- Skype with Dr. Weiss:
- Suggested that we ask Nikon sales rep. to download the Demo NIS-Element software to go with an updated Nikon microscope camera DS-Qi2, and DS-L3 controller. These new cameras are expected to arrive May 26th and Dr. Weiss wanted us to re-photograph images using the new cameras to fix the shadowing issues and to be able to stich less images together (maximum image panels would be 4×4)
- Learned and practiced how to use Color Threshold Plugin
- Segment the image on ImageJ based on the different colored dyes within the soybean by manipulating the Hue, Saturation, and Brightness or the Red, Green, Blue adjustments
- We were able to completely remove the pith using HSB
- Issue: There were areas within the stem that were the same color and were not able to be removed (also caused the segmented area to look faded).
- Learned and practiced how to use Trainable Weka Segmentation Plugin
- Worked on familiarizing using the plugin by watching and following along with a tutorial. This plugin “combines a collection of machine learning algorithms with a set of selected image features to produce pixel-based segmentations” (http://imagej.net/Trainable_Weka_Segmentation). In other words, this plugin uses algorithms and data analysis to predict segmentation of certain tissues based on the pixel density and color of our image.
- We had more success using Weka than the Color Threshold
- Issues: We were only able to segment the xylem and the fiber tissues.
- (I forgot to save the picture as an example). This segmentation made certain tissues disappear such as the pith, phloem and cortex leaving out only the xylem and fiber tissues. This is a problem because we would eventually need to measure the area of the lost tissues and we were not able to find a way to make them reappear.