A Novel Method for Single Infrared Dim Small Target Detection Based on ROI extraction and Matrix Recovery

Abstract

Low-rank and sparse matrix recovery method based on Robust Principal Component Analysis (RPCA) model are widely used in infrared small target detection. In order to solve the problem of time consuming and difficulty in parameter selection when using this method, a novel method for infrared dim small target detection under complex background based on Region of Interest (ROI) extraction and matrix recovery is presented. Calculate the Variance Weighted Information Entropy (VWIE) of every sub-block and extract the ROI firstly; then use Adaptive Parameter Inexact Augmented Lagrange Multiplier (APIALM) algorithm to recover target image from extracted ROI; finally segmenting and calibrating the target using an adaptive threshold method. Experiments results demonstrate that the proposed method can significantly decline the running time and retain most properties of traditional detection method based on low-rank and sparse matrix recovery.