Method for characterizing an object

Abstract

The invention relates to a method for characterizing an object having at least one reflective property, the method comprising: providing an optical system comprising a panel displaying a first model having asymmetric patterns and a second model having irregular patterns, and at least one sensor, the object to be characterized comprising a first surface; detecting, by the sensor, the first model and the second model in reflection on the first surface of the object; the detection being a single image capture by the sensor; determining coefficients of deformation of the first model reflected on the first surface and detected by the sensor, relative to the first model displayed on the panel; establishing a unique relationship between the patterns of the second model detected by the sensor after reflection on the first surface and the patterns of the second model displayed on the panel.

Description

[0001] METHOD FOR CHARACTERIZING AN OBJECT

[0002] technical field

[0003] The present invention relates to a method for characterizing an object.

[0004] Previous art

[0005] Numerous methods have been developed for the characterization of an object, as described in documents EP3012578B1, US7554678B2, US7602507B2, WO2015169730A1 and KR102015219B1. However, the solutions proposed in these documents are complex to implement.

[0006] There is a need for a simpler method of characterizing an object.

[0007] Description of the invention

[0008] To this end, the invention proposes a method for characterizing an object exhibiting at least one reflective property, comprising

[0009] The supply of an optical system comprising at least one light-diffusing panel on which are displayed a first model with asymmetrical patterns and a second model with irregular patterns, and at least one sensor,

[0010] The provision of the object to be characterized, the object comprising a first surface,

[0011] The sensor detects the first and second models reflected on the first surface of the object, the detection being a single image capture by the sensor.

[0012] The determination of deformation coefficients of the first model detected by the sensor relative to the first model displayed on the panel, the deformation occurring by reflection on the first surface,

[0013] Using deformation coefficients, a unique relationship is established between the patterns of the second model detected by the sensor after reflection on the first surface and the patterns of the second model displayed on the panel. According to one variant,

[0014] The panel emits light rays,

[0015] Sensor detection involves acquiring a light ray at a pixel of a sensor's pixel matrix, each light ray reaching the sensor in a specific orientation after reflection from the object.

[0016] Establishing the unique relationship involves establishing the relationship between the pixel on the sensor and each pattern of the second pattern displayed on the panel.

[0017] According to one variant, the method further includes, using the unique relationship and orientation of the light rays reflected towards the sensor,

[0018] The identification of a reflection point for each light ray on the first surface of the object, and the identification of a normal to the first surface at the reflection points, and the geometric characterization of the first surface using the reflection points and normals.

[0019] According to one variant, the optical system comprises a single sensor; the geometric characterization of the first surface is obtained by the sensor's detection, which includes the acquisition of three light rays in three neighboring pixels of the sensor's pixel matrix.

[0020] According to one variant, the optical system includes several sensors; the geometric characterization of the first surface is obtained by detection by the sensors, including the acquisition of each light ray in a pixel of the sensor pixel matrix.

[0021] According to one variant, the process also includes

[0022] The measurement of the light intensity of the first or second model displayed on the panel and the light intensity of the corresponding model detected by the sensor after reflection on the first surface,

[0023] Comparing light intensities,

[0024] The identification of an anomaly on the first surface as a function of a variation in light intensity.

[0025] According to one variant, the method is for characterizing an object that also exhibits a refractive property. According to another variant, the object provided includes a second surface, the method comprising

[0026] The sensor detects the first and second models reflected on the second surface of the object, the detection being a single image capture by the sensor.

[0027] The determination of the deformation coefficients of the first model detected by the sensor relative to the first model displayed on the panel, the deformation occurring by reflection on the second surface

[0028] Using the deformation coefficients, a unique relationship is established between the patterns of the second model detected by the sensor after reflection on the second surface and the patterns of the second model displayed on the panel.

[0029] According to one variant, the process further includes the separation of patterns captured on the single image by reflection on the first surface and on the second surface.

[0030] According to one variant, the object has a plurality of superimposed layers in an adjacent or non-adjacent manner, each presenting a first surface and possibly a second surface, the steps of the process being repeated for each layer.

[0031] According to one variant, the panel is a backlit screen.

[0032] According to one variant, the models are displayed on a plurality of panels, so that the entire surface of each object reflects both models back to the sensor(s).

[0033] According to one variant, the characterization of the object includes the determination of at least one characteristic including the shape, dimensions, curvature, thickness, wavefront regardless of wavelength, optical properties deducible from the wavefront, reflectance and transmittance of each surface.

[0034] In one variant, the method further includes determining the optical power of the object using its refractive index, thickness, and curvature. In another variant, a plurality of distinct objects of a structure are geometrically characterized and then assembled together.

[0035] According to one variant, the object is an object from the group including a vehicle window, a vehicle body part, a building window, a helmet window, a spectacle lens, a watch window, a diamond, contact lenses, an intraocular lens, an eye.

[0036] The use, in this document, of the verb "comprendre" (to understand), its variants, and its conjugations, cannot in any way exclude the presence of elements other than those mentioned. The use, in this document, of the indefinite article "un" (a), "une" (an), or the definite article "le" (the), "la" (the), or "l'" (it) to introduce an element does not exclude the presence of multiple such elements.

[0037] The terms "first >>", "second >>", "third >>, etc. are used in this document exclusively to differentiate different elements, without implying any order between these elements.

[0038] The different modes of implementation can be considered alone or in combination.

[0039] Brief description of the figures

[0040] Other features and advantages of the present invention will become apparent upon reading the detailed description that follows, for understanding of which reference should be made to the accompanying figures which show:

[0041] - Figure 1, a view of an example of implementation of the invention;

[0042] - Figure 2, a view of an example of implementation of the invention;

[0043] - Figure 3, a view of an example of implementation of the invention;

[0044] - Figure 4, a view of an example of implementation of the invention;

[0045] - Figure 5, a view of an example implementation of the invention.

[0046] The drawings in the figures are not to scale. Similar features are generally denoted by similar reference numerals in the figures. Within the scope of this document, identical or analogous features may bear the same reference numerals. Furthermore, the presence of reference numerals or letters in the drawings shall not be considered limiting, even when such numerals or letters are specified in the claims.

[0047] Detailed description of embodiments of the invention

[0048] The invention relates to a method for characterizing an object exhibiting at least one reflective property. The method includes a step of providing an optical system comprising a light-diffusing panel on which are displayed a first model with asymmetrical patterns and a second model with irregular patterns, and the system also comprising at least one sensor. The method includes the detection by the sensor of the first and second models reflected on a first surface of the object, the detection being a single image capture by the sensor. The method includes determining the deformation coefficients of the first model detected by the sensor relative to the first model displayed on the panel, the deformation occurring by reflection on the first surface.Using deformation coefficients, the process also establishes a unique relationship between the patterns of the second model detected by the sensor after reflection from the first surface and the patterns of the second model displayed on the panel. The process is simple because a single image is captured by the sensor, an image that includes both the first and second models. Processing a single image incorporating two types of models is simplified, thus simplifying the characterization process.

[0049] Figures 1 to 5 show examples of implementation of the invention. Figure 1 shows an object 3 to be characterized. Object 3 exhibits at least one reflective property. Object 3 is capable of reflecting a light ray. Object 3 comprises a first surface 3a capable of reflecting a light ray. This surface may be an opaque object or a mirror.

[0050] The figures show an optical system 10 for implementing the characterization method. The optical system 10 may include a central processing unit (not shown) for implementing the invention, enabling information processing and the characterization of the object 3. The optical system 10 includes at least one sensor 1. A plurality of sensors may be provided, for example, two sensors 1 and 6, as shown in Figure 1. The optical system 10 also includes at least one light-diffusing panel 2. This panel may be a surface of any type, shape, and material. It may be any surface that diffuses light, either from the rear, such as a backlit screen, or from the front, such as a surface illuminated by a light source or by natural light. The optical system 10 also includes models displayed on the panel 2. The optical system 10 includes at least two types of models (different and distinct) that are combined.There can be more than two types of patterns. A first pattern, 5, is displayed with asymmetrical patterns. The patterns have a single direction. In other words, the first pattern, 5, contains patterns that are individually asymmetrical. The patterns have no axis or center of symmetry. The patterns have a uniquely identifiable orientation. For example, the letter "P" can be a pattern of the first pattern, 5; this letter is asymmetrical because it has no axis or center of symmetry. According to Figure 1, the patterns of the first pattern, 5, are, for example, a vertical stroke and a horizontal stroke, separated by a space at the vertical stroke. These patterns have no symmetry. The patterns can be repeated (all identical). The patterns can be arranged with regular spacing. The patterns can be arranged periodically within the first pattern, 5. A second pattern, 25, is displayed with irregular patterns.The patterns are all different from each other and different from the patterns of the first model 5. The second model 25 also contains patterns lacking symmetry. The patterns of the second model 25 are also asymmetrical. The patterns of the second model 25 are unique and locally identifiable. The patterns have no axis or center of symmetry. For example, the letter "F" can be a pattern of the second model 25, as this letter has neither an axis nor a center of symmetry. According to Figure 1, the patterns of the second model 25 are random, irregular shapes. The patterns are arranged with irregular spacing. The patterns of the second model 25 are arranged in a mixed (but distinct) manner among the patterns of the first model 5. The method includes the detection by sensor 1 of the first model 5 and the second model 25 reflected on the first surface 3a of the object 3.The two patterns 5 and 25 displayed on panel 2 are reflected onto surface 3a and detected by sensor 1. The displayed (initial) patterns on panel 2 are detected (captured) by sensor 1. Since the patterns of the two patterns 5 and 25 are mixed, sensor 1 detects the patterns of both patterns simultaneously. The detection of the patterns of the two patterns 5 and 25 by sensor 1 constitutes a single image capture. A single image is captured by sensor 1. This reduces processing by the central processing unit and thus simplifies the process. It also allows the process to be applied quickly.

[0051] The process then involves determining the deformation coefficients (or parameters) of the first pattern detected (or captured) by sensor 1 relative to the first pattern 5 displayed (or initial) on panel 2, the deformation occurring through reflection on the first surface 3a. Sensor 1 captures an image of the patterns of the first pattern 5. The patterns detected (or captured) by sensor 1 are deformed relative to the patterns displayed (or initial) on panel 2 by reflection on the first surface 3a. The extent (or manner) by which the patterns of the first pattern 5 are deformed is determined. In other words, a deformation law is established between the patterns of the first pattern 5 and the patterns captured in the image by sensor 1. Since the patterns of the first pattern 5 are asymmetrical, it is possible to establish how the deformation occurs through reflection on the surface.Since all the patterns of the first model 5 are identical—but asymmetrical—it is possible to establish the deformation coefficients without determining which pattern of the first detected model corresponds to the pattern of the first model displayed on panel 2. The deformation of the patterns occurs according to coefficients that can be geometric. These coefficients may relate to the size and / or shape of the deformed patterns. The first model 5 and its patterns serve as orientation markers. The patterns of the first model 5 can have any shape, as long as they uniquely define a transformation (possibly geometric) between the patterns displayed on panel 2 and the captured patterns. Using the deformation coefficients, the process then re-establishes a unique relationship between the patterns of the second model 25 detected by the sensor after reflection on the first surface 3a and the patterns of the second model 25 displayed on panel 2.This involves a bijective relationship between the patterns of the second model 25 detected by the sensor after reflection on the first surface 3a and the patterns of the second model 25 displayed on panel 2. It is thus possible to determine which pattern of the initial second model 25 displayed on panel 2 corresponds to the pattern of the second detected model, captured by sensor 1 – and vice versa. Therefore, on a single image captured by the sensor, the process allows us to obtain both information on the deformation produced by the reflection and information on the position of a pattern displayed on the panel and its position on the captured image. This allows us to characterize the object; the characterization can be, for example, geometric (particularly for reconstructing the object), cosmetic, or optical.

[0052] Sensor 1 may include a pixel array for image capture. Sensor 1 may be a camera. Sensor 1 – for example, in the form of a camera – is calibrated before the process is initiated. Figure 3 schematically represents a pixel array 16 of pixels 17, 18, as well as a projection center 15 of sensor 1 (corresponding to the focal point of sensor 1 in the form of a camera).

[0053] Panel 2 diffuses light. Panel 2 emits light rays. The light ray from the panel is referenced as 26, and the light ray detected by sensor 1 is referenced as 261, after reflection from the first surface 3a. In the figures, the light ray 26 is shown from a position 1b on panel 2. For each ray 261 captured by the sensor, there is a one-to-one correspondence between the position on panel 2 and a pixel 17 of the 16-pixel matrix of sensor 1. The light ray 261 arriving at sensor 1 is in a certain orientation after reflection from the first surface 3a of object 3. The angle of penetration of the light ray 261 (more specifically, of its principal axis) into sensor 1 is known by calibrating sensor 1. In Figure 3, the light ray 26 follows the path 1b-1a-15-17, or in other words, a path from pattern 1b, reflection at 1a on the first surface 3a, center 15 of sensor 1 and pixel 17 on matrix 16.Establishing the unique relationship involves establishing the relationship between the pixel on sensor 1 and each motif of the second pattern 25 displayed on panel 2. There is a bijective relationship between the position of a given motif on the panel and the pixel on sensor 1. This allows us to know with certainty the starting point of the light beam 26 and the ending point of the light beam 261, and the link between the two points. The same applies to a point 1e. Point 1e is used to reconstruct a point 1d (point 1b is used to reconstruct point 1a).

[0054] Using the unique relationship and the orientation of the light rays detected towards the sensor, the reflection point 1a of each light ray 26 on the first surface 3a of the object 1 is identified. The 3D position of point 1a is calculated using coordinates (e.g., XYZ). Also, using the unique relationship and the orientation of the light rays reflected towards the sensor, the normal 28 (with vector n) to the first surface 3a at the reflection points 1a is identified. This allows us to determine the curvature of the first surface 3a of object 3. Identifying the coordinates of the reflection points 1a and the normals 28 at these points allows us to geometrically characterize the first surface 3a – which in turn allows us to reconstruct object 3. In Figure 3, knowing point 1b, pixel 17, and the orientation of ray 261 allows us to identify the coordinates of point 1a and the curvature of the first surface 3a – to characterize object 3.The identification of the XYZ coordinates of the points and normals of the 3a surface can be achieved using stereovision (two or more cameras viewing the same area of ​​the object) or using a single-camera system (in which case the object is assumed to be locally continuous). For objects with only a reflective property, the characterization (or reconstruction) process considers only the first 3a surface of the object. Even if such an object has thickness, only the first 3a surface is taken into account.

[0055] It is possible to implement a single sensor 1 in the optical system 10. The geometric characterization of the first surface 3a is obtained by detection by sensor 1, which includes the acquisition of at least three light rays at three neighboring pixels of the sensor's pixel array. The same measurement can be performed with four, five, six, or even more rays for increased measurement stability. Alternatively, the optical system 10 comprises several sensors 1, 6, each sensor acquiring a single image simultaneously.

[0056] The process also allows for the characterization of the object for cosmetic (but nonetheless technical) aspects. For example, it can characterize alterations to the object that are not visible to the naked eye. For instance, it can characterize a scratch on a surface that may have been treated to try to remove it. Cosmetic defects, surface quality, paint finish, etc., of the inspected object can alter the quality of the reflected pattern, creating a difference between the pattern displayed on the panel and the pattern measured by the sensor. This difference can be observed and measured to detect defects or provide information on the surface quality of the object. Thus, we can detect scratches, chips, dents; measure the quality of a paint job; and so on. Therefore, the process can be used to inspect car body parts.This application involves examining cars for damage, particularly for the rental market. For example, cars are inspected before and after rental to assess damage caused by a specific driver. If modifications have been made to the car, such as repainting a section to conceal damage or replacing a body panel, the process can detect these changes. More generally, an anomaly on surface 3a can be identified. The process may include measuring the light intensity of panel 2 and measuring the light intensity on sensor 1. The panel intensity can be measured as energy (luminous intensity) or as energy per unit area (luminous density).The light intensity can then be determined by the sensor, in the form of a camera, using two methods: calibration at the intensity level (using a mirror, for example) or by taking into account the characteristics of the detector, the lens, and the camera's exposure time. One measurable parameter is reflectance, defined as the ratio between the measured light intensity of a pixel on the camera and the intensity of the light source. The light intensity is measured in the "non-characteristic area of ​​the pattern," that is, where light can pass freely. The light intensity of the first pattern 5 or the second pattern 25 displayed on panel 2 is measured. The light intensity of the corresponding pattern detected by sensor 1 after reflection on the first surface 3a is also measured.More specifically, the light intensity of the light beam 26 from panel 2 and the intensity of the light beam 261 arriving at sensor 1 are measured. The light intensities are then compared. A variation in light intensity allows for the identification of an anomaly on the first surface 3a. Furthermore, the method also allows for the identification of the anomaly's location.

[0057] Object 3 may also exhibit refractive properties. Object 3 has a thickness within which the light ray can propagate. Object 3 comprises a first surface 3a and a second surface 3b, which define the thickness of object 3. The light ray reaching the first surface is partly reflected by the first surface 3a, partly transmitted through the first surface 3a, and partly absorbed by the object. The portion of the light ray transmitted through the first surface 3a and reaching the second surface 3b is partly reflected by the second surface 3b back towards the first surface 3a, partly transmitted through the second surface 3b, and partly absorbed by the object. The thickness defined between the two surfaces 3a and 3b is a layer, and object 3 may have several superimposed layers.The layers can be adjacent (the second surface of one layer is in contact with the first surface of a neighboring layer). The layers can also be non-adjacent, without contact (the second surface of one layer is separated from the first surface of a neighboring layer). When the layers are non-adjacent, they are separated, for example, by an air film. The process allows for the characterization of the second (or third, fourth, etc.) layer. In this case, the refractive index is calculated and is among the measurements that can be provided. The steps of the process are repeated for each layer.

[0058] The process also allows for the characterization of such an object 3, which also exhibits a refractive property. The steps described above are applied to the second surface 3b. Sensor 1 detects the first model 5 and the second model 25 in reflection on the second surface 3b of object 3, the detection being a single image capture by the sensor. Furthermore, this single capture applies to both surfaces 3a and 3b of the object. The distortion coefficients of the first model 5 detected by sensor 1 relative to the first model displayed on the panel are then determined. The distortion occurs through reflection on the second surface 3b. Using the distortion coefficients, a unique relationship is established between the patterns of the second model 25 detected by sensor 1 after reflection on the second surface 3b and the patterns of the second model 25 displayed on panel 2.This also allows us to characterize the second surface 3b, in the same way as the first surface 3a.

[0059] In the case of the second surface 3b, the light ray 26 from panel 2 is partially reflected at point 1a (to form the light ray 261 described previously) and partially transmitted to point 1a within the thickness of object 3. The transmitted light ray 262, refracted through the thickness, has its orientation altered and is partially reflected at point 1c to form the reflected light ray 263 within the thickness. This ray is then refracted to point 1d to form the light ray 264, whose orientation is also altered. The light ray 264 reaches sensor 1. The detection by sensor 1 includes the acquisition of the light ray 264 in a pixel 18 of the matrix 16 of pixels of sensor 1, the light ray 264 arriving at sensor 1 with a certain orientation after reflection on the second surface 3b of the object 3. The angle of penetration of the light ray 264 (of its principal axis) into sensor 1 is known, by calibration of sensor 1.In Figure 3, the light ray 26 follows the path 1b-1a-1c-1d-15-18, or in other words, the path is from pattern 1b, transmission at 1a (refraction) on the first surface 3a, reflection at 1c on the second surface 3b, transmission at 1d (refraction) on the first surface 3a, center 15 of sensor 1, and pixel 18 on the matrix 16. Establishing the unique relationship involves establishing the relationship between the pixel on sensor 1 and each pattern of the second pattern 25 displayed on panel 2. There is a bijective relationship between the position of a certain pattern on the panel and the pixel on sensor 1. This allows us to know the starting point of the light beam 26 and the ending point of the light beam 264 with certainty, and the connection between the two points.

[0060] Using the unique relationship and the orientation of the light rays detected towards the sensor, a reflection point 1c of the light ray 26 on the second surface 3b of object 1 is identified. The 3D position of point 1c is calculated using coordinates (e.g., XYZ). Also, using the unique relationship and the orientation of the light rays reflected towards the sensor, the normal 281 (with vector n') to the second surface 3b at the reflection point 1c is identified. This allows us to determine the curvature of the second surface 3b of object 3.Identifying the coordinates of reflection points 1c and normals 281 at these points allows us to characterize the second surface 3b in the same way as the first surface 3a - which allows us to reconstruct object 3 on figure 3, knowledge of point 1b, pixel 18 and the orientation of ray 264 allows us to identify the coordinates of point 1c and the curvature of the second surface 3b - to characterize object 3.

[0061] For objects exhibiting reflection and refraction properties, the characterization (or reconstruction) process considers the first surface 3a and the second surface 3b of the object. For such objects, since the process involves capturing a single image containing the patterns of the first model 5 and the second model 25 reflected on the first and second surfaces, the process may include a step for separating the detected information for each surface. To this end, the process for characterizing the first surface 3a and the second surface 3b of the object 3 (for example, to recover the shape and curvature) may include additional steps. These may include a step for estimating the parameters of a ghost image, a step for separating the ghost image, and a step for characterizing the surface. The term "ghost" refers to the reflected signal from the second surface 3b.

[0062] According to Figure 3, where sensor 1 is shown with its matrix 16 and its projection center 15, the step of estimating the parameters of the ghost image consists of identifying the characteristic differences on the matrix 16 of sensor 1 between the primary reflection of pattern 1b on the first surface 3a along rays 26 and 261 and the secondary reflection of pattern 1b on the second surface 3b along rays 26, 262, 263, 264. The primary reflection of pattern 1b at pixel 17 is characterized by the path of rays 26, 261 passing through the points 1b-1a-15-17. The secondary reflection of pattern 1b at pixel 18 is characterized by the deflected path 1b-1a-1c-1d-15-18. The two parameters measured are the pixel offset information between pixel 17 and pixel 18 on matrix 16 of sensor 1 and their relative intensity attenuation.

[0063] As an example, Figure 4 shows a typical detected image 7, acquired by sensor 1. A primary image shows the patterns of the first model 5 and the second model 25 of panel 2 in addition to a secondary image 8 which appears as a "shadow" of the deviated, reflected pattern of the models. In this example, the secondary image corresponds to the type of points such as the point at pixel 18, while the primary image, linked to the sole reflection on surface 3a, is characteristic of the point at pixel 17. For each subset 9 of the image, the step of estimating the parameters of the ghost image determines the pixel shift vector 14 between pixels 17 and 18, as shown in Figure 5, which displays the result 10 of the cross-correlation of subset 9. The data sketched in Figure 5 are plotted by highlighting the contour lines 13 of the cross-correlation of subset 9, the zero-shift peak 11, and the ghost peak 12.The sensor images 7 showing more than one ghost image will then exhibit more relevant peaks in the result 10. For example, if there are two distinct shadows in the image, resulting in two ghost images, cross-correlation of region 9 reveals at least three peaks. The central peak indicates that the two images are perfectly aligned. The second peak corresponds to the overlap of the primary reflection from the first image with the secondary reflection from the second image, while the third peak represents the overlap of the primary reflection from the first image with the tertiary reflection from the second image. Algorithms performing the ghost image estimation step can be implemented using artificial intelligence (AI), digital image correlation models, and similar tools.To simplify, the signal on the primary image is called primary reflection while the secondary image is called secondary reflection.

[0064] The ghost image separation step, in the case of image 7 in Figure 4, consists of separating the primary reflection from the secondary reflection, obtaining two distinct images: one containing only the reflection of the patterns from the first surface 3a, and the second containing only the patterns reflected from the second surface 3b. In other words, the process further includes separating the patterns captured in the single image by reflection from the first surface 3a and the second surface 3b. From this point onward, each separated image can be processed individually.

[0065] The surface characterization step (specifically reconstruction) is applicable to the characterization of surface 3a alone and any additional surfaces. This characterization step can be performed using phase measurement deflectometry (PMD) or similar techniques with one or more sensors (such as cameras) in stereovision. The reconstruction step can be applied to the first image (primary reflection) to recover the shape and curvature maps of the first surface 3a, knowing the resolution of the patterns on panel 2 and the location of point 1b on panel 2. The first and second models, combining unidirectional and irregular patterns, allow for object characterization (or reconstruction) with a single image detected by the sensor (i.e., a single camera shot).However, this process can also be performed using other techniques, such as phase-shift keying (PST) or similar methods. Once the relationship between point 17 and point 1b is established, the reconstruction is carried out using algorithms that implement either mono- or stereo vision of the sensor (e.g., a camera). In the case of mono-vision, the algorithm solves the localization of point 1a by locally fitting an analytical model to the surface and estimating its likelihood using neighboring points on the surface. In the case of stereo-vision, the algorithm solves the localization of point 1a by combining the sensor views.

[0066] Information about the shape and curvature of the second surface 3b is obtained by combining the shape and curvature information of the first surface 3a and processing the secondary image obtained after the ghost image separation step. To characterize the second surface 3b, the process combines the shape information of the first surface 3a, the normals, and the image offset identifying points 1a and 1d. For point 1d, in other words, point 1b on the screen reaches sensor 1 via two different paths: 1b-1a-15-17 and 1b-1a-1c-1d-15-18. Thus, pixel 18 will provide the superimposed information of the first reflection from point 1e and the second reflection from point 1b. Since the images are superimposed, the separation step is performed.

[0067] Images captured by sensors can be processed with AI-based algorithms that separate all internal reflections of the viewed object into ghost images on the sensor array. AI can initiate the process of determining the pixel offset of the ghost image on the acquired sensor array using sub-pixel precision algorithms such as digital image correlation or similar methods. The information from each ghost image can be used to separate the first reflection from the others. At least the external shape and curvature of the object are obtained through monovision or stereovision using the information from the first reflection.

[0068] Knowing the shape and curvature characteristics of the first surface 3a and the second surface 3b allows us to determine other characteristics of the object. For example, we can determine the object's optical power using its refractive index, thickness, and curvature. Knowing the object's complete geometry allows us to calculate its optical properties (such as focal length or the optical power of a lens), either by applying general formulas or using numerical optical simulations. We can also, for example, determine the object's thickness by knowing the 3D position of each point on the first surface 3a and the second surface 3b.

[0069] Furthermore, referring to Figure 3, from the image intensity of the point at pixel 18, it is possible to calculate the reflectance of the first surface 3a and, knowing the power density emitted by the screen 2 displaying the models, the transmittance of the first surface 3a can be obtained. Using similar considerations, it is possible to determine the reflectance and transmittance of the second surface 3b and, consequently, the transmittance of object 3. The reflectance and transmittance can be obtained with polarized or unpolarized light.

[0070] The object 3 to be characterized can be of any size. Object 3 can be, for example, a glazed or opaque surface. It could be a vehicle window (windshield, rear window, side window, or sunroof), a vehicle body panel, a building's glazed surface, a helmet's glazed surface, a pair of eyeglasses, a watch crystal, a diamond, contact lenses, an intraocular lens, or even an eye. The described method allows the object to be characterized. The characteristics include, for example, the shape, dimensions, curvature, thickness, wavefront at any wavelength, optical properties deducible from the wavefront (such as optical power, optical aberrations, etc.), reflectance, and transmittance of each surface of the object.The invention has diverse applications, such as the optical inspection of large curved surfaces like windshields, panoramic or sunroofs, and architectural glass, or for smaller objects like helmet visors, eyeglasses, screens, watch crystals, diamonds, contact lenses, and intraocular lenses (IOLs). The invention extends to medical applications such as the geometric detection of the cornea for glaucoma prevention, by measuring parameters such as central corneal thickness (CCT) and anterior chamber angle (ACA). Furthermore, with a sufficient number of multispectral cameras, it is also possible to measure the shape of implanted intraocular lenses. The method can also be applied to the characterization of a coating on a surface or material of the object.

[0071] Any combination of patterns with a single direction (asymmetric) and irregular patterns can be used. Similarly, any number of separate sensors and / or panels can be implemented for measurement, as shown by the second sensor 6 in Figures 1 and 2 and the second screen 2a in Figure 2. Generally, the patterns 5, 25 can be displayed on a single panel 2 or on a plurality of panels, for example, two panels 2, 2a in Figure 2. The patterns 5, 25 on the panel(s) 2, 2a are detected by sensor 1 or several sensors (for example, two sensors 1, 6) by reflection (or reflection and refraction) on the surface(s) 3, 3a. This increases the accuracy of the characterization of the object 3. The panel(s) and sensor(s) are arranged so that the entire surface of each object reflects both patterns back to the sensor(s).This is particularly useful for a highly curved (or domed) object where a single panel and sensor would not be sufficient to cover the entire surface (or surfaces). Sensor 6 detects a light ray 30 directed from a position 6b on panel 2 corresponding to a pattern of the second model 25 different from point 1b, as shown in Figure 1 or 4, for example. Detection by sensor 6 involves acquiring each light ray 301 as a pixel in the sensor 6's pixel matrix 16. Each light ray 301 reaches sensor 6 in a specific orientation after reflection from the first surface 3a of object 3. The same developments as for rays 26 and 261 apply here. Sensor 6 can also detect the light ray 30 directed from a position 6b on panel 2a corresponding to a pattern of the second model 25 on the other panel, as shown in Figure 2, for example.

[0072] It is also possible to characterize a structure based on the characterization of a plurality of objects. In other words, it is possible to geometrically characterize a larger structure based on the geometric characterization of a plurality of objects that are sub-parts of the structure. That is to say, the structure can be reconstructed by assembling the reconstructions of the objects. To do this, the structure moves in front of the optical system, and the final information about the structure (shape, curvature, thickness, reflectance, and optical power) is obtained by assembling each individual reconstruction of each sensor image corresponding to objects. The displacement and velocity vectors of the structure are obtained from the process of assembling the 3D reconstructed images of the objects. In other words, a plurality of distinct objects of a structure are geometrically characterized and then assembled together.The process is implemented to obtain several unique image captures of the structure, each corresponding to a 3D object forming a sub-part of the structure. Once each 3D object has been geometrically characterized (for example, its shape, dimensions, and curvature), the characterized 3D objects are combined into a single structure, allowing for the geometric characterization of the structure. This makes it possible, for example, to reconstruct all the faces of a vehicle in motion.

[0073] In general, the process allows for a simple and rapid characterization of an object. This makes it possible, for example, to implement the process in a high-speed production line.

[0074] The developments concerning the first surface 3a are applicable to the second surface 3b. In the case of a transparent object with a first surface 3a and a second surface 3b that are very different, it is possible to use S-polarized light and add a P-polarized filter in front of the sensor 1 (such as in a camera). The secondary image can be difficult to separate, and using a filter allows only the reflection from the first surface 3a to be captured.

[0075] In this document, the display, detection, reflection, etc. of a pattern or a model amounts to the same thing - whether it is the model (including the patterns) that is displayed, etc., or the patterns (included in a model) that are displayed, etc.

[0076] It will be obvious to those skilled in the art that the invention is not limited to the embodiments and examples illustrated and / or described above, but that its scope is more broadly defined by the claims introduced below. For example, it should be noted that the optical power and reflectance of the object can be obtained by combining information relating to the external shape and curvature of the object with the pixel offset of each ghost image.

Claims

Demands 1. A method for characterizing an object (3) exhibiting at least one reflective property, comprising - The supply of an optical system (10) comprising • At least one light-diffusing panel (2, 2a) on which are displayed a first model (5) with asymmetrical patterns and a second model (25) with irregular patterns, • At least one sensor (1, 6), - The provision of the object (3) to be characterized, the object comprising a first surface (3a), - The detection by the sensor (1, 6) of the first model (5) and the second model (25) in reflection on the first surface (3a) of the object (3), the detection being a single image capture by the sensor (1, 6), - The determination of deformation coefficients of the first model (5) detected by the sensor (1, 6) with respect to the first model (5) displayed on the panel (2), the deformation taking place by reflection on the first surface (3a), - Using the deformation coefficients, establishing a unique relationship between the patterns of the second model (25) detected by the sensor (1, 6) after reflection on the first surface (3a) and the patterns of the second model (25) displayed on the panel (2).

2. A method according to the preceding claim, in which - Panel (2) emits light rays, - Detection by the sensor (1, 6) includes the acquisition of a light ray in a pixel of a pixel matrix of the sensor (1, 6), each light ray arriving at the sensor (1, 6) in a certain orientation after reflection on the object (3), Establishing the unique relationship includes establishing the relationship between the pixel on the sensor (1, 6) and each pattern of the second pattern (25) displayed on the panel (2).

3. A method according to the preceding claim, further comprising, using the unique relationship and orientation of the light rays reflected towards the sensor (1, 6), - The identification of a point (1a) of reflection of each light ray on the first surface (3a) of the object (3), and - The identification of a normal (28) to the first surface (3a) at the reflection points, and - The geometric characterization of the first surface (3a) using reflection points and normals.

4. Method according to the preceding claim, wherein the optical system (10) comprises a single sensor (1), the geometric characterization of the first surface (3a) is obtained by the detection by the sensor comprising the acquisition of three light rays in three neighboring pixels of the sensor's pixel matrix.

5. Method according to claim 3, wherein the optical system (10) comprises several sensors (1, 6), the geometric characterization of the first surface (3a) is obtained by detection by the sensors comprising the acquisition of each light ray in a pixel of the pixel matrix of the sensors.

6. A method according to any one of the preceding claims, further comprising - The measurement of the luminous intensity of the first model (5) or the second model (25) displayed on the panel (2) and the luminous intensity of the corresponding model detected by the sensor (1, 6) after reflection on the first surface (3a), Comparing light intensities, The identification of an anomaly on the first surface (3a) as a function of a variation in light intensity.

7. A method according to any one of the preceding claims, the method being for the characterization of an object (3) further exhibiting a property in refraction.

8. A method according to the preceding claim, wherein the supplied object (3) comprises a second surface (3b), the method comprising - The detection by the sensor (1, 6) of the first model (5) and the second model (25) in reflection on the second surface (3b) of the object (3), the detection being a single image capture by the sensor (1, 6), - The determination of the deformation coefficients of the first model (5) detected by the sensor (1, 6) with respect to the first model (5) displayed on the panel (2), the deformation taking place by reflection on the second surface (3b) - Using the deformation coefficients, establishing a unique relationship between the patterns of the second model (25) detected by the sensor (1, 6) after reflection on the second surface (3b) and the patterns of the second model (25) displayed on the panel (2).

9. Method according to the preceding claim, further comprising the separation of the patterns captured on the single image by reflection on the first surface (3a) and on the second surface (3b).

10. A method according to any one of the preceding claims, wherein the object (3) has a plurality of superimposed layers in an adjacent or non-adjacent manner, each having a first surface (3a) and optionally a second surface (3b), the steps of the method being repeated for each layer.

11. A method according to any one of the preceding claims, wherein the panel (2) is a backlit screen.

12. A method according to any one of the preceding claims, wherein the models are displayed on a plurality of panels (2, 2a), such that the entire surface of each object (3) reflects both models to the sensor(s) (1, 6).

13. A method according to any one of the preceding claims, wherein the characterization of the object (3) includes the determination of at least one characteristic comprising the shape, dimensions, curvature, thickness, wavefront regardless of wavelength, optical properties deducible from the wavefront, reflectance and transmittance of each surface.

14. Method according to the preceding claim, further comprising determining the optical power of the object (3) using the refractive index of the object (3), its thickness and its curvature.

15. A method according to any one of the preceding claims when they depend on claim 3, wherein a plurality of objects (3) distinct from a structure are geometrically characterized and then assembled together.

16. A method according to any one of the preceding claims, wherein the object (3) is an object (3) from the group comprising a vehicle window, a vehicle body element, a building glass surface, a helmet glass surface, a spectacle glass surface, a watch glass, a diamond, contact lenses, an intraocular lens, an eye.

Images