MULTI SPECTRAL/PANCHROMATIC-BASED DATA FUSION: ANALYSIS OF TECHNIQUES TO ROAD NETWORK EXTRACTION, IN URBAN AREA, USING ETM+IMAGES

Labeling pixels by drawing inferences from several available sources of data is referd to as data fusion (Richard, 1999). The objective of this study is assess different Data Fusion (DF) techniques in order to increase spatial resolution, while protect spectral characteristics of fused images intend to road network extraction. ETM+ MS and ETM+ PAN images, (Tehran, Iran, 2000), have been used. In this reaserch three used Data Fusion (DF) models include Principal Component Transform (PCT), Intensity Hue Saturation (IHS), and Brovey Transform (BT) have been examined. The fused images were statistically analyzed and post classification accuracy assessment has been done. While, the results show PCT model has higher ability to protect the statistical characteristics of fused images followed by brovey and IHS. Also, In post classification accuracy assessment domain results show that PCT represent better results followed by Brovey and IHS.