HIST-SI: HYBRID IMAGE SEGMENTATION TECHNIQUE FOR SATELLITE IMAGERY TO DECREASE THE SEGMENTATION ERROR RATE

Abstract

Image segmentation is the process in which an image is split into separate parts having the same properties which belong to the same objects. Various segmentation approaches have been proposed recently by eminent researchers. But, after concluding the detailed research, the authors have analyzed that most of the existing methods does not decrease the segmentation error rate. The aim of this research paper is to propose a hybrid image segmentation technique for satellite imagery to decrease the segmentation error rate. In this research cluster based and threshold based segmentation techniques are merged together. The goal of this research is to show how combining these two techniques for decrease the segmentation error rates produce much better results than others individual techniques. For testing the proposed technique, a dataset of Bhuvan – a National Geo-portal developed and hosted by ISRO (Indian Space Research Organisation) is used. Experiments are conducted using Scikit-image & OpenCV tools of Python and performance is evaluated and compared over various existing image segmentation techniques for several metrices i.e. Mean Square Error and Peak Signal Noise Ratio