Waveform classification method based on strong background interference separated

Seismic waveform classification is a critical technology to identify the underground reservoir. Various methods have been proposed including supervised or unsupervised learning to describe the waveform classification. However, in some cases, the seismic response of the reservoir geologic body is seriously affected by the strong reflection (background interference) of the overlying stratum or the underlying strata in field seismic data, so that it is often difficult to identify the reservoir geologic body by applying the waveform classification method to original data directly. In this paper, a waveform classification method based on removing the strong background interference is proposed. Firstly, morphological component analysis (MCA) is applied to separate the background interference from the original data, then the SOM classification method is used to the data after removing the background interference. The proposed method is applied to the real seismic slice with channel structure. The classification results indicate that waveform classification based on removing strong background interference effectively delineate boundary of channels in seismic data, which is useful to enhance the accuracy of reservior interpretation.