COUNTRY SCIENTIST

BY FORREST M. MIMS III

How to Analyze Scientific Images

Digital cameras are among the most important instruments in my science tool kit. They have provided thousands of images of twilight glows, solar aureoles, clouds, tree rings, insects, vegetation, bacteria and mold colonies, and much more.

Most images speak for themselves and need no more processing than that applied by your eyes and brain. But what if you want to extract data from images? Analytical studies of photographs require more than simply describing, say, the color of a leaf or the brightness of a sunset — they require numbers.

Prior to the digital era, scientists relied on instruments called densitometers to extract data from photographs. Transmission densitometers convert the degrees of darkness (or density) of points on photographic film into a representative voltage or signal. One side of the film is illuminated by a light source, and the light that leaks through is detected by a photosensor on the opposite side. Photographic prints are digitized by placing both the light source and the detector on the same side of the image.

While densitometry is still used to extract data from photographic film and prints, these can now be easily digitized by what amounts to a new kind of densitometer: the flatbed scanner. And the data in digital images is already present in a form that can be easily analyzed by image processing software.

ImageJ: Image-Processing Software that's Powerful, Platform-Independent, and Free

Processing digital images to extract their data once required powerful, expensive software. Its high price prevented many students and amateur scientists from analyzing their images.

This changed in 1997 with the introduction of ImageJ, a public-domain image analysis program developed by Wayne Rasband at the National Institutes of Health (NIH). During its first decade, ImageJ became a powerful, platform-independent image analysis package that can be run on Linux, Macintosh, and Windows machines. I’ve run it on a variety of Windows machines, including a tiny Acer

42 Make: Volume 18

Aspire One running Win XP with only 1GB of RAM (and a 160GB hard drive).

ImageJ requires no license, and the program and its Java source code are freely available at the ImageJ homepage: rsbweb.nih.gov/ij/index.html.

Running ImageJ

After you download the version of ImageJ designed for your operating system, go ahead and select and run the program. A rectangular command bar will appear at the top of your computer screen (Figure A). This tiny but powerful startup menu is a toolbox that includes a set of 8 text selections over a row of 19 icons that point to menu options known as macros, some of which can be accessed immediately while others lead to still more macros.

ImageJ’s startup menu floats over the upper edge of whatever program was running when you clicked on it. You can use your mouse to drag it anywhere on your screen. If the toolbox disappears, simply click on the ImageJ icon on your task bar to place it back on your screen.

Doing Science with ImageJ

Let’s get started with ImageJ by analyzing a photograph taken with a digital camera. The image in Figure B reveals a glow around the sun known as the solar aureole. The sun itself is blocked by the occluder device I described in my previous column (see MAKE, Volume 17, page 48).

First, select File ⇒ Open and choose an image, in this case the JPEG in Figure B. After the image is displayed on the screen, maximize it. You’re now ready to analyze the image.

This image was taken on a clear day at solar noon, and it can tell us much about the presence of aerosols in the sky, such as smoke, dust, and haze. You can quickly see how — simply place the cursor over various parts of the image and watch the 5 numbers that appear below the row of toolbar icons.

The first 2 numbers indicate the x–y coordinates of the position of the cursor on the image. The next 3 numbers indicate the intensity of the red, green, and blue (RGB) wavelengths of light at that position, on a scale from 0 to 255. Placing the cursor over the