Scholarlink

<p>Traditional level set image segmentation methods rely solely on global information, performing well on images with significant differences between targets and backgrounds. However, due to this characteristic, they often fail to effectively segment images with intensity inhomogeneity. To address this limitation, this paper improves the optimized LFCV model by incorporating the difference between the original image and the fitted image into its energy functional as a correction, thereby proposing a multi-scale level set image segmentation model. Additionally, Gaussian kernel smoothing is applied to significantly counteract noise interference. Furthermore, a genetic algorithm is introduced to achieve adaptive parameter selection. Experimental results demonstrate that this method achieves remarkable segmentation performance on intensity-inhomogeneous images, with segmentation accuracy far surpassing traditional level set algorithms and optimized level set methods.</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

Multi-scale level set image segmentation model based on genetic selection

  • 期刊论文

  • 发表日期：

    2025年4月17日

查看全文
  • Weiyueheng Chen,
  • Yibin Lu,
  • Dean Wu,
  • Wenyu Yuan

摘要

Traditional level set image segmentation methods rely solely on global information, performing well on images with significant differences between targets and backgrounds. However, due to this characteristic, they often fail to effectively segment images with intensity inhomogeneity. To address this limitation, this paper improves the optimized LFCV model by incorporating the difference between the original image and the fitted image into its energy functional as a correction, thereby proposing a multi-scale level set image segmentation model. Additionally, Gaussian kernel smoothing is applied to significantly counteract noise interference. Furthermore, a genetic algorithm is introduced to achieve adaptive parameter selection. Experimental results demonstrate that this method achieves remarkable segmentation performance on intensity-inhomogeneous images, with segmentation accuracy far surpassing traditional level set algorithms and optimized level set methods.