Scholarlink

In the arena of pattern recognition, surgical facial recognition (SFR) is an important research topic. The surgical face, which is unaltered by plastic surgery, has been difficult for some algorithms to recognize because of differences in skin texture and illumination. The issue of enucleation operations is difficult, but it needs to be re-examined from both hypothetical and investigational perspectives. This study examines the effectiveness of support vector machines (SVM) for classification, PCA-based methods for dimensionality reduction, and textural image features for surgical face identification. This method is established mainly in two phases. First is all surgical training image converts into textural image features (i.e LBP and LTP) images. After that PCA and 2DPCA applied for dimension reduction as well as feature extraction. Textural characteristics are used in this investigation because they are strongly linked to improvements in surgical face recognition. Various global and local enucleation operations were among the techniques that were tried on 900 plastic surgery faces. The investigational results demonstrate that the LTP with 2DPCA procedure achieves competitive performance, outperforming other approaches in plastic surgery database. Moreover, our methods explain the strengths, restrictions, and probable areas for enhancement.

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

Performance Analysis on Classification of Surgical Face Using Textural Features and PCA

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Kakoli Dey,
  • Satadal Chakraborty,
  • Shiladitya Chowdhury,
  • Aniruddha Dey

摘要

In the arena of pattern recognition, surgical facial recognition (SFR) is an important research topic. The surgical face, which is unaltered by plastic surgery, has been difficult for some algorithms to recognize because of differences in skin texture and illumination. The issue of enucleation operations is difficult, but it needs to be re-examined from both hypothetical and investigational perspectives. This study examines the effectiveness of support vector machines (SVM) for classification, PCA-based methods for dimensionality reduction, and textural image features for surgical face identification. This method is established mainly in two phases. First is all surgical training image converts into textural image features (i.e LBP and LTP) images. After that PCA and 2DPCA applied for dimension reduction as well as feature extraction. Textural characteristics are used in this investigation because they are strongly linked to improvements in surgical face recognition. Various global and local enucleation operations were among the techniques that were tried on 900 plastic surgery faces. The investigational results demonstrate that the LTP with 2DPCA procedure achieves competitive performance, outperforming other approaches in plastic surgery database. Moreover, our methods explain the strengths, restrictions, and probable areas for enhancement.