Alternative differential gradient edge contrast operator suitable for use within high speed pyramid-based imaging systems

Numerous edge operators, kernels and directional filters have been proposed, and edge detection is a well-established and mature area of computer vision. Despite this, in developing a real-time, pyramid-based imaging system, inadequacies in the standard, commonly used techniques have lead us to implement an alternate set of simple directional filters that combine to act as an edge contrast operator fulfilling the familiar requirements for (1) speed, and (2) a high localization of response. Within our application, this operator set is proving superior to the classical kernels at preserving the key requirement (2) as the operator's edge response energies are propagated up to pyramid levels of lower resolution for further recursive processing. Our scheme is shown to provide a unified response to both step and line edge types, and acceptable responses in the presence of anticipated noise levels. The method is direct, of low complexity, and we hope of general utility.