A salient object detection framework using linear quadratic regulator controller

Morteza Moradi, Farhad Bayat, Mostafa Charmi

Abstract: Highlights•A salient object detection framework using optimal control theory is proposed.•Two LQR controller systems for optimizing background and foreground detection are employed.•Different initialization strategies are applied to background and foreground estimations.•The overall performance (in term of F-measure) is comparable to deep learning-based methods.