The classification of motion in image sequences using 3D recursive adaptive filters to obtain neural network input vectors

Author

Bruton, L.T. ; Bartley, N.R. ; Liu, Z.Q.

Author_Institution

Dept. of Electr. & Comput. Eng., Calgary Univ., Alta., Canada

Volume

4

fYear

1995

fDate

Nov/Dec 1995

Firstpage

1595

Abstract

A new application of neural networks is described that permits the selective classification of objects on the basis of their motion in digital image sequences. An adaptive 3D recursive linear-trajectory (LT) filter is employed to track a moving object on a smoothly-varying space-time trajectory. The trajectory information produced by the adaptive 3D LT filter is used as the input vector to a conventional multilayer perceptron (MLP) neural network to perform the classification of motion

Keywords

adaptive filters; filtering theory; image classification; image sequences; motion estimation; multilayer perceptrons; recursive filters; 3D recursive adaptive filters; adaptive 3D recursive linear-trajectory filter; digital image sequences; image sequences; motion classification; multilayer perceptron neural network; neural network input vectors; selective classification; smoothly-varying space-time trajectory; Adaptive filters; Digital images; Image sequences; Neural networks; Noise robustness; Nonlinear filters; Passband; Tracking; Trajectory; Vehicles;

fLanguage

English

Publisher

ieee

Conference_Titel

Neural Networks, 1995. Proceedings., IEEE International Conference on

Conference_Location

Perth, WA

Print_ISBN

0-7803-2768-3

Type

conf

DOI

10.1109/ICNN.1995.488856

Filename

488856