Search

EP-4737878-A1 - METHOD AND APPARATUS FOR ENGRAVING REMOVAL

EP4737878A1EP 4737878 A1EP4737878 A1EP 4737878A1EP-4737878-A1

Abstract

The present invention relates to a computer-implemented method for identifying cosmetic defects of a spectacle lens, comprising: identifying, if at least one detected object in a captured image of the spectacle lens is associated with a first plurality of features; and extracting, the at least one detected object associated with the first plurality of features from the captured image before a cosmetic defect detection is performed.

Inventors

  • Schoebel, Daniel
  • Sulugodu Arunachala, Vivek Bharadhwaj
  • KUNGEL, Christian

Assignees

  • Carl Zeiss Vision International GmbH

Dates

Publication Date
20260506
Application Date
20241031

Claims (15)

  1. A computer-implemented method for identifying cosmetic defects of a spectacle lens, comprising: identifying, if at least one detected object in a captured image of the spectacle lens is associated with a first plurality of features; and extracting, the at least one detected object associated with the first plurality of features from the captured image before a cosmetic defect detection is performed.
  2. The method according to claim 1, wherein the first plurality of features comprises intended engravings of the spectacle lens.
  3. The method according to claim 2, wherein the intended engravings comprise at least one of: - an orientation mark, - a code, - a structure suitable for myopia control, - an electronic structure, - an optical structure, - an individualized information or design pattern.
  4. The method according to one of the claims 1 to 3, further comprising: identifying, if the least one detected object in the captured image of the spectacle lens is associated with a second plurality of features.
  5. The method according to claim 4, wherein the at least one detected object, associated with a second plurality of features, is identified at a position in the captured image, where the at least one detected object associated with the first plurality of features has been extracted from the captured image.
  6. The method according to one of the claims 4 to 5, wherein the at least one detected object associated with the second plurality of features comprises a cosmetic defect of the spectacle lens.
  7. The method according to claim 6, wherein the cosmetic defect is transparent or opaque.
  8. The method according to one of the claims 4 to 7, wherein the at least one detected object associated with the second plurality of features is classified.
  9. The method according to one of the claims 6 to 8, wherein the detected cosmetic defect is classified based on its dimension.
  10. The method according to one of the claims 6 to 9, wherein the detected cosmetic defect is classified in a defect category of a plurality of defect categories.
  11. The method according to claim 10, wherein the plurality of defect categories comprises at least one of: - surface defects, - substrate defects, - finishing defects.
  12. The method according to one of the claims 1 to 11, wherein a pattern matching algorithm is used comprising an artificial neural network, wherein the artificial neural network is trained to identify, if the at least one object in the captured image is associated with the first plurality of features.
  13. An apparatus for identifying cosmetic defects of a spectacle lens, comprising: means for identifying, if at least one detected object in a captured image of the spectacle lens is associated with a first plurality of features; means for extracting, the at least one detected object associated with the first plurality of features from the captured image before a cosmetic defect detection is performed.
  14. A computer-implemented method for training an artificial neural network for identifying intended engravings in a captured image of a spectacle lens, the method comprising: receiving captured images of a spectacle lens comprising intended engravings, wherein each of the intended engravings comprise a plurality of circles; and training an artificial neural network based on received images to identify intended engravings in a captured image of a spectacle lens, wherein the artificial neural network is trained to use circlet transformation to detect the plurality of circles of the intended engravings.
  15. A computer program comprising instructions which, when executed by a computer, cause the computer to perform the steps of one of the methods of one of the claims 1-12 or 14.

Description

1. Technical field The present invention relates to a computer-implemented method and a corresponding apparatus for identifying cosmetic defects of a spectacle lens as well as a training method. 2. Prior art In the manufacturing of spectacle lenses, a cosmetic defect inspection is important to ensure a high quality of each spectacle lens. Thereby, the spectacle lenses must be tested for cosmetic defects, particularly for surface flaws such as scratches, smears, cracks, chips, stains, and for occlusions such as inclusions or streaks. Limit values of tolerable flaws are established in internationally recognized standards for precision optics, specifically DIN 3140 or MIL-0-13830 or less stringent standards used by the ophthalmic industry, such as DIN 58203. Conventionally, such an inspection of each spectacle lens is carried out by specially trained personnel, which determine whether certain defects (scratches, indentations, contaminations like for example dust, coating defects, etc.) are present. However, the use of specially trained personnel is cost-effective, time-consuming and may not guarantee an objective assessment without exception at any time. Cosmetic defect inspection is a process of finding manufacturing defects on a surface that affect the quality of a produced spectacle lens either visually or functionally. This may be done after every production step or at the end of the production chain. Potential defects may be distributed punctually or over a large area of the spectacle lens. In certain circumstances the defects may cover the entire spectacle lens. Further, these defects may vary not only in their dimension but also in their appearance, which turns their classification in a sophisticated process. Furthermore, defects may be transparent and/ or non-transparent, which makes their detection challenging, in particular, on a transparent surface. Moreover, spectacle lens may have intended engravings, which can serve as orientation marks, codes for traceability or structures suitable for myopia control. An overlap of these intended engravings with one or more defects may hinder a reliable defect detection. Especially, when the overlap causes that the intended engraving no longer appearing continuous. Recent developments have improved the ability of computers to detect and classify certain types of defects by using trained methods that have learned to detect and classify certain types of defects. However, in these methods, every engraving, even the intended engravings, has to be processed and evaluated in the entire image. This results in a high computational and time effort. Moreover, the ability of a trained method to detect cosmetic defects fails in the case of an overlap of an intended engraving and a cosmetic defect or reduces the reliability of cosmetic defect detection significantly. In addition, it is particularly difficult for those methods to differentiate between an intended engraving and a cosmetic defect, when the intended engraving is defective (for example when a part of an orientation mark is missing). It is therefore an object of the present invention to provide a computer-implemented method and a corresponding apparatus, which provide a more reliable automatic detection of cosmetic defects. It is a further object of the present invention to provide a training method, which improves the efficiency and reliability of intended engravings detection. H. Chauris et al.: "A robust tool for detecting features with circular shapes" (Computers & Geosciences 37 (2011) 331-342) proposes a method for detecting circles on digital images. Thereby, a circlet transformation is used, which takes the finite frequency aspect of the data into consideration. Further, the circlet transformation is coupled with a soft thresholding process and applied to a series of images. WO 2024/038170 A3 refers to a computer-implemented method for facilitating an identification of an optical mark in an image of a spectacle lens. Thereby, enhancement information based on the image of the spectacle lens is generated, wherein the enhancement information enhances the detectability of the optical mark in an automated identification process when impairing structures impair the detectability of the optical mark. WO 2021/137745 A1 relates to a method for detection of imperfections in products using image analysis on digital representations of the products, wherein images of the products are captured through digital camera means and imported to data processing means running a computer executable artificial neural network. US 2008/317329 AA is directed to method and apparatus which perform detection and classification of defect, such as a fine pattern defect and a foreign particle, based on an image of an object obtained using a light of a lamp, a light of a laser, or an electron beam, for thin film devices such as a semiconductor wafer, TFT, and a photomask. WO 2018/098551 A1 refers to a method and a system for automatic inspe