Scholarlink

<p>This study proposes a novel classification algorithm for image data with three significant improvements. The first improvement involves image preprocessing using the Gamma filter to enhance image quality, convert images to grayscale, and extract 24 representative features for recognition. The second improvement estimates the prior probabilities based on a fuzzy cluster analysis technique. The final improvement constructs a representative probability density function (PDF) for each group and applies the Bayesian classification principle, integrating the outcomes of the previous two stages. The proposed algorithm is described in detail and can be implemented by a Python-based procedure. It is applied to classify two image datasets, and results across all evaluation metrics demonstrate that the proposed algorithm outperforms several existing methods, including both statistical and machine learning approaches.</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

Developing Bayesian classifier algorithm for images based on fuzzy relationship of extracted features

  • 期刊论文

  • 发表日期：

    2026年3月16日

查看全文
  • Hung Tran-Nam,
  • Toan Huynh-Phung,
  • Tai Vo-Van

摘要

This study proposes a novel classification algorithm for image data with three significant improvements. The first improvement involves image preprocessing using the Gamma filter to enhance image quality, convert images to grayscale, and extract 24 representative features for recognition. The second improvement estimates the prior probabilities based on a fuzzy cluster analysis technique. The final improvement constructs a representative probability density function (PDF) for each group and applies the Bayesian classification principle, integrating the outcomes of the previous two stages. The proposed algorithm is described in detail and can be implemented by a Python-based procedure. It is applied to classify two image datasets, and results across all evaluation metrics demonstrate that the proposed algorithm outperforms several existing methods, including both statistical and machine learning approaches.