Method and arrangement for detecting surface and structural defects of a long moving product

• US 7,970,202 B2
• Filed: 06/12/2006
• Issued: 06/28/2011
• Est. Priority Date: 06/14/2005
• Status: Expired due to Fees

• Current Assignee: Clecim
• Original Assignee: Siemens Magnet Technology Limited (Siemens AG)
• Inventors: Janin, Pierre-Jean, Peyssard, Mathieu
• Primary Examiner(s): Mariam; Daniel G
• Application Number:
  US11/922,075
• Publication Number:
  US 20090245622A1
• Time in Patent Office: 1,842 Days
• Field of Search: 382/141, 382/149, 382/152, 382/224, 348/86, 348/92, 348/125
• US Class Current: 382/152
• CPC Class Codes: B21B 38/00 Methods or devices for measuring, detecting or monitoring specially adapted for metal-rolling mills, e.g. position detection, inspection of the product control devices or methods B21B37/00
• CPC Class Codes: G01N 21/8901 Optical details; Scanning details per se G02B
• CPC Class Codes: G01N 21/952 Inspecting the exterior surface of cylindrical bodies or wires G01N21/956 takes precedence
• CPC Class Codes: G01N 25/72 Investigating presence of flaws

Abstract

A method for detecting defects of a material moving in a rolling mill generally comprises illuminating product, obtaining (ACQ) image thereof in at least one spectral band, in pre-processing the formed image, in detecting and extracting (DTEXTR) eventually suspected areas of the pre-processed image, and in classifying (CLASS) the suspected areas into one or several defect or defect-free categories. For hot-rolled materials, the product is inspected in three disjointed spectral bands, including infrared, red and, for instance, green band in such a way that the suspected areas are classified into one or several defect-free or defect categories including the surface and structural defects.

Claims

1. A method for detecting defects of a product moving in a rolling mill, said method comprising the steps of illuminating the moving product at least once with a light source, forming by scanning a first image of the moving product as observed in a first spectrum band of light, producing from a first image of an observation area of the product a first digitized luminosity distribution of said area, using the first digitized luminosity distribution of the observation area of the product to detect the presence and location of any defects in said area, classifying suspect zones into at least one category of defects or non-defects by comparing morphological and/or photometric characteristics of said suspect zones with morphological and/or photometric characteristics of known defects and of known non-defects held in a pre-established database, applying said method to a product moving in a hot-rolling installation and spontaneously emitting a light radiation in a spontaneous emission spectrum including the infrared, said illuminating step being implemented by illuminating the moving product with a visible light source emitting at least outside the spontaneous emission spectrum, said scanning forming step comprising forming at least three images of the same observation area of said product in three respective non-contiguous spectral bands of which a first band lies at least in the infrared, a second band at least in the red, and a third band at least in part of the visible spectrum outside the spontaneous emission spectrum, said producing step being at least implemented on said three images to provide at least three respective luminosity distributions digitized over several bits performing a detection and extraction operation of suspect zones on the digitized luminosity distributions and said classifying step comprising classifying at least the suspect zones extracted from the digitized distributions into at least one category of inner defects of the product, surface defects of the product, and non-defects of the product at least by comparison of the morphological and/or photometric characteristics of said suspect zones and scattering of said suspect zones among the digitized distributions with the morphological and/or photometric characteristics and scattering among corresponding digitized distributions of known inner defects, known surface defects and known non-defects held in said pre-established database.

2. The method according to claim 1, comprising forming the images by means of respective sensors at least sensitive to the infrared and producing respective raw images forming the first image in the infrared and directly obtaining the first image by capture of the first raw image behind an optical filter opaque to radiations of wavelengths lower than the infrared, and obtaining the second and third images from the second and third raw images by subtracting the first raw image.

3. The method according to claim 1, wherein the illuminating step comprises illuminating the product with a green light source, and said forming step comprises forming said third image in the green.

4. The method according to claim 1, further comprising producing a fourth digitised luminosity distribution obtained by linear combination of the three original digitised distributions and using said fourth distribution for detection and extraction of said suspect zones and classifying said suspect zones.

5. The method according to claim 1, further comprising providing the product moving in the hot-rolling installation in the form of a long product, whose cross-section is contained in at least one of a polygon, a circle or an ellipse.

6. The method according to claim 5, comprising forming the observation area common to the three images so that said observation area extends over an entire perimeter of the product and over at least part of a length of the product, and forming each of the three images by joining side by side at least two elementary images formed in one spectral band of light over one part of the length of the product and over non-contiguous, complementary portions of the perimeter of said product.

7. An installation for implementing a method according to claim 1, said installation comprising at least one digital camera CCD provided with three linear detection arrays arranged transverse to a direction of travel of a moving product and sensitive to different colors of the spectrum of visible light.