Scholarlink

This study investigates the applicability of Multilevel Thresholding and Multi-Otsu segmentation algorithms in hyperspectral imaging with detailed analysis using Pines, Pavia University, Salinas, and Multispectral Landsat 8 OLI Imagery of the Saptamukhi region in Sundarbans, West Bengal. The results show that the Multi-Otsu and Multilevel Thresholding algorithms encounter the highest Peak Signal-to-Noise Ratio (PSNR) values 74.82 and 74.16 respectively for the Indian Pines hyperspectral dataset. The lowest Mean-Squared Error (MSE) and Root Mean-Squared Error (RMSE) values 3.36 MSE and 1.83 RMSE for Multi-Otsu segmentation, 19.28 MSE and 4.40 RMSE for Multilevel Thresholding, are obtained for Pavia University data. To achieve accurate evaluation in the specified spectral context, optimal threshold values have been defined for each spectral band for both algorithms, which creates a balance between the accuracy matrices applied. This work aims to compare and contrast the effectiveness of these algorithms hyperspectral and multispectral data with a discussion of relevant optimization techniques.

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

Enhancing Spatial-Spectral Analysis of Hyperspectral and Multispectral Satellite Imagery with Multilevel Thresholding and Multi-Otsu Segmentation

  • 章节

  • 发表日期：

    2025年

查看全文
  • Anindita Das Bhattacharjee,
  • Srijita Sarkar,
  • Snehil Dey

摘要

This study investigates the applicability of Multilevel Thresholding and Multi-Otsu segmentation algorithms in hyperspectral imaging with detailed analysis using Pines, Pavia University, Salinas, and Multispectral Landsat 8 OLI Imagery of the Saptamukhi region in Sundarbans, West Bengal. The results show that the Multi-Otsu and Multilevel Thresholding algorithms encounter the highest Peak Signal-to-Noise Ratio (PSNR) values 74.82 and 74.16 respectively for the Indian Pines hyperspectral dataset. The lowest Mean-Squared Error (MSE) and Root Mean-Squared Error (RMSE) values 3.36 MSE and 1.83 RMSE for Multi-Otsu segmentation, 19.28 MSE and 4.40 RMSE for Multilevel Thresholding, are obtained for Pavia University data. To achieve accurate evaluation in the specified spectral context, optimal threshold values have been defined for each spectral band for both algorithms, which creates a balance between the accuracy matrices applied. This work aims to compare and contrast the effectiveness of these algorithms hyperspectral and multispectral data with a discussion of relevant optimization techniques.