基于窄带反射滤光阵列的光谱偏振形貌多模态成像方法OA

To meet the demands for precise target recognition and multidimensional information acquisition in complex scenarios,this paper precents an integrated multispectral-polarimetric-topographic imaging system sharing a common optical path.The system employs a narrowband reflective filter array and a micro-polarization modulation array to capture multispectral,multi-polarization-state images.It is compactly integrated with the topographic channel of a near-infrared lidar,enabling simultaneous acquisition of multispectral-polarimetric information along with intensity and depth data.Building upon this foundation,an automatic tri-modal registration and fusion algorithm is propased for intensity-spectral-depth data.Through scale estimation,coarse registration via similarity transformation,fine registration based on mutual information,and edge-phase correlation,the algorithm achieves pixel-level alignment of the three modalities within a common field of view and enables multi-pseudo-color fused display.Indoor experimental results demonstrate stable imaging quality and registration accuracy sufficient for multimodal information fusion.Multispectral polarization and topographic data exhibit strong complementarity in material differentiation,edge enhancement,and spatial-structure representation.This study provides a theoretical basis and technical support for material identification,structural perception,and multidimensional collaborative detection of complex targets in the field of aerospace remote sensing and related applications.