No abstract

No abstract

Denoisers trained with synthetic noises often fail to cope with the diversity of real noises, giving way to methods that can adapt to unknown noise without noise modeling or ground truth. Previous image-based method leads to noise overfitting if directly applied to temporal denoising, and has inadequate temporal information management especially in terms of occlusion and lighting variation. In this paper, we propose a general framework for temporal denoising that successfully addresses these challenges. A novel twin sampler assembles training data by decoupling inputs from targets without altering semantics, which not only solves the noise overfitting problem, but also generates better occlusion masks by checking optical flow consistency. Lighting variation is quantified based on the local similarity of aligned frames. Our method consistently outperforms the prior art by 0.6-3.2dB PSNR on multiple noises, datasets and network architectures. Stateof-the-art results on reducing model-blind video noises are achieved.

The traditional model of grey nearness degree of incidence contains some inherent limitations in the calculation of data sequences. It does not consider the impacts of certain data on degree of incidence when there are significant differences in orders of magnitude between adjacent data in the same sequence, and big errors may occur in the calculation of some special oscillation sequences. In response to these problems, we propose a new improved method, which uses the characteristics of the model of grey nearness degree of incidence and introduces a neural network algorithm to define a grey neural network-nearness degree of incidence. Thereby, a model of nearness degree of incidence is established based on grey neural network. Then we apply a new model to the field of data mining. According to the clustering algorithm, we take all the degrees of incidence as the variables of the distance metric function, and use the clustering algorithm of data mining for data analysis. Finally, through simulation experiments, we verify the effectiveness of the clustering algorithm under the new distance metric definition. The experimental results show that, compared with other methods, the computational outcomes of the improved model are more consistent with the actual situation. The cluster algorithm with the model used can deliver results that have a high accuracy, so the new model can be applicated in a wide range of fields.INDEX TERMS Grey neural network, nearness degree of incidence, data mining, clustering algorithm.

Based on the theories of fuzzy set and fuzzy conversion, the method of fuzzy comprehensive appraisal is a decision-making process which combines qualitative analysis and quantitative analysis and can be used to forecast risk of electric power engineering projects. Using the method of AHP to establish risk early-warning indicators system and method of fuzzy comprehensive appraisal to establish risk early-warning model, the paper constitutes risk early-warning system of electric power engineering projects. A case from western China is applied to prove the validity of the risk early-warning system.