IRJET- Implementation of Histogram based Tsallis Entropic Thresholding Segmentation for Plasma Detection from Visible Images of TOKAMAK

This document discusses image segmentation techniques for plasma detection in visible images of tokamaks. It compares Gray Level Local Variance (GLLV), Gray Level Local Entropy (GLLE), and Gray Level Spatial Correlation (GLSC) based 2D histogram segmentation methods using Tsallis entropy thresholding. These methods construct 2D histograms using pixel gray levels combined with local variance, entropy, or spatial correlation features. The document implements these methods on visible tokamak images and evaluates the results using an unsupervised uniformity value metric. It finds that the GLSC method provides better segmentation in terms of uniformity value compared to the other techniques.