Beamlet: Principles, algorithms and applications

The concept of beamlet was first introduced by Donoho and Xiaoming Huo. As an efficient tool for multi-scale image analysis, beamlet dyadically organized the collection of line segments which exhibits a certain range of lengths, positions and orientations. The role played by line segments in beamlet analysis is analogous to which points play in wavelet analysis. This paper describes a multi-scale analysis framework based on beamlet analysis, presents the structure and principle of beamlet transform, and summaries the algorithms and the related research works of beamlet. Its development history, current and future challenges are also reviewed in details.