MD3: Model-Driven Deep Remotely Sensed Image Denoising

Remotely sensed images degraded by additive white Gaussian noise (AWGN) have low-level vision, resulting in a poor analysis of their contents. To reduce AWGN, two types denoising strategies, sparse-coding-model-based and deep-neural-network-based (DNN), are commonly utilized, which respective merits drawbacks. For example, the former pursue enjoyable performance with high computational burden, while latter powerful capacity completing specified task efficiently, but this limits application range. combine for improving paper proposes model-driven deep (MD3) scheme. solve MD3 model, we first decomposed it into several subproblems alternating direction method multipliers (ADMM). Then, replaced different learnable denoisers, plugged unfolded model to efficiently produce stable solution. Both quantitative qualitative results validate that proposed approach is effective efficient, has more ability generating preserving rich textures than other advanced methods.