针对高光谱图像全色锐化中的光谱失真和纹理细节提升不足问题,结合交叉皮层神经网络模型(intersecting cortical model,ICM),提出一种自适应高光谱图像全色锐化算法.该算法采用ICM分割,先将高光谱图像与空间分辨率较为接近的多光谱图像进行匹配融合,再将结果与高分辨率的全色图像融合,以获得同时具有全色图像的高空间分辨率和高光谱图像的光谱分辨率融合结果.同时,在锐化融合中采用灰狼优化算法(grey wolf optimizer,GWO)自适应优化ICM模型参数,生成最优非规则分割区域,为高光谱图像提供更精准全面的细节和光谱信息.采用 2 组资源一号 02D卫星高光谱数据集进行实验验证,结果表明,提出的新的锐化融合算法在空间细节和光谱信息评价指标上均表现最优,验证了该算法有效性.
An ICM-based adaptive pansharpening algorithm for hyperspectral images
Considering spectral distortion and insufficient texture details in the pansharpening of hyperspectral images,this study proposed an adaptive pansharpening algorithm for hyperspectral images based on the intersecting cortical model(ICM)for image segmentation.First,hyperspectral images were matched and fused with multispectral images with similar spatial resolution.Then,the matching and fusion results were fused with high-resolution panchromatic images,obtaining the fusion results possessing both the high spatial resolution of panchromatic images and the spectral resolution of hyperspectral images.Moreover,the grey wolf optimizer(GWO)was employed in sharpening fusion to adaptively optimize ICM parameters,generating the optimal irregular segmentation regions,thus providing more accurate and comprehensive details and spectral information for hyperspectral images.Finally,experiments were conducted on the proposed algorithm using two hyperspectral datasets from the ZY-1 02D satellite.The experimental results demonstrate that the proposed algorithm manifested the optimal performance in the evaluation indices of spatial details and spectral information,substantiating its effec-tiveness.
pansharpeningintersecting cortical modelhyperspectral imagegrey wolf optimizerremote sensing image fusion
万方数据
维普
