Sparse regularization image denoising based on gradient histogram and non-local self-similarity in WMSN

Recently WMSN (wireless multimedia sensor networks) owing to the unique advantage of rapid deployment, flexible networking and perceiving multimedia information, plays an important role in environment monitoring. However, adverse weather or severe environment often leads that WMSN video image is corrupted by much noise and fail to meet the quality requirements. To address this problem, a sparse regularization denoising method based on gradient histogram and non-local self-similarity (NSS) is proposed. Firstly, the denoising model containing image gradient prior and NSS prior is build. Then, image blocks that are similar in structure are clustered, and for each image blocks, we use the Sparse K-SVD dictionary instead of PCA dictionary to run dictionary learning independently. Finally, the iterative histogram specification algorithm is adopted to solve the denoising model. Experimental results showed that the method can achieve better visual quality while removing much noise and further reduce the computational complexity, suitable for the WMSN video image denoising.