Scholarlink

<p>Accurate estimation of weed diversity is crucial for agricultural sustainability and biodiversity conservation. Traditional morphological and ecological analyses often fail to detect subtle structural differences between species. This study proposes an integrative multivariate approach that combines Principal Component Analysis (PCA), Structural Similarity Index (SSIM), and Fast Fourier Transform (FFT) to enhance morphotaxonomic analysis of herbarium images. The PCA revealed that the first four principal components captured over 68.32% of the total variance, enabling effective dimensionality reduction. SSIM values across pairs of herbarium images ranged from 0.6594 to 0.7953, indicating moderate to high structural similarity. FFT-based phase analysis further distinguished weed species, with mean phase differences ranging from 0.0594 to 0.1396 radians, capturing fine-scale morphological variation. By integrating morphological, structural, and periodic information, the proposed method provides a more holistic and objective framework for weed identification and ecological studies.</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

Advancing weed morphotaxonomy through multivariate image-based analysis

  • 期刊论文

  • 发表日期：

    2025年7月25日

查看全文
  • S. N. Dhawale,
  • A. N. Deore,
  • N. B. Bhagat

摘要

Accurate estimation of weed diversity is crucial for agricultural sustainability and biodiversity conservation. Traditional morphological and ecological analyses often fail to detect subtle structural differences between species. This study proposes an integrative multivariate approach that combines Principal Component Analysis (PCA), Structural Similarity Index (SSIM), and Fast Fourier Transform (FFT) to enhance morphotaxonomic analysis of herbarium images. The PCA revealed that the first four principal components captured over 68.32% of the total variance, enabling effective dimensionality reduction. SSIM values across pairs of herbarium images ranged from 0.6594 to 0.7953, indicating moderate to high structural similarity. FFT-based phase analysis further distinguished weed species, with mean phase differences ranging from 0.0594 to 0.1396 radians, capturing fine-scale morphological variation. By integrating morphological, structural, and periodic information, the proposed method provides a more holistic and objective framework for weed identification and ecological studies.