Method for detecting optical defects in a windshield glass

Abstract

A method for detecting optical defects within a windshield. The method takes as input a digital image map of optical power intensity of the windshield and provides as output a digital image map of optical defects. The method comprises the following steps: (a) image processing of said digital image map of optical power for detecting and delimiting zones differing in optical power intensity; (b) computing for each detected zone representative values of representative geometric distance and optical power; (c) computing an image map of the detected zones for which the product between the representative values of representative geometric distance and optical power is equal to or greater than 2.9.10 ‑4 ​

Description

TECHNICAL FIELD

[0001] The present invention relates to a computer-implemented method for detecting optical defects within a windscreen. BACKGROUND

[0002] Windshields are well known in the transportation industry, for example in automotive, rail transportation and aviation. They are usually made of two curved glass sheets laminated with a polymeric interlayer.

[0003] A windshield is a glass window through which a driver looks at things in front (e.g. road, track, landscape...). Therefore, for safety reasons, the distortion of the objects seen through the windshield should be as low as possible, at least it should not bother the driver. In this context, the optical quality of the windshield must meet specific requirements detailed in section 9.2 of annex 3 of regulation No. 43 of the United Nations Economic Commission for Europe (UN / ECE).

[0004] Several methods and instruments are described in the prior art, which can aim at measuring the optical distortion of a windshield under the framework of regulation No. 43.

[0005] EP0463940A1 describes a method for measuring the optical quality of a windshield based on shadow lighting.

[0006] WO 9817993A2, GB2152210A and EP1061357A2 describe methods to measure the optical distortion of a windshield by image analysis of a transmitted or reflected pattern.

[0007] WO 2017 / 008159A1 describes a method to detect optical defects in a windshield by analysis of a synthetic image of chromatic aberrations. SUMMARY

[0008] TECHNICAL PROBLEM

[0009] While they can be efficient, the described prior art methods can fail to detect some optical defects, however these optical defects can be detected by the human eye. Since human vision is generally more flexible than most inspection systems implementing the described methods, sometimes optical defects that can only be visible on the driver and / or passenger side of the windshield at a specific angle or different angles can be completely overlooked by the inspection systems on the manufacturing line. The immediate and negative consequence is that a windshield initially considered to meet the technical specifications can be rejected by the customer later. Complaints and production losses can occur.

[0010] Moreover, it has been found that some of these overlooked or undetected optical defects can not have the same features or signatures, so that the inspection systems can behave differently from each other in their detection of them. With some specific settings, certain inspection systems can be able to detect them, while other systems can fail. It is thus challenging to find settings and / or criteria to make the inspection systems detect them regardless of the characteristics of the inspection systems. Further still, even if such settings or criteria can be found, it still requires to remain compatible with the whole configuration of the manufacturing line.

[0011] There is thus a need for an innovative, easy to implement method for detecting those specific optical defects that can not be detected by current inspection systems but that can still be visible to a human driver.

[0012] Solution to the technical problem

[0013] In a first aspect of the application, a computer-implemented method for detecting optical defects within a windshield is provided, other advantageous embodiments being proposed.

[0014] In a second aspect of the application, a data processing system, a computer program and a computer readable medium to implement the method are provided.

[0015] In a third aspect of the application, a process for detecting optical defects within a windshield is provided, aspects of the application being advantageous embodiments.

[0016] Both the method and the process can be used in a manufacturing process of a windshield.

[0017] Advantages of the application

[0018] A first outstanding benefit of the application is that it allows to detect optical defects in a windshield that can still not be detected by most common inspection systems but that can still be seen by a human eye, for example the eyes of a driver.

[0019] A second advantage is that the application is relatively easy to implement in an existing manufacturing process, so that it requires little adaptation, if any. More precisely, the computer-implemented application and the process according to the application can benefit from equipment already available in the manufacturing line and / or in the control quality process for acquiring digital image maps of the optical power.

[0020] A third advantage is that the application indirectly provides an insight into the relative size at which the optical defect can appear to the driver when looking through the windshield at a distant object. It then allows to assess whether the optical defect can impact the ease of viewing. BRIEF DESCRIPTION OF DRAWINGS

[0021] Figure 1 is a data flow diagram of a computer-implemented method for detecting optical defects according to embodiments of the first aspect of the invention.

[0022] Figure 2 is an example of a gray-scale digital image map of optical power strength of a windshield.

[0023] Figure 3 is a data processing system according to the second aspect of the invention. DETAILED DESCRIPTION

[0024] With reference to Figure 1 In embodiments of the first aspect of the invention, a computer-implemented method 1000 for detecting optical defects within a windshield is provided, wherein the method takes as input I1001 a digital image map of optical power strength of the windshield and provides as output O3001 a digital image map of optical defects, wherein the method 1000 comprises the following steps:

[0025] (a) image processing 1001 of the digital image map of optical power for detecting and delimiting zones differing in optical power strength;

[0026] (b) for each detected zone, computing 1002 representative values representative of geometric distance and optical power;

[0027] (c) computing 1003 an image map of detected zones for which the product between the representative values representative of geometric distance and optical power is equal to or greater than 2.9 x 10 -4 .

[0028] The method 1000 takes as input I1001 a digital image map of optical power strength of the windshield, Figure 2 Examples thereof are provided in the description. Optical power refers to the ability of a lens to focus light. Depending on how a lens refracts light, the light can diverge or converge. The SI unit for optical power is the meter -1 ), also known as diopters (dpt).

[0029] In Figure 2 In the example of Fig. 1, the optical power strength is represented by a gray-scale pattern over the entire windshield, the white framed zone being discussed later.

[0030] The measurement of the optical power in a windscreen is well described in the prior art (e.g. EP3012619A1, EP0463940A1, EP0342127A1, EP0685733A1). Alternatively or in addition to the measurement, the optical power can also be simulated, e.g. EP3756114A1. All these methods can directly provide or can be adapted to obtain a digital map of the optical power of the windscreen.

[0031] In some embodiments, the digital map can be obtained at a specific angle or for a given angular range with respect to the normal of the windscreen. In preferred embodiments, the digital map can be obtained at or represent an angle or angular range corresponding to the angle at which the driver can look through the windscreen when in use to view an object. Indeed, as the angle of inclination of the windscreen with respect to the frame of the vehicle typically varies from one vehicle to another, the viewing angle of the driver and thus the likelihood of viewing an optical defect can also vary depending on this angle of inclination. It can thus be advantageous to take this effect into account when obtaining the digital map of the optical power. The accuracy and reliability of the method can be improved.

[0032] The image processing 1001 of step (a) to detect and delimit zones of the digital map of the optical power that differ in optical power intensity can be any adapted image processing method for object detection. The image processing method can be a neural network or a non-neural network method.

[0033] In preferred embodiments, the image processing 1001 can be blob detection, in particular by computing the Gaussian Laplacian of the digital map of the optical power, the Gaussian difference of the digital map of the optical power, or the Hessian determinant of the digital map of the optical power. Blob detection can be simpler to implement compared to neural network methods, while still providing valuable results for most kinds of windscreen and applications.

[0034] In step (b), the representative geometric dimensions of the detected zones can depend on the shape chosen at step (a) to represent their boundaries. However, as a general rule of thumb, in order to be representative, the value of the calculation of the geometric dimensions should advantageously show little variation regardless of the chosen shape, as long as the shape is relevant to delimiting the detected zone.

[0035] Many shapes can be suitable and can be more or less complex, depending on the image processing method used to detect and delimit the zones and on the degree of proximity and accuracy that can be sought for the boundaries of the zones. For example, they can be convex shapes, such as convex polygons, e.g. squares or rectangles, or curved figures, e.g. circles or ellipses, or concave shapes, such as concave polygons or concave curved figures.

[0036] Since optical defects within the scope of the present application can often have regular, relatively circular or elongated convex shapes, in certain advantageous embodiments the detected areas at step (a) can be delimited with an ellipse, and wherein the representative geometric distance of the detected areas calculated at step (b) is the minor axis of said ellipse.

[0037] The calculated representative geometric dimension of each detected area at step (b) has a length unit, for example in meters according to SI units, and the calculated representative value of the optical power has an inverse length unit, for example in m -1 or dpt.

[0038] The representative value of the optical power of each detected area can be calculated by different methods. In some embodiments, the representative value of the optical power of each detected area is the average optical power, the median optical power, the maximum optical power or the maximum / minimum optical power difference in said detected area.

[0039] The product value calculated at step (c) is a dimensionless number. In certain embodiments, this number can be considered as a representative of an angle which can be expressed in radians (rad). In this respect, the minimum value of 2.9 x 10 -4 provided at step (c) can then be considered as a representative of the minimum distortion factor under which the optical defect can appear to the human eye, i.e. the driver, when looking through the windscreen at an object located at a given distance D from the windscreen. The product of this minimum value expressed in radians and the distance D can provide the apparent size of the optical defect with respect to the distance D at which the driver can look at the object. -4

[0040] At step (c), for the detected areas for which the product between the representative geometric distance and the representative value of the optical power is equal to or greater than 2.9 x 10 -4 , the image mapping of the detected areas is calculated. The minimum value of 2.9 x 10 -4 may then be considered as a low boundary value for the semi-infinite interval. Since any optical defect can in practice have any size greater than the size which can be perceived by the human eye, any attempt to define a high boundary value can be considered as purely artificial and arbitrary. In practice, the high boundary value can be defined depending on the requirements which are desirable for greater optical distortions.

[0041] In some embodiments, the product value calculated at step (c) can be comprised between 5 x 10 -4 and 2 x 10 -3 ; preferably between 7 x 10 -4 and 1.5 x 10 -3 ; more preferably it can be 1 x 10​-3 The interval of these values is greater than 2.9xl0 -4 The interval of these values is greater than 2.9xl0 -4 They can therefore appear less stringent than the minimum value of 2.9xl0

[0042] The method provides as output O1001 an image map of the zones for which the product between the representative value representative of the geometric distance and the optical power is equal to or greater than 2.9xl0 -4 An example of such an image map is provided in Figure 2 , in which the white boxed zones represent the detected zones and are overlaid to the digital map of optical power.

[0043] In certain embodiments, the detected zones at step (b) can further be such that their apparent size in the visual decimal acuity scale is between 0.5 and 3; preferably between 0.67 and 1.25, more preferably 1. Visual acuity in the human visual range is well known in the art and is fully described in the EN ISO 8596:2018 standard. In some applications where a high degree of optical quality is required (i.e. aviation, racing cars, high-end vehicles), a criterion on the apparent size can be an advantageous complement. Such a criterion can ensure that optical defects for which the human eye can be sensitive can be detected with a high degree of reliability.

[0044] The method 1000 according to the first aspect of the application can advantageously be used in the manufacturing process of a windshield. Since the manufacturing process can already comprise the instrument for acquiring the digital map of optical power, little, if any, adaptation of the method can be required to implement the method.

[0045] In a second aspect of the application, with reference to Figure 3 , there is provided a data processing system 3000 comprising means for performing the method 1000 according to any one of the embodiments of the first aspect of the application; and a computer program I3001 comprising instructions which, when executed by a computer, cause the computer to perform the method according to any one of the embodiments of the first aspect of the application.

[0046] The data processing system 3000 comprises means 3001 for performing the method according to any embodiment of the first and second aspects of the application. An example of means 3001 for performing the method can be a device capable of being instructed to automatically perform a sequence of arithmetic or logical operations to perform a task or action. Such a device, also referred to as a computer, can comprise one or more central processing units (CPUs) and at least one controller device adapted to perform those operations. It can further comprise other electronic components, like input / output interfaces 3002, non-volatile or volatile storage devices 3003, and a bus as a communication system for data transfer between components within the computer or between computers. One of the input / output devices can be a user interface for human-machine interaction, e.g. a graphical user interface to display human understandable information.

[0047] As the computations can require a large amount of computing power to process a considerable amount of data, the data processing system can advantageously comprise one or more graphic processing units (GPUs), whose parallel structure makes them more efficient than CPUs, especially for image processing in ray tracing.

[0048] As to the computer program I 3001, the steps of the method of the application can be implemented using any kind of programming language (compiled or interpreted). The computer program I 3001 can be part of a software solution, i.e. a collection of executable instructions, code, scripts, etc. and / or a database.

[0049] In certain embodiments, a computer readable storage or medium 3003 comprising instructions which, when executed by a computer, cause the computer to perform the method according to any embodiment of the first aspect of the application is also provided.

[0050] The computer readable storage 3003 can preferably be a non-volatile storage or memory, e.g. a hard disk drive or a solid state drive. The computer readable storage can be a removable storage medium or a non-removable storage medium as part of the computer.

[0051] Alternatively, the computer readable storage can be a volatile memory within a removable medium.

[0052] The computer readable storage 3003 can be part of a computer used as a server, the executable instructions can be downloaded from the server and, when they are executed by the computer, cause the computer to perform the method according to any embodiment described herein.

[0053] Alternatively, the program I3001 can be implemented in a distributed computing environment, for example in cloud computing. The instructions can be executed on a server, a client computer can be connected to the server and provide the encoded data as input to the method. Once the data is processed, the output can be downloaded and decoded on the client computer or sent directly, for example as instructions. Such an implementation can be advantageous as it can be implemented in a distributed computing environment such as a cloud computing solution.

[0054] In embodiments of the third aspect of the application, a process for detecting optical defects within a windshield is provided, wherein the method comprises the steps of:

[0055] (a) acquiring a digital image map of the optical power intensity of the windshield;

[0056] (b) processing the digital image map with a computing system for detecting and delimiting zones differing in optical power intensity;

[0057] (c) calculating with the computing system an image map of the detected zones for which the product between a representative value of the geometric distance and a representative value of the optical power is equal to or greater than 2.9 x 10 -4

[0058] The technical aspects and features of the different embodiments detailed in the context of the first and second aspects of the application can also be applied to the third aspect of the application. It is within the reach of the skilled person to change, transform or adapt them in the process according to the third aspect of the application.​

Claims

1. A computer-implemented method for detecting optical defects within a windscreen, wherein the method takes as input a digital image map of the intensity of the optical power of the windscreen and provides as output a digital image map of the optical defects, wherein the method comprises the following steps: (a) image processing of the digital image map of the optical power for detecting and delimiting zones differing in intensity of optical power; (b) computing for each detected zone a representative value of the representative geometrical dimension and of the optical power; (c) calculating a product between a representative value representative of the geometrical size and of the optical power for which the product is equal to or greater than 2.9 x 10 -4 the image map of the detected zones, wherein the optical power refers to the windscreen's ability to focus light, and wherein the representative geometrical dimension is a specific length determined depending on the shape of the boundary of the detected zone.

2. The method of claim 1, wherein the product in step (c) is between 5 x 10 -4 and 2 x 10 -3 .

3. The method of claim 1, wherein the product in step (c) is between 7 x 10 -4 and 1.5 x 10 -3 .

4. The method of claim 1, wherein the product in step (c) is 1 x 10 -3 .

5. The method according to any one of claims 1 to 4, wherein the image processing of step (a) is blob detection.

6. The method according to claim 5, wherein the blob detection is by computing a blob detection of the digital image map of the optical power as a Gaussian Laplacian, a Gaussian difference of the digital image map of the optical power, or a Hessian determinant of the digital image map of the optical power.

7. The method according to any one of claims 1 to 4, wherein at step (b) the representative value of the representative geometrical dimension and of the optical power is computed for detected zones such that their apparent size in the visual decimal acuity scale is between 0.5 and 3.

8. The method according to any one of claims 1 to 4, wherein at step (b) the representative value of the representative geometrical dimension and of the optical power is computed for detected zones such that their apparent size in the visual decimal acuity scale is between 0.67 and 1.

25.

9. The method according to any one of claims 1 to 4, wherein at step (b) the representative value of the representative geometrical dimension and of the optical power is computed for detected zones such that their apparent size in the visual decimal acuity scale is 1.

10. The method according to any one of claims 1 to 4, wherein the detected zones at step (a) are delimited with an ellipse, and wherein the representative geometrical dimension of the detected zones computed at step (b) is the minor axis of the ellipse.

11. The method according to any one of claims 1 to 4, wherein the representative value of the optical power of each detected zone is the mean optical power, the median optical power, the maximum optical power or the maximum / minimum optical power difference in the detected zone.

12. Use of the method according to any one of claims 1 to 11 in a manufacturing process of a windscreen.

13. A data processing system comprising means for performing the method according to any one of claims 1 to 11.

14. A computer program product comprising instructions which, when the program of the computer program product is executed by a computer, cause the computer to carry out the method according to any one of claims 1 to 11.

15. A computer-readable medium comprising instructions which, when executed by a computer, cause the computer to carry out the method according to any one of claims 1 to 11.

16. A method for detecting optical defects within a windshield, wherein the method comprises the steps of: (a) acquiring a digital map of the intensity of the optical power of the windshield; (b) processing the digital map with a computing system for detecting and delimiting zones differing in intensity of optical power; (c) using the computing system to calculate a product between a representative value for the geometric dimension and a representative value for the optical power for which the product is equal to or greater than 2.9 x 10 -4 of the detected regions, wherein optical power refers to the ability of the windshield to focus light, and wherein a representative geometric dimension is a specific length determined depending on the shape of the boundary of the detected zone.

17. The method of claim 16, wherein the product in step (c) is between 5 x 10 -4 and 2 x 10 -3 .

18. The method of claim 16, wherein the product in step (c) is between 7 x 10 -4 and 1.5 x 10 -3 .

19. The method of claim 16, wherein the product in step (c) is 1 x 10 -3 .

Images