A new edge detection algorithm using data fusion approaches

A robust edge detection algorithm for gray scale noisy images using data fusion approaches is presented. Two sets of multi-scale linear weighted gradient operators that show different directional properties with respect to impulse-type noise are utilized in the proposed algorithm. The complementary responses of the used operators are combined effectively to obtain a high level of edge localization and insensitivity with respect to noise. This is in contrast to conventional edge detection algorithms employing a single gradient operator that are very sensitive to noise. Simulation results obtained using a variety of noisy images to demonstrate the robustness of the proposed algorithm are presented.