基于单色与全光谱照明的高光谱成像系统及图像融合

Hyperspectral Imaging System Based on Active Band-pass and Full-spectrum Illumination and Image Fusion

高光谱成像是一种同时采集目标物体空间与光谱信息的实用技术。由于照明条件与采集方法的限制,高光谱成像技术采集的图像通常低于全波段照明下的全色图像的空间分辨率。本文提出了一种主动式照明的高光谱成像方式,通过产生共光路的单色光和全光谱白光进行目标物体的照明,采集共配的高光谱图像与全色图像,通过图像融合的方式提升高光谱系统的空间分辨率。针对系统采集的高光谱图像与全色图像,提出提升小波自适应梯度加权图像融合的方法,提升了空间分辨率,同时,相比于传统Mallat小波融合方法节省50%的计算时间,具有工程应用意义。提出的高光谱成像系统能够用于实验室常规目标物体的照明与成像,并给出了一种图像融合的思路以弥补高光谱图像低空间分辨率的缺陷,有助于提升后续特征提取和图像分类等分析过程的准确度。

Hyperspectral imaging(HSI)is a practical technique which simultaneously collects spatial and spectral information of a target.Due to the limitations of illumination condition and acquisition methods,hyperspectral images usually have lower spatial resolution than panchromatic images under full-band illumination.In this paper,a HSI system with active lighting is proposed,which illuminates the target by monochromatic light and full-spectral white light in the same optical path,then collects a set of co-registered hyperspectral images and a panchromatic image,and finally improves the spatial resolution of the HSI system by image fusion.For the fusion of hyperspectral images and the panchromatic image,lifting scheme wavelet adaptive weighted fusion operates well in improving spatial resolution.Meanwhile,lifting scheme saves 50%of time compared to traditional Mallat wavelet transformation,having practical application significance.The proposed HSI system can be used for illumination and imaging of objects in lab,and the idea of image fusion is proposed to compensate for the low spatial resolution of hyperspectral images,contributing to improving the accuracy of subsequent analysis processes such as feature extraction and image classification.