Scholarlink

<p>Extracting waterbodies from satellite images is essential for environmental monitoring, hydrology and urban planning. Traditional algorithms for this task are often slow and inefficient, prompting the shift towards machine learning (ML) techniques. This study introduces a novel PO-SVM model that combines Support Vector Machine (SVM) with Particle Swarm Optimization (PSO) for enhanced waterbody classification in satellite imagery. Using LANDSAT images from South India, preprocessing is applied to improve quality, followed by splitting into training and testing datasets. Features are extracted using the Gray-level Co-occurrence Matrix (GLCM), and the SVM classifier is initialized with parameters such as kernel and regularization values. The PSO algorithm fine-tuned these parameters based on fitness criteria. The PO-SVM effectively classified water, achieving an impressive accuracy of 91.67%. It outperformed Cost-sensitive SVM, Potential SVM, Random and Grid search models, and LBP-SVM by significant margins, demonstrating its superior performance and efficiency.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Refining Hyperparameters of SVM for Enhancing Waterbody Classification in Satellite Imagery

  • 期刊论文

  • 发表日期：

    2025年10月13日

查看全文
  • S. Rajeswari,
  • P. Rathika

摘要

Extracting waterbodies from satellite images is essential for environmental monitoring, hydrology and urban planning. Traditional algorithms for this task are often slow and inefficient, prompting the shift towards machine learning (ML) techniques. This study introduces a novel PO-SVM model that combines Support Vector Machine (SVM) with Particle Swarm Optimization (PSO) for enhanced waterbody classification in satellite imagery. Using LANDSAT images from South India, preprocessing is applied to improve quality, followed by splitting into training and testing datasets. Features are extracted using the Gray-level Co-occurrence Matrix (GLCM), and the SVM classifier is initialized with parameters such as kernel and regularization values. The PSO algorithm fine-tuned these parameters based on fitness criteria. The PO-SVM effectively classified water, achieving an impressive accuracy of 91.67%. It outperformed Cost-sensitive SVM, Potential SVM, Random and Grid search models, and LBP-SVM by significant margins, demonstrating its superior performance and efficiency.