Seismic Random Noise Remove Based On The Bandelet Transform

For the seismic data collected in the field usually needs to be explain onlyafter data processing. And the seismic data processing has three high requirements,that is, high signal-to-noise ratio, high resolution and high fidelity. Thereforesuppress the noise in seismic sources is an important sector of data processing.Noise in seismic signal has been divided into two categories, nonrandom andrandom noise. the random noise in seismic records randomly appear and no unifiedrule, it is more difficult to be removed. Now to suppress the random noise inseismic data becomes a research hotspot. This paper mainly studies seismicrandom noise removal method based on the Bandelet transform.At first, this paper briefly describes several kinds of seismic random noisesuppression method that are commonly used, and illustrate their advantages anddefect; then list several super wavelet analysis theory, explained their thinkingtheory, implementation process and respective advantages also the problems theyfaced; then elaborated two generations Bandelet transform theory and its algorithm;Then create a dictionary denoising model based on the second Bandelet transform,and point out choosing the optimal basis from the second generation Bandeletbasis dictionary can achieve the optimal approximation of the original image,compared to the second generation Bandelet transform it simplify the operationeffect, the denoising algorithm achieved by combining the model selection theoryand Bandelet dictionaries can effectively remove the Gaussian white noise in theimage; Finally, this method is applied to two seismic image denoising, throughcompared with the two-dimensional wavelet threshold denoising verify thesuperiority of the proposed method.