A flexible and accurate digital volume correlation method applicable to high-resolution volumetric images

Digital volume correlation (DVC) is a powerful technique for quantifying interior deformation within solid opaque materials and biological tissues. In the last two decades, great efforts have been made to improve the accuracy and efficiency of the DVC algorithm. However, there is still a lack of a flexible, robust and accurate version that can be efficiently implemented in personal computers with limited RAM. This paper proposes an advanced DVC method that can realize accurate full-field internal deformation measurement applicable to high-resolution volume images with up to billions of voxels. Specifically, a novel layer-wise reliability-guided displacement tracking strategy combined with dynamic data management is presented to guide the DVC computation from slice to slice. The displacements at specified calculation points in each layer are computed using the advanced 3D inverse-compositional Gauss-Newton algorithm with the complete initial guess of the deformation vector accurately predicted from the computed calculation points. Since only limited slices of interest in the reference and deformed volume images rather than the whole volume images are required, the DVC calculation can thus be efficiently implemented on personal computers. The flexibility, accuracy and efficiency of the presented DVC approach are demonstrated by analyzing computer-simulated and experimentally obtained high-resolution volume images.

数字体相关（DVC）是一种用于量化固体不透明材料和生物组织内部变形的强大技术。在过去的二十年中，人们付出了巨大努力来提高DVC算法的准确性和效率。然而，仍然缺乏一种灵活、稳健且准确的版本，能够在内存有限的个人计算机上高效实现。本文提出了一种先进的DVC方法，该方法能够实现适用于高达数十亿体素的高分辨率体图像的精确全场内部变形测量。具体而言，提出了一种新颖的分层可靠性引导位移跟踪策略与动态数据管理相结合的方法，以引导DVC逐层计算。每层中指定计算点的位移是使用先进的三维逆合成高斯 - 牛顿算法计算的，该算法根据已计算的计算点准确预测变形矢量的完整初始猜测。由于只需要参考体图像和变形体图像中有限的感兴趣层，而不是整个体图像，因此DVC计算可以在个人计算机上高效实现。通过分析计算机模拟和实验获得的高分辨率体图像，证明了所提出的DVC方法的灵活性、准确性和效率。