The SLIC (simple linear iterative clustering) superpixel algorithm is an efficient and fast algorithm for segmentation. The algorithm is inherently designed on three band color images. Hyperspectral imaging, which is a relatively new remote sensing technology, contains hundreds of bands which carry rich spectral and spatial information. In this study, the SLIC algorithm is modified according to the structure of hyperspectral images. In addition to that, the similar superpixels are merged with the DBSCAN (density-based spatial clustering of applications with noise) algorithm. As a novel inspired approach, the spectral similarity index between the superpixels are computed based on the universal image quality index. The contribution of the obtained segmentation maps to the classification performance is presented comparatively. With the approaches presented in the study, the accuracy of the Pavia University dataset was increased from 86.85% to 96.66% using the SLIC algorithm by performing dimension reduction. In the Indian Pines dataset, the traditional SLIC approach contributed 89.06% to the classification, while the proposed approach contributed 97.26%.