Edge selection on gradients based on dome extraction

The phenomenon of multiple responses to a single edge often occurs in detecting noisy ramp edges. This paper investigates three properties of dome extracted by grayscale reconstruction and then presents a novel approach to select edges from gradients suppressing the multiplicity of response. The approach extracts domes in gradients, and then threshold them to get edges. Its computation time and memory requirements are not more than the nonmaxima suppression algorithm obviously. Experiment results show that our approach suppresses the multiplicity of response effectively.