Title :
Data fusion with neural networks
Author :
Wang, Yifeng ; Goodman, Stephen D.
Author_Institution :
Titan Ind. Inc., Atlanta, GA, USA
Abstract :
This paper studies neural data fusion (NDF) from the viewpoint of histogram features. Recurrent and multiple layer perceptron neural networks are trained and tested for comparison and contrast. The fusion performance as seen by the reduction of the histogram variance is improved by neural network processing as well as by the design of the neural network architecture. Signal filtering and image restoration find applications in NDF. The performance of NDF under different signal-to-noise ratio conditions is studied, outperforming weighted average data fusion by 7.62 to 10.8% with a multiple layer perceptron and by 3.2% to 22.5% with a recurrent neural network
Keywords :
filtering theory; image restoration; multilayer perceptrons; neural net architecture; recurrent neural nets; sensor fusion; histogram features; histogram variance; image restoration; multilayer perceptron; multiple layer perceptron neural networks; neural data fusion; neural network architecture; recurrent neural networks; signal filtering; signal-to-noise ratio; Artificial neural networks; Filtering; Histograms; Logic programming; Network topology; Neural networks; Neurons; Nonlinear filters; Sensor fusion; Sensor systems;
Conference_Titel :
Systems, Man, and Cybernetics, 1994. Humans, Information and Technology., 1994 IEEE International Conference on
Conference_Location :
San Antonio, TX
Print_ISBN :
0-7803-2129-4
DOI :
10.1109/ICSMC.1994.399912
