Neural network approaches to fractal image compression and decompression

In image compression technologies, fractal image compression/decompression has the advantages of a high compression ratio and a low loss ratio. Howeve r, it requires a great deal of computation, which limits its applications, and so far, no parallel processing technique has been designed and implemen ted. In this study, we use neural networks to perform a large number of com putations in fractal image compression and decompression in parallel. The s imulation results show that the quality of images generated by neural netwo rks is similar to that produced using traditional methods, which verifies t he high value of our research, which has shown that the neural network tech nology is useful and efficient when applied to fractal image compression an d decompression. (C) 2001 Elsevier Science B.V. All rights reserved.