Overlapping Cervical Cell Image Segmentation Based on Bottleneck Detection and Watershed Algorithm

This study proposes an image segmentation method based on bottleneck detection and watershed algorithm to solve the problem of overlapping cervical cell image. First, we use polygon approximation to get all feature points on the cell contour and then use bottleneck detection and ellipse fitting to obtain the correct split point pairs. Therefore, the approximate range of the overlapping region was determined. The watershed algorithm was used to obtain the internal boundary information for the gradient image of the region. Finally, the segmentation results of the overlapped cells were obtained by superimposing with the outer contour. The experimental results show that this algorithm can segment the contour of a single cell from the overlapping cervical cell images with good accuracy and integrity. The segmentation result is close to that of doctors' manual marking, and the segmentation result is better than other existing algorithms.