Abstract A new interpolation algorithm based on intensity normal distribution of matters in medical CT slices was presented in this paper. First, the CT slices were classified into different matters, next, the best matching pairs belonging to the same matter was searched for every interpolated pixel. After that the weighs based on intensity normal distribution were used to calculate the gray level for the interpolated pixels. The traditional linear interpolation could introduce artifacts between the best matching pairs whose gray level differed significantly. The new algorithm could avoid those artifacts in these cases. For the four groups of experimental slices which had a lot of the best matching pairs whose gray level differed significantly, experimental results showed that the new algorithm decreased the average errors of the marked pixels 29.55% averagely more than the traditional linear interpolation between the best matching pairs.
|
|
|
|
|