Projects‎ > ‎Root Gravitropism‎ > ‎

Image Segmentation

One of the problems arising in processing the many images generated in the root growth studies is identifying the root or roots in a particular image.  A second problem is the large amount of disk space required to story image sequences, since each file is of order 200 MB.  We have developed an algorithm and code that can identify parts of an image containing a root and create a time sequence of individual root images from a sequence of scanner images.

Figure 1. Original color scanner image

Figure 2. Grey scale scanner image

Figure 3. Segmented root images for left side of scanner image.
Figures 1-3 show the result of processing by the code.  The code starts with the original color scanner image, converts it to grey scale, then identifies the individual roots on the left and right sides placing each root in its own file.  The code will operate on a time sequence of images, placing the time sequence of individual root images in their own folder.  The final output is an archival file (zip) with individual folders for each root.

The code is written in Python, and is available under the GNU GENERAL PUBLIC LICENSE Version 3.  The file is available below. 

Last updated: 12/28/2012
Chris Wentworth,
Aug 2, 2013, 12:00 PM