Description
The work for this thesis is to develop an image rectification application that performs automatic image ortho-rectification for images taken by the Rotomotion robotic UAV helicopter. The application also geo-references the images with the geo-location and attitude data for use in other geographic image system (GIS) software. This thesis is in support of the "Robotic Helicopter for Monitoring Pollution and Habitat Restoration at Tijuana Estuary" project, which goal is to build and test a robotic helicopter that can fly to locations at Tijuana Estuary that are too dangerous or sensitive to reach on foot or on a boat, and capture high-resolution, geo-referenced images for use in monitoring pollution and habitat condition. Although there are many image processing applications available that can perform image rectification, most are based on non-parametric methods, which require users to input reference control points for each image. In addition, the proprietary format of the Rotomotion data and the large number of image files call for the need of an automated application that can perform image rectification and image geo-reference. Hence, the automatic image rectification application in this thesis has been developed to resolve these problems. For this thesis, a parametric based ortho-rectification algorithm is developed. The algorithm models the attitude error of the robotic helicopter and rectifies the geometric distortions. This process uses the GPS geo-location data and Inertial Measurement Unit (IMU) attitude data obtained from the helicopter's onboard flight computer during flight. In addition to the algorithm development, a windows application has been developed by implementing the parametric ortho-rectification algorithm. The image rectification application is fully automated and can process multiple images without any user intervention. Finally, several test flights were conducted near the Tijuana Estuary to validate the performance of the robotic helicopter and image rectification applications. The application provided a modest improvement on the rectification of the images, due to sensor in-accuracy and timing synchronization of the system.
