Catálogo de publicaciones - libros

High-resolution digital photography and acquisition equipment is becoming widely available and cost-effective. Additionally, SEM and related acquisition technologies provide the means for obtaining images with resolution containing both structural and elemental information. Despite the development of new acquisition techniques and hardware, selecting an appropriate imaging solution for a research problem remains a critical issue for sedimentary research. Moreover, despite the differences between acquisition technologies, all share common approaches and considerations in order to obtain the highest quality images for quantitative sedimentary analysis. Consideration should be made to account for artefacts in the acquisition process, particularly that sample shape and illumination variations are accounted for. Additionally, color or equivalent image standards should be included in each image to permit quantifying the properties between images. Finally, image resolution and file storage protocols should be appropriate for the research objective and preserve the original image information (i.e., no compression). Technological advances will undoubtedly create new image acquisition opportunities for sedimentary researchers. Additionally, evolving approaches for the automation of image acquisition will provide increasingly efficient methods to obtain data from long sedimentary sequences or with higher spatial resolution.

In this chapter we discussed image calibration, filtering, and processing techniques, which are used to prepare an image for subsequent data extraction and analysis. Size measurements from a digital image are calibrated by imaging objects with a known size. Pixel intensity is a measure for the composition of the imaged object and can be calibrated by imaging objects with known composition. Methods depend on the type of material and imaging technique. We discuss colour calibration, as colour is one of the most widely used types of data in image analysis. Filtering is performed on an image to remove artefacts that are unrelated to the object of study. The challenge is to find the best filter, one that removes all noise with minimum change to the actual information in the image. Described are techniques to remove the effects caused by uneven illumination during imaging, and methods to filter camera related noise. Image processing involves modification and/or enhancement of the image in such a way that the required numerical data can be extracted more easily. Processing techniques that are outlined include edge detection, segmentation, and processing of binary images.

Image analysis is probably among the most innovative tools of recent years and has gained major importance because of its wide circulation. A large set of tools for addressing image quantification problems is now available and helps solve problems in quantitative sedimentology, such as in the analysis of grains, matrices and porous networks. Nevertheless a sound use of the technique requires better education and a wider circulation of the mathematical background that is behind most concepts. Of particular interest are Stereology, Mathematical Morphology, Stochastic Geometry, Spatial Statistics, etc. This chapter explained basic notions of sampling theory linked to the field of image acquisition, and reviewed some of the most essential parameters available for proper characterization of the image content. A particular emphasis was put on the analysis of image intensities (grey levels, colors) and individual objects (size, shape, orientation). More advanced concepts related to structural or textural analysis could not find place in this chapter and are only briefly commented. The readers are referred to more specialized publications for the discussion on microstuctural analysis, 3-D measurements or stereological estimations.

Because image analysis is subject to errors, this chapter advocates for the systematic performance of quality controls of the results. Errors can be reduced to a minimum, but to do so it is necessary to identify their cause. Since an image is the result of a long series of sampling and operations, many factors can introduce error. Each parameter can be tested by slightly varying that parameter while maintaining the other constant and monitoring their impact on the final measurement. Preparation (or acquisition) errors are easy to test but require performing some empirical trials. Integration errors (or sampling errors) can be substantially limited if basic rules are respected (Table 6). Analysis errors (or errors due to processing the images) require that the transformation applied to the image is fully understood by the operator. There is a great risk of bias if a processing algorithm is used as a magic black box. Finally, efforts should also be made to try to validate the results with external methods and correctly educate geoscientists in image processing and analysis.

This chapter discusses procedures to obtain sediment colour time series from digital images of split core surfaces. This method is ideally suited to obtain sub-mm-scale colour records from laminated sediments. The most essential part of data acquisition to obtain good quality images, is that sediment surfaces are scraped carefully to remove any surface irregularities. After acquisition, line scans have to be collected through all images along the stratigraphic axis. The line scan data are subsequently corrected for uneven light distribution of the light source. Various methods for light correction are described. Examples are presented to illustrate application of the techniques especially to laminated sediments, and to compare the results with those obtained from photospectrometer data and X-radiographs.

Grayscale image analysis provides a useful means of extracting both stratigraphic depth series and quantitative data useful for compositional analysis of sedimentary cores and lithologic sections. For quantitative application of the method, the user should always consider the context of the complete imaging system (acquisition, processing, storage, output) from which the data arose. Our results demonstrate that bias of mid-tone grayscale values is more sensitive than bias of highlights or shadows. When corrections are applied to the grayscale values that take this nonlinearity into account, the resulting grayscale values compare favorably with sediment carbonate content, a strong influence on sediment brightness in the sediments that were studied. Future implementation of grayscale image analysis will benefit from comparative measurement of sample and grayscale standards to account for this source of bias.

High-resolution time-scales are important for the precise correlation of spatially distributed geological records, and further development of process-oriented models used to predict climate change and other terrestrial processes. The extraction of digital line-scan data from images of laminated sediments provides a tool for the rapid and non-invasive analysis of sedimentary records, including sediment and ice cores, and tree ring growth patterns.