Segmentation of Overlapping Elliptical Objects Using K-Mean Clustering Method

Main Article Content

Sukhjinder Kaur
Puneet Mittal

Abstract

The aim of the Segmentation of overlapping objects is to direct the issue of representation of multiple objects with partial views. In different applications, overlapping or impeded articles happen, for example, morphology analysis of atomic or cell objects in biomedical and industrial imagery in which by the size and shape of quantitative investigation of individual items is desired. We propose an algorithm specialized in ellipse detection technique using k-mean clustering to detect elliptical objects. The techniques used in this new method are very simple and intuitive. The new technique mainly consists of four steps: (1) Selecting input image; (2) detect the centre of the ellipse by using bounded erosion and fast radial symmetry called as detection of seed points (3) find the edge points on the ellipse by contour evidence detection using specialized ellipse detection technique with K-Means clustering; and (4) performing contour estimation method. The proposed work is performed using two different images which give promising results. The number of detected objects in image 1 using previous method is 26 objects and with proposed method are 16 objects. Whereas the number of detected objects in image 2 using previous method is 14 objects and with proposed method is 20 objects.

Downloads

Download data is not yet available.

Article Details

Section
Articles