Contents I. INTRODUCTION
Cone beam computer tomography (CBCT) imaging technique with a flat panel detector has a capability for the 3-dimentional imaging of large volumes with isotropic resolution [1]. But the main limitation of CBCT is that CBCT is much less immune to x-ray scatter than fan-beam CT. The x-ray scatter may enhance the noise in the reconstructed images [2]. Thus it is an important issue to control and reduce the x-ray scatter in x-ray projection images. To date, many hard or soft based methods [3]–[5] have been proposed for reduction of x-ray scatter, such as air gap, anti-scatter grid, and filtering. In this paper, we propose a new method based on independent component analysis (ICA) [6] for removing the x-ray scatter in the observed x-ray images. The observed x-ray images can be considered as a linear mixture of scattered x-ray and primary x-ray. We show that the scattered x-ray and the primary x-ray can be blindly separated from two observed x-ray images by the use of ICA. The results indicate that the proposed method is effective for removing x-ray scatter from the x-ray images.
