Scholarlink

<p>Leaf recognition plays a crucial role in agricultural research and is a significant area of study in image processing. This study presents a leaf recognition system that leverages Principal Component Analysis (PCA) and morphological features. The system consists of five key phases: image acquisition, preprocessing, feature extraction, feature selection, and classification. During the image acquisition phase, leaf images are captured using a digital camera against a white paper background. The preprocessing phase applies various techniques to enhance image quality for feature extraction. Several morphological features, including average intensity, centroid, major axis length, minor axis length, and solidity, are extracted from different leaf categories. PCA is then employed to refine the feature vectors by retaining essential information. For classification, the system utilizes three classifiers: k-Nearest Neighbors (k-NN), Decision Tree, and Random Forest, with the AdaBoost technique applied to improve accuracy. The system was evaluated on a dataset of 400 images representing 10 leaf types, achieving a highest precision rate of 96.32%. Performance was further assessed using metrics such as Root Mean Square Error (RMSE) and False Acceptance Rate (FAR).</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

Leaf Recognition Using Principal Component Analysis and Morphological Features

  • 期刊论文

  • 发表日期：

    2026年3月6日

查看全文
  • Sakinah Mohd Shukri,
  • S. Srinadh Raju,
  • H. S. Shreenidhi,
  • Meenakshi Garg,
  • Mandeep Kaur Chohan,
  • Abhilasha Jadhav,
  • Ahmed Alkhayyat,
  • Sanjeev Kumar Shah

摘要

Leaf recognition plays a crucial role in agricultural research and is a significant area of study in image processing. This study presents a leaf recognition system that leverages Principal Component Analysis (PCA) and morphological features. The system consists of five key phases: image acquisition, preprocessing, feature extraction, feature selection, and classification. During the image acquisition phase, leaf images are captured using a digital camera against a white paper background. The preprocessing phase applies various techniques to enhance image quality for feature extraction. Several morphological features, including average intensity, centroid, major axis length, minor axis length, and solidity, are extracted from different leaf categories. PCA is then employed to refine the feature vectors by retaining essential information. For classification, the system utilizes three classifiers: k-Nearest Neighbors (k-NN), Decision Tree, and Random Forest, with the AdaBoost technique applied to improve accuracy. The system was evaluated on a dataset of 400 images representing 10 leaf types, achieving a highest precision rate of 96.32%. Performance was further assessed using metrics such as Root Mean Square Error (RMSE) and False Acceptance Rate (FAR).