Vision Aided Autonomous Forced Landing Site Selection for UAV Using Hyperspectral Image Processing

The crucial phases in aircraft navigation are takeoff, cruise and landing. Most of the aircrafts accidents (around 25%)happen during landing phase, that’s why it’s such an important phase of aircraft cruise. So to increase the reliability in aircraft cruise, we propose an idea for landing of aircraft using Hyperspectral imaging (HSI) sensor, which has spectral richness than spatial. Unmanned air vehicle is not currently designed to process contextual information (land use, building types, and water bodies) to assist in landing and navigation. Instead, a human operator often provides manually gathered and synthesized contextual input through control commands. We need to detect safe-landing site in unstructured terrain where the key problem is for the onboard vision system to detect a suitable place to land without the aid of a structured landmark such as a helipad. This paper is intended to provide a Hyperspectral image vision based safety mechanism for UAVs to use in case of emergency landing in safe areas. The image for landing can be obtained using onboard Hyperspectral sensor known as spectrometer. The spectra of the known object can be compared with the object in current image and continue for landing. Here we use various algorithms for classification which are used to locate the landing site.