Images acquired by the Moderate Resolution Imaging Spectroradiometer (MODIS) aboard Terra and Aqua suffer from the common problem of the stripe noise, which heavily limits the application of the MODIS data. In this paper, we present an analysis of the shortcoming in recent variation-based destriping algorithm and also the characteristic of the stripe image, and then automatically detect the stripe lines, finally, we introduce several novel terms within this probabilistic model that are inspired by our analysis, therefore, a maximum a posteriori (MAP)-based destriping algorithm is proposed. A model of the high order spatial randomness of noise is incorporated into the data fidelity term, and a unidirectional adaptive total variation prior is incorporated into the prior term. What is more, the algorithm can automatically detect the stripe lines. An adaptive matrix obtained by the stripe detection is used to distinguish the stripe pixel and non-stripe pixel and then incorporated into the data fidelity term and the prior term to improve the destriping performance. To settle the nondifferentiability property of the TV and reduce the computational load in the process of destriping process, the split Bregman iteration algorithm is employed. As a result of these steps, we are able to produce high quality destriping results in low computation time. A number of comparative experiments including quantitative and qualitative analysis were performed to verify the superiority of the proposed algorithm.
|