Scholarlink

<p>A RANSAC estimator works by continuously generating the sample sets, and comparing the quality among the models computed from them. After multiple iterations, a so-far-the-best model will be chosen as the final output. In this paper, we re-examine the RANSAC used in image matching application, more precisely, in calculating homography matrices, and propose a RANSAC based algorithm, named LS-RANSAC (Local spatial structure-RANSAC). It innovatively introduces information from image pairs directly into the RANSAC estimator and extracts spatial structural connections between points from it. The proposed algorithm works as the local-optimization step, we evaluate our method on the Extreme view dataset (EVD Dataset) and it shows a competitive performance.</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

LS-RANSAC: find spatial structures directly from the image

  • 期刊论文

  • 发表日期：

    2025年3月20日

查看全文
  • Jiming Sa,
  • Liwei Ding,
  • Yechen Huang

摘要

A RANSAC estimator works by continuously generating the sample sets, and comparing the quality among the models computed from them. After multiple iterations, a so-far-the-best model will be chosen as the final output. In this paper, we re-examine the RANSAC used in image matching application, more precisely, in calculating homography matrices, and propose a RANSAC based algorithm, named LS-RANSAC (Local spatial structure-RANSAC). It innovatively introduces information from image pairs directly into the RANSAC estimator and extracts spatial structural connections between points from it. The proposed algorithm works as the local-optimization step, we evaluate our method on the Extreme view dataset (EVD Dataset) and it shows a competitive performance.