Research on critical issues of bi-dimensional empirical mode decomposition

• Main Author: Huang, Yujia
• Other Authors: Qian Kemao
• Format: Final Year Project
• Language: English
• Published: 2015
• Subjects: DRNTU::Engineering::Computer science and engineering::Computing methodologies::Image processing and computer vision
• Online Access: http://hdl.handle.net/10356/62697
• Institution: Nanyang Technological University

In this report, we investigate the critical issues of bi-dimensional empirical mode decomposition (BEMD). BEMD has three major components – extrema detection, surface interpolation, and sifting iteration stoppage criterion. Multiple algorithms are available for each component and they can lead to totally different decomposition results. In this report, we will first analyse the logic behind each algorithm, compare these methods, and then reach a final recommendation. Furthermore, a new concept called Block-based BEMD is proposed in this report. Block-based BEMD splits the original 2D data into multiple blocks for surface interpolation, then combines them again for envelope processing. It improves the intrinsic mode function (IMF) quality by limiting the impact of noise and intermittence to a smaller area and also speeds up the decomposition process.