RESOLUTION ENHANCEMENT OF MULTISPECTRAL IMAGE DATA TO IMPROVE CLASSIFICATION ACCURACY

This study evaluated several methods for enhancing the spatial resolution of multispectral images using a higher resolution panchromatic image. A new method that relies on the local correlation between the low resolution superpixel and the corresponding mean panchromatic radiance is presented. High resolution multispectral images are produced based on the local correlation value and the high resolution panchromatic brightness levels. This method allows for within-pixel local adaptation of the correlation values to account for mixed pixel effects. The resultant hybrid, high resolution, multispectral data set has increased visual interpretability and improved classification accuracy while preserving the radiometric integrity of the original multispectral data. These methods can be applied to create simulated high resolution multispectral data to enhance existent image analysis techniques such as land-cover classification.