احمد کشاورز

In Hyperspectral images, segmentation is important as a pre-processing. In this paper, the segmentation is done in two steps, taking into account the pixel spectrum in all bands. In the first step, using the region-zoning method, the image is converted to a number of sub-regions, and then a more complete segmentation is performed using the method of estimating the minimum path on the regions. The proposed method creates new parts by using the appropriate selection of seeds and taking into account the minimum pixel-to-seed path. The proposed method is implemented on the AVIRIS image and shows the number of regions and frontiers more favorable than other methods.