Cell Classification Using Shape Analysis
In this paper, we present a new method of cell classification using geometrical features of shape analysis. Firstly, the contour of cell is detected and approximated by one polygon. Secondly, by using the bounded polygon, the shape of cell is normalized. Then, the cell is classified into three kinds of cells: bull sperm cell (B-SC), human fibroblast cell (H-FC), and human mesenchymal stem cell (H-MSC) by using the structure of bounded polygon. Our approach overcomes the difficulty caused by the similarity between three kinds of cells when their shapes are unpredictably deformed in microscope image. We have applied our proposed method into data collected by one microscope with 40x eyepiece lens.
KeywordsBinary image Otsu method contour Douglas Peuker algorithm
Unable to display preview. Download preview PDF.