Abstract

Satellite-based resource classification is a critical task in remote sensing, enabling efficient resource management and monitoring for environmental and developmental applications. Traditional classification techniques often struggle to balance accuracy and computational efficiency due to the complexity and high dimensionality of satellite imagery data. The proposed approach integrates the Hybrid Curvelet Transform (HCT) with the K-Nearest Neighbor (KNN) algorithm to address these challenges effectively. The Hybrid Curvelet Transform is utilized to enhance image feature extraction by capturing both multiscale and multidirectional information, enabling better edge preservation and noise reduction. Subsequently, the K-Nearest Neighbor algorithm is employed for classification due to its simplicity and effectiveness in handling non-linear data patterns. The methodology was tested on a satellite image dataset comprising 1000 samples, categorized into five resource classes: water bodies, vegetation, urban areas, barren lands, and snow cover. The proposed approach achieved an accuracy of 96.8%, outperforming traditional methods such as Principal Component Analysis (PCA) and Support Vector Machines (SVM), which achieved 88.5% and 92.3% accuracy, respectively. Additionally, the hybrid approach demonstrated a classification precision of 95.4%, recall of 96.2%, and F1-score of 96.1%. The computational time for classification was reduced by 15%, indicating the approach's efficiency in processing large satellite datasets.

Authors

Priusha Narwariya, Susheel kumar Tiwari
Madhyanchal Professional University, India

Keywords

Satellite Imagery, Curvelet Transform, K-Nearest Neighbor, Resource Classification, Remote Sensing