Physical Analytics - Geospatial UAV


Description: The Geospatial Unmanned Air Vehicle (UAV) project seeks to explore the potential for data acquisition from semi-professional drones for geospatial applications. The drones used in this project are equipped with standard RGB cameras, GPS modules and various sensors. The high resolution images acquired by the drone can be processed in multiple ways. 

One application explored thus far is high resolution image stitching.  Standard open source satellites have a maximum resolution of approximately 20m. The use of UAVs can allow for the acquisition of high-resolution georeferenced images. These images can be used for multiple applications that involve monitoring micro temporal changes over a large area.  This is directly related to precision agriculture and its variants. These images can be accurately georeferenced and incorporated into PAIRS.

 

Stitched Image

 

The high resolution stitched images acquired can be used for a second application which is incorporating object recognition techniques to detect objects and quantities present in the image. An example is determining the numbers of trees present in a given area. This application can be expanded to integrating the detection process with advanced deep learning algorithms to recognize complex objects.

 

Trees_Detected

Another application being explored is the development of accurate 3D models from 2D drone images. A generated 3D model from drone image using photogrammetry can be seen below.

3D model