Fusion of Hyperspectral and High resolution imagery based on different level of HAAR DWT.

Image fusion is the process of integrating the information from a set of images in an image, as the fused image contains more useful information than any input data. The aim of remotely sensed image fusion is integration of information that obtained from sensors with different spatial, spectral and temporal resolution in order to get an image with more detail than any individual data. In the fusion process, output image is a combination of important features of two or more input data. The aim of this study was to evaluate the performance of discrete wavelet transform in fusion of hyperspectral and high resolution images. For this purpose, a window of images of Hyperion, ALI and OrbView3 sensors was selected. First, the Hyperion image was corrected for unusable bands and strip noise. Panchromatic band of ALI sensor was used for geometric correction and registration of hyperion image. The hyperion image transformed into a 10 m pixel image using the sampling operation and fused with the ALI image using the Gram-Schmit algorithm. Using the OrbView3 image, the results was captured on the fused image, then both images were converted to 4-pixel pixel size using the resampling operation. The OrbView3 image was decomposed into four levels using a HAAR wavelet and used for fusion procedure. The results showed that with increased level of image decomposition, the accuracy and precision of the integration increases.