GLASS INSPECTION

Abstract

A method of detecting a defect in a sheet of glass includes: (i) directing a converging beam from a source of illumination onto a surface of the sheet of glass to illuminate the defect; (ii) focussing an image capture device onto a first plane to image the defect in the sheet of glass; (iii) capturing a first image of the defect; (iv) carrying out an adjustment step; and (v) capturing a second image of the defect. Each of the first and second images has a respective first portion from the illuminated defect and a respective second portion due to reflection of a portion of the beam from the glass surface. In the first image of the defect the first portion is brighter than the second portion, and in the second image of the defect the first portion is darker than the second portion. An apparatus for carrying out the method is also provided.

Claims

1.-43. (canceled)

44. A method of detecting a defect in a sheet of glass, the sheet of glass having a first major surface and a second opposing major surface, the method comprising the steps: (i) directing a beam from a source of illumination onto the first major surface of the sheet of glass to illuminate the defect in the sheet of glass, the beam that strikes the first major surface of the sheet of glass being a converging beam having a first focal point; (ii) focusing an image capture device onto a first plane to image the defect in the sheet of glass, the image capture device being at a first position relative to the defect; (iii) using the image capture device to capture a first image of the defect, the first image comprising a first portion from the illuminated defect and a second portion due to reflection of at least a portion of the beam from the first major surface of the glass sheet; (iv) carrying out an adjustment step; (v) using the image capture device to capture a second image of the defect, the second image comprising a first portion from the illuminated defect and a second portion due to reflection of at least a portion of the beam from the first major surface of the glass sheet; wherein in the first image of the defect the first portion is brighter than the second portion, and in the second image of the defect the first portion is darker than the second portion.

45. The method according to claim 44, wherein the adjustment step comprises at least one of (i) adjusting the first focal point of the converging beam to a second focal point so that the beam striking the first major surface is a converging beam having the second focal point; (ii) focussing the image capture device onto a second plane to image the defect with the image capture device focussed onto the second plane; and (iii) moving the image capture device to a second position relative to the defect.

46. The method according to claim 44, wherein the first plane is aligned with the first major surface of the sheet of glass.

47. The method according to claim 44, wherein at step (i) positive optical power is added to the beam before the beam strikes the first major surface of the sheet of glass; and/or wherein at step (i) negative optical power is added to the beam before the beam strikes the first major surface of the sheet of glass.

48. The method according to claim 44, wherein at step (iv) negative optical power is added to the beam after the beam strikes the first major surface of the sheet of glass.

49. The method according to claim 44, wherein at step (iv) the image capture device remains focussed onto the first plane.

50. The method according to claim 44, wherein at step (iv) the image capture device remains at the first position relative to the defect such that at step (v) the second image of the defect is taken with the image capture device at the first position relative to the defect.

51. The method according to claim 44, wherein the defect comprises a localised shape change of the first major surface of the sheet of glass, the localised shape change of the first major surface of the sheet of glass being caused by a particle that was deposited onto the first major surface of the glass sheet during the formation of the glass sheet; and/or wherein the particle is spherical.

52. The method according to claim 51, wherein the particle is at least partially submerged beneath the first major surface of the sheet of glass or wherein the particle that caused the localised shape change of the first major surface of the sheet of glass is not in the sheet of glass when any or all of the steps (i), (ii), (iii), (iv) or (v) are carried out.

53. The method according to claim 44, wherein the first image is a monochromatic image; and/or wherein the second image is a monochromatic image.

54. The method according to claim 44, wherein the defect has a major axis having a length less than 200 m; and/or wherein the defect being detected has a major axis having a length greater than 0.5 m.

55. The method according to claim 44, wherein the first plane is aligned with the first major surface of the sheet of glass, and wherein prior to step (ii) the method includes a distance measuring step to determine a position of the first major surface of the sheet of glass relative to the image capture device so that during step (ii) the image capture device can be focussed onto the first major surface of the sheet of glass.

56. The method according to claim 44, wherein the defect being detected is a first defect of a plurality of defects in the sheet of the glass, the plurality of defects also comprising at least a second defect, the first defect being detected initially by carrying out steps (i) (ii) and (iii) and then the second defect being detected by carrying out steps (i) (ii) and (iii), thereafter steps (iv) and (v) being carried out to detect the first defect, followed by steps (iv) and (v) being carried out to detect the second defect.

57. The method according to claim 44, wherein the beam that illuminates the first major surface shares an optical axis with at least a reflected ray from the first major surface of the sheet of glass to the image capture device.

58. The method according to claim 44, wherein the sheet of glass has been produced using a float process, and wherein the first major surface has not been in contact with molten tin when the sheet of glass was formed.

59. The method according to claim 44, wherein the beam is a stopped beam from the source of illumination, the stopped beam passing through a least a first aperture positioned between the source of illumination and the first major surface of the sheet of glass.

60. An apparatus for determining the presence of a defect in a sheet of glass, the sheet of glass having a first major surface and a second opposing major surface, the apparatus comprising a source of illumination for illuminating a portion of the surface of the sheet of glass containing the defect with a converging beam having a first focal point; an image capture device for taking a first image of the illuminated defect when the image capture device is at a first position relative to the defect and focussed onto a first plane; and at least one of an adjustment means for adjusting the focal point of the converging beam of light to a second focal point while the image capture device remains focussed onto the first plane, a focussing means for focussing the image capture device onto a second plane to image the defect with the image capture device focussed onto the second plane and a moving means for moving the image capture device to a second position relative to the defect.

61. The apparatus according to claim 60, wherein adjustment means for adjusting the focal point of the converging beam to a second focal point while the image capture device remains focussed onto the first plane comprises a lens; and/or wherein the focussing means for focussing the image capture device onto a second plane to image the defect with the image capture device focussed onto the second plane comprises a lens.

62. The apparatus according to claim 60, further comprising a controller for controlling at least one of the source of illumination, the image capture device and the means of adjusting the focal point of the converging beam to a second focal point while the image capture device remains focussed onto the first plane; and/or further comprising a computer, wherein images taken by the image capture device are processed by software installed on the computer to determine a parameter related to the defect.

63. The apparatus according to claim 60, wherein the apparatus comprises moving means for moving the image capture device from the first position relative to the defect to the second position relative to the defect, further wherein the moving means can move the image capture device in a direction parallel to a normal extending from the first major surface of the sheet of glass being measured to be closer or further away from the first major surface of the sheet of glass being measured; and/or wherein the moving means is configured to move the image capture device from the first position to detect a first defect in a sheet of glass being measured to the second position relative to the first defect to detect a second defect in the sheet of glass.

Description

[0133] Embodiments of the present invention will now be described by way of example only with reference to the accompanying drawings, in which:

[0134] FIG. 1 shows a schematic cross-sectional representation of an apparatus for carrying out a method in accordance with the present invention;

[0135] FIG. 2 shows a schematic representation of a first image (FIG. 2a) and a second (2b) of a defect taken with the apparatus shown in FIG. 1;

[0136] FIG. 3 shows a sequence of images of a defect taken with the apparatus shown in FIG. 1 with different levels of optical power added to the system; and

[0137] FIG. 4 shows a schematic isometric view of the apparatus shown in FIG. 1 used to scan a surface of a glass sheet.

[0138] FIG. 1 shows a schematic cross-sectional representation of an apparatus for carrying out a method in accordance with the present invention.

[0139] The apparatus 1 comprises a light emitting diode (LED) light source 3 producing a diverging light beam 5 that is reflected through 90 by a beam splitter 7. The reflected light passes through a lens 9 and a suitable lens 9 is a Nikon EL-NIKKOR lens 50 mm F/2.8. The lens 9 converges the reflected light beam onto major surface 11 of the glass sheet 13. The major surface 11 of the glass sheet 13 is opposite the major surface 15 of the glass sheet 13. Before striking the major surface 11 the beam of light focussed by lens 9 also passes through a computer adjustable lens 17. A suitable computer adjustable lens 17 is an Optotune EL-16-40-TC-VIS-5D-C.

[0140] As can be seen from FIG. 1, the apparatus 1 is arranged such that the lens 9 is between the beam splitter 7 and the computer adjustable lens 17. The lens 9 is between the major surface 11 and the beam splitter 7. The computer adjustable lens 17 is between the major surface 11 and the beam splitter 7.

[0141] The beam 5 has an optical axis that strikes the major surface 11 at an angle of incidence greater than 45, preferably between 50 and 140, more preferably between 80 and 100. Preferably the optical axis of the beam strikes the major surface at normal, or substantially normal incidence. Shallow angles of incidence are not preferred because such illumination conditions may illuminate defects such as dust and scratches on the major surface 11 to a greater extent thereby reducing sensitivity of the method to defects of interest.

[0142] The light reflected by the major surface 11 of the glass sheet 13 passes back through the computer adjustable lens 17 and the lens 9, through the beam splitter 7 and travels on to a digital camera 19. In this example the digital camera 19 is an area camera, a suitable example thereof being a Teledyne Dalsa Genie Nano M1920 digital camera.

[0143] If an infrared emitting solid state device (i.e. an infrared emitting LED) is used instead of the LED light source 3, a suitable camera sensitive to infrared radiation emitted by the infrared emitting solid state device is used to capture images of the illuminated defect. As will be readily apparent, an image capture device that is suitable for capturing an image of a defect illuminated by a beam from the particular source of illumination is used to carry out the method of the present invention.

[0144] To image defects on or close to the major surface 11 of the glass sheet 13, the digital camera 19 is focussed onto the major surface 11.

[0145] In order to focus the digital camera 19 onto the major surface 11 the distance between the digital camera 19 and the major surface 11 is required and this distance is measured using a laser triangulation sensor such as a Micro-Epsilon optoNCDT ILD1420 to allow accurate focussing.

[0146] The distance to the major surface 11 from the digital camera 19 is then used to adjust the computer adjustable lens 17 so that the major surface 11 the glass sheet 13 is kept precisely in focus.

[0147] A controlled amount of negative optical power (for example about 0.2 dioptres) is then added by the computer adjustable lens 17 as it has been found that this makes the optical system sensitive to distortion of the glass surface around the defect of interest.

[0148] It has been found that the use of converging illumination from lens 9 with the addition of optical power by the computer adjustable lens 17 causes the image of certain defects obtained by digital camera 19 to appear bright compared to the surrounding image of the glass. It has been found that certain particles such as tin containing particles cause a concave-like localised distortion of the glass surface 11 around the particle that may be used to determine the presence of such defects optically, even when the particles themselves have small dimensions, such as less than 50 m.

[0149] It is thought particles that affect the surface profile of the glass sheet locally around the particle will have been deposited onto the glass surface when the glass sheet was at a sufficiently high temperature so that the glass viscosity was low enough for the deposition of the particle to alter the surface shape or profile around the particle. One such example is a particle containing tin, or a particle of tin, which may be formed in the float bath when a glass sheet is formed using a float process. As the glass viscosity increases as the glass is cooled during the forming process, the surface profile around the particle becomes frozen-in to cause the defect in the sheet of glass. Even if the particle that caused the surface profile of the glass sheet to change locally around the particle is later removed (for example because the particle was only weakly attached to the surface of the sheet of glass), a defect is still present in the sheet of glass due to the aforementioned localised surface profile change.

[0150] Dust and dirt that may be on the major surface 11 typically appear darker than the surrounding image of the glass. Such defects are typically deposited onto the major surface 11 when the glass has been cut into sheets, for example during handling and/or transportation of the glass sheet.

[0151] By scanning the glass sheet 13, regions of the surface 11 with a bright defect compared to the image of the surrounding glass sheet may be identified using suitable image processing software.

[0152] Advantageously the apparatus 1 is arranged to have a short depth of field so that the image captured by the digital camera 19 does not show defects in the body of the glass or on the opposite major surface 15 for a glass sheet having a thickness of at least about 1.5 mm. Consequently, using the system as described above may be used to detect the presence of defects on or close to the surface of the major surface 11. Such defects may be at least partially below the major surface 11.

[0153] Once an image has been taken that shows a defect that appears bright compared to the surrounding glass image, the image may be further processed to help further determine the type of defect and to differentiate the defect from other types of defect.

[0154] For example, under the illumination conditions described above it has been found that scratches in the major surface 11 also show up as brighter than the image of the surrounding glass sheet. To discriminate scratches from other types of defect, further image processing may be used to determine at least one parameter related to the shape of the defect in the image captured by the digital camera 19.

[0155] One particularly useful parameter related to the shape of the defect is the aspect ratio. For example, a scratch is typically narrow and long. A defect such as a tin containing particle, or a particle of tin, is typically spherical, so has a circular profile in the image captured by the camera. The degree of circularity of the defect may also be used as a parameter to discriminate between different types of defect, for example by using a ratio of the area of the defect to the perimeter of the defect in the image obtained by the digital camera 19. One such measure of circularity is:

[00001]$\mathrm{circularity}=\frac{\left({\mathrm{perimeter}}^2\right)}{\left(4\mathrm{area}\right)}$

which for a perfect circle has a circularity of 1 and the higher the circularity, the less circular the defect is. For a digital image of the defect, the perimeter and area of the defect can be determined by simply counting pixels.

[0156] To further help reduce any mis-classification of a particular defect, the method of the present invention includes a step where a second image of the defect is taken with the controlled level of negative optical power provided by the computer adjustable lens 17 removed and preferably adding a predetermined amount of positive optical power such as 0.2 dioptres using the same computer adjustable lens 17.

[0157] The first and second images containing the same defect are then compared. As discussed above, certain defects such as tin, or tin containing defects, that have been deposited on the surface of the glass during forming cause the glass surface to be distorted locally (i.e. from flat to concave-like) cause the light reflection therefrom to be adjusted depending upon the degree of focussing. In the defocussed image when negative optical power was applied the image of the defect appears bright compared to the image of the surrounding glass, whereas when the defect is in focus, or positive optical power is applied using the computer adjustable lens 17, the image of the defect appears dark compared to the image of the surrounding glass.

[0158] The apparatus 1 can then be used, together with a suitable image processing step, to be insensitive to dust, dirt, scratches and all other defects apart from defects that have locally affected the surface of the glass sheet when deposited thereon, such as tin defects, or tin containing defects.

[0159] The method of the present invention can be applied to glass moving relative to the optical system by using a short exposure time on the camera and/or a short illumination time on the light source to freeze the relative motion. To cover a large area of glass the optical system can be raster scanned over the glass, or a number of optical systems can be used in combination to cover the required area.

[0160] In an alternative to the apparatus shown in FIG. 1, the computer adjustable lens 17 may be positioned in between the beam splitter 7 and the lens 9.

[0161] In another alternative to the apparatus shown in FIG. 1, the computer adjustable lens 17 may be positioned in between the beam splitter 7 and the digital camera 19.

[0162] In another alternative to the apparatus shown in FIG. 1, the computer adjustable lens 17 may be positioned in between the beam splitter 7 and the light source 3.

[0163] The computer adjustable lens 17 provides a quick way to add negative optical power to the light beam that illuminates the major surface 11 and is a faster way of controlling the degree of defocus of the imaging system i.e. digital camera 19.

[0164] To add negative optical power instead of using a computer adjustable lens 17, the camera may be focussed into the body of the glass sheet 13 instead. However, to adjust the focus to carry out the further distinguishing step where positive optical power is added, the camera focus is then adjusted to be above the major surface 11. Moving the camera focus is a way to add negative or optical power but is slow compared to using the computer adjustable lens 17.

[0165] A similar effect can be obtained by using lenses having fixed focal lengths, but this takes time to replace lenses.

[0166] In other embodiments the digital camera 19 may be movable towards or away from the first major surface 11.

[0167] In other embodiments the light source may be movable towards or away from the beam splitter 7 such that the optical path from the light source to the first major surface 11 is adjustable. In the embodiment shown in FIG. 1, the apparatus 1 includes a housing 21 inside which the LED 3, the beam splitter 7, the lens 9, the computer adjustable lens 17 and the digital camera 19 are contained. Also contained inside the housing 21 is a controller 23 that is used to control the operation of the LED 3 and the computer adjustable lens 17. The controller 23 is also used to control the operation of the digital camera 19.

[0168] The controller 23 is in electrical communication with the LED 3 via cable 25. The controller 23 is in electrical communication with the computer adjustable lens 17 via cable 27. The controller 23 is in electrical communication with the digital camera 19 via cable 29. The controller 23 is also in electrical communication with a computer 31 via cable 33.

[0169] The computer 31 can be provided with software to control the operation of the LED 3, the computer adjustable lens 17 and the digital camera 19 via the controller 23. Images taken by the digital camera 19 may also be sent to the computer 31 via the controller 23 along cables 29, 33. The software may include image processing software to help identify and discriminate the defect as described above, for example by determining a shape parameter of the defect such as circularity.

[0170] FIG. 2a shows a schematic representation of part of an image taken by the digital camera 19 of a portion of the surface 11 where a spherical tin particle was deposited during forming the glass sheet, the spherical tin particle having caused a local shape change of the glass surface around the tin particle from flat to concave-like. In the image the defect 40 has a circular profile. The glass surrounding the defect 40 is also included in the image and labelled 42. In FIG. 2a, the defect 40 is bright compared to the glass 42 surrounding the defect and is from an image taken with the apparatus 1 when the digital camera 19 is focussed on the major surface 11 of the glass sheet 13 and a controlled amount of negative optical power of about 0.2 dioptres is added by computer adjustable lens 17.

[0171] Using the image of FIG. 2a, the computer 31 of FIG. 1 is used to carry out the additional image processing as discussed above, for example to determine the circularity of the defect in the image. The image processing step is able to determine that the defect is a certain type and not another type that is not of interest. For example, when looking for certain defects due to tin or tin-containing particles that have a spherical shape and have been deposited on the glass surface, the image processing step is used to determine that the defect has a degree of circularity that indicates the defect is circular (and so spherical) and that further inspection is required. If the image processing step determines that the defect was a scratch, further inspection may not be required.

[0172] The apparatus 1 is then used to take another image of the same defect with the computer adjustable lens 17 used to provide positive optical power of about 0.2 dioptres (instead of being used to add about 0.2 dioptres of negative optical power).

[0173] FIG. 2b shows a schematic representation of part of the image taken by the digital camera 19 of the circular defect 40 when the apparatus 1 is used with about 0.2 positive optical power provided by the computer adjustable lens 17. All other illumination conditions were the same as used to take the image shown in FIG. 2a.

[0174] The circular defect in the image is now dark compared to the surrounding glass. In FIG. 2b, the glass surrounding the defect is still labelled 42 but the defect is labelled 40.

[0175] FIG. 3 shows a sequence of images (a)-(g) of another defect that has been identified using the apparatus 1.

[0176] In FIG. 3(a) the apparatus 1 is configured with the digital camera 19 focussed on the major surface 11 and 0.3 dioptres negative optical power added to the system using the computer adjustable lens 17. As can be seen, the defect is circular and is brighter (a lighter shade of grey) than the image of the glass surrounding the defect.

[0177] In FIG. 3(b) the apparatus 1 is configured in the same way as when the image in FIG. 3(a) was taken with the exception that 0.2 dioptres negative optical power was added to the system using the computer adjustable lens 17 (instead of 0.3 negative optical power). As can be seen, the defect is brighter (a lighter shade of grey) than the image of the glass surrounding the defect.

[0178] In FIG. 3(c) the apparatus 1 is configured in the same way as when the image in FIG. 3(a) was taken with the exception that about 0.1 dioptres negative optical power was added to the system using the computer adjustable lens 17 (instead of 0.3 negative optical power). The defect is in focus and appears dark compared to the image of the glass surrounding the defect.

[0179] In FIG. 3(d) the apparatus 1 is configured in the same way as when the image in FIG. 3(a) was taken with the exception that no optical power was added to the system using the computer adjustable lens 17. As can be seen, the defect is darker (a darker shade of grey) than the image of the glass surrounding the defect.

[0180] In FIG. 3(e) the apparatus 1 is configured in the same way as when the image in FIG. 3(a) was taken with the exception that 0.1 dioptres positive optical power was added to the system using the computer adjustable lens 17. As can be seen, the defect is darker (a darker shade of grey) than the image of the glass surrounding the defect and is slightly larger than the image of the defect in FIG. 3(d) because the amount of defocus from the glass surface 11 has increased.

[0181] In FIGS. 3(f) and 3(g) the apparatus 1 is configured in the same way as when the image in FIG. 3(a) was taken with the exception that 0.2 dioptres positive optical power and 0.3 dioptres positive optical power respectively was added to the system using the computer adjustable lens 17. As can be seen in both images, the defect is darker (a darker shade of grey) than the image of the glass surrounding the defect. Also, in each image the image of the defect is slightly larger than the image of the defect in FIG. 3(e) because the amount of defocus from the glass surface 11 has increased. The image of the defect in FIG. 3(g) is larger than the image of the defect in FIG. 3(f) because more positive optical power was added to the system using the computer adjustable lens 17 when the image in FIG. 3(g) was taken compared to when the image in FIG. 3(f) was taken.

[0182] The sequence of images 3(a) to 3(g) illustrate how a defect of interest may be differentiated from other defects that do not distort the surface of the surrounding glass, so do not appear bright or dark relative to the image of the surrounding glass when the amount of optical power is adjusted from a negative amount of optical power to a positive optical power.

[0183] As mentioned above, optical power added to the system using the computer adjustable lens 17 is more convenient and quicker than adjusting the point at which the digital camera 19 focusses.

[0184] FIG. 4 is a schematic representation showing how the apparatus 1 may be used to scan an entire glass sheet 13 for defects on or in the major surface 11. In this example the glass sheet 13 is rectangular and is static relative to the apparatus 1. In this example the apparatus 1 is as shown in FIG. 1.

[0185] In this example the glass sheet 13 is flat and has a major surface 11, but the glass sheet 13 may be curved. On the major surface 11 is a first axis 47 and a second axis 48 orthogonal thereto. A third axis 49 extends from the major surface 11 and is orthogonal to both the first and second axes 47, 48.

[0186] The optical axis of the beam 5 striking the major surface 11 is preferably substantially parallel to the third axis 49.

[0187] The apparatus 1 can be moved using a suitable X-Y stage or robot arm (not shown) to move the apparatus in the direction of arrow 50 to obtain images of the glass sheet 13 using the digital camera 19. The direction of arrow 50 in this example is parallel to a lateral edge of the glass sheet 13 and parallel to the first axis 47.

[0188] The entire glass sheet 13 may be scanned by moving the apparatus 1 in raster fashion, for example when the apparatus 1 reaches the end of the first pass (shown there with the label 1), the apparatus may be moved in the direction of arrow 52 to the position shown as 1 to scan along the path shown by arrow 54, and so on. The path of arrow 50 and the path of arrow 54 are parallel to each other and both are parallel to the first axis 47.

[0189] In the images acquired with negative optical power added to the system, regions where a bright defect compared to surrounding glass (for example as shown in FIGS. 2 and 3), are further analysed to determine how circular the image is, and to determine if the image of the defect changes as illustrated in FIGS. 3(a)-3(g). This can be done when a defect of interest is identified, or after the entire glass sheet has been scanned. In this latter embodiment, the position of each image where a bright defect was identified is recorded so the apparatus can go back to that position to carry out the steps shown in FIGS. 3(a)-3(g).

[0190] To speed up the operation, it is possible to only use the computer adjustable lens 17 to provide a predetermined negative optical power in the first scan (for example 0.3 dioptres) and to then use a single predetermined positive optical power (for example +0.3 dioptres) to see if the image of the defect changes as illustrated in FIGS. 2a and 2b.

[0191] Instead of the apparatus 1 moving across the surface 11 of the glass sheet 13, the glass sheet 13 may be positioned on a suitable movable support to move the glass sheet 13 relative to the apparatus 1. Each of the apparatus 1 and the glass sheet 13 may move during the scanning process.

[0192] In another embodiment, the apparatus 1 only traverses across the glass sheet along a fixed path and the glass sheet 13 is moved along a path perpendicular to the path traversed by the apparatus 1. Such an embodiment is particularly useful when the glass sheet 13 is in the form of a moving ribbon of glass as is formed in a float process on a bath of molten tin, whereby the apparatus may be positioned downstream of the bath of molten tin and scanned thereafter, preferably before the glass ribbon has been cut into individual sheets. The apparatus 1 may scan across the ribbon width at a rate commensurate with the ribbon speed so that lengths of the ribbon may be scanned because the ribbon moves in a direction of conveyance after a scan across the ribbon width has been carried out.

[0193] In another embodiment, two or more apparatus as shown in FIG. 1 are used to scan different parts of the glass sheet or glass ribbon.

[0194] In another embodiment, the apparatus 1 is also movable in a direction parallel to the axis 49. In such an embodiment it is not essential to use the computer adjustable lens 17 to provide positive and negative optical power, instead the apparatus 1 can be moved toward and away from the major surface 11 in a direction parallel to the third axis 49 to position the image capture device to acquire first and second images of a defect in the sheet of glass 13.

[0195] The present invention is particularly useful to determine the presence of defects in a sheet of glass where a particle has been deposited on the sheet of glass when the glass was at a sufficiently low viscosity (i.e. during forming the glass sheet) to locally distort the shape of the surface of the glass around the deposited particle. The local shape distortion of the surface of the glass around the particle adjusts the path of light illuminating the defect differently depending upon the degree of convergence of the incident light beam and this may be used to discriminate this type of defect. The local shape distortion of the surface of the glass around the particle remains even if the particle subsequently becomes removed from the sheet of glass. Using the optical effect associated with local shape distortion of the surface of the glass around the particle also allows smaller defects to be identified because the optical effect is large even when the particle creating the local shape distortion of the surface of the glass around the particle is small.