The image transformation coding is classified into two categories:one-dimension transformation coding and two-dimension transformation coding. One-dimension transformation coding need little computation

proceeds fast

but its effect is dependent on the image scanning techniques. The central problem of the image scanning techniques is how to decrease the undulation of the generated data. In this paper

based on the deep researches

and objective quantity exponent is defined to measure the undulation of the scanning techniques. By it

with the help of computer experiments

it is proved that the Hilbert scanning technique is superior to the other scanning techniques and the data it generates is the most continuous based on the hypothesis that the images are continuous in the small range. The effect to the DCT coding of the scanning techniques is analyzed by computer experiments. Finally

a fast recurrent algorithm to construct the Hilbert scanning matrixes is given out for easy to use this scanning technique to code the images and to go on more researches.