开源工具Pyvale可快速处理千兆像素图像的二维数字图像相关分析。

Background: Digital Image Correlation (DIC) is a widely used full-field measurement technique, but both open-source and commercial packages often have limitations such as operating-system restrictions, lack of support for deployment on computing clusters, and poor scalability to gigapixel-scale images common in Scanning Electron Microscopy DIC (SEM-DIC). Objective: Pyvale is an open-source software package designed for sensor simulation, uncertainty quantification, placement optimization, and calibration/validation. A key component of this is the development of a dedicated 2D DIC module intended for standalone use and integration within broader workflows. Methods: Pyvale provides a user-friendly Python interface with performant compiled routines underneath. At its core is a multithreaded, reliability-guided DIC algorithm. Its open-source MIT license enables wide deployment, including on computing clusters and in automated pipelines. Results: Benchmarking with the publicly available 2D DIC challenge 2.0 dataset shows that Pyvale achieves metrological performance comparable to existing commercial and open-source DIC codes. It can correlate gigapixel-scale image pairs in under 5 minutes on high-specification desktop workstations, with memory peaking at approximately 50 GB. Conclusions: Pyvale's strong metrological foundation, coupled with its scalability for SEM-DIC, positions it as a platform for sustained, community-driven development. Its design and licensing provide a foundation for future improvements in open-source DIC and integration into experimental design and validation workflows.