Abstract:

While hyperspectral (HS) images have high spectral resolution, their spatial resolution is low due to technological limitations. In order to be able to meet requirements such as increasing classification success or obtaining more detailed content, it is beneficial that these types of images have a high spatial resolution in addition to high spatial resolution. Therefore, HS images are being combined with Multispektral (MS) images as a topic that has been studied to retain popularity in recent years. In literature, Matrix Division (MA) based imaging was not met with a study taking into account the effects of neighboring pixels. Therefore, this study suggests a new coating approach that takes into account the effects of spectral separation and spatial neighbourhood. First of all, the ultimate element and coating rates are removed from the hyperspectral image using spectral separation. The output of the final elements is transferred to the multi-spectrum spectral resolution using the spectral response of the multi-spectrum image detector. Then, for each pixel in the multispectural image, the combination rates of the latter elements and neighboring pixels are cut. Finally, the multi-spectral image with the cutting-edge map, the ultimate elements of the hyperspectral image and neighboring pixels are used to obtain high-resolution both spatial and spectral image. The suggested coating method has been tested on real hyperspectral images and experimental studies show that its success is higher than the studies in literature.

Keywords: