Multiresolution filtering with application to image segmentation

In this paper, the problem of segmentation of a smooth image has been studied using multiresolution analysis. The analysis has been carried out with the aid of a new orthonormal wavelet basis introduced in this paper. A procedure has been developed to approximate an image at a coarse resolution by dropping its components at finer resolutions. The proposed wavelet basis has been used to represent the fine resolution components of the image. An algorithm is proposed which performs an initial segmentation on the coarse approximation of the image, and a region information table is formed. The table is used and updated repeatedly via a 'region refinement procedure' with the introduction of the image information at finer resolution. The procedure progresses until all the information has been taken into account. The proposed algorithm has been tested on a variety of real images such as human faces, natural scenes, and medical images.