Research On Segmentation Technology Of Wall-pasted Cells Image

The automatic segmentation technology of cell image supplies a strong tool for clinic, but the phenomena of adhesive cells often appear . It is important for successive analysis and statistic to segment image . The segmentation methods of the wall-pasted cells image are studied. The scheme of Watershed algorithm is introduced to image segment according to the characteristic of wall-pasted cells.The distance transformation is introduced to the binary image given to look for the nearest distance between each pixel and the background pixels. We choose the Euclidean and use the distance transformation based on edge stripped. The situation information is transformed to gray-level information, and the binary image is transform to gray-level image.Watershed algorithm is applied to the distance image to search the adhesive pixels on the edge in order to realize the image segmentation.In order to test the superiority of the algorithm, the edge detection and the transformation of Top-hat are used to the image of the wall-pasted cells separately. Compared with the segmentation images, we get the conclusion that the watershed algorithm is qualified, steady and quick. It proves that the algorithm can separate the adhesive cells acutely.