Weighted RPCA based Background Subtraction for Automatic Berthing
Yinxiao Zhan, Ting Liu · 2019
Automatic berthing system is a kind of environment through the information collection around to find suitable berthing vessels, automatically enter the position control system. In recent years, background modeling, as an effective computer vision technology, has been widely studied to provide a rapid way to detect the ships entering the imaging area. Recently, robust principal component analysis has attracted much attention in the field of image analysis, it can also be applied to background modeling and is more robust than traditional models in diverse scenes. Alternating direction method of multipliers is used to solve this model for its high precision and speed. However, the process is not fast enough, since the singular values are treated equally during the minimization of nuclear norm and it is not consistent with physical meanings. In this paper, a weighted robust principal component analysis based background modeling method for automatic berthing has been proposed, where different weights are assigned to different singular values. Specifically, the designation of weights has been analyzed in detail to acquire a fast and accurate background modeling. Experimental results clearly show that the proposed method outperform the other state-of-the-art ADMM based algorithm in terms of both qualitative and quantitative analysis.