Automatic Tracking Horizons Method Based On Structure Tensor And Its Application

Accurate picking of seismic horizons plays an important role in processing and interpreting the seismic exploration data.Accurate horizon information can provide better constraints and guidance for seismic data processing and interpretation,making the results more refined and reliable.To the literature,a variety of auto-tracking technologies have been developed,such as methods based on correlation,dip,and neural networks.However,as the difficulty of exploration increases and the processing objects more and more tend to pre-stack datasets,the data volume increases sharply and there are problems such as more noise,lower signal-to-noise ratio,false frequency and cross-interaction.For this reason,the study of phase tracking technology based on tilt angle constraint has been proposed.Based on the post-stack and pre-stack data,a set of stable,efficient,and adaptable seismic event automatic tracking technology was developed and studied.The use of multi-shot tracking in front multi-shots.The main research contents and achievements of the paper are as follows:1.A structural tensor algorithm based on data envelopment was developed to overcome the problem of conventional structure tensor to some extent,and to improve the efficiency and reliability of dip angle information estimation.There are phenomena such as noise,aliasing and cross-interaction of the seismic data.Bilateral filtering and other processing are performed on the data to enhance its edge features and reduce random noise;the data is then envelope-extracted,and the structural tensor algorithm is used to estimate the dip angle information based on the envelope data.Envelope data can effectively highlight data low-frequency information and weaken the effect of aliasing on the angle.2.A phase search and tracking algorithm based on tilt angle constraint is formed,which has strong adaptability and stability to post-stack and pre-stack data.Select a seed point on the target event and automatically correct it as a peak(trough)point.With the inclination information,the next spatial point of the events is obtained.At this point,a certain time window is selected to perform feature extraction such as similarity on the points within the window range,and the best matching point is automatically obtained as the final layer site.Tilt constraint can effectively reduce the time window range,thereby improving the calculation efficiency.At the same time,the addition of constraint optimization conditions makes the tracking results more credible and reliable.Compared with the waveform-similar and coherent auto-tracking techniques,the poststack and pre-stack shots are compared to verify the stability and adaptability of the method on the auto-tracking events.3.A multi-gun multi-layer automatic tracking strategy was designed for prestacked guns,and the tracked travel time information was used for tomographic inversion,which improved the reliability of travel time tracking.Firstly,several seed points will be given on the near-offset profile of the target event,the events on the offset will be automatically tracked,and the tracking results will be used as seed points for all single shot sets.Finally,based on these seed points,the event tracking of all single shots is automatically implemented.This makes tracking more intelligent and reduces labor costs.Comparing the tracking results with the theoretical results,the reliability of the tracking results was verified.The information of the reflected wave travel time of all the acquired shots was used for the pre-stack reflection wave travel time tomographic inversion,and the underground velocity parameters were estimated.