Abstract
Haar transform is computationally very efficient. Its local nature which allows for the possibility of permuting the input date is exploited to reduce the number of non-vanishing coefficients in the transform domain. For exact reconstruction, the inverse Haar transform of the transform domain data is first computed and then restored in its original order by operating with the inverse of the permutation matrix. This inverse is easily computed using the periodicity property of such matrices belonging to GF(2).