Fly-Inspired Edge Detection: Architecture and Reconfigurable Embedded Implementation

This work presents a reconfigurable embedded implementation of a fly-inspired edge detection algorithm that allows for run-time alteration of the numerical format in response to constraints on accuracy and resources. The low-resource, high-performance system can effectively detect edge orientation and location. This work illustrates the advantages of run-time reconfiguration technology for the implementation of fly-inspired vision algorithms whose capabilities can surpass or supplement traditional image processing systems.