Video Inpainting And Three Dimentional Sesmic Signal Denoising Based On Surfacelet

The upcoming needs in practice and breakthroughs in theoretical research has pushed forward the development of Signal Processing. In the past decade, the research focus of the approaches of representing signals has changed from Fourier transform with the Orthogonal transforms nature to multi-scale/resolution wavelet transform, then to the current multi-dimensional/scale redundant super wavelet transform. Super wavelet can better display high-singular geometric qualities such as borderline, outline and texture, so it has become a hot research recently. Among them, the Surfacelet has become a sufficient tool of three dimensions processing by capturing the camber singularity of three dimensions. This paper has built its study and discussion around the theory and application of Surfacelet.First, we have introduced the origins, basic principle and nature of Surfacelet. We will make analytical comparison between the DTWT,3-D Curvelet and Surfacelet.To the quality of small scale video damage inpainting, the paper based on the sparse feature of Surfacelet and sparse restructuring theory.It implements a video inpainting algorithm based on the sparse reconstruction of sur facelet. According to different scales in different directions and efficient tree stru cture of three-dimensional sparse characteristics of surfacelet transform, the surfa ce of the video can be captured effectively by the algorithm. Consequently, the o riginal video signal can be sparser in surfacelet transform domain. The sparse obj ective function of the global optimization is constructed and solved using the rela xation algorithm. The experimental results show that, compared with the existing video inpainting algorithm, the new algorithm not only can improve inpainting eff ects of texture and structure of the video with superior inhibition of the virtual m ovement but also not need any the new algorithm not only can improve inpainting effects of texture and structure of the video with superior inhibition of the virtual movement but also not need any the new algorithm not only can improve inpainti ng effects of texture and structure of the video with superior inhibition of the virt ual movement but also not need any complex pretreatment such as segmentation, edge detection and so on.Finally,according to the three high requirement of prestack denoising in low SNR enviroment and to solve the contradiction of two-dimensional method based on three-dimentional sesmic signals, this paper implements the sesmic signal random denoising algorithm based on Surfacelet. Under the sparse feature of Surfacelet transformation, This algorithm removes correlation of effective signal and noise in transform domain and estimate the weights and multi-dimensional criterion on each subbands of transform domain coefficient through the Monte Carlo method to realize inverse transform of sesmic signals. Compared with the current algorithm, its effect is better and preserves the effective signals better. What’s more, it works remarkably better when processing low SNR data.