Python parallel processing for hyperspectral image simulation: based on distance functions

The hyperspectral image consists of a high number of bands with low bandwidth which gives the advantage in the identification and detection of the features in the level of mineral and chemical composition. But the availability of hyperspectral data is very less and is highly expensive when compared to multispectral data. Simulation of hyperspectral data with the existing hyperspectral and multispectral data can be used as an alternative if data availability is less and is cost-effective. A new method is proposed for hyperspectral image simulation with Chebyshev and Spectral Angle Mapper (SAM) distance functions using python programming and its libraries. The process is selecting similar spectra of each pixel. Using normal processing, the data simulation is very time-consuming. By increasing the cores employed with parallel processing in python programming, the hyperspectral data simulation time is decreased exponentially from 19 to 1 h 21 min. The study clearly explains the logic and open source python code for the simulation of hyperspectral data. Data can be simulated with python code by just giving the paths of test Sentinel-2 and reference Sentinel-2, AVIRIS data. The simulated image gave better normalized cross-correlation values when compared with the original Airborne Visible/Infrared Imaging Spectrometer-Next Generation (AVIRIS-NG) data.