池宇希, 杨昊天, 潘兵. 增量数字图像相关方法的误差分析及应用考量. 力学与实践, 2023, 45(6): 1217-1226. doi:10.6052/1000-0879-23-275
引用本文: 池宇希, 杨昊天, 潘兵. 增量数字图像相关方法的误差分析及应用考量. 力学与实践, 2023, 45(6): 1217-1226.doi:10.6052/1000-0879-23-275
Chi Yuxi, Yang Haotian, Pan Bing. Error analysis and practical considerations of incremental digital image correlation. Mechanics in Engineering, 2023, 45(6): 1217-1226. doi:10.6052/1000-0879-23-275
Citation: Chi Yuxi, Yang Haotian, Pan Bing. Error analysis and practical considerations of incremental digital image correlation.Mechanics in Engineering, 2023, 45(6): 1217-1226.doi:10.6052/1000-0879-23-275

增量数字图像相关方法的误差分析及应用考量

ERROR ANALYSIS AND PRACTICAL CONSIDERATIONS OF INCREMENTAL DIGITAL IMAGE CORRELATION

  • 摘要:增量数字图像相关方法(digital image correlation, DIC)常应用于变形后图像中被测物体出现大变形、环境光变化剧烈以及采用激光散斑追踪变形等容易发生图像退相关的场景。增量DIC充分利用局部图像序列的相关性,通过更新参考图像计算增量变形,再进行累加以获得整体变形,可解决采用固定参考图像的传统DIC方法在出现图像退相关时计算失效的问题。但是,由于DIC计算变形时存在系统误差,在增量计算以累加增量位移的过程中会导致误差累积。为实现更准确高效的增量DIC分析,需明确这一误差的累积规律并提出抑制误差的方法。本文通过模拟和真实实验研究了采用不同增量策略的增量DIC方法在不同场景下的误差累积规律。结果显示:为减少参考图像更新带来的累积误差,应采用条件增量计算策略以尽量减少参考图像的更新次数;同时,自适应参考图像子区平移策略和图像高斯低通预滤波方法可以显著降低插值带来的系统误差,应与增量DIC方法配合使用。

    Abstract:Incremental digital image correlation (DIC) method is commonly used in scenarios prone to image decorrelation, such as when the measured object undergoes large deformations, strong illumination changes exist or laser speckle is used for deformation tracking. Incremental method utilized the correlation within the image sequence and update the reference image to calculate the incremental deformation, which is accumulated to obtain the total deformation, thus overcoming the failure of traditional methods that use a fixed reference image when decorrelation occurs. However, the systematic error exists when computing deformation using DIC, accumulating incremental displacements leads to error accumulation in incremental calculation. To perform efficient and accurate measurement incremental DIC analysis, it is necessary to clarify the accumulation law and propose the suppression methods of the cumulative error. In this work, through simulations and experiments, we investigate the cumulative error of incremental method with different incremental strategies in different application scenarios. Results show that to minimize the cumulative error caused by reference image updating, conditional updating strategy should be adopted to reduce reference image updating frequency. Additionally, the adaptive reference subset shifting strategy and the classic Gaussian low-pass pre-filtering can significantly reduce the systematic error caused by interpolation and should be used in conjunction with the incremental method.

/

    返回文章
    返回
      Baidu
      map