Image compression using haar wavelet based tetrolet transform

Abstract

Tetrolet Transform, a wavelet based efficient and effective transform, utilizes supports formed by connecting four identical squares known as tetrominoes such that each is connected to at least one other square along its boundary. Haar wavelets are defined on these tetrominoes so as to form an localized orthonormal basis. The procedure is applied on small 4×4 partitions of the low pass image and the sparsest covering from each partition is stored. The non-redundancy in the wavelet basis results in sparse image representation. Consequently high compression ratio is achieved after the application of a wavelet shrinkage procedure on the tetrolet coefficients. In order to reduce the processing cost certain modifications in the original Tetrolet Transform are also discussed.