Infrared small target detection based on non-convex Lp-norm minimization
Detection of target using infrared images is much demanding now days in different areas of applications. The properties of small target and its presence in the noisy background makes detection task very difficult. So, to mitigate this issue, this paper proposed a novel method using robust principal component (IPNCWLP-RPCA) analysis and non-convex lp-norm minimization. In the existing infrared patch image (IPI) model, because of nuclear norm minimization, the edges in the background were missdetected as a target because of more shrinkage of singular values. To address this issue, proposed method uses non-convex lp-norm (0$<$p$<$1) minimization of singular values, where the weights are assigned to singular values based on p values instead of assigning equal weights to all. The proposed method is finalized using the alternating direction method (ADMM) of the multiplier. The results from the simulation depicts that the presented approach outperformed the base-line methods.