An Adaptive Edge Detection Algorithm Using Two Power Spectrum Attributes

Through the Fourier analysis of input image, we could obtain its power spectrum which implies image texture. Further the ring and the wedge attributes of this spectrum could be defined. The ring attribute could describe the density of image texture. The wedge attribute could describe the direction of image texture. Therefore, the algorithm proposed could detect image edges based on texture adaptively, by using these two attributes as the criteria of template selection. The experiment shows this algorithm could detect image edge accurately and has the function of anti-noise.