Efficient algorithms of local discrete wavelet transform with Haar-like bases

In this paper, two new fast algorithms for calculating a local discrete wavelet transform for a one-dimensional signal based on the example of Haar’s wavelet basis are presented and their computational complexities are evaluated. The algorithms are compared with each other and with well-known algorithms of fast wavelet transform. Employment recommendations are given for each algorithm presented. Among other things the preference domains of these algorithms are shown, i.e., the parameters of wavelet transform calculating problem are specified so that these algorithms are computationally efficient. Conclusions regarding the advantages of the recursive algorithm over the alternative one and over the well-known algorithm of the fast wavelet transform are reached using the analysis of algorithmic complexities as the base and with regard to additional possibilities of the recursive algorithm. The extension of the algorithms considered to a two-dimensional case is presented. Vasilii N. Kopenkov. Was born in 1978. He was graduated in Samara State Aerospace University in 2001. At present time he works as an assistant on chair of geoinformatics in SSAU and as a junior research assistant in Institute of Image Processing Systems RAS. Head interests: image processing, earth remote sensing data and pattern recognition, geoinformatic systems. Author of 17 publications, including 7 articles. Member of the Russian Association Pattern Recognition and Image Analysis.