A millimeter-wave image denoising method based on adaptive sparse representation

In this paper, based on passive millimeter-wave (PMMW) imaging system, we apply a novel image representation theory - the sparse representation to PMMW image denoising procedure. It is proved that PMMW image can have spare representation based on overcomplete dictionary, and the sparsity of image plays a remarkable role in our denoising method. By choosing a reasonable threshold, we use K-SVD algorithm to learn a overcomplete dictionary based on image itself adaptively. Within the application of sparse representation on learned overcomplete dictionary, this method can restore our PMMW image effectively and efficiently. Experiments demonstrate good robustness and practicality both in synthetic PMMW images and actual PMMW images.