Abstract:

A novel algorithm for extracting texture and detecting edge is presented based on sparse representations and halfquadratic regularization, which is a direct extension of the recently developed sparse-representation-based image decomposition method called morphological component analysis (MCA). The basic idea presented in this method is the use of two appropriate dictionaries, one for the representation of texture parts——the dual tree complex wavelet transform and the other for the cartoon parts——the second generation of curvelet transform. It results in a new decomposition model. Then by using the method of half-uadratic regularization, a new method combining texture extraction with edge detection is derived based on the new decomposition model. Finally, numerical experiments show that the new method can decompose a given image into a cartoon and an oscillatory component better while detecting contour well and truly.