Introductory study on the Discrete Trigonometric Transform (DTT) wavelet-like decomposition-reconstruction procedure

In the paper the results of investigations indicating computational efficiency of Discrete Trigonometric Transforms (DTTs) used to design filters by means of the complete wavelet-like decomposition/reconstruction procedure, are presented. The study indicates flexibility when comparing DTT selection from the set of 16 transforms, and the sampling type used. The results, obtained for the block length up to N=1024, should encourage applications of the suggested solutions to all areas where the classical wavelet filters are utilized, while the reconstruction error at the Peak Signal to Noise Ratio (PSNR) level of 37.5 dB can be accepted to balance the benefit of reduction of computations up to 68 times, obtained after including the Hamming window in the filtering process. The comparisons are based on the error measure derived from the signal covariance matrix.