A new robust approach to diffuse edge detection

The authors outline an edge-prediction-adjustment strategy for the detection of diffused edges using an image pyramid structure. An initial edge map is obtained from an appropriate high-level image of the pyramid using a conventional edge detection method. Next, using a simple linear interpolation, the predicted edge map for the adjacent low-level image is determined. Based on the information of the predicted edges, edge adjustment is carried out using a sequential search in a small neighborhood of each of the predicted edges with the aid of a dynamic programming based method. Simulation results on both synthetic and medical images indicate that the performance of the proposed approach is much better than that of the commonly used differential of a Gaussian-based approach.<<ETX>>