AN ATTENUATION COEFFICIENT RECONSTRUCTION ALGORITHM FOR DUALPROJECTION DIGITAL RADIOGRAPHY VEHICLE INSPECTION SYSTEM
-
- Zhongwei Zhao
- Tsinghua University
-
- Yuewen Sun
- Tsinghua University
-
- Xuerui Chen
- Tsinghua University
-
- Guangchao Li
- Tsinghua University
-
- Peng Cong
- Tsinghua University
Abstract
Digital Radiography (DR) is an effective method for nuclear facility security inspection. However, most of traditional DR systems employ one radiation source and one set of detectors, which can only obtain a projection image from one view of the object and possibly lose key information from other views. Most recently, dual-projection digital radiography inspection system has been developed to improve the ability of nuclear facility security to find dangerous materials, such as hidden nuclear waste and explosive of entry-exit vehicles. The attenuation coefficient is an important property to distinguish these dangerous materials, which can be obtained by a DR system under some conditions. Thus in this paper, we proposed a gradient descent iteration algorithm to reconstruct the shape and attenuation coefficient of the objects using two projection images from two different views under a certain constraint. The constraint requires that the cross section (geometry) in the ray plane of the object should be a convex polygon. The feasibility of the method has been proved by simulation experiments, while the application experiments based on the dual-projection digital radiography vehicle inspection system have also been taken. The Experiments show that the proposed method can effectively distinguish polycarbonate, graphite, aluminum, and water, of which the relative error of the attenuation coefficient and density can be less than 10%.
Journal
-
- The Proceedings of the International Conference on Nuclear Engineering (ICONE)
-
The Proceedings of the International Conference on Nuclear Engineering (ICONE) 2019.27 (0), 2209-, 2019
The Japan Society of Mechanical Engineers
- Tweet
Details 詳細情報について
-
- CRID
- 1390565134810598272
-
- NII Article ID
- 130007773834
-
- ISSN
- 24242934
-
- Text Lang
- en
-
- Data Source
-
- JaLC
- Crossref
- CiNii Articles
-
- Abstract License Flag
- Disallowed