Enhanced one-bit transform for single-frame small target detection

In this paper, an enhanced one-bit-transform-based region of interest (ROI) method is proposed for the infrared small target detection under complex background. This novel method, named as En1BT for short, leverages the local variance weighted information entropy filter to increase the signal-to-noise ratio (SNR), followed by the one-bit transform method to automatically compute a valid threshold value for region of interest extraction in an infrared image. Numerical experiments are conducted, where small targets are immerged by the cluttered background. The validity of the enhanced one-bit transform method, comparing other popular ROI extraction methods, is verified in the experiments. Additional, the performance of our enhanced one-bit-transform approach as the detection procedure of a detecting before tracking method is demonstrated in the experiments too.