多光谱成像可以从背景中凸显出目标物, 但微型多通道滤光片在不同通道之间存在光谱串扰问题, 影响目标的识别。通过光谱串扰分析, 建立光谱串扰模型, 提出光谱串扰系数, 通过灰度值统计的方法定量评价光谱串扰。对已加工的一种光谱串扰程度约为16.74%的微型多通道滤光片获取的多光谱图像进行了对比分析, 经过串扰模型处理后的图像与原始图像相比, 其对比度有了显著的提高。实验分析结果表明, 该模型对评价多通道微型多光谱滤光片的光谱串扰具有很强的针对性, 对提高多光谱图像目标识别效果明显。

The target can be highlighted from background by multispectral imaging, while spectral crosstalk between different channels of micro-arrayed multichannel optical filter affects the target recognition. The extent of spectral crosstalk is evaluated by building up the model of spectral crosstalk and proposing the coefficient of spectral crosstalk. The reliability of the model is evaluated by the method of gray value statistics. By contrastive analysing the multispectral images from the processed micro-arrayed multichannel optical filter, the degree of spectral crosstalk of which is 16.74%, the contrast of the processed image is improved obviously compared with the original one. Experimental analysis indicates that this model has a strong pertinence in evaluating the spectral crosstalk of the micro-arrayed multichannel optical filter and obvious effect in improving the target recognition of the multispectral image.