基于IHS_NSCT多光谱图像融合算法的研究

Research on IHS_NSCT Based Multispectral Image Fusion Algorithm

基于无人机多光谱影像的大豆生产状况信息获取是近年来农业信息技术研究的主要方向之一,但大豆生长过程中无人机多光谱图像不同通道间图像存在重叠且合成的多光谱图像分辨率较低。为此,针对传统的IHS变换只能进行3个波段的图像融合且易出现光谱扭曲失真现象,将源图像配准操作消除重叠影响,提取图像中光谱信息更为丰富的强度分量单位,并利用NSCT变换的多尺度、平移不变特性结合分量单位进行高、低频子带的提取和再融合操作,完成MS图像与PAN图像的融合。同时,以信息熵(EN)、相关系数(CC)等参数作为评价指标,得到高分辨率的融合图像。实试结果表明:算法熵值信息和相关系数参数明显优于另两种算法结果,其信息熵值达到7.79,相关系数达到0.97,可见融合算法能够表达作物光谱信息且可为作物营养成分监测任务提供理论与技术支撑。

The acquisition of soybean production information based on UAV multispectral images is one of the main directions of agricultural information technology research in recent years,as there is overlap between different channels of UAV multispectral images and the resolution of the synthesized multispectral images is low during the soybean growth process,the traditional IHS transform can only fuse images in three bands and is prone to spectral distortion and distortion.The NSCT transform’s multi-scale and translation-invariant characteristics are used to extract and re-fuse the high and low frequency sub-bands by combining the component units to complete the fusion of MS images with PAN images.The parameters such as information entropy(EN)and correlation coefficient(CC)are also used as evaluation indicators to obtain high-resolution fused images.The entropy information and correlation coefficient parameters of the algorithm are significantly better than the results of the other two algorithms,with the information entropy value reaching 7.79 and the correlation coefficient reaching 0.97.It is evident that the fusion algorithm can express crop spectral information as well as provide theoretical and technical support for crop nutrient monitoring tasks.