|Affiliation(s)||PI||Project period||Funded by|
|DEES||McGwire, Ken||05/01/2004 - 11/01/2009||NSF Office of Polar Programs|
This project has been developing novel methods for digital image analysis of glacial ice cores that are stored at the National Ice Core Laboratory (NICL) in Denver, Colorado. Imaging hardware and software techniques have been developed to meet specific needs in the scientific analysis of cores.
The NICL Optical Scanning System
The scanning camera operates on the core processing line within the NICL cold-room facility (-25°C) and is capable of scanning a typical ice core with an approximate resolution of 0.025 millimeters. This linear scanning system (Figure 1) solves numerous problems that have been experienced with simple photography of cores such as parallax in images collected with a stationary lens viewing over many degrees and the difficulty of uniformly lighting an entire core at once.
A touch-screen enabled front-end application was developed that allows the functions of the separate software packages for image scanning (ViewFinder, BetterLight, Inc.), the NICL inventory database (FileMaker Pro), image processing checks for image quality (Java-based image processing routines), and file management to be seamlessly integrated.
This project has also created an image processing software system specifically for analyzing ice core data. The image processing tools and user interface are modifications of the ImageJ image processing software that was developed for the National Institutes of Health. The ImageJ structure has been enhanced with processing functions that are specific to ice core analysis. This includes methods of metadata and image file organization, allowing the user to easily browse through the thousands of images that may be acquired for multiple ice core projects.
Examples of functions that specifically address the characteristics of ice core imagery include automatic mapping of fractured ice in images. This mask is used to screen "bad" areas out in other image processing routines (Figure 2).
Other feature extraction capabilities include methods for calculating the angle of ice layer inclination in the image and a melt-layer finder that automatically scans ice images for fine-scale melt features. An annotation tool (Figure 3) provides an efficient mechanism for sharing information among scientists. Image processing functions are tightly integrated with the annotation capabilities, so that a user can manually choose annotation features from the image, or plot a brightness profile of the ice in a separate window.
The software developed for this project is a modified form of the ImageJ program that was developed for the National Institutes of Health by Wayne Rasband. In addition, there are statistical components that are drawn from Michael Flanagan's Java Scientific Library. To run, install ImageJ and then replace the ij.jar file in the distribution with the version below.
The ImageJ 1.34 distribution for Windows is: ij134-setup.exe
Download the following jar file to replace ij.jar: ij.jar
A demonstration image database with images (1 mm resolution) from WAIS Divide and the Norwegian-American Traverse: Temporarily unavailable (expected online by 2/3/10).
Funding for this project was provided through a grant from the Office of Polar Programs of the National Science Foundation (NSF Grant OPP-0230149) in order to support the existing efforts of the staff at the U.S. Geological Survey's National Ice Core Laboratory in Denver, Colorado. This effort is also supported through collaborative efforts with Dr. Richard Alley of Pennsylvania State University.
Kenneth McGwire (DRI)
Kendrick Taylor (DRI)
Richard Alley (PSU)
Geoffrey Hargreaves (USGS NICL)