基于拮抗视觉特性的多波段彩色夜视融合方法研究

以差分高斯函数(DOG)模拟人眼视网膜中的同心圆拮抗特性,分析了基于DOG的ON-中心OFF-周边分离网络(CSSN)平衡方程,发现其有增强反差、突出边缘和压缩动态范围的特点.将CSSN平衡方程应用于夜视图像的增强与融合,取到了保留图像间共有信息、增强互补信息的作用.提出了以CSSN平衡方程为基础的多种彩色夜视融合结构,包括双波段微光、微光/长波红外、中/长波红外、微光/中/长波红外,其中的彩色映射规则用色彩的冷暖突出红外图像的冷热目标、细节丰富的微光图像作为绿色背景.仿真结果表明,这些融合结构均获得了色彩自然的彩色融合效果.

RESEARCH ON FUSION SCHEMES OF MULTI-BAND COLOR NIGHT VISION IMAGES BASED ON OPPONENT VISION PROPERTY

Equilibrium equation of on-center off-surround shunting network(CSSN) based on difference of Gaussians(DOG) that simulated concentric circle pattern opponent receptive field in retina was analyzed and utilized to enhance and fuse night vision images,which could improve contrast,pop out edge,compress dynamic range,reserve common information and enrich complementary information.Multi-band fusion architectures for color night vision images built on CSSN equilibrium equation were proposed,which included multi-spectrum low-level light visible(LLLV),LLLV(long-wave infrared(LWIR),LWIR/media-wave infrared(MWIR),and LLLV/LWIR/MWIR images.Color mapping rules in these architectures used cool and warm color to present cold and warm targets in infrared,and map LLLV image with plenty of details into green channel as green background.Emulating results show that all architectures obtain natural color appearance.