A turbulence image restoration approach for visual inspection of nuclear power plants

摘要

Visual inspection is a common procedure during outages of nuclear power plants. For the underwater visual inspection of the nuclear plant reactor after fuel reloading, the water turbulence generated by nuclear fuel assemblies can seriously degrade the quality of video. Online image restoration is required in order to meet the need of minimizing the duration of visual inspection. The paper proposes a new method to solve the image degradation and to realize online image restoration when visual inspection. First, the image degradation model is founded. In the model that water turbulence weakly satisfies a Laplacian distribution, it is demonstrated in the paper that the geometric distortion can be removed and a corrected image can be recovered. Then the image is partitioned into small patches which have partly overlapping between adjacent areas. Image quality assessment is used to make phases of image patches homomorphism. Image quality index method is used to image quality measurement in practice. Moreover, the phase average patches combine into a new image. At last the wiener filter is used to estimate the image which would have been observed without turbulence. The experimental result shows that the method can well realize restoration of images affected by turbulence and obtain a satisfactory effect, which can help the operator to carry out the visual inspection which underwater camera is used to achieve more accurate operation information of the fuel reloading.