Intelligent seismic horizon picking based on Gabor wavelet transform of seismic attribute

Seismic horizon picking is the foundation of structural interpretation and seismic inversion. Current intelligent seismic horizon picking methods have not utilized the inherent characteristics of seismic data and their attibutes to supervise the crucial step. Therefore, we propose a new intelligent seismic horizon picking method based on seismic signal feature extraction. This method does not use the peaks or troughs of the seismic signal and instead focuses on the energy change in the seismic signal at the stratum interface, extracting the characteristics of the energy change to horizon picking. We apply sliding window preprocessing to the seismic signal to convert the seismic signal to a spectrum analysis energy view by performing operations such as short-time Fourier transforms (STFTs) on the preprocessed results. On this basis, the local texture information of the spectrum analysis view can be processed by the Gabor wavelet transform to obtain the Gabor attribute features of the seismic signal and seismic Gabor attribute solves the problem of single feature of amplitude data. Finally, the support vector machine, random forest, extreme gradient boosting models and deep residual shrinkage network are used to horizon picking. The results show that accuracy and root mean square error of horizon picking are satisfactory.