AIC's 44th Annual Meeting

Reflectance Transformation Imaging (RTI) has become an important part of the documentation repertoire of many conservation laboratories. The ability to enhance details of surface shape and color helps in discerning surface information not otherwise easily visible. RTI is usually used to obtain qualitative data, such as reading difficult-to-see inscriptions and decorative details. We have been experimenting with combining RTI with image analysis for quantitative applications. Image analysis starts with algorithms that enhance visual separation of different features in an image and mark for analysis (in a process called ‘segmentation’) features of a specific color, contrast, size range, and/or morphology. Satisfactory segmentation is the core requirement for successful image analysis. Once this is achieved, a variety of quantitative data on those highlighted regions can be collected simultaneously. RTI plus image analysis is a natural coupling. Since quantitative analysis of surface features first requires the best possible segmentation, the enhanced surface detail produced by RTI is a clear advantage. One application we have been experimenting with is the use of RTI plus image analysis to obtain quantitative data on surface corrosion. The technique has been applied to coupons from Oddy tests, coated coupons artificially aged in a weatherometer, and metal sheets used for rapid corrosion tests. Oddy tests are used to assess compatibility of storage and display materials with metals found in collections. The test provides qualitative data as to whether a material is advisable for long-term use, for short-term exhibitions, or not at all. Reading the results of these exposure tests on coupon surfaces, however, can be tricky. For example, the British Museum has recommended that to reduce surface reflections from silver and copper coupons, a sheet of white paper should be held at an angle of approximately 60° to horizontal over the coupons while making assessments. Another difficulty is that control coupons themselves can change due to the elevated RH of the test. These changes have to be accounted for in making judgements about the degree of change in non-control coupons. RTI can improve the test by allowing qualitative assessments to be made under the best standardized viewing conditions. Adding image analysis allows surface effects on the controls to be subtracted from all other coupon images, and can add quantitative data on percentage of surface tarnish. We applied this process to weatherometer tests of coatings recommended for outdoor architectural brass. Image analysis gives the percentages of pitting and corrosion products present. However, using images obtained through RTI, rather than through typical photographic or scanning methods, provides more satisfactory results. We also used this approach to assess the results of rapid corrosion tests developed in industry to test the efficacy of corrosion inhibitors. Two indicators are important but cannot be assessed easily in one image: the percentage of surface area covered by corrosion products, and the degree of pitting attack, which has the effect of darkening the shiny, polished metal surface. Using the RTI viewer followed by image analysis these two indicators can be separated and quantified.