Computer vision strain sensors typically require the camera position to be fixed, limiting measurements to surface deformations of structures at pixel‐level resolution. Also, sensors have a service term significantly shorter than the designed service term of the structures. This paper presents research on a high durable computer vision sensor, microimage strain sensing (MISS)‐Silica, which utilizes a smartphone connected to an endoscope for measurement. It is designed with a range of 0.05 ε, enabling full‐stage strain measurement from loading to failure of structures. The sensor does not require the camera to be fixed during measurements, laying the theoretical foundation for embedded computer vision sensors. Measurement accuracy is improved from pixel level to sub‐pixel level, with pixel‐based measurement errors around 8 µε (standard deviation approximately 7 µε) and sub‐pixel calculation errors around 6 µε (standard deviation approximately 5 µε). Sub‐pixel calculation has approximately 30% enhancement in measurement accuracy and stability. MISS‐Silica features easy data acquisition, high precision, and long service term, offering a promising method for long‐term measurement of both surface and internal structures.

中文翻译：

---

基于智能手机的高耐用应变传感器，具有亚像素级精度和可调节的相机位置


计算机视觉应变传感器通常需要固定相机位置，从而将测量限制在像素级分辨率下结构的表面变形。此外，传感器的服务期限明显短于结构的设计服务期限。本文介绍了一种高耐用性计算机视觉传感器——显微图像应变传感 （MISS）-二氧化硅的研究，该传感器利用连接到内窥镜的智能手机进行测量。其设计范围为 0.05 ε，可实现从负载到结构失效的全阶段应变测量。该传感器在测量过程中不需要固定相机，为嵌入式计算机视觉传感器奠定了理论基础。测量精度从像素级到亚像素级都有所提高，基于像素的测量误差约为 8 με（标准偏差约为 7 με），亚像素计算误差约为 6 με（标准偏差约为 5 με）。亚像素计算的测量精度和稳定性提高了大约 30%。MISS-Silica 具有易于数据采集、精度高、使用寿命长等特点，为表面和内部结构的长期测量提供了一种很有前途的方法。