Contour Detection of Gradient Images Using Morphological Operator and Transform Domain Filtering

In this work, an improved computational approach based on gradient image and transform domain filtering has been proposed. The proposed technique effectively detects contours of objects while suppressing the low frequency edges in natural scene images. To accomplish this, a novel morphological operator has been designed which generates the gradient image. This image is filtered using transform domain filtering by spectrum distribution of low frequency edges and contours. The proposed approach can effectively detect contours of objects and able to reduce irrelevant information generated from background with textures. The results of qualitative evaluation clearly validate the efficiency of the proposed method.