Path: utzoo!utgpu!news-server.csri.toronto.edu!rpi!uwm.edu!bbn.com!archive.bbn.com!aboulang From: aboulang@bbn.com (Albert Boulanger) Newsgroups: comp.compression Subject: Re: Fractal Image Compression (and wavelets) Message-ID: Date: 22 Jun 91 16:42:00 GMT References: <1991Jun20.193546.10135@uicbert.eecs.uic.edu> Sender: news@bbn.com Reply-To: aboulanger@bbn.com Distribution: comp Organization: BBN, Cambridge MA Lines: 39 In-reply-to: hart@uicbert.eecs.uic.edu's message of 20 Jun 91 19:35:46 GMT In article <1991Jun20.193546.10135@uicbert.eecs.uic.edu> hart@uicbert.eecs.uic.edu (John C. Hart) writes: The only methods I am aware of are by Arnaud Jacquin and Ed Vrscay. These two methods will be summarized at SIGGRAPH '91 in the Fractal Models in Computer Graphics course. Jacqin's method is a block-coding, Vrscay's method uses power moments. Is anybody aware of any others? Others besides of course the famous Barnsley top-secret method. Moments were used in Barnsley's Royal Society paper. A neural net to compute the Hutchinson metric used in the closeness calculation in the inverse problem is described in: "A Neural Network to Compute the Hutchinson Metric in Fractal Image Processing" J. Stark, IEEE Trans. on Neural Networks, Vol 2 No 1, January 1991, 156,158 A novel optimization method used in the inverse problem, which has some of the features of "tabu" search, because it involves marking points in search space with positive and negative affinities is described in: "Chaotic Optimization and the Construction of Fractals: Solution of an Inverse Problem", Giorgio Mantica & Alan Sloan, Complex Systems 3(1989) 37-62. Interestingly, IFS Fractals and Wavelets share many properties. Some of which is explored in: "IFS Fractals and the Wavelet Transform", G.C. Freeland & T.S. Durrani, 1990 ICASSP proceedings, 2345-2348 Wavelets are good for measuring empirical fractal properties of signals in general. Recuse, of course, Albert Boulanger aboulanger@bbn.com