資源開発技術

[3510] 画像処理技術を用いた反射法地震探査データのS/N比改善

^○Sajjad AMANI¹, 三ケ田均¹, 武川順一¹ (1.京都大学大学院)

司会:児玉淳一(北海道大学)

キーワード：地震探査, 画像処理技術, S/N比

In order to interpret the seismic data precisely, it is very important to increase the signal to noise ratio (SNR). These data are usually contaminated with both random and coherent noises. In this study, we introduce a new nonlocal transform-domain method called “3 Dimensional Block Matching (3DBM)” which can be used in both pre-stacked and post-stacked seismic data to attenuate the random noises. This method is a combination of nonlocal and transform-domain denoising methods which uses some degree of redundancy within the data due to the repetition of geological structures such as the curvature of an anticline or the lineation of a fault. We compare 3DBM with the state-of-the-art curvelet-based seismic denoising method (CT) which is one of the most powerful and fast methods in this area. We show that the proposed method outperforms the CT both visually and in the sense of signal/noise. In addition, we demonstrate that the 3DBM is significantly faster than CT.