A comparison of 2D fourier and spatial techniques for removing noise in real time x-ray radiography

When x-raying rocket propulsion systems during testing, periodic noise can corrupt the overall quality of an x-ray image. This periodic noise results from vibration due to the rocket motor firing and/or from the pulsing of the x-ray source. A method of removing band-limited noise for enhancement of real-time radiographic images of propulsion systems under test conditions is therefore desired to improve data analysis.

This investigation compares the Fourier and spatial techniques relative to their ability to eliminate random and periodic noise. Fast Fourier Transform (FFT) techniques are discussed with respect to efficiency and ease of implementation. Methods of removing low, and high frequency, and band-limited noise are demonstrated using both Fourier and spatial techniques. The radix-2 method of performing an FFT was selected for this investigation because of ease of implementation and reasonable computation time for a data transfer processor. The Fourier domain techniques were found to be better suited for removing band-limited periodic noise because extensive spatial filtering by means of frame averaging destroys the time resolution of the image.