Digital Radiography with a Consumer Camera: Image Denoising and Deblurring

Abstract

Projection radiography is the most popular medical imaging modality due to its simple design and ease of use. To maintain quality, direct flat-panel detectors must be used to collect patient images, thereby providing a digital radiography (DR) imaging system. In this paper, we investigate an imaging system utilizing a DSLR camera to capture images from a scintillator illuminated by X-rays passing through the object to be imaged. However, maintaining the quality of the image from the proposed system is difficult since it contains impulsive-style noise, and blur varies based on distance. To get rid of unwanted radiographic impulsive noise, we use a statistical-based impulsive noise removal method. To handle the blurring issue in our proposed imaging system, we first use a systematic method to estimate the X-ray source and detector point spread functions (PSFs) from the radiographic image. The PSFs are then used in a non-blind deconvolution algorithm to deblur the image. Experiments on a variety of DR images show that with the proposed denoising and deblurring approaches, our imaging system can provide DR images with fine detail, and it can potentially be used for basic radiography.