Arbitrary internal depth slice imaging is an effective way to realize rapid industrial scanning and reconstruction method in x-rays imaging. This paper proposes a novel arbitrary internal depth slice imaging (AIDSI) method for cone beam x-rays imaging. AIDSI builds the relationships between voxels of depth slices and pixels of sampling images under the cone beam x-ray source, which serves to refocus sampling data to specified depth slices. To remove artifacts of direct refocusing, a 3-point weighted practical ramp filtering method is used for sparse sampling data. The side lobes of the filter converge fast, artifacts caused by Gibbs effects are effectively reduced compared with common Shepp-Logan and Ram-Lak filters. Additionally, a weighting function is designed to reduce artifacts caused by structures of uninteresting layers. The feasibility and effectiveness of AIDSI is evaluated through simulation and real experiments. Results show that the proposed method has the ability to directly reconstruct arbitrary depth slice images based on random sparse sampling data, which is expected to be used for rapid detection of a large number of samples in industry.

任意内部深度切片成像是实现X射线成像中快速工业扫描和重建方法的有效途径。本文提出了一种新的任意内部深度切片成像 (AIDSI) 方法用于锥束 X 射线成像。AIDSI在锥束X射线源下建立深度切片的体素与采样图像像素之间的关系，用于将采样数据重新聚焦到指定的深度切片。为了消除直接重新聚焦的伪影，对稀疏采样数据使用 3 点加权实用斜坡滤波方法。该滤波器的旁瓣收敛速度快，与常见的 Shepp-Logan 和 Ram-Lak 滤波器相比，有效减少了由 Gibbs 效应引起的伪影。此外，加权函数旨在减少由无趣层的结构引起的伪影。通过模拟和真实实验评估AIDSI的可行性和有效性。结果表明，该方法具有基于随机稀疏采样数据直接重构任意深度切片图像的能力，有望用于工业上大量样本的快速检测。