An End to End Hyperspectral Scene Simulator with Alternate Adjacency Models and Its Comparison with CameoSim
posterposted on 19.11.2019 by Usman Zahidi
Poster sessions are particularly prominent at academic conferences. Posters are usually one frame of a powerpoint (or similar) presentation and are represented at full resolution to make them zoomable.
In this research we developed a rendering based End to End Hyperspectral scene simulator CHIMES (Cranfield Hyperspectral Image Modelling and Evaluation System), which generates NADIR images of passively illuminated 3-D outdoor scenes in visible and reflective infrared region i.e.360 nm to 2520 nm. CHIMES is capable of generating 3D geometry based on Digital Elevation Model (DEM) data. The atmospheric data is generated by radiative transfer code, MODTRAN, which is used to generate sky-dome environment map that also includes polarisation effect of sky due to Rayleigh scattering. CHIMES generates materials’ optical radiance by employing Metropolis Light Transport (MLT) method to simulate light. It includes render-time scattering phase function, adjacency-effect, sky-dome emitter and sensor modelling. CHIMES also includes adjacency-effect model similar to commercial Hyperspectral Image (HSI) simulator CameoSim, developed by Lockheed Martin, UK. This similarity is premised by the results of rigorous comparison of both simulators under several atmospheric conditions which shows that both simulators are in agreement when executed under same atmospheric and similar rendering configurations. An improved adjacency-effect model is also incorporated which is again compared with CameoSim. Results manifest that CHIMES together with this adjacency model perform better than CameoSim with respect to ground truth data.
Keywords: Hyperspectral Imaging, Synthetic Image Generation, Adjacency-Effect, CameoSim.