Fractal-based image coding with fast decoder convergence

摘要: Abstract In this paper, a new class of fractal image coders is described. As in previously published coding algorithms, the encoder models each to be coded as an attractor simple affine mapping, and finds parameters such mapping by blockwise analysis. The code consists parametric description which selected provide best fit l 2 sense. decoder uses it receives synthesize through iterative procedure. Although resulting are strongly related coders, they modified way that gives exact convergence towards lowest possible number iterations — typically three or less. contrast what has been case with previous independent. can implemented computationally very efficient pyramid structure. Also, coder offering non-iterative decoding included special case. paper describes modifications known necessary obtain fast convergence. It shown quality remains unimpaired these modifications. computational complexity algorithm analyzed, structure for outlined. Finally, example given.