.png){kind=icon}
.png){kind=icon}
Return
MCCR: Multi-scale cross-modal cascade registration method for water surface scenes
.png){kind=icon}
PRE
AI
Y
Q
M
DOI：10.1007/s11042-026-21786-6.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Addressing the challenges of field-of-view (FOV) discrepancies, scale inconsistencies across multi-sensor imaging systems, and feature sparsity in infrared images of water surface scenes, this paper proposes a multi-scale cross-modal cascaded registration method. First, a multi-scale cross-modal FOV alignment model is developed to eliminate scale deviations induced by FOV discrepancies among sensors. Then, phase consistency (PC) feature detection is integrated with channel features of oriented gradients (CFOG) descriptors to enhance the density of reliable feature matches in weakly textured aquatic regions. Finally, a non-rigid transformation model based on thin plate splines (TPS) is introduced, which corrects non-rigid deformations through an energy minimization mechanism. Experimental results demonstrate that our method overcomes feature sparsity and geometric distortion in multi-modal images of water scenes. Through a cascaded optimization of FOV alignment, coarse registration, and non-rigid transformation, it achieves high-precision cross-modal image alignment.
Keywords:
Image registration
Cross-modality
Multi-scale analysis
Feature point matching
Image fusion
Water surface scenes
Journal
IF:
3
Papers:
1.9W
Citations:
3.2W
Organization
Cited Papers