Investigations on Color Microfilm as a Medium for Long-Term Storage of Digital Data

Digital data storage on microfilm is a promising alternative for long-term storage of digital data. Its estimated lifetime of up to 500 years and the availability of reading devices allow entirely new migration-free storage concepts. This paper presents investigations on the suitability of color microfilm as a medium for digital data storage being an alternative to conventional black-and-white film material. The main question we address is whether the advantage of three color channels justifies higher efforts and expenses related to this material. Therefore, an analysis based on several exposed test patterns has been performed. It turned out that the regarded film in combination with the employed exposure setup is very differently capable of storing data points depending on the color layer. Although black-and-white film material has several advantages in our opinion, special cases are pointed out where the use of color microfilm for digital data storage is attractive. Introduction In the past few years, microfilm has become an attractive medium for long-term storage of digital data. As an alternative to conventional storage media such as DVDs, CDs, hard drives, or magnetic tapes it offers estimated lifetimes of up to 500 years, depending on the specific film material and storage conditions (see, e.g., [1, 2]). Further advantages of this novel technology are the possibility to use common reading devices (film scanners) as well as the high data integrity level due to the genuine WORM (write once read many) character of the medium itself. There have already been various investigations on microfilmbased digital data storage with respect to signal and information processing [3], hardware aspects [4], channel modelling [5], modulation coding [6], as well as storage capacities and error correction coding [7, 8]. A detailed review of this technology including an overview of relevant microfilm standards and related systems is provided in [8]. Laser recording technology (see, e.g., [4, 9, 10]) is widely used for digital data storage on microfilm to expose the film material. Systems involving three separately modulated laser beams can be employed to expose both black-and-white as well as color microfilms. However, recent scientific contributions mostly focus on black-and-white microfilm material for digital data storage applications. Furthermore, the laser recording system described in [4] has been developed for high-density exposure of digital data on blackand-white microfilm with high data transfer rates. Of course, this material has several advantages compared to color microfilm, such as the lower price or the less complex photochemical processing. On the other hand, there are also various motivations for the use of color film material for data storage on microfilm. Firstly, all three color layers can be utilized leading to possible storage capacity improvements. Secondly, for hybrid storage of digital and analog data on the same medium (see [8]), the digital data can be stored along with color images on the same film. This is also an interesting solution for the storage of digital metadata. Some approaches to digital data storage on microfilm already make use of the advantages of color microfilm. Concerning the storage capacity for digital information we encounter the practical problem that the resulting characteristics of the color layers cannot be regarded as independent. The reason for this phenomenon is spectral overlap of the employed dyes leading to mutual influences, similar to so-called crosstalk in a communications system. Also, the optical properties can be different for each color layer. This contribution provides an analysis dealing with the possibilities offered by color microfilm for digital data storage on film. Therefore, as a starting point, the relevant basic principles as well as the physical background of color microfilm are regarded. Important differences to monochrome film material are identified with respect to data storage applications. The second step is an analysis of dedicated test patterns that have been exposed to color microfilm material by means of a laser recorder. Bit error rates are measured and serve as an objective criterion to compare different alternatives for exposing the data points. A research microscope equipped with a high-resolution camera allows accurate imaging of these samples at various resolutions and therefore serves as a reference reading device. Finally, aim of this contribution is a set of recommendations regarding the use of color microfilm for digital data storage on microfilm based on the described analysis and experiments. Data Storage on Microfilm Current approaches to data storage on microfilm are based on laser recording technology (see, e.g., [5, 8]). Therefore, tiny data points are exposed to the film by means of a modulated laser beam that is moved over the film material. These data points represent logical ones or zeros, respectively. The grid space d, i.e., the distance between the exposure points is a crucial factor for the storage capacity. Furthermore, amplitude modulation can be used to store more than one bit within a single data point. A detailed analysis concerning the influence of these factors on the storage capacity and a discussion of the resulting trade-offs can be found in [7, 8]. These investigations show that a small grid space d and binary modulation is a reasonable choice. Accordingly, we focus on this type of modulation in this paper. Several laser recording devices exist for microfilm featuring different technical specifications (see, e.g., [4, 9, 10]). All film samples used for the analysis in this publication are exposed with the Arche laser recorder [10] as a state-of-the-art laser recoding device that is capable of handling both black-and-white as well as color microfilm. Besides laser film recording, there are also alternative technologies to computer output microfilm devices (COM) that are 142 Society for Imaging Science and Technology not in further focus of this paper (see, e.g., [11]). The photosensitive layer of black-and-white film material (see, e.g., [1, 2]) basically consists of silver halide crystals (so-called grains) in gelatine. Simplified, during the exposure process, the laser beam causes photochemical reactions in which parts of the silver halide ions react to metallic silver and halide atoms. A chemical development process serves to transform the silver halide crystals containing such metallic silver atoms completely to metallic silver whereas the other grains remain unaffected – at least in the ideal case. As the remaining silver halide grains are still sensitive to light, a fixing process is required to remove them from the film material. However, the silver grains are not – also at least in the ideal case – affected by the fixing process and form the stable photographic image. For a detailed description of the photographic process please see [11, 12]. When regarding these photochemical reactions, three fundamental facts should be emphasized with respect to digital data storage on black-and-white microfilm: First of all, the photographic image consists of metallic silver grains ensuring a high degree of stability. As an example, the materials described in [1, 2] are expected to achieve a life expectancy of about 500 years. Secondly, the underlying photographic image formation process is well-understood and extensively investigated. Finally, it should be noted that both the material and the photographic process will presumably be available in the foreseeable future. However, no color can be reproduced with this kind of film material. Also, the image formation during this process is negative, i.e., formerly exposed places appear black on the film and unexposed places appear white or transparent, respectively. Anyway, for storing data points in the context of data storage on film applications this is actually not relevant. The photographical reproduction of color requires both more complex film materials and chemical processes as described in the next section. Color Microfilm As opposed to black-and-white microfilm, color microfilm generally consists of several color layers. A widespread positive color microfilm material is Ilfochrome R © Micrographic [13] that can be exposed, e.g., with the Arche laser recorder. It is available in two versions with different contrast, Type M and Type P, respectively, both being positive films. The Type M material exhibits a higher contrast compared to Type P and is the basis for the investigations within this paper. Both types of this positive film are processed in the P-5 process that mainly consists of three baths: developer, bleach, and fix [13] (as opposed to two baths, developer and fix, for the abovementioned black-and-white process). Although the image formation process for Ilfochrome R © Micrographic also involves silver halides, it is much more complex since the actual image finally consists of organic dyes and the silver is removed during the bleaching process (see, e.g, [12, 13] for more details on color film processing). Basically, there are three sensitive color layers for reproducing the colors blue, green, and red. Although the image for the Ilfochrome R © Micrographic film is no longer composed of silver atoms – as it is for the above-mentioned black-and-white microfilm – but merely of organic dye molecules, it also provides an excellent long-term stability (see, e.g., [14]). For traditional archiving of analog images, color microfilm has the clear advantage of preserving color information of colored documents, such as paintings, photographs, sketches, or drawings. When Table 1: Normalized spectral dye densities Dr(λ),Dg(λ), and Db(λ) at different wavelengths λ (values coarsely reconstructed from a figure in [13]). Db(λ ) Dg(λ ) Dr(λ ) λ = λb = 425nm 1.00 0.20 0.05 λ = λg = 570nm 0.00 1.00 0.45 λ = λr = 635nm 0.00 0.05 1.00 λ = λr2 = 685nm 0.00 0.00 0.95 λ = λFb = 429nm 1.00 0.20 0.05 λ = λFg = 529nm 0.05 0.80 0.20 λ = λFr = 672nm 0.00 0.00 0.90 using laser film recorders or other COM devices, these images can be directly exposed to the film. On the other hand, compared to black-and-white microfilm, color microfilm is more expensive and – as already described – also the chemical processing is more complicated [13]. Accordingly, if color is not relevant, it is reasonable to use black-and-white microfilm instead. When storing digital data in form of data points on color film it is straightforward to exploit the multiple color layers and use blue, green, and red data points simultaneously. However, it has to be taken into account that the color layers cannot necessarily be assumed as independent. When regarding the normalized spectral dye densities Db(λ ), Dg(λ ), and Dr(λ ) (with Db(λ ),Dg(λ ),Dr(λ ) ∈ [0,1]) of the Ilfochrome R © Micrographic film depending on the light wavelength λ in [13], it is obvious that there is a certain overlap of the spectral characteristics. The blue-sensitive dye has its maximum density at a light wavelength of about λb = 425nm, the greensensitive at approximately λg = 570nm, and the red-sensitive dye has its maximum density at around λr = 635nm as well as a second local maximum with a slightly lower density level at about λr2 = 685nm. All values Db(λ ), Dg(λ ), and Dr(λ ) are provided in Table 1 for these wavelengths. Especially the blue-sensitive color layer is influenced by green and also red. Furthermore, the green characteristics are significantly influenced by the red-sensitive dye. This effect is similar to crosstalk in a communications system [15] and is referred to as spectral overlap in the following.