Vast amount of imageries from different remote sensing sensors are now available. Imageries have different characteristic and properties distinction in term of spatial resolution and spectrum resolution Image fusion technique can be applied to merge these imageries covering the same geographic extent One of the goals of this research is to investigate fusion techniques for generating new image containing geometric and spectral information for supporting image interpretation In this study live techniques for image fusion have been used : color composite, Brovey transformation, IHS transformation, principal component analysis (PCA), and arithmetic method. Tests have been done on three key areas: a rural area of Rung - Kaben Gulf in Chantaburi province, mixed types of landuse in Amphoe Muang Chantaburi and a dense metropolitan area surrounding Chulalongkorn University in Bangkok. The dataset comprises of LANDSAT 7 ETM + band 3, 4 and 5 images (ETM/B3, ETM/B4, ETM/B5) with 30 meter ground resolution and higher spatial resolution, SPOT - 2 panchromatic Ullages with 10 meter resolution, sharpened panchromatic SPOT -2 images (SPOT/SH) with 10 meter resolution, LANDSAT 7 ETM + panchromatic images with 15 meter resolution, ADEOSI band 3 images with 16 meter resolution and IRS-IC panchromatic images with 5.8 meter resolution. Geometric correction must have been done between images before fusion. The polynomial equation with 2nd degree was chosen and applied with test areas resulting in maximum discrepancies up to 0.14 pixel. The investigation in area of Kung -Kaben Gulf for dataset of LANDSAT 7 ETM + band 3 ,4 and 5 images and SPOT/SH panchromatic images indicates that arithmetic method {R = 05*[(05)*(SPOT/SH) +(05)* (EIM/B4)]+ 127, G = 05*[(05)* (SPOT/SH)+(05)*(ETM/B5)]+127, B = 05*[(05)* (SPOT/SH)+(05)*(EIMB3)]+127} is superior flan other selected fusion techniques. The result image distinguishes in-land forest and mangrove forest The length of digitized road and river network on fused image still has the same length before fusion. The fused image can be used more conveniently for visual interpretation supporting classification of in-land forest, mangrove forest, shrimp-form, road and river. In area of Amphoe Muang Chantaburi for dataset of LANDSAT 7 ETM+ band 3,4 and 5 images and SPOT/SH panchromatic images indicates that arithmetic method ; {R = 127+05 * [(SPOT/SH) * (ETM/B5)]1/2 , G = 127+05 * [(SPOT/SH)* (EIM/B4)]12, B = 127+05*[(SPOT/SH)*(ETM/B3)]1/2} issuperion than other techniques. This resultimage distinguishes boundary of built-up area , block of buildings and roads from green area and other surroundings. Evidently tile length of digitized road on fised image increases 30.73 percent. In area of Chulalongkorn University for dataset of LANDSAT 7 ETM+ band 3, 4 and 5 images and IRS-IC panchromatic images fused with tile afore-mentioned technique resulting also superior color composite image. Important land uses and land covers are visually enhanced, which implies better visual interpretation.
