A new edge detection for multispectral images

Edge detection is the process of localizing pixel intensity transitions. The goal of this paper is to find ways to extend gravitational method to multispectral images. This paper considers the edge detection problem in feature space. Different from the old gravitational method, in feature space, each image pixel is considered as a celestial body with a mass represented by its spectral intensity instead of grayscale intensity. All these body attract each other by the gravity force, and hence, the edge information is measured by the gravitational force exerts on by its neighboring pixels. The proposed algorithm was tested and compared with some methods such as proposed by Kang. Experiments indicate that the approach is effective for multispectral images.