Research On Improvement And Application Of Robust Principal Component Analysis Model

With the continuous development of information processing technology,data collection technology and the arrival of the era of big data,people have more and more ways to obtain information,and the amount of information is also increasing.As one of the effective methods for dimensionality reduction of high-dimensional data,the robust principal component analysis model is widely used in the field of image processing.This paper will mainly study the improvement of the robust principal component analysis model and its application in the separation of video foreground and background.The main work of the paper is as follows:Firstly,the basic principle of RPC A model and the research status at home and abroad are introduced.At the same time,the model solving algorithm used in this paper is introduced,and the basic framework of the algorithm is given.Secondly,to solve the problem that the RPCA model based on standard kernel norm has poor performance in the background foreground separation under the condition of irregular object motion or complex dynamic background,an improved RPCA model combining weighted kernel norm with 3D-total variation is proposed.Firstly,the weighted kernel norm is used to approximate the rank function to apply the low rank constraint to the video background.Secondly,the full variational regularization term is introduced to constrain the continuity of foreground moving objects in time and space,and to suppress the random disturbance caused by the dynamic background effectively.The augmented Lagrange method is used to solve the improved model,and it is applied to the video foreground and background separation under dynamic background.Thirdly,for the existing RPCA model based on non-convex rank approximation due to the random disturbance of the dynamic background,the extracted foreground part is incomplete and the background part is too noisy,a combination of non-convex gamma norm and An improved model of 3D-total variation technology.The augmented Lagrangian method was used to solve the improved model and applied to the separation of the foreground and background of the video under dynamic background.The effectiveness of the algorithm was verified through experiments.Finally,it summarizes the work done in this article and proposes further research directions.