With the development of society, China has gradually strengthened the research on address work. In the existing seismic layer tracking methods, the traditional formation exploration method has a large demand for manpower, which is easy to have a great impact on the efficiency of seismic exploration. Therefore, it is necessary to study the automatic horizon tracking method. We carry out the following visual processing on the seismic layer data:(1) the raw data of the three-dimensional seismic layer are processed by wavelet de-noising record, and the effective signal is retained as much as possible while the noise is removed to the maximum extent. (2) The correlation-based seismic layer automatic tracking model is established, and the BP neural network system can continuously train the deviation between the real data and the experimental data, so as to achieve the model Del fitting, so the BP neural network is used to track the correlation-based seismic data. (3) To establish the seismic automatic tracking model based on waveform characteristics, this paper chooses to build the GA-BP neural network automatic tracking model, that is, the application of genetic algorithm to optimize the BP neural network, to obtain the best weight, so that the neural network stable training results. (4) Establish a 3D horizon automatic tracking model based on correlation and waveform features. The three-dimensional automatic tracking model is established from two angles, and the horizon tracking and fault data are identified and analyzed.