Computational Imaging, Centrum Wiskunde en Informatica, Amsterdam, The Netherlands.
Faculteit Wiskunde en Informatica, Technical University Eindhoven, Eindhoven, The Netherlands.
J Xray Sci Technol. 2024;32(4):1099-1119. doi: 10.3233/XST-230389.
X-ray imaging is widely used for the non-destructive detection of defects in industrial products on a conveyor belt. In-line detection requires highly accurate, robust, and fast algorithms. Deep Convolutional Neural Networks (DCNNs) satisfy these requirements when a large amount of labeled data is available. To overcome the challenge of collecting these data, different methods of X-ray image generation are considered.
Depending on the desired degree of similarity to real data, different physical effects should either be simulated or can be ignored. X-ray scattering is known to be computationally expensive to simulate, and this effect can greatly affect the accuracy of a generated X-ray image. We aim to quantitatively evaluate the effect of scattering on defect detection.
Monte-Carlo simulation is used to generate X-ray scattering distribution. DCNNs are trained on the data with and without scattering and applied to the same test datasets. Probability of Detection (POD) curves are computed to compare their performance, characterized by the size of the smallest detectable defect.
We apply the methodology to a model problem of defect detection in cylinders. When trained on data without scattering, DCNNs reliably detect defects larger than 1.3 mm, and using data with scattering improves performance by less than 5%. If the analysis is performed on the cases with large scattering-to-primary ratio (1 < SPR < 5), the difference in performance could reach 15% (approx. 0.4 mm).
Excluding the scattering signal from the training data has the largest effect on the smallest detectable defects, and the difference decreases for larger defects. The scattering-to-primary ratio has a significant effect on detection performance and the required accuracy of data generation.
X 射线成像广泛应用于输送带工业产品的无损检测。在线检测需要高精度、鲁棒性和快速的算法。当有大量标记数据可用时，深度卷积神经网络（DCNN）满足这些要求。为了克服收集这些数据的挑战，考虑了不同的 X 射线图像生成方法。
根据对真实数据的相似程度的要求，可以模拟或忽略不同的物理效应。已知 X 射线散射的模拟计算成本很高，并且这种效应会极大地影响生成的 X 射线图像的准确性。我们旨在定量评估散射对缺陷检测的影响。
使用蒙特卡罗模拟生成 X 射线散射分布。在有和没有散射的情况下对 DCNN 进行训练，并将其应用于相同的测试数据集。计算概率检测（POD）曲线以比较它们的性能，其特征是最小可检测缺陷的大小。
我们将该方法应用于圆柱体缺陷检测的模型问题。当在没有散射的情况下对数据进行训练时，DCNN 可以可靠地检测到大于 1.3 毫米的缺陷，而使用带有散射的数据集只能提高不到 5%的性能。如果对散射与初级比（1<SPR<5）较大的情况进行分析，性能差异可能达到 15%（约 0.4 毫米）。
从训练数据中排除散射信号对最小可检测缺陷的影响最大，而对于较大的缺陷，这种差异会减小。散射与初级比对检测性能和数据生成的精度要求有重大影响。
