A gradient and Laplacian based reaction-diffusion filter for hyperspectral image denosing

Noise reduction is an active research area in hyperspectral image processing due to its importance in improving the quality of image for the subsequent applications. To improve the accuracy and efficiency of object recognition and classification using hyperspectral imagery (HSI), we propose a novel smoothing algorithm by coupling a Laplacian-based reaction term to a classical anisotropic diffusion partial differential equation (PDE). In addition, an adaptive parameter is introduced to regularize the proposed reaction-diffusion PDE by explicitly integrating the interband correlations with the noise level of each band in hyperspectral images. As a result, our algorithm is more effective at controlling the behavior of the diffusion function when compared to previous multi/hyperspectral diffusion algorithms.