An Edge Detection Approach for Conscious Machines

We propose a novel edge detection methodology for conscious machines. We show that summing the outputs of multiple pathways (the decisions of several constructive kernel equations of edge detection techniques) enhances the perception of visible edges. Unlike previously published research, which has emphasized differences in the efficiencies of particular kernel equations, here we apply a linear summation of the outputs of diverse kernel equations. Despite the simplicity of this approach, our edge detection approach performs better than the individual pathways. More important, our proposed approach has biological plausibility in that human vision depends on parallel computation across diverse spatial frequency channels. We hope that this concept, along with other computational, behavioral, and neuroscientific concepts, will eventually assist in building better conscious machines.