A statistical correction method for minimization of systemic artefact in a continuous-rotate X-ray based industrial CT system

Abstract

The use of a linear detector array (LDA) in X-ray computed tomography (CT) imaging is well established. Generally the CT system using an LDA operates in a fan-beam configuration. In a non-medical tomography set-up, the X-ray source and detectors are stationary and the object is rotated for scanning. Equi-spaced angular projections over a complete 360° object rotation are required for CT image reconstruction using standard Convolution Back Projection (CBP) algorithm as applied to the fan-beam scanning geometry. If there is a lack of timing synchronization between the LDA data acquisition system and the rotary motion in a continuous-rotate system, the incremental angle for the acquired projections may not be known exactly. The error may cause artefacts and blurring in the CT image. The present paper describes a possible way of numerically tuning the acquired transmission data matrix for minimization of such artefacts. It is based on finding out statistically similar projections, which represent 0° and 360° angular views. The correction method has been developed to avoid modification in detector hardware and interface devices. The experimental CT system for industrial applications has been developed by making use of an independent scintillator-based linear detector array, commonly used for on-line radiography of low-density specimens. The proposed data pre-processing algorithm with some experimental results as well as limitations of the correction method is discussed.