Author
Jianping Fan, Walid G. Aref, Mohand-Said Hacid, and Ahmed K. Elmagarmid
Abstract
In many image processing applications, edge detection is a useful method for obtaining a simplified image that preserves the domain geometric structures and spatial relationships among variant image components. For providing automatic edge detection, two problems should be solved: one is feature extraction for calculating the edge strength, another is feature classification for automatic edge detection. For solving these two problems, we propose an improved automatic edge detection technique. Three color components of a color image and more potential edge patterns are integrated for solving the feature extraction problem. A fast entropic thresholding technique is also developed for solving the feature classification problem. Experimental results have confirmed that this proposed edge detector can provide more reasonable results as compared with the traditional isotropic edge operators, and its calculation cost has been reduced as compared with the complex edge detectors. Good balance between the calculation cost and the edge detection accuracy is achieved.
