PDE-based video image deblurring model and its parallel computation with multithreads

In recent years, the total variational model proposed by Rudin, Osher and Fatemi (ROF) has been often used in image restoration. As the numerical iterations increase, ROF model may produce the staircase effect while deblurring the noise effectively. Another fourth-order PDE model proposed by Lysaker, Lundervold and Tai (LLT) can protect more the fine image textures, but its deblurring result and boundary protection are lower effectively than ROF model. Combine these two models by using the weight function may create more advantages in remove the noise, enhence the boundary, protect the smooth region and fine texture. In this paper, we first analyze ROF model and its diffusion behavior, then combine ROF and LLT model to form an integrated model by choosing the weight parameter, detail the parallel implemenation with multithreads programming using OpenMP. Some experimental results show that this model has a better deblurring for remove the noise and protect the texture.