The present invention relates to a method for detecting a defect on a surface.
It also relates to an associated sensing device.
One current method is to acquire an image or film from the surface. This image or film are then viewed by an operator indicating the defects he identified.
However, the operator may not detect a fault by lack of attention. In addition, the criteria for judging a failure are subjective, that is to say that the definition of what a default is likely to vary according to the operator.

A solution that is often implemented to overcome these problems is viewing the film or image by several operators. In addition, surface image examples are used to show what is a defect or not.

However, the examples shown are neither exhaustive nor reproducible.

In the event of a default, experts are consulted. Many trips between experts are then made, experts do not necessarily agree with each other.

The analysis of a surface is likely to be long and complex.

The invention particularly aims to overcome this disadvantage by providing a simple and reliable method to detect a defect on a surface.

To this end, the invention in particular relates to a fault detection process on a surface, the method comprising the steps of:

- acquiring a plurality of images of the surface by an optical device having an optical axis, each image being acquired with illumination of the surface in a given direction of illumination to each point of the surface and with the optical axis the optical device according to a given optical direction, the images being acquired with different illumination directions or different combinations of illumination directions and / or different optical directions;

- for each point of the surface, computing from the images acquired from a plurality of parameters, the parameters including coefficients of an equation characterizing the response of said point on the surface depending on the direction of illumination and a direction of observation;

- deduction of parameters calculated if the surface has a said point defect.

The detection method according to the invention may further comprise one or more of the following characteristics, taken alone or in all technically possible combinations:

- the optical direction is the same for the plurality of images;

- the optical direction is substantially perpendicular to the surface;

- lighting is achieved by a lighting device, the lighting device comprising a first number of light sources, the first number being greater than six, each light source having a different illumination direction, for each image acquired, a single light source or a combination of defined sources being turned on, the single light source or combination being different for each acquired image;

- the light sources have an identical light intensity;

- the light sources are arranged in a hemisphere surrounding the surface;

- the illumination device comprises one or more light sources being adapted (s) to move into a first number of positions relative to the surface;

- for each point of the surface, calculation of parameters comprises the steps of:

• definition of a quantity changes as a function of the acquired images, · choosing a surface of model-dependent or direction (s) of lighting and / or the viewing direction for the evolution of variable, the model comprising coefficients, and

• calculating the coefficients regression evolution of greatness;

- the quantity corresponding to the gray intensity from the point of the surface, the model depending on the illumination direction;

- a fault is detected at said point when at least one said point parameter is not within a defined interval;

- a fault is detected at said point when at least one said point parameter is not included in a range centered around the mean value of said parameter on the set of points of the surface; and

- a fault is detected at said point when at least one item to said parameter different from a value selected as a detection threshold with respect to the background noise of said parameter at a point adjacent the surface having no defect at the point adjacent.

The invention also relates to a failure detection device on a surface, the sensing device comprising:

- an optical device having an optical axis, the optical device being capable of acquiring an image of the surface in a given optical direction,

- a lighting device having a plurality of different illumination directions, and

- an electronic computing device, the electronic computing device configured to:

• acquiring a plurality of images of the surface by the optical device, each image being acquired according to the optical direction given with different illumination directions or different combinations of illumination directions for each image,

· For each point of the area calculated from the acquired images a plurality of parameters, the parameters including coefficients of an equation characterizing the response of said point on the surface depending on the direction of illumination and direction of observation; and

• deduct the calculated parameters if the surface has a defect in that point.

The invention will be better understood from reading the description which follows, given by way of example and with reference to the attached figures:

- Figure 1 is a schematic view of an exemplary device used in the detection method of the invention,

- Figure 2 is a diagram of steps of an embodiment of the inventive method,

- Figure 3 is a sample image of the surface on which is applied the method of detection,

- Figure 4 is an example of the evolution of three parameters on a line from the surface of Figure 3,

- Figure 5 is an example for detecting defects on the surface of Figure 3, and

- Figure 6 is an example of parameters at a point having a defect and a point having no defect.

An example of a detection device 10 of a defect in a surface 12 is shown in Figure 1.

The detection device 10 comprises an optical device 14, a lighting device 15 comprising a first number of source (s) of light 16 and an electronic computing device 18.

The optical device 14 has an optical axis aligned with a given optical direction O. It has a field of acquisition.

The optical device 14 is adapted to acquire an image of the surface 12 in the optical direction O Data. The acquisition field coincides with the surface 12 to be inspected, so that the captured image represents the surface 12.

For each point of the surface, defining a viewing direction B, B 'between the optical device 14 and the point on the surface 12.

The acquisition field is fixed here.

The optical direction O is typically substantially perpendicular to the surface 12. Alternatively, the optical direction O is not perpendicular to the surface 12.

The optical device 14 is, for example, a camera.

Alternatively, the acquisition field coincides only with a portion of the surface 12 and is adapted to move relative to the surface 12 so as to acquire all of the surface 12 into several parts.

Each acquired image comprises a matrix of pixels of size i * j, i with the number of horizontal lines and j the number of vertical lines of the matrix, with at least one of the two parameters (i, j) greater than 1 . The pixel is, for example, a size of rectangle denoted h * L, with its height h L and its width. The height h is equal to the width L. In a variant, the height is different from the width

subsequently called interchangeably with "surface point," one pixel of an acquired image or a portion of the surface acquired in a pixel.

The light sources 16 are able to illuminate the surface 12.

For each point of the surface, the light sources 16 each have a lighting direction E, E 'given connecting the light source to the point of the surface 12. The illumination direction is different for each light source.

The light sources 16 are typically arranged in a semi-sphere 20 surrounding the surface 12.

The first number is, for example, more than six, more preferably greater than twenty. The first number is for example between 32 and 128, more particularly equal to 96.

The light sources 16 are, for example, light emitting diodes

(OF THE).

They can be switched off or on. They are configured to be lit independently of each other.

The light sources 16 typically have the same light intensity.

Alternatively, the illumination device 15 comprises one or more light sources being adapted (s) to move, for example on the half-sphere 20, in a first number of positions relative to the surface.

Alternatively, the illumination device 15 comprises one or more optionally non-identical lighting. When the lighting device is fully on, it performs, for example, a non-uniform illumination on the acquired surface. The acquisition system and the surface have relative movement.

The electronic computing device 18 is, for example, a computer, a computer, a computing unit, at least one programmable logic device such as FPGA (English Field-Programmable Gate Array) or at least an integrated circuit dedicated as ASICs (English application-specific integrated circuit).

The electronic computing device 18 is, for example, connected to the optical device 14 and the lighting device 15.

The electronic computing device 18 is adapted to turn on or off each of the light sources 16 independently of each other.

The electronic computing device 18 is further arranged to trigger the acquisition of at least one image of the surface by the optical device 14.

It is thus configured to drive the acquisition of a plurality of images of the surface by the optical device 14, each image being acquired according to the optical direction given O and with a single light source or a defined combination of unique sources lit different for each image.

For each point on the surface, the electronic computing device 18 is configured to calculate from the images acquired a plurality of parameters, the parameters including coefficients of an equation characterizing the response of said point on the surface depending on the direction of illumination and an observation direction and then to deduct the calculated parameters if the surface has a said point defect.

This last point is detailed below in connection with a fault detection method.

A fault detection method in a surface will now be described with reference to Figure 2.

The method here is implemented by the device described previously, in particular thanks to the electronic computing device 18.

The method comprises the steps of:

- acquisition 100,

- calculation parameters 102, and

- deduction of a fault 104.

During acquisition 100, a plurality of images 200 of the surface is acquired by the optical device 14. Such an image is shown in Figure 3.

Each image 200 is acquired with illumination of the surface 12 in a given direction of illumination E, E 'for each point of the surface and with the optical axis of the optical device 14 according to a given optical direction O.

The images 200 are acquired with different illumination directions E, E 'or combinations of illumination directions.

Alternatively, the images 200 are acquired with different optical direction O.

Alternatively, the images 200 are acquired with different illumination directions E, E 'or combinations of illumination directions and with different optical direction O.

In the embodiment shown, the lighting directions E, E 'are different for each acquired image and the optical direction O is invariable and substantially perpendicular to the surface 12.

For each acquired image, a single light source or a defined combination of single sources 16 of the lighting device is turned on, the single light source or the combination being different for each acquired image.

All light sources or combinations is alternately turned on, so acquiring an image by light source or by combination. Alternatively, some sources or combinations are not lighted.

Then, during the calculation parameters 102, 202 for each point of the surface, that is to say, here for each pixel of the acquired images, a plurality of parameters are calculated from the acquired images.

The parameters are calculated from the coefficients of an equation characterizing the response of said point on the surface depending on the lighting direction E, E 'and a viewing direction B, B'.

For each point 202 of the surface 200, the calculation parameters 102 here comprises the steps of:

- definition of a quantity 106,

- choice of model 108, and

- computing weightings 1. 10

When defining a size 106, a size changing according acquired images is defined.

For example, the size here is the gray intensity from the point of the surface, the gray intensity being variable according to the acquired images.

In parallel a model representing the evolution of the magnitude is selected 108. The model depends on the one or more direction (s) of light E, E 'and / or the viewing direction B, B'. The model includes factors.

Here the model depends only on the direction of illumination. It has, for example, the following form: G (S) = a 0 x D u 2 + a t x D v 2 + a 2 x D u x D v + a 3 x D u + a 4 x D v + is 5 , with the magnitude G, E the illumination direction, a 0 , a 1; a 2 , a 3 , a 4 and a 5 coefficients and D u and D v being the coordinates of the vector connecting the light source to the point of the surface.

The coefficients a 0 , a 1; a 2 , a 3 , a 4 and 5 are here undetermined.

Then, when calculating the coefficients 1 1 0, an application of size on the template is carried out so as to calculate the values of a 0 , a 1; a 2 , a 3 , a 4 and a 5 .

The settings here are equal to the coefficients a 0 , a 1; a 2 , a 3 , a 4 and 5 thus calculated. For each point of the surface, the same size and the same model are defined. Thus, for each point of the surface, parameters obtained has 0 , ai, a 2 , a 3 , a 4 and a 5 . Is obtained, for example, a surface mapping representing parameter values for each point of the surface.

In Figure 4 is shown the evolution of three parameters along the line 204 surface the points shown in Figure 3.

Alternatively, other parameters are values ​​calculated from the coefficients.

When inferring the presence of defect 104, it is deducted from the calculated parameters if the surface has a said point defect.

In Figure 6, for example, the parameters are represented with the curve C at a point having no defect and the curve C 2 at a point having a defect.

The result obtained in step 1 04 for each point is, e.g., a Boolean binary representing if a fault is detected at said point.

In Figure 5 is shown the boolean along line 204: If the boolean is 0, no fault is detected; If the boolean is equal to 1, a fault is detected.

Thus, in Figure 5, a single fault is detected 300 and extends over several adjacent points.

One or more criteria are used to deduce the presence of a defect from the parameters. These criteria are described below and are used alone or in combination to detect a fault. For example, a fault is detected when two criteria are validated.

A first criterion is that a fault is detected at said point when at least one said point parameter is not within a range.

More particularly the range is centered around the average value of said parameter on the set of points of the surface.

Alternatively, the interval is a specified interval.

Alternatively, a fault is detected at said point when at least a given number of points to said parameters are not included in a respective interval.

A second criterion is that a fault is detected at said point when at least one item to said parameter different from a value selected as a detection threshold with respect to the background noise of said parameter at a point adjacent the surface having no defect at the point adjacent.

Thus, step by step, the surface is analyzed.

It is then necessary to define at least a first point on the surface to which a defect is not present.

Alternatively, the electronic computing device 18 considers that a location of the surface is defect-free and, step by step, detects defects in the surface. If the result is considered unacceptable, for example, if a majority of the surface is detected as a defect, then the step is iterated by considering another location, different from above, is free from defects.

A third criterion is based on the geometry of a signal.

The signal here is the evolution of one or more parameters on a set of defined points. The points of the set are adjacent here. The set of points is, for example, an area centered about a central point.

Are defined patterns corresponding to cases where a fault is detected and / or in which no fault is detected. The shape of the signal is compared to the grounds, so decide if a fault is detected. The fault is detected at the central point, in an area including the center point and included in the set of points or all points.

Thus, for each point on the surface, it is deducted parameters if a defect is present or not.

This is, e.g., representable on a binary mapping of the surface to easily visualize the defects.

Alternatively, the light sources 16 does not have the same light intensity. A calibration of the device is then carried out.

In another embodiment, the optical direction of the optical device varies, the surface and remaining fixed illumination source between each acquired image.

Alternatively, the optical direction and the illumination direction are movable relative to the surface.

Alternatively, the calculation of the parameters 102 based on a different mathematical model such as the distribution function of the bidirectional reflectance (or BRDF) or discrete modal decomposition (or DMD) or the polynomial texture mapping (or TMP).

The direct deduction from parameters calculated from images acquired notably allows to overcome the susceptibility to defects in the presence different for different operators. The method thus provides a unique result and objective. Thus, the process is simplified and safer.

CLAIMS
1. - A method for detecting a defect on a surface (12), the method comprising the steps of:

- acquiring (100) a plurality of images (200) of the surface (12) by an optical device (14) having an optical axis, each image (200) being acquired with illumination of the surface in a direction lighting (E, E ') provided for each point of the surface (12) and with the optical axis of the optical device according to an optical direction (Y) data, images (200) being acquired with different illumination directions ( E, E ') or different combinations of illumination directions and / or different optical directions (O);

- for each point (202) of the surface (200), calculation (102) from the images (200) acquired from a plurality of parameters, the parameters including coefficients of an equation characterizing the response of said point (202) of the surface depending on the direction of illumination (E, E ') and an observation direction (B, B');

- deducting (104) the parameters calculated if the surface (12) has a defect at said point (202).

2. - A method of detection according to claim 1, characterized in that the optical direction (O) is the same for the plurality of images (200).

3. - A method of detecting according to claim 2, characterized in that the optical direction (Y) is substantially perpendicular to the surface (12).

4. The detection method according to any one of the preceding claims, characterized in that the illumination is realized by an illumination device (15), the illumination device (15) comprising a first number of light sources (16), the first number being greater than six, each light source (16) having a different direction of illumination (E, E '), for each acquired image, a single light source or combination of sources defined ( 16) being turned on, the single light source or combination (16) being different for each image (200) acquired.

5.- A method of detecting according to claim 4, characterized in that the light sources (16) have an identical light intensity.

6. The detection method according to claim 4 or 5, characterized in that the light sources (16) are arranged in a semi-sphere (20) surrounding the surface (12).

7. The detection method according to any one of claims 4 to 6, characterized in that the illumination device (15) comprises one or more light sources being adapted (s) to move in a first number of positions relative to the surface.

8. A detection method according to any one of the preceding claims, characterized in that, for each point of the surface, calculation of parameters comprises the steps of:

- definition of a quantity (106) evolve depending on the acquired images,

- choice of a model of the surface (108) depending from the or direction (s) of light (E, E ') and / or the observation direction (B, B') for the evolution of size, the model comprising coefficients, and

- calculation of the coefficients (1 10) by regression of the changing magnitude.

9. - A method of detection according to claim 8, characterized in that the magnitude corresponding to the gray intensity from the point (202) of the surface (12), the model dependent on the direction of illumination (E, E ' ).

10. - A method of detection according to any one of the preceding claims, characterized in that a fault is detected at said point (202) when at least one said point parameter is not within a specified interval.

1 1 - The detection method according to any one of the preceding claims, characterized in that a fault is detected at said point (202) when at least one said point parameter is not included in a range centered around the value average of said parameter of the set of points of the surface (12).

12. The detection method according to any one of the preceding claims, characterized in that a fault is detected at said point (202) when at least one parameter at said point differs from a value selected as a detection threshold in relation to background noise of said parameter at a point adjacent the surface (12) having no defect at the point adjacent.

13. A sensing device (10) of a defect on a surface (12), the detection device (10) comprising:

- an optical device (14) having an optical axis, the optical device (14) being adapted to acquire an image (200) of the surface (12) in an optical direction (Y) given,

- a lighting device (15) having a plurality of different illumination directions (E,

E "), and

- an electronic calculation means (18), the electronic computing device configured to:

• acquiring a plurality of images (200) of the surface (12) by the optical device (14), each image (200) is acquired according to the optical direction (O) Data with different illumination directions (E, E ' ) or different combinations of illumination directions for each image (200),

• for each point (202) of the surface (12), calculated from the images (200) acquired a plurality of parameters, the parameters including coefficients of an equation characterizing the response of said point (202) of the surface (12 ) depending on the direction of illumination (E, E ') and an observation direction (B, B');

• deduct the calculated parameters if the surface (12) has a defect at said point.