WO2025219683A1 - Calibration chart for an optical sensor - Google Patents

Abstract

The present disclosure relates to a calibration chart (1) comprising a face (10) having a first zone (11) and a second zone (12), the first zone (11) being defined by a pattern configured to be detectable by a camera (2), the first zone having a reflectivity factor of at least 94% for which at least 90% of the reflection is diffuse, the second zone having an emissivity factor of at least 50% for an emission wavelength range of between 2 and 12 μm.

Description

Description

Title of the invention: Calibration target for optical sensor Technical field

[0001] The present invention relates to the field of optical calibration and more particularly to the calibration of infrared cameras.

[0002] For the purposes of various applications, it is known to equip a robot arm with one (or more) camera(s) equipped with a matrix optical sensor for the purpose of acquiring data. As with other effectors equipping the robot, it is necessary to know precisely the actual position and orientation of the camera, which serves as a tool reference for the robot's computer and allows it to correctly position the camera for the needs of the operations (generally control) to be carried out.

[0003] A common example of application is that of non-destructive testing (NDT) by thermography of a part (e.g. a sheet metal) having a complex surface. On large parts, it is then necessary to carry out several optical acquisitions from different points of view using an infrared camera positioned on the mobile arm of a robot. It is also possible for the part to be movable by a robot arm, so as to offer several points of view to a fixed camera.

[0004] This is also the case for smaller parts, where we are trying to detect smaller defects. For example, to check defects of the order of ten microns, it is necessary to perform a zoom operation, which reduces the camera's field of vision. For example, if we consider that the camera has a rectangular field of vision, this can measure a few tens of centimeters on each side if we do not use a zoom. The same camera, with a zoom, will see its field of vision reduced to, for example, 1 or 2 centimeters on each side. It is therefore sometimes necessary to assemble several hundred acquisitions to cover a part.

[0005] Using the information available on the position and orientation of the robot, for each measurement point, the images from the camera can be projected onto a 3D virtual model of the part. However, ordinary procedures only provide a very approximate projection of the images onto the three-dimensional virtual model of the part. In fact, the different projected images do not perfectly superimpose on the virtual model of the part and discontinuities are observed between them. These discontinuities result, for example, from a gap between two acquisitions, induced by a lack of precision in the knowledge of the position of the camera relative to the subject. [0006] In the case of a non-destructive testing operation, these discontinuities may be taken for defects by the person or the algorithm in charge of checking the image, which may slow down the diagnostic operation or even lead to erroneous conclusions about the state of the part in question.

Prior art

[0007] In order to limit the shifts between the images, a theoretical solution consists of precisely determining the dimensions of the tooling used to fix the camera to the robot in order to obtain better control of its positioning. However, the low tolerance of the assembly to angular positioning makes its implementation complex and drastically increases the cost of the tools for an uncertain effect.

[0008] It is therefore preferable to use a calibration target to precisely determine the position of the camera in relation to the subject. For example, in the case of non-destructive testing, this allows the actual position of the camera to be known in real time in order to precisely assemble several images and thus limit the risk of errors due to imprecise connections.

[0009] For example, it is known from document US2012069193 to use a calibration target comprising a pattern visible by an infrared camera. The pattern being, in one example, a checkerboard whose squares have a different emissivity from the other squares. This difference in emissivity is sufficient, with the addition of heat, for the checkerboard pattern to appear in the infrared image and be used for calibrating the camera.

[0010] In theory, using two different emissivities is satisfactory, especially for large targets. However, it is difficult to maintain a satisfactory level of precision with this solution on small targets, for example of the order of a few centimeters on each side. In addition, such a calibration target is difficult to manufacture, since it involves applying a plurality of squares of each of the two materials to create the checkerboard. In short, such a solution is not sufficiently precise for industrial needs.

[0011] It is also known from the publication by R. Usamentiaga et al. “Highly accurate geometric calibration for infrared cameras using inexpensive calibration targets”, to print a high-emissivity pattern in ink on a low-emissivity metal support. The printing principle described is not entirely satisfactory, particularly for small-sized targets. In fact, the position of the parts of the printed pattern is not sufficiently precise to meet industrial needs.

[0012] Finally, state-of-the-art test patterns are not fully satisfactory in terms of image quality. Indeed, it is essential for test patterns, particularly at high precision, to provide a very high contrast image, that is to say on which the area of high emissivity is very “clearly” distinguished from the area of lower emissivity.

[0013] To increase the contrast of the image, it is inevitable to increase the reflectivity factor of the area of lower emissivity, so that it has more contrast in the infrared. The choice is therefore systematically made for materials which have a so-called "specular" reflection, because they offer the highest reflectivity and thus allow better levels of contrast to be achieved.

[0014] On the other hand, such targets are difficult to use because they provide very “noisy” images, due to the shine of the materials used. These images then require significant digital processing before being used, which slows down control operations and increases their cost.

[0015] The state of the art is therefore not fully satisfactory in that it is systematically confronted with an inevitable compromise: choosing between a material with a high reflectivity factor, but generating restrictive noise, or a material with a lower reflectivity factor, generating little noise, but producing low contrast images that are unusable for high-precision applications such as industrial non-destructive testing.

[0016] The present disclosure therefore aims to at least partially overcome the drawbacks of the state of the art cited above.

[0017] In particular, an objective of the present disclosure is to propose a solution which makes it possible both to provide a high-contrast image, making it possible to meet the precision requirements of an industrial application, and at the same time to reduce the noise in the image, hitherto unavoidable on state-of-the-art test charts.

Summary

[0018] The above-mentioned objectives are achieved in particular by a calibration target comprising a face comprising a first zone and a second zone, the first zone being defined by a pattern configured to be detectable by a camera, the first zone comprising a reflectivity factor of at least 94% for which at least 90% of the reflection is diffuse, the second zone comprising an emissivity factor of at least 50% for an emission wavelength range included in the medium and long infrared spectrum, i.e. between 2 and 12 pm.

[0019] In other examples, the second zone has an emissivity value (or factor) of at least 75%, or at least 90%, or at least 95%.

[0020] Indeed, the inventors determined during research work that the calibration target generates significantly less “noise” in the image when the reflection of the first zone is diffuse, in particular 90% diffuse, and that it retains its highly reflective character, in particular that it includes a reflectivity factor greater than 94%.

[0021] The features set out in the following paragraphs may, optionally, be implemented, independently of one another or in combination with one another:

[0022] According to examples, the calibration target comprises a plate, one face of which comprises the first zone and the second zone, the second zone being defined by a plurality of unitary patterns, preferably circular, produced on said face of the plate, the first zone being defined as the part of the face which is not covered by the plurality of unitary patterns.

[0023] According to examples, the calibration target comprises a pattern formed from at least three unitary patterns.

[0024] According to examples, the unitary patterns are regularly distributed in a plurality of rows and columns.

[0025] According to examples, each unitary pattern is a few mm, or a few cm apart from the other unitary pattern(s).

[0026] According to examples, the first zone comprises a material from among: a metal, a polymer.

[0027] According to examples, the first zone comprises aluminum.

[0028] The present disclosure further comprises a method of manufacturing the calibration target described above, comprising:

- Supply of a plate comprising a face defining a first zone comprising a reflectivity factor of at least 94%, and for which at least 90% of the reflection is diffuse,

- Printing, in ink, one or more unitary patterns to form a pattern on said first area of said plate, said pattern defining a second area having an emissivity factor of at least 50%,

[0029] According to examples, the ink used for printing one or more unitary patterns defining the second zone is an ink polymerized under UV radiation.

[0030] According to examples, each unitary pattern has a height, when measured perpendicular to the plate, which is less than or equal to 200 microns, preferably less than or equal to 50 microns. Such a thickness may correspond to the thickness of the ink. This makes it possible to avoid shadow phenomena generating location errors when taking images.

[0031] According to examples, the unitary patterns have an elliptical shape, more advantageously a circular shape. [0032] The present disclosure further relates to a method of using a calibration target described above, comprising:

- Supply of a robot arm carrying a camera,

- Provision of a calibration target as described above, the pattern of said calibration target comprising a plurality of distinct unitary patterns,

- Acquisition, by means of the camera, of a plurality of images of the target according to various orientations and spatial positions of the robot arm,

- Calculation, from said plurality of images and the position of said plurality of unitary patterns, of the calibration parameters of said camera.

[0033] The present disclosure further relates to a thermography installation capable of operating in a mid-infrared wavelength range of between 2 and 12 pm comprising a calibration target as described above, a robot arm and a camera, one or other of the calibration target or the camera being mounted on the robot arm.

[0034] According to examples, the camera is an infrared camera, the wavelength of which is between 2 and 5 pm or between 7 and 12 pm.

Brief description of the drawings

[0035] Other characteristics, details and advantages will appear on reading the detailed description below, and on analyzing the attached drawings, in which:

[0036] [Fig.l] shows a schematic representation of a robot arm supporting a camera, oriented towards a calibration target according to an example of the present disclosure.

[0037] [Fig.2] shows several views 2a, 2b, 2c, 2d illustrating different examples of possible patterns for a calibration target.

[0038] [Fig.3] shows a section of a calibration target according to an example of the present disclosure.

[0039] [Fig.4] shows a first representation 4a of a calibration target whose first zone does not exhibit a diffuse reflection, and a second representation 4b of a calibration target according to an example of the present disclosure whose first zone exhibits a diffuse reflection.

Description of the embodiments

[0040] The drawings and the description below contain, for the most part, elements of a certain character. They may therefore not only serve to better understand the present disclosure, but also contribute to its definition, where appropriate.

[0041] In the various figures, the same references designate identical or similar elements. For the sake of brevity, only the elements which are useful for understanding of the described embodiment are shown in the figures and are described in detail below.

[0042] Reference is now made to [Fig. 1], which shows an example of a thermography installation. In this example, the aim is to know precisely the position of a camera 2 on a robot arm 3 using a calibration target 1 according to an example of the present disclosure. As explained above, it is essential to know precisely the position parameters of the camera 2 in order to be able, in particular during a non-destructive testing operation, to reconstruct a faithful view of the object being tested by assembling several acquisitions with each other. Inaccuracies in the position of the camera result in inaccuracies at the junctions between the images glued together which can distort or slow down the testing operation.

[0043] In the solution proposed in the present document, the calibration target 1 comprises a pattern to be detected in the infrared domain by the camera 2, in a medium (2 to 5 pm) or long (7 to 12 pm) infrared wavelength range. It is however possible to adapt the calibration target to other wavelength ranges without departing from the scope of the present disclosure.

[0044] Such detection can be carried out by an image processing algorithm, executed by a processing unit comprising a processor (not shown). This pattern to be detected can take several forms depending on the camera used: checkerboard, network of points, etc. as shown by way of example in [Fig.2]. The calibration target can be composed of several materials (paper, plastic, metal, wood, etc.). In the context of an application using an infrared camera presented here by way of non-limiting example, the pattern is different from the checkerboard encountered conventionally, and comprises a network of unitary patterns P of circular shape.

[0045] The calibration target 1 illustrated in [Fig.l] comprises a plate 4, having a first zone 11 of a first emissivity value, the plate 4 also comprising a pattern defining a second zone 12 of a second emissivity value.

[0046] To better understand the concept of emissivity, it can be considered as inversely proportional to reflectivity. If a body is particularly emissive, it is relatively non-reflective, and vice versa.

[0047] In our case, according to examples, the first zone 11 can then be characterized in that it is particularly reflective, and the second zone 12 in that it is particularly emissive. For example, the first emissivity value can be of the order of 30%, while the second emissivity value can be of the order of 90%. Although the present disclosure focuses on configurations in which the first zone 11 is reflective and the second zone 12 is emissive, it is also possible to provide an inverse configuration, in which it is the first zone

11 which is particularly emissive, and the second zone 12 which is particularly reflective. It is therefore implicit in the description that the reverse configuration is possible. The essential thing is that the first and second zones 11, 12 have different emissivity values. Advantageously, they are significantly different: the greater the difference, the more the calibration target 1 will make it possible to obtain contrasting images in the infrared.

[0048] In other examples, it is possible for the first emissivity value to be between 5 and 10%, and for the second emissivity value to be between 70 and 80%.

[0049] In other examples, the second zone has an emissivity value of at least 75%, or at least 90%, or at least 95%.

[0050] In order to obtain a high contrast for the camera, it is advantageous to use a highly reflective material for the first zone 11 of the calibration target 1. This type of material has, according to DIN 5036, part 3, a total reflection factor of at least 85%, preferably at least 90%. The first zone 11 of the calibration target 1 according to the present disclosure advantageously has a reflectivity factor, in other words a total reflection factor, of at least 94%.

[0051] In examples, the first zone 11 is complementary to the second zone

12 on the face 10 of the plate 4. In this way, regardless of its shape, the second zone 12 is always included in the outline of the first zone 11. The second zone 12 may typically comprise a plurality of regions distinct from each other. In examples, each region is materialized by a unitary pattern P of circular shape. In other examples already discussed, the regions will be squares, possibly in contact by their corners, distributed in a regular checkerboard. By “regular” pattern, we mean that it is constructed from one or more repetitions of a unitary pattern. The pattern may comprise unitary patterns which may be identical, i.e. of the same 3D conformation.

[0052] In examples shown in [Fig. 1] and in Figures 2a and 2b, the second zone 12 is defined on the plate 4 by unitary patterns P, regularly spaced on the plate 4 so as to form a pattern. The background of the plate 4 on which the second zone 12 is defined then defines the first zone 11. [Fig. 2] shows different, non-limiting examples of possible patterns. Conventionally, the network of unitary patterns P can comprise a number M of columns and a number N of rows. It is also possible for one or more column(s), or row(s), to have a different number of unitary patterns P. Since the invention is not limited to infrared camera checks, it is possible to use a target formed by any pattern on a background so that there is a contrast between the background and the pattern. [0053] It is finally possible to adapt the embodiment of the calibration target 1 according to the application cases: the type of camera, the type of operation, etc. It is however advantageous for locating the position of the camera to be calibrated, that the second zone 12 which forms a pattern on the first zone 11 makes it possible to easily and precisely locate the majority of the points appearing in the image.

[0054] The plates shown by way of non-limiting example in [Fig.l] and 2 are substantially flat, and of generally rectangular or square shape. It is however possible, in other examples, for the plate to be curved. Furthermore, it is also possible for the plate to be of a shape other than a rectangle, for example an ellipse or any shape including or encompassing the pattern.

[0055] In one example, the plate 4 is made of a highly reflective material as defined above, and the second particularly emissive zone 12 is made by adding material to this plate 4, the surface of the plate 4 not covered by the second zone 12 constituting the first zone 11. Such an example is notably shown [Fig. 3]. The plate 4 can be fixed on a more rigid support 5, for example made of a plastic with high mechanical resistance, in particular to bending, or of any other material chosen more for its mechanical properties than for its optical properties. According to a preferred example, the plate 4 comprises aluminum, the face on which the pattern is made is thus made of aluminum. According to another example, the plate 4 comprises a polymer.

[0056] According to an example of a method for manufacturing the calibration target 1, of which [Fig. 3] represents an example of the product obtained, a plate 4 is provided comprising a face defining a first zone 11 having a reflectivity factor of at least 94%, and for which at least 90% of the reflection is diffuse, and on which one or more unitary patterns P are printed, for example with ink, to form a pattern on said first zone 11 of said plate 4, said pattern defining a second zone 12 having an emissivity factor of at least 50%,

[0057] In one example, the plate 4 has a thickness of less than 1 mm, and is glued to a thicker support 5 which makes it possible to stiffen the assembly. It may be advantageous to stiffen the calibration target 1 to prevent it from deforming and losing its flat shape. Indeed, it is not essential but it remains advantageous for the plate to be substantially flat, because this can facilitate image acquisition.

[0058] The support 5 may also comprise a heating means 6 which makes it possible to further increase the contrast of the image by adding heat, for example during acquisition by the camera 2.

[0059] The pattern which defines the second zone 12 can be produced by printing an ink, preferably black, on the plate 4. The black color makes it possible in particular to obtain the highest emissivity factor. Such a method using an operation printing is particularly suited to the high precision required for this type of target, particularly for application in the inspection of industrial parts. However, it is possible to produce the pattern defining the high emissivity zone in any other manner known to those skilled in the art.

[0060] Printing can be carried out using an ink that hardens upon exposure to UV rays, but can also be carried out with any ink known to those skilled in the art.

[0061] It may be advantageous for the pattern to have a height of less than 200 micrometers, advantageously less than 50 micrometers. This ensures that no shadow obscures parts of the first zone 11 and makes the learning of the position of the unitary patterns by the camera 2 imprecise.

[0062] Reference is now made to [Fig. 4]. According to the present disclosure, the first zone 11 allows a so-called “diffuse” reflection. Such a reflection is characterized by the fact that it reflects light in a large number of directions. Ideally, a reflection that is 100% “diffuse” reflects light, regardless of its direction of incidence, uniformly in all directions. Such a reflection is opposed to a so-called “specular” reflection, in which the light is reflected in a direction that depends on the direction of incidence of the light on the reflecting surface, obeying the Snell-Descartes law. In other words, a plane surface reflecting in a perfectly specular manner will return an incident ray in a direction symmetrical with respect to a plane of incidence, normal to the reflecting surface. This is the case, for example, of a mirror, or known calibration targets, an example of which is shown in part 4a on the left of [Fig. 4]. Conversely, a surface reflecting in a “diffuse” manner will return an incident ray in a multitude of directions, so that it will not be possible, by perceiving one of the reflected rays, to determine where the source of the incident ray comes from. In other words, it is a visual rendering of the type obtained by a surface with a so-called “matte” appearance. For example, a sheet of paper, or a brushed metal coating. However, the materials used for this type of rendering do not have a sufficient reflectivity factor. Furthermore, on materials that are originally sufficiently reflective, the surface treatments used to obtain this appearance degrade their reflectivity factor and therefore do not constitute a solution to the problem posed. The calibration target 1 according to the present disclosure proposes a solution making it possible to overcome this compromise. A representation of such an example of calibration target 1 is shown in part 4b on the right of [Fig.4],

[0063] The unitary patterns P are an example of a pattern that the camera 2 must detect, and are in principle highly emissive. In particular, a unitary pattern P comprises a width 1 of the order of a millimeter. Preferably, the width 1 is comparable to a diameter of the unitary pattern P if it is circular, the height H of this pattern corresponding to a thickness of ink necessary for the formation of the unitary pattern (We then understand that the surface of the pattern is not smooth) and is controlled so as not to generate shadow phenomena when taking images.

[0064] [Fig.l] thus represents a method of using a calibration target 1 according to an example of the present disclosure, the method comprising:

- The supply of a robot arm 3 carrying a camera 2, the camera possibly being an infrared camera,

- The provision of a calibration target 1 according to the present disclosure, the pattern of said calibration target 1 comprising a plurality of distinct unit patterns P,

- The acquisition, by means of camera 2, of a plurality of images of the target

1 according to various orientations and spatial positions of the robot arm 3,

- Calculating, from said plurality of images and the position of said plurality of unitary patterns P, the calibration parameters of said camera

2 in order to know its precise position.

Claims

Claims [Claim 1] Calibration target (1) comprising a face (10) comprising a first zone (11) and a second zone (12), the first zone (11) being defined by a pattern configured to be detectable by a camera (2), the first zone comprising a reflectivity factor of at least 94% for which at least 90% of the reflection is diffuse, the second zone comprising an emissivity factor of at least 50% for an emission wavelength range of between 2 and 12 pm. [Claim 2] Calibration target (1) according to claim 1, comprising a plate (4) one face (10) of which comprises the first zone (11) and the second zone (12), the second zone (12) being defined by a plurality of unitary patterns (P), preferably circular, produced on said face (10) of the plate (4), the first zone (11) being defined as the part of the face (10) which is not covered by the plurality of unitary patterns (P). [Claim 3] Calibration target (1) according to one of claims 1 to 2, in which the first zone (11) comprises a material from among: a metal, a polymer. [Claim 4] Calibration target (1) according to claim 3, in which the first zone (11) comprises aluminum. [Claim 5] Method of manufacturing the calibration target (1) according to one of claims 1 to 4 comprising: Provision of a plate (4) comprising a face defining a first zone (11) having a reflectivity factor of at least 94%, and for which at least 90% of the reflection is diffuse, Printing, in ink, one or more unitary patterns (P) to form a pattern on said first zone (11) of said plate (4), said pattern defining a second zone (12) comprising an emissivity factor of at least 50%, [Claim 6] Method according to claim 5, in which the ink used for printing the one or more unitary patterns (P) defining the second zone (12) is an ink polymerized under UV radiation. [Claim 7] Method according to one of claims 5 to 6, in which each unitary pattern (P) has a height (H), when measured perpendicular to the plate (4), which is less than or equal to 50 microns. [Claim 8] Method of using a calibration target (1) according to one of claims 1 to 4, comprising: Supply of a robot arm (3) carrying a camera (2), Provision of a calibration target (1) according to one of claims 1 to 5, the pattern of said calibration target (1) comprising a plurality of distinct unitary patterns P, Acquisition, by means of the camera (2), of a plurality of images of the target (1) according to various orientations and spatial positions of the robot arm (3), Calculation, from said plurality of images and the position of said plurality of unitary patterns (P), of the calibration parameters of said camera (2). [Claim 9] Thermography installation capable of operating in a wavelength range between 2 and 12 pm comprising a calibration target (1) according to claims 1 to 5, a robot arm (3) and a camera (2), one or other of the calibration target (1) or the camera (2) being mounted on the robot arm (3).