First

a new class of gray scale transformation equation composed of the direct 1-order polynomial mixed mapping are proposed

and it improves the existing image information restoration algorithm based on long-range correlation.The improved method not only overcomes the poor smoothness performance caused by the direct mapping

but also solves the possible serious bug brought by 1-order mapping when the denominator of the polynomial equals to zero.Second

a novel improved genetic algorithm used in the search of the remote window in the image restoration is presented.The new algorithm avoids the laboriousness of ergodic search and overcomes the blindness and instability of the random searching.Compared with the simple genetic algorithm

the improved genetic algorithm speeds up the convergent process and improves the accuracy.The experimental results show that the satisfyactory performances can be obtained by means of the two improvements above.Especially the novel genetic algorithm demonstrates good robustness in the optimal search of the whole.