A Segmentation Method Based on Standard Difference Gradient with Dual-Threshold for Urinary Sediment Visible Components

As the urinary sediment images have low contrast and blur edges meanwhile the visible components are complicated and there are a lot of overlapped cells in them, it is difficult to perform accurate and efficient segmentation. This paper proposes a urinary sediment image segmentation method based on standard deviation gradient with dual-threshold. First the morphological filtering is done to the image, then the segmented image is obtained by the standard deviation gradient segmentation based on the double threshold, at last the add operation is executed between the segmented image and the edges image got by the edge detection to achieve adhesion cell division. The actual urinary sediment image segmentation results show that this method can divide all the physical components of adhesion cell division more effectively and completely, and can avoid over-segmentation as while, lay the foundation for the subsequent component identification.