Title :
A new method for selecting gradient weight in incremental eigen-background modeling
Author :
Lv, Xiaolong ; Zhao, Guoliang ; Meng, Hao
Author_Institution :
Coll. of Autom., Harbin Eng. Univ., Harbin, China
Abstract :
Background modeling is an important approach used in motion detection. The background model should adapt to dynamic change of the environment and update quickly. In this paper, some improvements on incremental eigen background modeling are made. A new method of selecting weight according to reconstruction error is proposed. Compared with existing means, this method can not only model background effectively but also more efficiently.
Keywords :
eigenvalues and eigenfunctions; gradient methods; image motion analysis; object detection; gradient weight selection; incremental eigen background modeling; motion detection; Automation; Covariance matrix; Gaussian processes; Gray-scale; Histograms; Kernel; Lighting; Matrices; Object detection; Principal component analysis;
Conference_Titel :
Information and Automation, 2009. ICIA '09. International Conference on
Conference_Location :
Zhuhai, Macau
Print_ISBN :
978-1-4244-3607-1
Electronic_ISBN :
978-1-4244-3608-8
DOI :
10.1109/ICINFA.2009.5205030