In order to provide high accuracy in shape measurement with multicore optical fibres, characterization and calibration procedures are an important part of sensor preparation. Some procedures can be considered mandatory for adequate shape reconstruction, while others can help to enhance the measurement accuracy. Several of such procedures are discussed and experimentally applied for demonstrating the possible performance enhancement of curvature sensing, a fundamental phase of the shape reconstruction process. The maximum error observed in curvature calculation for a test object has been proved to be almost halved, decreasing from 2.48% to 1.36%, by applying such calibration corrections. The overall average relative accuracy of curvature measurement was improved from 0.89% to 0.5% (an improvement of 44%).

为了使用多芯光纤在形状测量中提供高精度，表征和校准程序是传感器准备工作的重要组成部分。对于适当的形状重建，可以认为某些程序是必需的，而其他程序可以帮助提高测量精度。讨论了几种这样的程序，并在实验中应用了这些程序，以演示可能的性能增强（形状重构过程的基本阶段）曲率感测。通过应用此类校准校正，已证明在测试对象的曲率计算中观察到的最大误差几乎减少了一半，从2.48％降至1.36％。曲率测量的整体平均相对准确度从0.89％提高到0.5％（提高了44％）。