Separating contours from texture edges in natural images

The main purpose of this paper is to develop a contour extraction scheme based on suppression of texture-like edges. Traditional edge detection methods are unable to distinguish between contours of objects and edges originating from textured regions. This difficulty is resolved by the proposed scheme. Three modules of contour attributes describing the independence of the detected edges are established. A procedure of eliminating texture-like edges by applying the contour attributes is performed through processes of contour decision, generation of model parameters, and contour integration. From experimental results, it has shown that the developed contour detection scheme produces reliable object boundaries.