Share Email Print
cover

Proceedings Paper

Moving object detection under complex background using radial basis function neural network
Author(s): Zuomei Lai; Jingru Wang; Qiheng Zhang
Format Member Price Non-Member Price
PDF $14.40 $18.00

Paper Abstract

It is well known that moving object detection under complex background becomes more difficult because of moving cameras. According to the fact that background and objects have different motion, the moving scene can be decomposed into different regions with respect to their motion by means of a radial basis function(RBF) learning scheme. After global background motion compensation, five-dimensional (5-D) feature vectors include pixel intensities, current pixel coordinates and pixel dense optical flow field extracted from image sequences are treated as the inputs of the RBF network. The learning algorithm for the network minimizes a cost function derived from the Bayesian estimation theory. Each output unit of the network is associated to a moving object. Experimental results indicate the algorithm's validity after many complex sequences are tested.

Paper Details

Date Published: 29 January 2007
PDF: 6 pages
Proc. SPIE 6279, 27th International Congress on High-Speed Photography and Photonics, 62794U (29 January 2007); doi: 10.1117/12.725457
Show Author Affiliations
Zuomei Lai, Institute of Optics and Electronics (China)
Chinese Academy of Sciences (China)
Jingru Wang, Institute of Optics and Electronics (China)
Qiheng Zhang, Institute of Optics and Electronics (China)


Published in SPIE Proceedings Vol. 6279:
27th International Congress on High-Speed Photography and Photonics

© SPIE. Terms of Use
Back to Top