Least-Squares Interpretation of Minimum Entropy Deconvolution
IEEE Signal Processing Letters · 2026 · 33 · 2715-2719
DOI: 10.1109/lsp.2026.3707809 ↗Kurz gefasst
Fachartikel in der Zeitschrift IEEE Signal Processing Letters, 2026. Verfasst von Yu Wang und 3 weiteren Autor:innen. Die Arbeit nennt 21 Literaturangaben. Gefördert von: National Natural Science Foundation of China.
Abstract (Originalsprache)
Minimum entropy deconvolution (MED) is a widely used blind deconvolution method for sparse component recovery, but its underlying optimization mechanism remains insufficiently understood. This letter shows that MED admits a least-squares (LS) interpretation. Specifically, the objective value of MED is proved to be equivalent, up to constant factors, to that of an induced LS problem that drives a filtered output toward a signed canonical vector. This result reveals that MED is intrinsically driven toward a feasible filtered output that best approximates an isolated impulse in the LS sense, rather than directly enhancing an existing sparse component. As a tractable approximation to MED, the induced LS problem also admits a closed-form global solution and further leads to a $\frac{3}{2}$-approximation guarantee for MED in its equivalent gap-minimization form. Furthermore, the proposed interpretation explains the effect of filter length and the extraction of a spurious impulse from white noise, thereby providing further insight for understanding and analyzing MED.
Förderung
- National Natural Science Foundation of China
Zitieren
APA
Wang, Y., Wang, D., Smith, W., & Peng, Z. (2026). Least-Squares Interpretation of Minimum Entropy Deconvolution. IEEE Signal Processing Letters, 33, 2715-2719. https://doi.org/10.1109/lsp.2026.3707809
Vancouver
Wang Y, Wang D, Smith W, Peng Z. Least-Squares Interpretation of Minimum Entropy Deconvolution. IEEE Signal Processing Letters. 2026;33:2715-2719. doi:10.1109/lsp.2026.3707809
BibTeX
@article{Wang2026,
author = {Wang, Yu and Wang, Dong and Smith, Wade and Peng, Zhongxiao},
title = {Least-Squares Interpretation of Minimum Entropy Deconvolution},
journal = {IEEE Signal Processing Letters},
year = {2026},
volume = {33},
pages = {2715--2719},
doi = {10.1109/lsp.2026.3707809}
}
EUVIMED