Bent beamlets - efficient tool in image coding

Abstract

In recent years spectral methods, especially wavelets theory, have gained wide popularity in digital image processing because they allow for very sparse and efficient image coding. Although they are good in analyzing one dimensional signal, they cannot properly catch line discontinuities, so they are often present in two dimensional signals, that is in images. To avoid these problems the theory of geometrical wavelets has been created in recent years, which have all advantages of wavelets and moreover, allows to catch line discontinuities properly. Beamlets are those of a wide spectrum of the new theory of geometrical wavelets. They are successfully used in many areas of digital image processing, particularly in multiresolution image coding. In the paper presents the improvement of the beamlets theory, which allows to code images in a more efficient way than in the case of the classical beamlets. Also in other areas of image processing this improved theory can be successfully used. The experiments performed on a wide spectrum of test images have confirmed great usefulness of the improved - bent beamlets. In the paper the examples of isobar image coding are also presented.