| Author: | Feng-I Liu |
| Advisor: | George C. Stockman |
| Email: | liufeng@cse.msu.edu; http://www.cse.msu.edu/~liufeng |
The microbes show different color when stained with special fluorescein. The color represents the current status of microbes for some important activity, for example, the metabolic activity and fission. Position, shape and area of those color regions are important for further biological analysis. The backgrounds of such images are usually very complex and noisy. The goal of this work is to classify pixels using the color information from a digitized color image, allowing the identification of semantically different regions. Since the input images are typically noisy, we first apply filters to smooth the images. Instead of providing a simple threshold, the proposed system offers an interactive environment to choose some sample points of interest. Based on both the color distance and the spatial distance from the points of interest, the system segments the microbes from the background. The system uses region-growing method for the foreground to find the boundary of the microbes. Some other methods are also applied to enhance the result. With the interactive environment, the user can easily select their target and the purposed system can segment the regions of interest in a confusing background.
Return to Workshop Listing