Journal of Microscopy, Vol. 196, Pt 3, December 1999, pp. 347±351. Received 3 March 1999; accepted 17 May 1999
SHORT TECHNICAL NOTE
Stereoscopic display of atomic force microscope images using anaglyph techniques J. R. SMITH,* S. D. CONNELL* & J. A. SWIFT² *Scanning Probe Microscopy Laboratory, School of Pharmacy and Biomedical Sciences, University of Portsmouth, St. Michael's Building, White Swan Road, Portsmouth PO1 2DT, U.K. ²Department of Textile Design and Production, De Montfort University, The Gateway, Leicester LE1 9BH, U.K.
Key words. Anaglyph methods, atomic force microscopy (AFM), cuticle, hair, stereo microscopy.
Summary This paper describes the use of a standard stereo-pair image display method for presenting the three-dimensional relief information found in atomic force microscope (AFM) images. The method makes use of commercially available image processing software packages. The techniques are illustrated on AFM images of the cuticle structure of a human hair ®bre.
Introduction While stereoscopical methods have been widely used in a vast range of microscopical techniques (Hudson, 1972), their introduction to atomic force microscopy (AFM) has received very little attention. It is further surprising that AFM images, while stored as three-dimensional datasets, continue to be presented in two-dimensional formats, such as perspectives, contours and shaded relief. This problem has been recognized by Shao & Somlyo (1995), who have described the use of stereo pair atomic force micrographs to gain further structural insights of biological macromolecules, such as cholera toxin B-oligomer (Shao & Somlyo, 1995). More recently, Raspanti (1999) reviewed a wide range of methods, such as stereoscopy, stereo-animation, holograms and lenticular prints, speci®cally for threedimensional visualization of AFM images.
Correspondence: Dr James R. Smith. Tel/Fax: 44 (0)1705 842556; e-mail:
[email protected] Presented in part at the Eleventh International Hair-Science Symposium, HairS'98, Deutches Wollforschungsinstitut an der RWTH Aachen e.V., Maastricht, The Netherlands, 9±11 September 1998. q 1999 The Royal Microscopical Society
This paper describes in detail the methods necessary to produce a three-dimensional AFM image using the wellknown anaglyph technique (Barber & Brett, 1982; Glen, 1985; Cherkasov, 1997). The anaglyph method makes use of complementary coloured images and lenses to display stereo pairs of images containing horizontal parallax (Usery, 1993; Turnnidge & Pizzanelli, 1997). By using the anaglyph approach we also demonstrate the greater wealth of valuable visual information available to the atomic force microscopist by using such stereoscopical methods than with conventional two-dimensional images.
Materials and methods Atomic force micrographs were obtained using a TopoMetrix Discoverer TMX2000 scanning probe microscope, described in detail elsewhere (Smith, 1997, 1998; Smith et al. 1997). Image processing was carried out using a Dell Dimension XPS D233 Pentium II microcomputer with 64 MB RAM, initially using TopoMetrix image analysis software (TopoMetrix SPM Lab., Version 3´06´06, TopoMetrix Corporation, Santa Clara, CA, U.S.A., 1996). Further processing was performed using Paint Shop Pro, Version 3.0 (Minnetonka, MN, U.S.A.). In the same manner as that adopted to produce side-by-side stereo pairs, two slightly offset renditions of a single dataset were required. These were produced by tilting both images (datasets), with respect to one another, so that the right-hand-side of the left image and the left-hand-side of the right image were conceptually at the greatest distance from the observer. However, before stereo pair images could be tilted, it was necessary to rotate each rendition. This was because the software only permitted vertical tilts of images, with top 347
348
J. R. SMI TH E T AL .
Fig. 1. Image processing steps required to produce an anaglyph AFM image. (a) A raw AFM image where height is encoded as grey level. In this example, the range of grey levels is 218 covering a range of 2652 nm in height difference. Processing steps b±c, d±e and f±g are described in the text. The ®nal anaglyph image (h) encodes the separate left±right views (f) and (g) into a single image. This image can be viewed using low-cost red/green glasses from Agar Scienti®c.
edges retracting away from the observer. A detailed description of the image processing methodology used to produce a three-dimensional anaglyph from a single AFM image now follows. Two copies of the original unprocessed image (Fig. 1a),
required for the left and right renditions, were separately rotated in a clockwise direction through 2708 and 908, respectively (Fig. 1b,c). Both images were then shaded with a hypothetical light source positioned immediately above and perpendicular to the image plane and separately q 1999 The Royal Microscopical Society, Journal of Microscopy, 196, 347±351
S T E R E O S C O P I C D I SP LAY OF A F M I M AG E S
349
Fig. 2. An AFM image of a human hair shown in various formats. (a) A top-shaded view generated by arti®cially shading the image with a hypothetical light source in order to reveal the ®ne surface architecture of the cuticle, (b) an isomorphic projection which gains three-dimensional height information at the expense of introducing distortions in the relative distances between features, and (c) the anaglyph image that reveals the true three-dimensional relief of the surface, including good visualization of the ®ne surface detail of the sample. x y 20 mm, z 2´7 mm. q 1999 The Royal Microscopical Society, Journal of Microscopy, 196, 347±351
350
J. R. SMI TH E T AL .
displayed as isomorphic representations (Fig. 1d,e). This pseudo `three-dimensional' format was produced using a rotation angle of 2708 (clockwise) and an image tilt of 48, which had the effect of producing a top view format. The 48 tilt of each rendition allows for a total of 88 tilt in the ®nal anaglyph (stereo) image. In addition, the z-axis scaling was sometimes increased with respect to x and y to further enhance the ®nal perceived stereo effect. For the anaglyph images presented in this paper, a ratio of 5z: x: y was found to be most suitable. We believe that the perceived depth in three-dimensional images, such as anaglyphs, need not be the correct depth. Rather, it is more important to use suf®cient depth to illustrate a point in question, although any enhancement in the z-height should be announced. The left and right `isomorphic projections' of the images (datasets) were then rotated clockwise back through 908 and 2708, respectively, to the original orientation. These procedures produced left and right stereo pairs, which could be viewed stereoscopically when placed side-by-side (Fig. 1f,g). Anaglyph images (Fig. 1h) were produced using commercially available software, in this case ANAGLYPHÒ Version 0.1 (Junichi Masano, Japan). The software used the stereo image pairs to create red/green combination anaglyph images directly without the requirement of the user to have a detailed knowledge of colour lookup tables (Hodges & McAllister, 1985). Anaglyphs were examined using red/ green stereo viewers (Agar Scienti®c, Stansted, Essex, U.K.).
a large degree on details of the sample being imaged. The isomorphic projection in Fig. 2b goes some way to alleviating the step-direction problem, although this is at the expense of distorting within plane distances. The corresponding red/green combination anaglyph image, Fig. 2c, allows true three-dimensional observation of the ®ne surface architecture while preserving the relative height information necessary for correct interpretation of cuticle step-direction. A great wealth of ®ne surface detail can be readily appreciated from this single image (Swift & Smith, unpublished data), although the extent of this lies beyond the scope of this short note. While the perceived image is in many ways comparable to that observed from the alternative technique of viewing laterally separated stereo pairs with the aid of a stereoscope (not shown), the anaglyph method does have several advantages. For example, stereoscope observations require the two pictures to be suitably spaced and sized. In addition, anaglyphs can be viewed directly on a PC monitor, and in this way images can also be animated to provide virtual reality-type observation.
Acknowledgements We would like to thank Dr C. L. Gummer, Procter & Gamble Technical Centres Ltd, Egham, Surrey, and Mr D. Burder, 3D Images Ltd, Grange Park, London, for constructive discussions. J.R.S. also acknowledges The Royal Society for the provision of a conference travel grant.
Results and discussion
References
Fundamentally, the output from an AFM imaging experiment is a two-dimensional x,y-image containing an array of pixels that represent height in the z-direction. Often, the heights are encoded as grey levels, with brighter values proportional to greater heights. A typical `raw' AFM image is shown in Fig. 1a. There are several standard twodimensional `tricks' for showing three-dimensional (height) information from the raw AFM data. For the purposes of highlighting the problems caused by displaying AFM images in two-dimensional formats, Figs 2a and b show images of a human hair viewed directly from overhead and as an isomorphic projection, respectively. With this sample it was necessary to arti®cially shade the image with a hypothetical light source in order to reveal the ®ne surface architecture of the cuticle (Swift, 1991). However, the use of this shading causes the image to appear rather ¯at and in some cases it becomes dif®cult to see whether cuticular sheets overlap from top-left to bottom-right (correctly) or vice versa. Prior knowledge that rough endocuticle deposits (Fig. 2a, centre, bottom) sometimes remain close to the scale edge after the upper exocuticle has been removed help to solve the problem, although such features are not always observed. Thus, the shaded view approach is dependent to
Barber, V.C. & Brett, D.A.L. (1982) Colour bombardment: a human visual problem that interferes with the viewing of anaglyph stereo material. Scanning Electron Microsc. P2, 495±498. Cherkasov, Y.A. (1997) Optical terms and concepts for the section on photography (photographic processes and effects). J. Opt. Technol. 64, 872±879. Glen, R.C. (1985) In¯uence of background colour on the chirality of anaglyph (red/green) stereo diagrams. J. Mol. Graphics, 3, 135. Hodges, L.F. & McAllister, D.F. (1985) Stereo and alternating pair techniques for display of computer-generated images. IEEE Comput. Graph. Appl. 5, 38±45. Hudson, B. (1972) The application of stereo-techniques to electron micrographs. J. Microsc. 98, 396±401. Raspanti, M. (1999) Advanced visualization techniques for scanning probe microscopy. Microsc. Anal. 69, 13±15. Shao, Z. & Somlyo, A.P. (1995) Stereo representation of atomic force micrographs: optimizing the view. J. Microsc. 180, 186± 188. Smith, J.R. (1997) Use of atomic force microscopy for highresolution non-invasive structural studies of human hair. J. Soc. Cosmet. Chem. 48, 199±208. Smith, J.R. (1998) A quantitative method for analysing AFM images of the outer surfaces of human hair. J. Microsc. 191, 223±228. q 1999 The Royal Microscopical Society, Journal of Microscopy, 196, 347±351
S T E R E O S C O P I C D I SP LAY OF A F M I M AG E S
Smith, J.R., NahleÂ, A.H. & Walsh, F.C. (1997) Scanning probe microscopy studies of EbonexÒ electrodes. J. Appl. Electrochem. 27, 815±820. Swift, J.A. (1991) Fine details on the surface of human hair. Int. J. Cosmet. Sci. 13, 143±159. Turnnidge, R. & Pizzanelli, D. (1997) Methods of pre-visualizing
q 1999 The Royal Microscopical Society, Journal of Microscopy, 196, 347±351
351
temporal parallax suitable for making multiplex holograms. Part II: Greyscale and colour anaglyphs made in photoshop. Imaging Sci. J. 45, 43±44. Usery, E.L. (1993) Virtual stereo display techniques for threedimensional geographic data. Photogram. Engng. Remote Sens. 59, 1737±1744.
