A 64-bit Quadratic Approximation of an Orthorectification Algorithm

Aerial image orthorectification is a useful tool that is currently being applied to many applications. However, orthorectification is a computationally complex algorithm that requires parallelization, interpolation, and divisions. When dealing with onboard processing of aerial imagery for real-time feedback, another constraint arises, power consumption. Field Programmable Gate Arrays (FPGAs) are a low power solution, however, FPGAs are limited to fixed-point processing. This paper develops and describes a fixed-point integer orthorectification algorithm that uses a quadratic approximation for the division inherent in the orthorectification algorithm. The new algorithm is then applied to aerial data and compared to the floating-point version using 64-bit data types. The results show an increase in computational speed of over 2x and a mean projected pixel position difference of 2% of a pixel size over the floating point algorithm.