Post-processing of UAV-captured images for enhanced mapping by image stitching

This report introduces 3 post-processing techniques that can be used on images captured from an unmanned aerial vehicle (UAV) to distort the images in such a way as to make all the images appear as though they were taken from a uniform altitude, pitch, roll and yaw and from a rectilinear lens, meaning that when the images are stitched together the resulting map is accurate. The recent increase in demand for UAV technologies in the marketplace has been illustrated by a host of new products from major companies. The work in this report adds to the field of imaging from UAV´s - for example, the creation of an aerial map. A current limitation in technology is the distortion of images captured by UAVs. This distortion is caused by turbulence as well as the necessary pitch and roll rotations required to move horizontally. This report introduces techniques which make images captured from fisheye lenses appear rectilinear; make all images appear of uniform scale, regardless of the height at which they were captured; and correct for any skew distortions caused by the pitch, roll and yaw of the UAV. All algorithms were developed in Matlab and tested on actual UAV images. This report focuses on rotary wing UAV´s such as quad-copters and hexa-copters, although some of the principles discussed could be applied to fixed-wing aircraft as well.