Sine and cosine functions based edge detection operators

In this paper, a new method for extracting edges in images is proposed. The local pixel values are normalized to make their sum become unity. Then, a nonlinear filtering process is applied to the normalized pixel values by using the sine or cosine function. Two new edge detection operators, which exploit the local convexity of the sine and cosine functions, are thus defined. The edges are detected at the places where these operators yield large values. The new approach exhibits desirable results in both effect and efficiency compared with some classical methods.