ARCHIVES
VOL. 12, ISSUE 6 (2025)
Enhancing cyclone monitoring through intensity interaction hypergraph segmentation and PSNR-optimized forecasting
Authors
Vandana Kumari, Ashish Chourey, Dr. Ritu Shrivastava, Dr. Rajiv Srivastava
Abstract
This study presents a novel hypergraph-based framework for the
segmentation and prediction of tropical cyclones using satellite imagery. Two
custom constructs—Intensity Neighborhood Hypergraph (INHG) and Intensity
Interaction Hypergraph (IIHG)—were developed to capture higher-order spatial
and intensity-based relationships within cyclone cloud structures. To address
uncertainties and overlapping cloud formations, a Dichotomous Logistic
Regression-Based Fuzzy Hypergraph (DLR-FH) classifier was implemented. The
proposed framework was evaluated using real satellite cyclone images sourced
from the publicly available INSAT-3D Infrared and RAW Cyclone Dataset
(2013–2021). The DLR-FH model was benchmarked against traditional Deep
Convolutional Neural Networks (CNN) and the Deviation-Angle Variance Technique
(DAV-T). It achieved superior results, with prediction accuracy up to 91%, a
PSNR improvement of 4.7 dB, a false positive rate reduction of ~19%, and a 35%
decrease in execution time compared to CNN.
Download
Pages:100-106
How to cite this article:
Vandana Kumari, Ashish Chourey, Dr. Ritu Shrivastava, Dr. Rajiv Srivastava "Enhancing cyclone monitoring through intensity interaction hypergraph segmentation and PSNR-optimized forecasting". International Journal of Multidisciplinary Research and Development, Vol 12, Issue 6, 2025, Pages 100-106
Download Author Certificate
Please enter the email address corresponding to this article submission to download your certificate.