Scholarlink

<p>The existing image segmentation methods have problems such as poor noise processing ability, long time consumption, and inaccurate segmentation. To address these issues, this paper proposes an active contour model (ACM) that combines differential optimization algorithms and nonlinear Poisson’s equation, aiming to enhance the speed and accuracy of target contour segmentation. This model constructs a new nonlinear Poisson’s equation using the grayscale morphology method and optimizes the second-order edge difference term using the original image gradient pattern. This method enhances the reliability and segmentation accuracy of target edge detection, especially in the presence of non-uniform and intense noise. Experimental comparisons with deep learning-based models have demonstrated the superiority of our algorithm in edge extraction and robustness. In addition, this algorithm performs well in the component segmentation task of printed circuit boards.</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

Enhanced active contour model with adaptive thresholds and noise reduction for robust image segmentation

  • 期刊论文

  • 发表日期：

    2026年3月11日

查看全文
  • Guina Wang,
  • Xirui Feng,
  • Guirong Weng,
  • Yiyang Chen

摘要

The existing image segmentation methods have problems such as poor noise processing ability, long time consumption, and inaccurate segmentation. To address these issues, this paper proposes an active contour model (ACM) that combines differential optimization algorithms and nonlinear Poisson’s equation, aiming to enhance the speed and accuracy of target contour segmentation. This model constructs a new nonlinear Poisson’s equation using the grayscale morphology method and optimizes the second-order edge difference term using the original image gradient pattern. This method enhances the reliability and segmentation accuracy of target edge detection, especially in the presence of non-uniform and intense noise. Experimental comparisons with deep learning-based models have demonstrated the superiority of our algorithm in edge extraction and robustness. In addition, this algorithm performs well in the component segmentation task of printed circuit boards.