Hyperspectral imaging has emerged as a powerful tool for remote sensing providing detection and identification of objects of interest using their unique spectral signature. Accurate information obtained can reveal details about the physical properties of materials that are relevant to intelligence gathering. Those interesting characteristics coupled with Unmanned Aerial Vehicles (UAVs) offer the perspective of easier detection of camouflaged army gear on the battlefield. This paper presents some aspect of hyperspectral measurements and processing and focuses on workload reduction and latent data representation using Principal Component Analysis (PCA) and autoencoder techniques. Both techniques are compared through a simple segmentation method that shows their efficiency in reducing the dimensionality of the hyperspectral data, spectrum-wise and image-wise.