the taking of two images.

[0012] The number of reference points is more than 50, more than 100 or more than 200. Tests on ordinary wire springs used in commercially available bedding and seating appliances have shown that a reasonable balance between the amount of processing and the modeling accuracy can be achieved when the number of reference points is in the range of 100 to 200.

[0013] The parameterized three-dimensional shape model can be based on a three-dimensional spline curve whose knots correspond to the reference points. The fitting can be based on minimizing a deviation of the parameterized three-dimensional shape model from the trajectory of the wire spring in the image. In some embodiments, the fitting is performed iteratively based on a gradient of deviation between the reference points and the position of the wire spring in the image.

[0014] In some embodiments, the fitting includes fitting a plurality of transition parameterized three-dimensional shape models to the image. The parameterized three-dimensional shape model is then generated by merging the transition parameterized three-dimensional shape models. The merging is achieved, for example, by selecting, averaging, and / or concatenating. In some cases, this operation can include fitting each of the transition parameterized three-dimensional shape models to a different subset of the image. Alternatively or additionally, the fitting is based on a different set of initial conditions for each of the transition parameterized three-dimensional shape models. In this way, the overall fitting process can flexibly take into account different types of assumptions and conditions.

[0015] In some embodiments, the number of parameters of the parameterized three-dimensional shape model can be adjusted during the fitting process. For example, the parameterized three-dimensional shape model can be relatively simple with fewer parameters at the beginning of the fitting process. The number of parameters can then be increased, for example, when the fitting process has reached a given degree of convergence based on the fewer parameters. The parameters of the relatively simple parameterized three-dimensional shape model can then be used as initial values for fitting a parameterized three-dimensional shape model with more parameters and more detail. As an example of such adaptive adjustment of parameters, the number of reference points can be adjusted.

[0016] In some cases, the wire spring can be considered to have intermediate portions and end portions, and such intermediate portions and end portions can differ from each other in geometry. For example, the end portions can have a smaller or larger diameter than the intermediate portions. In some embodiments, the fitting can include fitting the parameterized three-dimensional shape model to the intermediate portions first, and then extending the fitting of the parameterized three-dimensional shape model to the end portions. Since the geometry of the intermediate portions of the wire spring is more regular in most cases, this approach is expected to achieve convergence of the fitting process more quickly. The at least partial convergence achieved for the intermediate portions can then be used as an initial point for the fitting of the end portions by extending the fitting process to the end portions.

[0017] The fitted parametric three-dimensional shape model can be used in various ways. For example, it can be used to evaluate the quality of wire springs made by a coiling machine. This evaluation can be done either on-site at the coiling machine, for example as part of manufacturing quality control, or at some other location, for example at an assembly site where wire springs provided by a third party are assembled into a product such as a spring mattress or some other bedding or seating appliance. In such cases, the measurement method can be performed independently of the coiling machine. In some embodiments, the fitted parametric three-dimensional shape model can also be used as input for controlling the coiling machine, for example to set or readjust spring coiling parameters.

[0018] According to another embodiment, a device for measuring the shape of a wire spring is provided. The device comprises processing circuitry for acquiring images of the wire spring from a plurality of cameras installed at different locations. Further, the processing circuitry is configured to fit a parametric three-dimensional shape model to the acquired images. The parametric three-dimensional shape model is defined by a plurality of reference points which are fitted to align with the trajectory of the wire spring in the images. The device can be used to perform the wire spring measurement method described above. In some embodiments, the device can further comprise the cameras for acquiring the images.

[0019] According to another embodiment, a computer program or computer program product is provided, for example in the form of a non-transitory storage medium. The computer program or computer program product comprises program code which, when executed by a computer, causes the computer to perform the method described above. BRIEF DESCRIPTION OF DRAWINGS

[0020] Embodiments of the present application are described below, by way of example only, with reference to the accompanying drawings.

[0021] Figure 1A and Figure 1B Schematic diagram of a measurement system according to an embodiment.

[0022] Figure 2A , Figure 2B and Figure 2C Schematic diagram of a parametric three-dimensional shape model fitting process according to an embodiment.

[0023] Figure 3 Schematic diagram of a model fitting process according to an embodiment.

[0024] Figure 4 Schematic diagram of a transition model merging process according to an embodiment.

[0025] Figure 5 Flowchart of a method according to an embodiment of the present application. DETAILED DESCRIPTION

[0026] In the following, exemplary embodiments of the inventive concept are described with reference to the accompanying drawings. In particular, in the following detailed description, exemplary embodiments are described in connection with a measurement system 100 and a measurement method which can be performed by the measurement system 100. In the illustrated embodiments, it is assumed that the measurement system is used for measuring a wire spring 10 having a generally helical geometry. However, it should be noted that the inventive concept can also be applied to various other spring geometries, in particular to various types of spring geometries which can be obtained by bending a wire or other type of elongate object or otherwise deforming it.

[0027] Figure 1A Fig. 1 shows a schematic diagram of components of a measurement system 100. Figure 1B Fig. 2 shows a schematic diagram of a functional architecture of the measurement system 100. As shown, the measurement system 100 can comprise a measurement table 110 for placing a wire spring 10 to be measured. Furthermore, a number of cameras 150 are provided for taking images of a measurement space extending above the measurement table 110. The cameras 150 are positioned and oriented such that they take images from different angles. To this end, the measurement system 100 can for example comprise a stand for mounting the cameras 150 and / or the measurement table 110. In some embodiments, the position and / or orientation of the cameras 150 can be adjusted, for example by manual and / or electronic control. In some embodiments, the measurement table 100 can comprise optical markers which can for example assist in the evaluation of images taken by the cameras 150 and / or for the calibration of the cameras 150. The cameras 150 can be commercially available industrial cameras.

[0028] As Figure 1BAs further shown in FIG. 1, the measurement system includes a processing and control subsystem 160. The camera 150 is coupled to the processing and control subsystem 160 so that images taken by the camera 150 can be transmitted to the processing and control subsystem 160, e.g., in the form of image data files, each corresponding to a separate image and, optionally, further corresponding to metadata associated with the image such as a timestamp, camera identification information, etc. In some cases, however, the camera 150 can also provide raw image data (e.g., pixel data) to the processing and control subsystem 160, and the processing and control subsystem 160 can then generate images from the raw image data and, optionally, further generate metadata associated with the image such as a timestamp, camera identification information, etc. The processing and control subsystem 160 can also control the camera 150, e.g., by providing control data to the camera 150. Such control can include, e.g., triggering the camera 150 to take an image, and / or setting one or more image acquisition parameters such as exposure time, aperture, white balance, optical zoom, or digital zoom. In addition, such control can further include electronically adjusting the position and / or orientation of the camera 150. The processing and control subsystem 160 can further include a user interface through which an operator can interact with the measurement system 100 and control various operations as described herein. Such a user interface can be based, e.g., on a graphical user interface (GUI) and user input from various types of human interface devices (HIDs) such as computer mice, touchscreens, keyboards, tablets, or styluses. The processing and control subsystem 160 can be implemented by general-purpose computer hardware, special-purpose processing hardware (e.g., based on one or more application-specific integrated circuits (ASICs)), and / or a combination of general-purpose computer hardware and special-purpose processing hardware, programmed in an appropriate manner.

[0029] As further shown in the figure, the measurement system 100 can further comprise an illumination subsystem 170. The illumination subsystem 170 can comprise one or more light sources for illuminating the wire spring 10 when taking the images. In this way, control and stabilization of the image taking light conditions can be achieved compared to the case of using ambient light. For example, the one or more light sources can be mounted on the same stand as the camera 150 and / or the measurement table 110. In some cases, the illumination subsystem 170 can be used for taking images in a back-lighting mode, e.g. by placing the light sources and the camera 150 on opposite sides of the measurement space. In other cases, the illumination subsystem 170 can be used for taking images in a front-lighting mode, e.g. by placing the light sources and the camera 150 on the same side of the measurement space. In some cases, the illumination subsystem 170 can change its configuration, e.g. by changing the position of the one or more light sources and / or by switching between different light sources, to switch between the back-lighting mode and the front-lighting mode. Such switching between different light sources can also be achieved in accordance with control data provided by the processing and control subsystem 160. When using the back-lighting mode, the quality of the images taken by the camera 150 can be improved, e.g. by reducing reflections and / or image contrast, while when using the front-lighting mode, a simplification of the hardware setup can be achieved (e.g. suitable for cases where the available space is not sufficient to support mounting light sources on opposite sides of the camera 150). The processing and control subsystem 160 can also control the illumination subsystem 170, e.g. by providing control data to the illumination subsystem 170. Such control can for example comprise turning the light sources on or off, controlling the light intensity, or switching between different light sources. Furthermore, such control can further comprise electronically adjusting the position and / or attitude of the light sources, and / or controlling the switching between one or more of the back-lighting mode, the front-lighting mode, and the ambient lighting mode.

[0030] As further shown in the figure, the measurement system 100 can further comprise an actuation subsystem 180. The actuation subsystem 180 can comprise one or more actuators for adjusting the position or attitude of the camera 150. Furthermore, when the measurement system 100 comprises the measurement table 110 described above, the actuation subsystem 180 can comprise one or more actuators for adjusting the position or attitude of the measurement table 110. Furthermore, when the measurement system 100 comprises the illumination subsystem described above, the actuation subsystem 180 can comprise one or more actuators for adjusting the position or attitude of the light sources of the illumination subsystem 170. Furthermore, the actuation subsystem 180 can comprise one or more actuators of a robotic system, such actuators being for example used for automatically placing the wire spring 10 on the measurement table 110, or for automatically feeding the wire spring 10 into the measurement system 100.

[0031] As further shown in the figure, the processing and control subsystem 160 can also be provided with an interface for interacting with the spring winding machine 200. In some cases, the measurement system 100 can be co-located or integrated with the spring winding machine 200. In such cases, the interface for interacting with the spring winding machine can be used to provide evaluation results to the spring winding machine, such evaluation results for example for adjusting spring winding parameters or other configurations of the spring winding machine. Furthermore, the interface can be used to provide information about the wire spring 10 to be measured to the measurement system 100. Such information can for example include type information of the wire spring 10 or nominal spring geometry information of the wire spring 10. Subsequently, the measurement system 100 can use such information, for example by selecting corresponding evaluation input conditions, to improve the measurement process.

[0032] In the present concept, the measurement system 100 measures the wire spring 10 by fitting a parametric three-dimensional shape model to the image provided by the camera. The associated data processing can be done by the processing and control subsystem 160. The parametric three-dimensional shape model is based on a plurality of reference points, in the following also referred to as "model". The plurality of reference points is aligned with the trajectory of the wire spring 10 in the image provided by the camera 150 in the fitting process. Figure 2A Figure 2B Figure 2C A schematic representation of one embodiment of such a parametric three-dimensional shape model 20 is shown. In the illustrated embodiment, it is assumed that the wire spring to be modeled has a generally helical geometry and for example resembles the wire spring 10 described above. Figure 2A A result of the fitting process is shown, which is a schematic representation of the parametric three-dimensional shape model 20 in perspective view, wherein the reference points are indicated by solid dots. In this example, the reference points correspond to the nodes of a three-dimensional spline curve approximating the wire spring trajectory observed in the image provided by the camera 150. For each reference point, the parameters of the parametric three-dimensional shape model 20 can for example include the three-dimensional coordinates of the reference point, the tangent of the spline curve at the reference point, and / or the curvature of the spline curve at the reference point.

[0033] Figure 2B Figure 2C As further shown, the processing and control subsystem 160 can also be provided with an interface for interacting with the spring winding machine 200. In some cases, the measurement system 100 can be co-located or integrated with the spring winding machine 200. In such cases, the interface for interacting with the spring winding machine can be used to provide evaluation results to the spring winding machine, such evaluation results for example for adjusting spring winding parameters or other configurations of the spring winding machine. Furthermore, the interface can be used to provide information about the wire spring 10 to be measured to the measurement system 100. Such information can for example include type information of the wire spring 10 or nominal spring geometry information of the wire spring 10. Subsequently, the measurement system 100 can use such information, for example by selecting corresponding evaluation input conditions, to improve the measurement process. Figure 2B Figure 2C As further shown, the processing and control subsystem 160 can also be provided with an interface for interacting with the spring winding machine 200. In some cases, the measurement system 100 can be co-located or integrated with the spring winding machine 200. In such cases, the interface for interacting with the spring winding machine can be used to provide evaluation results to the spring winding machine, such evaluation results for example for adjusting spring winding parameters or other configurations of the spring winding machine. Furthermore, the interface can be used to provide information about the wire spring 10 to be measured to the measurement system 100. Such information can for example include type information of the wire spring 10 or nominal spring geometry information of the wire spring 10. Subsequently, the measurement system 100 can use such information, for example by selecting corresponding evaluation input conditions, to improve the measurement process. Figure 2B Figure 2C In the illustrated embodiment, it is assumed that each reference point of the model defines a reference sphere 21, the three-dimensional coordinates of the reference point corresponding to the center of the reference sphere 21. As further shown, the reference sphere defines a corresponding circle when projected onto the two-dimensional image of the wire spring. Figure 2B ​​​​​In the case where the fitting process has not yet converged, the circle defined by the reference sphere 21 is assumed to deviate from the trajectory of the wire spring. In particular, the circle defined by the reference sphere 21 is at least partially located outside the boundary of the wire spring. In Figure 2C In the case where the fitting process has converged, the circle defined by the reference sphere 21 is assumed to no longer deviate from the trajectory of the wire spring. In particular, the circle formed by the reference sphere 21 is located within the boundary of the wire spring.

[0034] From Figure 2B and Figure 2C It can further be seen that the diameter of the reference sphere 21 is preferably set to correspond to the diameter of the wire forming the wire spring. However, it should be noted that, in order to ensure that the two-dimensional projection of the reference sphere 21 has a suitable size in the image, it can be necessary to calibrate the camera 150 and / or to perform a calibration process on the image provided by the camera 150.

[0035] Figure 3 For a further illustration of an exemplary implementation of the fitting process, e.g. for generating the model 20 shown in Figure 2A , Figure 2B and Figure 2C .

[0036] As shown, the fitting process is based on a set of model parameters. Such model parameters can for example include: the overall course of the wire spring; the overall position of the wire spring; the number of reference points; and the reference point three-dimensional coordinates, the course of the spline curve at the reference point and / or the curvature of the spline curve at the reference point for each reference point. Furthermore, one or more other model parameters can be defined with one or more other model parameters of the variation freedom. For example, the amount of variation of the reference point coordinates from initial values can be limited to a selectable tolerance range.

[0037] Each reference point can correspond to a reference sphere, and the diameter of the reference sphere can be selected to correspond to the diameter of the wire of the wire spring being modeled. The initial values of the model parameters can be predetermined, and / or can be selected according to known information of the wire spring being modeled. For example, for a wire spring having a substantially cylindrical helical geometry, a particular set of initial values for modeling can be predetermined, and for a wire spring having a barrel-shaped helical geometry, another set of initial values for modeling can be predetermined. Subsequently, a suitable set of initial values can be selected according to known information of the wire spring being modeled. Similarly, the initial values of the model parameters can also be selected according to other known characteristics of the wire spring being modeled, such as the size or the number of turns. For example, for a wire spring having a larger size, more reference points can be selected; for a wire spring having a smaller size, fewer reference points can be selected. Furthermore, for a wire spring having a larger number of turns, more reference points can be selected; for a wire spring having a smaller number of turns, fewer reference points can be selected.

[0038] At block 320, a subset of the model parameters 310 to be used for the fitting process or for a given stage of the fitting process can be selected. For example, the selection at block 320 can comprise selecting only the reference points to be optimized at the fitting process or at the current stage of the fitting process, and / or selecting only the three-dimensional coordinates of such reference points. In this way, a parameter subset 330 is obtained. However, it should be noted that in some cases, all model parameters can also be selected at block 320, such that the fitting process is based on the complete set of model parameters.

[0039] As mentioned above, the images 340 provided by the cameras 150 are used as input for the fitting process. The images 340 are taken from different perspectives of the wire spring that is modeled. In this way, it can be avoided that certain parts of the wire spring are not visible in the images or only visible in a blurred manner, or that the wire spring needs to be moved in order to take images of all parts of the wire spring. In this way, the fitting process can be performed on the basis of the taken images in the free state of the wire spring, such that deformations of the wire spring during such movements are avoided. Furthermore, in this way, it is also possible to make use of images that are taken simultaneously and cover different perspectives of the wire spring. In addition, by performing the fitting on the basis of images that are taken simultaneously at a given time, it is also possible to perform a time-resolved measurement.

[0040] At block 350, the images can be pre-processed. Such pre-processing can for example comprise ensuring that features in images taken from different perspectives, i.e. images taken by different cameras 150, have a consistent representation with respect to each other by scaling and / or warping. Such scaling can for example be done by calibrating the cameras 150. Furthermore, by scaling and / or warping, consistency of the conversion between image features and model features can be ensured.

[0041] Furthermore, in the illustrated embodiment, it is assumed that the above-mentioned pre-processing comprises thresholding and / or filtering, for example by binarization, such that the wire spring profile in the image is sharpened and wire spring corresponding pixels are separated from background pixels. In such a binarized image, the pixel values corresponding to the wire spring trajectory can be "1", while the pixel values corresponding to the background can be "0". In the illustrated embodiment, such a binarized image is distance transformed, for example by replacing the pixel values corresponding to the background by values representing the distance between the nearest pixel corresponding to the wire spring trajectory. In the image resulting from such a distance transformation, the pixels on the wire spring trajectory generally have the maximum value, while the pixels in the adjacent regions generally decrease from the maximum value to zero.

[0042] At block 360, a consistency measure is estimated that represents the deviation of the model from the wire spring trajectory in the image. In particular, the reference points, e.g., the reference spheres, can be projected onto the pre-processed image of block 350. From this projection, a consistency measure can be computed that represents the deviation of the reference points from the wire spring trajectory. For example, for each pixel that corresponds to a projection of a reference sphere, a pixel matching process can be performed to determine whether the pixel matches a pixel in the image that represents the wire spring. For example, if the reference sphere pixel matches a pixel in the distance transformed image that is above a threshold, the result of the index process can be "1", otherwise "0". In this way, a value can be obtained that represents the degree of overlap of the projection of the reference sphere with the wire spring trajectory. Figure 2B Examples can achieve only partial overlap, while Figure 2C Examples can achieve complete overlap. Subsequently, the consistency measure can be obtained by averaging over the reference points and averaging over the images.

[0043] In addition, the image gradient of the distance transformed image can be computed and indexed or pixel matched in a similar manner as described above. That is, if the reference sphere pixel matches a pixel in the image gradient of the distance transformed image that is above a threshold, the result of the index process can be "1", otherwise "0". The resulting value corresponds to a gradient measure that represents the proximity of the projection of the reference sphere to the wire spring trajectory and decreases with increasing distance. This gradient measure can be used as a further input to the iterative optimization process of block 370. In some cases, the computed gradient measure can have a directionality that represents the proximity of the projection of the reference sphere to the wire spring trajectory in a particular direction. In this case, for each two-dimensional image, two independent directions can be considered.

[0044] At box 370, optimization is performed to minimize the deviation between the model and the linear spring trajectory. In the illustrated embodiment, it is assumed that the optimization aims to minimize the reciprocal of the consistency metric. This involves adjusting the current values ​​of the model parameters, particularly the coordinates of each reference point based on the gradient metric. The larger the gradient, the smaller the adjustment. Similarly, the overall orientation and / or coordinates of the model can be adjusted by performing 3D translation and / or 3D rotation. Some model parameters can also be used as constraints on other model parameters. For example, the spline orientation and / or curvature of a reference point can limit the adjustment of the coordinates of that reference point or adjacent reference points. This optimization updates the model parameters selected at box 320, resulting in an updated parametric 3D shape model 380. This updated parametric 3D shape model 380 can be used as the basis for recalculating the reference sphere projection in the next iteration, and the operations described above in conjunction with boxes 360 and 370 can be repeated in each iteration. However, it should be noted that the calculation of the image gradient only needs to be performed in the first iteration.

[0045] After performing fitting processing on the selected portion of the model parameters at box 320, the fitting process can be repeated for the other portion of the model parameters. Alternatively, interpolation or extrapolation can be used to estimate the other portion of the model parameters.

[0046] It should be noted that Figure 3 The fitting process in the example can be modified in various ways. For example, the degree and manner of preprocessing at box 350 can vary depending on the characteristics of the image 340 used as input. For instance, in some cases, the camera can provide an image 340 already in binary format, which can significantly simplify preprocessing. In addition, different types of distance transforms can be employed. In some cases, machine learning (ML) can assist the fitting process. For example, the initial values ​​of the model parameters can be determined at least partially based on a supervised machine learning algorithm. Furthermore, the optimization parameters at box 370 can be tuned using machine learning algorithms (e.g., optimizing the convergence process through reinforcement learning).

[0047] In some implementations, the parametric 3D shape model can also be determined by first defining a set of transitional 3D shape models (which can also be referred to as "transitional models" in this article).

[0048] Figure 4 The illustration shows an example of using this type of transition model. As shown, in this example, a multinomial fitting process 400 is performed. This fitting process 400 can be executed in parallel or not. For example, as combined with the above... Figure 3Each fitting process 400 is based on a respective set of initial conditions 410, which are used as input for the model fitting 420. The fitting processes 400 can use different initial conditions 410 and / or can be based on different subsets of images among the images provided by the camera 150. The subsets of images can for example correspond to partial camera shots or to images taken with back lighting and to images taken with front lighting, respectively. Moreover, the subsets of images can use different image acquisition parameters.

[0049] Each fitting process 400 thus provides a respective transition model 430. At block 440, a merged model 450 is obtained by merging the transition models 430. Such merging can include selecting among the transition models 430 and / or averaging at least some of the transition models 430 and / or concatenating at least some of the transition models 430 to each other. In some cases, the transition models 430 can correspond to different parts of the wire spring being modeled, such as middle and end parts, and the merging can include concatenating the transition models 430 to each other such that the merged model 450 is a model of all parts of the wire spring. The merging can be performed according to a level of quality of the transition models 430, for example according to a value of the consistency measure achieved by the fitting process.

[0050] In one embodiment, the fitting processes 400 can use different initial conditions 410. Accordingly, the transition models 430 are obtained by iteratively performing the fitting process until a given number of iterations is reached, for example 100 iterations. Figure 3 The illustrated fitting processes maximize a level of consistency measure and / or minimize a deviation for each model, respectively. The iterations can be stopped after a given number of iterations is reached, for example 100 iterations. Subsequently, one or more of the transition models 430 are selected according to the level of consistency measure. Among others, a transition model 430 can be selected for which the consistency measure is above a threshold or for which the level of consistency measure is the highest. When more than one transition model 430 is selected, they can be subsequently merged by averaging.

[0051] In some implementations, the fitting process can be performed by using a different set of initial conditions 410 for each iteration. In some implementations, the fitting process can be performed by using a different set of initial conditions 410 for each iteration. Figure 3The parameterized three-dimensional shape model is determined in a manner that the number of model parameters is adjusted during the fitting process. This approach can also be referred to as "adaptive resolution" of the fitting process. For example, at the beginning of the fitting process, a smaller number of model parameters can be used or processed in the update process at block 370. Subsequently, the number of model parameters can be increased, for example, after a certain number of iterations is completed, or when it is found that the level of consistency measure values between two iterations increases by less than a threshold. By adaptively adjusting the number of model parameters in this way, the number of degrees of freedom of the model can be gradually increased to obtain a high-resolution model. At the same time, the fitting process can be kept at a reasonable speed.

[0052] In addition, in some implementations, the fitting process can include different stages related to different parts of the wire spring as the modeling object. In an initial stage, a global parameterized three-dimensional shape model of the entire wire spring can be determined, for example, with a smaller first number of model parameters and Figure 3 The fitting process shown above determines a global parameterized three-dimensional shape model of the entire wire spring. Subsequently, in a next stage, the global parameterized three-dimensional shape model can be locally refined in an iterative manner starting from the middle part of the wire spring and extending towards both ends of the wire spring. In this case, the model parameters of the extended model are in the form of additional reference points at the edges of the middle part currently being modeled, and the initial values are the model parameter values of the previous iteration. The iterative extension process can end when the extended model reaches the endpoints of the wire spring.

[0053] As mentioned above, by calibrating the camera 150 or the image acquisition process of the camera 150, benefits can be obtained in terms of accuracy and / or speed of the fitting process. However, it should be noted that in some implementations, the fitting process can be at least partially self-calibrating by including one or more calibration parameters in the model parameters to be optimized. In either case, by the calibration or self-calibration process, a less sophisticated camera, such as one based on a relatively simple wide-angle lens, can be used compared to a camera based on an optically calibrated lens. The calibration process can be based on known camera calibration algorithms that allow the estimation of a camera matrix used to convert between object coordinates and image coordinates. Such a calibration process can also be based on markers arranged in the measurement space, such as on the measurement table 110.

[0054] After obtaining the parameterized three-dimensional shape model of the wire spring, it can be evaluated in terms of spring quality, for example, in terms of compliance with a nominal shape or geometry, and / or specific characteristics such as diameter, length (especially the length of the wire spring when not under stress), curvature, overall diameter, diameter at the ends of the wire spring, diameter at intermediate positions of the wire spring, etc.

[0055] Figure 5 Flowchart of a method of measuring a wire spring, such as the wire spring 10 described above. Figure 5The illustrated method can be performed by or in connection with the above-described measurement system 100. In some cases, the method can be performed by way of executing program code on a computer-based device. The program code can be executed, for example, by the above-described processing and control subsystem 160. Moreover, the program code can be provided within a storage medium or by way of download or streaming.

[0056] At block 510, one or more cameras are calibrated. The calibration process can include determining at least part of a camera matrix of the cameras. The operations at block 510 can include, for example, calibrating the above-described cameras 150.

[0057] At block 520, images of the wire spring under test are obtained from a plurality of cameras at different positions. Such cameras can correspond to or include the one or more cameras calibrated at block 510. The obtaining of the images can include receiving the images or image data corresponding to the images from the cameras. Moreover, the obtaining of the images can include triggering the image capture and / or otherwise controlling the image capture, for example, in terms of image capture parameters and / or illumination. The image capture can be in backlit or frontlit fashion. Moreover, it is within the contemplation of the present disclosure to employ both images captured in backlit fashion and images captured in frontlit fashion. In some cases, the images can be captured in a stationary state of the wire spring. However, one or more of the images can also be captured during movement of the wire spring. The number of cameras can be three or more. However, it is also possible to use only two cameras, or more than three cameras.

[0058] At block 530, a parameterized three-dimensional shape model is fitted to the obtained images. This operation can be accomplished by the above-described fitting process. Such parameterized three-dimensional shape model is, for example, the above-described model 20; 380; 430, 450. The parameterized three-dimensional shape model is defined by a plurality of reference points. Each reference point is fitted to align with a trajectory of the wire spring in the images (e.g., as described above in connection with Figure 3 The fitting process described above can be accomplished. Such parameterized three-dimensional shape model is, for example, the above-described model 20; 380; 430, 450. The parameterized three-dimensional shape model is defined by a plurality of reference points. Each reference point is fitted to align with a trajectory of the wire spring in the images (e.g., as described above in connection with Figure 2B and Figure 2C Each reference point can define a reference sphere that is fitted to align with the trajectory of the wire spring in the images. The number of reference points can be 50 or more, 100 or more, or 200 or more. The parameterized three-dimensional shape model can be based on a three-dimensional spline curve for each node corresponding to each reference point. The three-dimensional spline curve can be formed by linear spline segments or higher order spline segments.

[0059] The fitting process at block 530 can be accomplished by way of minimizing a degree of deviation of the parameterized three-dimensional shape model from the trajectory of the wire spring in the images (e.g., as described above in connection with Figure 3The fitting process can be performed in an iterative manner according to a gradient of deviation between the reference points locations and the wire spring locations in the image (e.g., according to the gradient metric values described above, which are based on image gradients of the distance transform image).

[0060] In some cases, the fitting process described above can include fitting a plurality of transition parametric 3D shape models to the image, and then generating the parametric 3D shape model by merging the transition parametric 3D shape models. Each of the transition parametric 3D shape models can be fitted to a different subset of the image, and / or each of the transition parametric 3D shape models can be fitted according to a different set of initial conditions. Such a merging of transition parametric 3D shape models is, for example, described above in connection with the method of FIG. 4. Figure 4 As can be seen, in such a merging of transition parametric 3D shape models, one transition parametric 3D shape model can be provided for each subset of the image, one transition parametric 3D shape model can be provided for each set of initial conditions, one transition parametric 3D shape model can be provided for each portion of the wire spring, and / or one transition parametric 3D shape model can be provided for each camera.

[0061] In some cases, the number of parameters of the parametric 3D shape model can be adjusted during the fitting process. For example, a smaller number of parameters can be initially provided, and then the accuracy of the parametric 3D shape model can be improved by increasing the number of parameters. For example, the number of reference points can be adjusted during the fitting process.

[0062] In some cases, the wire spring can have or be considered to be formed of a middle portion and end portions. In such cases, the fitting process described above can include fitting the parametric 3D shape model to the middle portion, and then extending the fitting process of the parametric 3D shape model to the end portions, for example by adding additional reference points at the edges of the parametric 3D shape model that has been fitted to the middle portion, in an iterative manner.

[0063] At block 540, the quality of the wire spring can be evaluated according to the fitted parametric 3D shape model obtained at block 530. Such a quality evaluation can be performed as part of a manufacturing process of the wire spring in a coiling machine (such as the coiling machine 200 described above), and can include performing such a quality evaluation for all wire springs manufactured by the coiling machine or for a selected sample of wire springs. In such cases, the wire springs to be evaluated can be automatically fed from the coiling machine 200 to the measurement system 100. In other cases, the quality evaluation can be performed independently of the manufacturing of the wire spring. For example, the quality evaluation can be performed for wire springs purchased from a supplier. In the latter case, the quality evaluation can be performed, for example, at a site where the wire springs are assembled into a product, such as a spring mattress or other type of bedding and seating appliance.

[0064] At block 550, a wire former (e.g., the wire former 200 described above) used to manufacture the wire spring can be controlled in accordance with the fitted parametric three-dimensional shape model obtained at block 530. For example, spring winding parameters of the wire former 200 can be set or adjusted in accordance with the fitted parametric three-dimensional shape model to, for example, improve the match of the manufactured wire spring to the nominal spring shape or to adjust the characteristics of the manufactured wire spring.

[0065] It should be understood that the above-described methods, devices, and systems can be varied without departing from the spirit of the present application. For example, the present application can be applied to the measurement and modeling of various types of wire springs, not limited to helical spring geometries. Furthermore, it should be noted that the present application can be applied to various types of deployment, such as in a stand-alone measurement system, a measurement system integrated within a wire former, a measurement system co-located with a wire former, or a measurement system integrated with or co-located with a machine used to assemble wire springs into other products.

Claims

1. A method for measuring the shape of a wire spring (10), characterized in that, The method includes: Multiple cameras (150) mounted at different locations acquire images (340) of the wire spring (10). The parametric 3D shape model (20; 380; 430, 450) is fitted to the acquired image (340). The parametric three-dimensional shape model (20; 380; 430, 450) is defined by a plurality of reference points, wherein the plurality of reference points are fitted to the trajectory of the linear spring (10) in the image (340).

2. The method according to claim 1, characterized in that, Each reference point defines a reference sphere (21), wherein the reference sphere (21) is aligned with the trajectory of the linear spring (10) in the image (340) by the fitting.

3. The method according to claim 1 or 2, characterized in that, The number of cameras (150) is three or more.

4. The method according to any one of the preceding claims, characterized in that, The image (340) was taken when the wire spring (150) was in a free state.

5. The method according to any one of the preceding claims, characterized in that, The number of reference points is 50 or more, 100 or more, or 200 or more.

6. The method according to any one of the preceding claims, characterized in that, The parametric 3D shape model (20; 380; 430, 450) is a 3D spline curve based on nodes corresponding to the reference points.

7. The method according to any one of the preceding claims, characterized in that, The fitting is based on minimizing the deviation between the parametric three-dimensional shape model and the trajectory of the linear spring (10) in the image (340).

8. The method according to claim 7, characterized in that, The fitting is performed iteratively based on the deviation gradient between the reference point and the position of the linear spring (10) in the image (340).

9. The method according to any one of the preceding claims, characterized in that, The fitting includes: fitting multiple transitionally parameterized 3D shape models (430) to the image (340); and The parameterized three-dimensional shape model (450) is generated by merging the transition parameterized three-dimensional shape model.

10. The method according to claim 9, characterized in that, Each of the transition parameterized 3D shape models (430) is fitted to a different subgroup of the image (340).

11. The method according to claim 9 or 10, characterized in that, For each of the aforementioned transition parameterized 3D shape models (430), the fitting is based on a different set of initial conditions.

12. The method according to any one of the preceding claims, characterized in that, The number of parameters in the parametric 3D shape model (20; 380; 430, 450) is adjusted during the fitting process.

13. The method according to any one of the preceding claims, characterized in that, The number of reference points is adjusted during the fitting process.

14. The method according to any one of the preceding claims, characterized in that, The wire spring (10) includes a middle portion and an end portion, wherein the fitting includes: first fitting the parametric three-dimensional shape model (20; 380; 430, 450) to the middle portion; and then extending the fitting of the parametric three-dimensional shape model (20; 380; 430, 450) to the end portion.

15. The method according to any one of the preceding claims, characterized in that, include: The quality of the line spring (10) is evaluated based on the fitted parametric three-dimensional shape model (20; 380; 430, 450).

16. The method according to any one of the preceding claims, characterized in that, include: The spring winding machine (200) is controlled based on the fitted parametric three-dimensional shape model (20; 380; 430, 450).

17. A device (100) for measuring the shape of a wire spring (10), characterized in that, The device includes a processing circuit, wherein the processing circuit is used for: Multiple cameras (150) mounted at different locations acquire images (340) of the wire spring (10); and The parametric 3D shape model (20; 380; 430, 450) is fitted to the acquired image (340). The parametric three-dimensional shape model is defined by a plurality of reference points, wherein the plurality of reference points are fitted to the features of the linear spring (10) in the image (340).

18. The apparatus (100) according to claim 17, characterized in that, The device (100) is used to perform the method according to any one of claims 1 to 16.

19. The apparatus (100) according to claim 17 or 18, characterized in that, The device (100) also includes the camera (150).

20. A computer program or computer program product comprising program code, characterized in that, When the program code is executed by a computer-based device (100), it causes the computer-based device (100) to perform the method according to any one of claims 1 to 16.

Images