Local threshold and Boolean function based edge detection

Abstract

Localization of edge points in images is one of the most important starting steps in image processing. Many varied edge detection techniques have been proposed. Different edge detectors present distinct and different responses to the same image, showing different details. We present a new approach for edge detection. The actual gray level image is locally thresholded using the local mean value to make a binary image. The binary image is checked for edges by comparison with the known edge like patterns, utilizing Boolean algebra. This approach recognizes nearly all-actual edges and edges due to noise. For removing edges due to noise, we adopt another approach. This time the actual image is globally thresholded by the variance value of the image. The two resulting images are logically ANDed to get the final edge map.