TY - GEN
T1 - GPR Missing Data Reconstruction via Curvelet-Domain Optimization
AU - Li, Kunao
AU - Li, You
AU - Wang, Zexi
AU - You, Tingjie
AU - Wen, Jing
AU - Lan, Tian
N1 - Publisher Copyright:
© 2026 IEEE.
PY - 2026
Y1 - 2026
N2 - Data missing in Ground Penetrating Radar (GPR) severely hinders subsurface imaging and target recognition. To address this, we propose a reconstruction framework that integrates adaptive preprocessing with curvelet-domain optimization for two distinct undersampling scenarios. For known missing traces, a local adaptive thresholding method is developed to localize gaps. For unknown missing traces, fracture positions are identified via adjacent trace cross-correlation, while the specific number of missing traces is estimated through phaseshift analysis of the normalized mutual power spectrum to restore spatial scales. The reconstruction utilizes the curvelet transform's sparse representation of hyperbolic features, employing multiscale masks and the Nesterov Accelerated Gradient (NAG) algorithm to restore signals through constrained optimization. Simulation examples with complex clutters demonstrate that the proposed method effectively recovers target continuity and suppresses clutter.
AB - Data missing in Ground Penetrating Radar (GPR) severely hinders subsurface imaging and target recognition. To address this, we propose a reconstruction framework that integrates adaptive preprocessing with curvelet-domain optimization for two distinct undersampling scenarios. For known missing traces, a local adaptive thresholding method is developed to localize gaps. For unknown missing traces, fracture positions are identified via adjacent trace cross-correlation, while the specific number of missing traces is estimated through phaseshift analysis of the normalized mutual power spectrum to restore spatial scales. The reconstruction utilizes the curvelet transform's sparse representation of hyperbolic features, employing multiscale masks and the Nesterov Accelerated Gradient (NAG) algorithm to restore signals through constrained optimization. Simulation examples with complex clutters demonstrate that the proposed method effectively recovers target continuity and suppresses clutter.
KW - Curvelet transform
KW - Ground penetrating radar (GPR)
KW - Local adaptive threshold
KW - Missing data reconstruction
KW - Phase correlation
UR - https://www.scopus.com/pages/publications/105051482826
U2 - 10.1109/GPR69898.2026.11678053
DO - 10.1109/GPR69898.2026.11678053
M3 - Conference contribution
AN - SCOPUS:105051482826
T3 - 2026 21st International Conference on Ground Penetrating Radar, GPR 2026 - Proceedings
BT - 2026 21st International Conference on Ground Penetrating Radar, GPR 2026 - Proceedings
A2 - Warren, Craig
PB - Institute of Electrical and Electronics Engineers Inc.
T2 - 21st International Conference on Ground Penetrating Radar, GPR 2026
Y2 - 23 June 2026 through 26 June 2026
ER -