Abstract: Traditional seismic data interpretation or velocity analysis usually depends on manual identification and picking up, which not only has heavy workload, but also has very low efficiency. Therefore, the industry began to use a variety of algorithms are used in production to automatically identify and pick up seismic events, but these algorithms have many defects and low accuracy. Seismic profiles can be regarded as the convolution of seismic wavelet and reflection coefficient. The existence of wavelet and noise makes it difficult to pick up profiles automatically. Through feature extraction of seismic wavelet and sparse expression of seismic profiles, the influence of wavelet and noise on automatic pickup is reduced, and the number of data sampling points is reduced to improve the calculation efficiency. By introducing vector distance and combining with dynamic waveform matching algorithm to calculate the minimum distance of characteristic vector data, the automatic tracking of events. The validity and anti-noise ability of the method proved by the test of theoretical data and the validity of the method in this paper have been proved by the automatic pickup of the actual data in an eastern exploration area.

Key words: seismic wavelet, waveform matching, wavelet feature vectors, automatic pick-up, wavelet