Segmentation Of Interested Objects In Botanic Somatic Embryo Images

In the bioinformatics, we usually need to quantitatively analyze the changing pattern of specific macromolecules content in the development of embryonic at each period, therefore, automatically and accurately extracting the macromolecules from the plant cell image becomes necessary. Combining previous methods of extracting Region of Interest (ROI) in the biological cell image and considering the characteristics of plant cell image, here, it proposed a new polysaccharides segmentation method, which employed the Pulse-Coupled Neural Networks (PCNN) and the Vector Field Convolution (VFC) active contour model for rough and accurate segmentation respectively. In the paper ROI is polysaccharides in Botanic Somatic Embryo Images. This paper mainly focuses on the following three aspects:First, according to the shape and contour characteristics of polysaccharides, a method of roughly extracting ROI, based on the PCNN, is proposed. These images are firstly segmented by PCNN and labeled, and then morphologically processed to extract the approximate locations and contours of polysaccharides.Second, the locations and contours of polysaccharides are used to initialize VFC. Then the polysaccharides are accurately detected by the active contour model. This algorithm fully utilizes the characteristics of polysaccharides and the advantages of both PCNN and VFC attaining a good performance.Finally, to validate the algorithm, this study did simulation experiments on several different kinds of plant cell images. The missing detect rate, false detect rate, correlation coefficient, and root mean square error are used to evaluate the algorithm. The results show that the proposed method can better extract the polysaccharides with low missing rate and false rate.