Geoscience Reference
In-Depth Information
above revealed significant peaks at about 13.1, 3.2, 2.2 and 1.0 yrs, suggesting
both ENSO and TAV inl uences in the area at around 33,000 years ago (see
Chapter 5 and Fig. 5.1).
8.10 Grain Size Analysis from Microscope Images
Identifying, measuring and counting particles in an image are the classic
applications of image analysis. Examples from the geosciences include
grain size analysis, counting pollen grains, and determining the mineral
composition of rocks from thin sections. For grain size analysis the task is to
identify individual particles, measure their sizes, and then count the number
of particles per size class. h e motivation to use image analysis is the ability to
perform automated analyses of large sets of samples in a short period of time
and at relatively low costs. h ree dif erent approaches are commonly used to
identify and count objects in an image: (1) region-based segmentation using
the watershed segmentation algorithm, (2) object detection using the Hough
transform, and (3) thresholding using color dif erences to separate objects.
Gonzalez, Woods and Eddins (2009) describe these methods in great detail
in the 2nd edition of their excellent topic, which also provides numerous
MATLAB recipes for image processing. h e topic has a companion webpage
at
http://www.imageprocessingplace.com/
that of ers additional support in a number of important areas (including
classroom presentations, M-i les, and sample images) as well as providing
numerous links to other educational resources. We will use two examples
to demonstrate the use of image processing for identifying, measuring, and
counting particles. In this section we will demonstrate an application of
watershed segmentation in grain size analysis and then in Section 8.9 we will
introduce thresholding as a method for quantifying charcoal in microscope
images. Both applications are implemented in the MATLAB-based RADIUS
sot ware developed by Klemens Seelos from the University of Mainz (Seelos
and Sirocko 2005). RADIUS is a particle-size measurement technique, based
on the evaluation of digital images from thin sections that of ers a sub-mm
sample resolution and allows sedimentation processes to be studied within
the medium silt to coarse sand size range. It is coupled with an automatic
pattern recognition system for identifying sedimentation processes within
undisturbed samples. h e MATLAB code for RADIUS can be downloaded
from
http://www.particle-analysis.info/
Search WWH ::




Custom Search