Method of checking the appearance of the surface of a tyre

• US 9,189,841 B2
• Filed: 09/27/2012
• Issued: 11/17/2015
• Est. Priority Date: 09/30/2011
• Status: Active Grant

• Current Assignee: Compagnie Generale DES Etablissements Michelin SCA
• Original Assignee: Compagnie Generale DES Etablissements Michelin SCA, Michelin Recherche et Technique SA (Compagnie Generale DES Etablissements Michelin SCA)
• Inventors: Noyel, Guillaume, Jeulin, Dominique, Parra-Denis, Estelle, Bilodeau, Michel
• Primary Examiner(s): Shah, Utpal
• Assistant Examiner(s): Duffy, Kate R
• Application Number:
  US14/348,156
• Publication Number:
  US 20140233841A1
• Time in Patent Office: 1,146 Days
• US Class Current: 1/1
• CPC Class Codes: G06F 18/2132 based on discrimination criteria, e.g. discriminant analysis
• CPC Class Codes: G06F 18/2135 based on approximation criteria, e.g. principal component analysis
• CPC Class Codes: G06F 18/2411 based on the proximity to a decision surface, e.g. support vector machines
• CPC Class Codes: G06T 2207/20021 Dividing image into blocks, subimages or windows
• CPC Class Codes: G06T 7/0004 Industrial image inspection
• CPC Class Codes: G06T 7/001 using an image reference approach
• CPC Class Codes: G06V 10/764 using classification, e.g. of video objects
• CPC Class Codes: G06V 10/7715 Feature extraction, e.g. by transforming the feature space, e.g. multi-dimensional scaling [MDS]; Mappings, e.g. subspace methods
• CPC Class Codes: G06V 2201/06 Recognition of objects for industrial automation

Abstract

A tire appearance detection method includes:

• • capturing an original grey-level image of an inner surface of a tire, and transforming the original image into an orthonormal space with an x-axis (OX) representing a circumferential direction and with a y-axis (OY) representing a radial direction;
  • applying a series of filters to the original image, to obtain a multivariate image;
  • splitting the multivariate image according to a predefined tiling in axial and circumferential directions, to obtain multivariate sub-images of the inner surface of the tire;
  • transforming each of the multivariate sub-images into one-dimensional vectors using selected descriptors, to obtain a simplified multivariate image of the inner surface of the tire;
  • transforming the simplified multivariate image into a common reduced factorial space; and
  • locating sub-images of the inner surface of the tire containing an anomaly using a classifier suitable for identifying an area of the common reduced factorial space containing an anomaly.

Claims

1. A method of detecting an anomaly on an internal surface of a tyre through digital processing of an image of the internal surface of the tyre, the method comprising steps of:

capturing a grey-level original image (f) of the internal surface of the tyre and transforming the original grey-level image into an orthonormal space, in which an x-axis (OX) represents a circumferential direction, and in which a y-axis (OY) represents a radial direction;

transforming the original grey-level image using a predetermined series of filters (F₁^f, F₂^f, . . . F_L^f), to obtain a multivariate original image;

splitting the multivariate original image into multivariate sub-images (W_F^k), to obtain predefined tiling in the axial direction and the circumferential direction;

transforming each of the multivariate sub-images is transformed into one-dimensional vectors using a given descriptor (H, G, . . . N), to obtain a simplified multivariate image (K^WF, G^WF . . . , H^WF) formed from a collation of the vectors;

reducing a number of dimensions of the simplified multivariate image data using a data analysis tool, to obtain a reduced factorial space; and

,

in the reduced factorial space, evaluating a distribution of sub-images containing an anomaly in relation to a distribution of sub-images containing no anomaly, in order to assess a separating power of the given descriptor.

2. The method according to claim 1,

wherein the original grey-level image is captured using a sensor capable of evaluating a relief of the surface, and

wherein the method further comprises a step of, for each pixel of the original grey-scale image, assigning a grey-level value proportional to a topographical elevation of a point corresponding to the pixel, to obtain a two-dimensional original image.

3. The method according claim 1, further comprising a step of, after the step of capturing the grey-level image, carrying out a pre-processing of the grey-level image using a polynomial interpolation of grey levels by line or by column, in order to lessen effects of lighting variations during capture of the grey-level image.

4. The method according to claim 1, wherein the series of filters is formed from morphological operators corresponding to swelling, wear, opening, and closure operations, or from residuals of one or more of the operations.

5. The method according to claim 1, wherein the tiling is performed in the circumferential direction and the radial direction in such a way that the multivariate sub-images have a shape of a square.

6. The method according to claim 5, wherein a surface of the multivariate sub-images is greater, by a factor of between five and ten times, than a Representative Elementary Surface (RES) of a textured base of a tyre having no anomalies.

7. The method according to claim 6, wherein sides of a square of a multivariate sub-image are formed by a number of pixels corresponding to a length of between 0.01*Φ

and 0.1*Φ

, with Φ

corresponding to a diameter at a seat of the tyre.

8. The method according to claim 1, wherein the data analysis tool is chosen from a set of factorial analysis tools that include:

a principal component factorial analysis (PCA) tool,

a correspondence factorial analysis (CFA) tool, and

a independent component analysis (ICA) tool.

9. The method according to claim 1, wherein a plurality of relevant descriptors is selected by performing the method multiple times utilizing a descriptor chosen from a set of descriptors that includes:

a histogram (H^WF(x_k)) of grey levels,

a covariance

10. The method according to claim 9, wherein the reduced factorial spaces of the relevant descriptors of which the separating power has been evaluated are combined into a common reduced factorial space (c^HWF, c^GWF, . . . , c^NWF).

11. The method according to claim 10, wherein

the grey-level original image (f) of the inner surface of a tyre is captured and is transformed into the orthonormal space in which the x-axis (OX) represents the circumferential direction and the y-axis (OY) represents the radial direction,

the predetermined series of filters (F₁^f, F₂^f, . . . F_L^f) is applied to the grey-level original image to obtain the multivariate original image,

the multivariate original image is split according to the predefined tiling in the axial direction and the circumferential directions, to obtain multivariate sub-images (W_F^k) of the inner surface of the tyre,

each of the multivariate sub-images is transformed into one-dimensional vectors, using each of the selected relevant descriptors (H, G, . . . N), to obtain a simplified multivariate image of the inner surface of the tyre,

the simplified multivariate image is transformed into the common reduced factorial space (c^HWF, c^GWF, . . . , c^NWF), and

the sub-images containing an anomaly are located using a classifier suitable for identifying areas of the common reduced factorial space containing anomalies.

12. The method according to claim 11, wherein the classifier is constructed using a Linear Discriminant Analysis (LDA) analysis tool or a Separators with Vast Margins (SVM) analysis tool.