Scholarlink

<p>Digital image correlation (DIC) is a fundamental technique for measuring deformations and analyzing material textures, particularly nano-textures. Traditional algorithms have demonstrated limited efficiency in the presence of complex deformations and high noise levels. To address these challenges, Frequency-adapted local correlation for nano-textures Model (FALCON) algorithm was developed, combining local Fourier transform with Hanning windows and adaptive frequency weighting. FALCON enables precise nano-texture analysis even under demanding conditions, providing more robust results compared to traditional methods. The experimental evaluation was conducted on standardized datasets with varying levels of noise and deformation, including translation, rotation, and scaling. Performance was measured using SNR, PSNR, SSIM, and Pearson correlation coefficients, while execution times allowed for a comparison of efficiency. Results demonstrated that FALCON maintains stable performance and accuracy, whereas ZNCC shows limitations at higher noise levels and with more complex deformations. An open-access GitHub repository has been provided to ensure reproducibility and facilitate further research.</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

Frequency-adapted local correlation for nano-textures model (FALCON)

  • 期刊论文

  • 发表日期：

    2025年9月26日

查看全文
  • Ratko Ivković,
  • Stefan Panić

摘要

Digital image correlation (DIC) is a fundamental technique for measuring deformations and analyzing material textures, particularly nano-textures. Traditional algorithms have demonstrated limited efficiency in the presence of complex deformations and high noise levels. To address these challenges, Frequency-adapted local correlation for nano-textures Model (FALCON) algorithm was developed, combining local Fourier transform with Hanning windows and adaptive frequency weighting. FALCON enables precise nano-texture analysis even under demanding conditions, providing more robust results compared to traditional methods. The experimental evaluation was conducted on standardized datasets with varying levels of noise and deformation, including translation, rotation, and scaling. Performance was measured using SNR, PSNR, SSIM, and Pearson correlation coefficients, while execution times allowed for a comparison of efficiency. Results demonstrated that FALCON maintains stable performance and accuracy, whereas ZNCC shows limitations at higher noise levels and with more complex deformations. An open-access GitHub repository has been provided to ensure reproducibility and facilitate further research.