Scholarlink

<p>Achieving precise 3D pose estimation in real-world scenarios using open-source stereo calibration techniques remains a significant challenge due to external factors such as camera positioning, calibration target size, and environmental noise. Existing calibration methods often rely on specialized hardware, controlled environments, or computationally intensive processes, limiting their practicality in low-resource settings. This paper proposes an optimized stereo calibration framework designed for conventional RGB cameras and standard calibration targets, ensuring both accuracy and efficiency in resource-constrained environments. We systematically analyze the influence of key calibration parameters, demonstrating that factors such as a 45<InlineEquation ID="IEq1"> <InlineMediaObject> <ImageObject Color="BlackWhite" FileRef="11042_2025_20846_Article_IEq1.gif" Format="GIF" Height="7" Rendition="HTML" Resolution="72" Type="Linedraw" Width="9" /> </InlineMediaObject> <EquationSource Format="TEX">\(^\circ \)</EquationSource> <EquationSource Format="MATHML"><math> <mmultiscripts> <mrow /> <mrow /> <mo>∘</mo> </mmultiscripts> </math></EquationSource> </InlineEquation> camera pair angle, A2 calibration board size, and still-image capture yield optimal results. Additionally, we introduce an improved evaluation method that produces a more accurate metric by implementing 3D point projection and 2D back-projection, both widely used in practical applications. This approach results in more consistent and reliable calibration estimations. Experimental results show a substantial reduction in 3D projection and 2D back-projection errors, significantly enhancing the accuracy of 3D pose estimation dataset generation, where these factors are crucial. Our method effectively identifies the best frame combinations for stereo calibration, ensuring precise 3D reconstruction and 2D back-projection from the world coordinate system (WCS). These advancements are particularly impactful for applications in robotics, augmented reality, and computer vision, where accurate and efficient pose estimation is critical, especially in low-resource environments that rely on standard RGB cameras and calibration targets.</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

Robust stereo calibration for improved 2D-3D projection in real-world pose estimation

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年5月6日

查看全文
  • Chris-Mari Schreuder,
  • Rensu P. Theart

摘要

Achieving precise 3D pose estimation in real-world scenarios using open-source stereo calibration techniques remains a significant challenge due to external factors such as camera positioning, calibration target size, and environmental noise. Existing calibration methods often rely on specialized hardware, controlled environments, or computationally intensive processes, limiting their practicality in low-resource settings. This paper proposes an optimized stereo calibration framework designed for conventional RGB cameras and standard calibration targets, ensuring both accuracy and efficiency in resource-constrained environments. We systematically analyze the influence of key calibration parameters, demonstrating that factors such as a 45 \(^\circ \) ∘ camera pair angle, A2 calibration board size, and still-image capture yield optimal results. Additionally, we introduce an improved evaluation method that produces a more accurate metric by implementing 3D point projection and 2D back-projection, both widely used in practical applications. This approach results in more consistent and reliable calibration estimations. Experimental results show a substantial reduction in 3D projection and 2D back-projection errors, significantly enhancing the accuracy of 3D pose estimation dataset generation, where these factors are crucial. Our method effectively identifies the best frame combinations for stereo calibration, ensuring precise 3D reconstruction and 2D back-projection from the world coordinate system (WCS). These advancements are particularly impactful for applications in robotics, augmented reality, and computer vision, where accurate and efficient pose estimation is critical, especially in low-resource environments that rely on standard RGB cameras and calibration targets.