Adaptive Bilateral Filter for Infrared Small Target Enhancement

Abstract

When applying a basic bilateral filter for an infrared small-target detection, it’s traditional structure and implementation should be changed in order to recover the background that is covered by small target. This paper presents an adaptive bilateral filter incorporating a surrounding block-variance analysis and an adaptive standard deviation based on this variance, which is able to obtains and analyzes more information from the vicinity of both the target and the background. This concept enables a bilateral filter to predict a background image excluding a small target because the surrounding block variance is adaptively mapped to the standard deviations of bilateral filter according to the target and background. Finally, a small target can be detected by subtracting the predicted background from the original image. In order to compare existing target detection methods against the proposed adaptive bilateral filter, the signal-to-clutter ratio gain and background suppression factor are employed. Experimental results show that the proposed adaptive bilateral filter method has a superior target enhancement performance compared to existing methods.