Local spectral similarity (LSS) algorithm has been developed for detecting homogeneous areas and edges in hyperspectral images (HSIs). The proposed algorithm transforms the 3-D data cube (within a spatial window) into a spectral similarity matrix by calculating the vector-similarity between the center pixel-spectrum and the neighborhood spectra. The final edge intensity is derived upon order statistics of the similarity matrix or spatial convolution of the similarity matrix with the spatial kernels. The LSS algorithm facilitates simultaneous use of spectral-spatial information for the edge detection by considering the spatial pattern of similar spectra within a spatial window. The proposed edge-detection method is tested on benchmark HSIs as well as the image obtained from Airborne-Visible-and-Infra-RedImaging-Spectrometer-Next-Generation (AVIRIS-NG). Robustness of the LSS method against multivariate Gaussian noise and low spatial resolution scenarios were also verified with the benchmark HSIs. Figure-of-merit, false-alarm-count and miss-count were applied to evaluate the performance of edge detection methods. Results showed that Fractional distance measure and Euclidean distance measure were able to detect the edges in HSIs more precisely as compared to other spectral similarity measures. The proposed method can be applied to radiance and reflectance data (whole spectrum) and it has shown good performance on principal component images as well. In addition, the proposed algorithm outperforms the traditional multichannel edge detectors in terms of both fastness, accuracy and the robustness. The experimental results also confirm that LSS can be applied as a pre-processing approach to reduce the errors in clustering as well as classification outputs.