SEGMENTATION OF ANTI-NUCLEAR ANTIBODY IMAGES BASED ON WATERSHEDS AND FAST REGION MERGING

In this paper, segmentation based on watersheds and fast region merging is proposed to be the main segmentation technique in this project. Segmentation of Anti-Nuclear Antibody (ANA) images is the data or input of this project by concentrating in four different patterns of ANA images. Watershed segmentation algorithms are combining edge and region based techniques. Furthermore, watershed segmentation algorithm has several drawbacks which are the image to be over segmented and separation of homogenous region. Therefore, fast region merging techniques are proposed and implement to overcome the problem. In addition, this technique is used to merge homogenous region based on several criteria which is intensity and some other homogenous parts. Moreover, in this paper, basic Gaussian filter is been implement to enhance the watershed segmentation by reducing some of the unwanted noise inside the images and highlighting the edge of cells. Gaussian filter is the fundamental technique in segmentation process of an image. This paper describes the watershed and fast region merging segmentation techniques to be implement in ANA images and the result will be the behaviour of each ANA images based on the selected pattern to be use in this paper.