An Approach towards Improved Hyperspectral Image Denoising

Amount of noise included in a hyperspectral image limits its application and has a negative impact on hyperspectral image classification, unmixing, target detection, and so on. The data that are contaminated with noise can cause a failure to extract valuable information and hamper further interpretation. The presence of noise in the image, extraction of all the useful information becomes difficult and noise can lead to artifacts and loss of spatial resolution. So to overcome this problem there should some method/system which removes noise to improve performance of subsequent application. It has been proved that the proper and joint utilization of global and local redundancy and correlation (RAC) in spatial/spectral dimensions gives better result in HSI denoising. Thus for removing noise we are going to use proper and joint utilization of global and local redundancy and correlation (RAC) in spatial/spectral dimensions and data representation scheme as sparse representation to capture local and global RAC in spatial and spectral domains. Especially it uses local RAC in the spectral domain and global RAC in the spatial domain which utilized in the framework of sparse representation. Here we can use image patches of few continuous bands with dictionary learning from the noisy HSI. In this case spectral distortion will introduce in global RAC, to reduce this proper regularization of low rank is helpful which make ill-posed denoising problem solvable. The low rank of HSI is also helpful to reduce error by enforcing low rank on the denoised data, which is introduced in the process of sparse coding and dictionary learning.