An Innovative Runway Landing Path Detection using UAV Implementation of the K-Means Clustering Algorithm

Abstract

Objective: To provide a novel approach for automatic Unmanned Aerial Vehicle (UAV) runway detection, leveraging remote sensing data and advanced image processing techniques. Methods: The methodology encompasses Gaussian filter-based despeckling and histogram equalization for preprocessing, followed by Independent Component Analysis (ICA) for feature extraction and segmentation using the K-means clustering algorithm. Findings: The research demonstrates successful UAV runway detection, even with unlabeled datasets, underscoring the efficacy of the proposed methods. Notably, the study contributes to automatic target recognition, specifically in Synthetic Aperture Radar (SAR) data analysis, where K-means clustering outperforms Korn B and morphological algorithms. Novelty : The K-means algorithms works by clustering the datasets obtained by integrating all the data collected from various sensors that are placed at specific positions in the runway. This work holds significance in facilitating immediate runway identification during emergencies and finds applications in military operations, surveillance, and remote sensing domains. Keywords: Runway detection, Unmanned Aerial Vehicle, Histogram Equalization, Gaussian filtering, Independent Component Analysis, K-means clustering based segmentation