A circular smoothing edge detector

A method of detecting step edges based on an edge-preserving circular smoothing technique is presented. The detection procedure consists of two stages: circular smoothing and thresholding. To achieve better results, multiple windows are used. Experimental results utilizing real images show that using multiple windows can more efficiently remove noise, reduce edge width, and detect weak edges. Based on the edge quality, the method is comparable to the commonly used edge detectors, but it requires fewer steps. Since the edge detector can be implemented in parallel, it is suitable for VLSI implementations leading to a real-time edge detector chip.<>