G. E. Øien, S. Lepsøy, T. Ramstad. Reducing the complexity of a fractal-based image coder. In Proceedings of the Vth European Signal Processing Conference EUSIPCO'92, Pages 1353-1356, 1992.
The basis principles of a block-oriented, fractal-based image coder are described. Two procedures for reducing its computational complexity are suggested. The first is an orthogonalization procedure serving to decouple a set of normal equations which is solved for every image block. This removes a significant part of the computations from a blockwise search loop used in the method. In addition, we introduce an automatic block classification procedure, in which block decimation and clustering is combined to greatly reduce both the necessary search for each image block, as well as the computational burden associated with each search. Combining orthogonalization and classification reduces the encoder complexity by a factor of about 500. The loss in quality is negligible for the images used.
@InProceedings{OiLeRa92,
Author = {Øien, G. E. and Lepsøy, S. and Ramstad, T.},
Title = {Reducing the complexity of a fractal-based image coder},
BookTitle = {Proceedings of the Vth European Signal Processing Conference EUSIPCO'92},
Pages = {1353--1356},
Year = {1992}
}