Automatic Multihorizons Recognition for Seismic Data Based on Kalman Filter Tracker

Abstract

Horizon recognition is an important but difficult task for a seismic interpreter. A novel automatic recognition method based on the Kalman filter tracker is proposed. According to the properties of the seismic reflection profile, a pair of new linear state and measurement equations is built. Combined with three aided steps including the threshold detection, which finds the candidate measurements, the logic method, which automatically starts a new possible horizon, and the probabilistic data association, which judges the probability of each candidate measurement belonging to each of all the existing horizons, the Kalman filter tracker can reduce the noise involved in the candidate measurements and recognize the multihorizons automatically and simultaneously. Unlike the common picking methods, the proposed method can get the detailed time-space location data for each identified horizon, which is very important for the followed processing and interpretation. The experimental results on the synthetic and real data show that the proposed method can not only track the faint horizons, but also improve the continuity of horizons. The proposed method outperforms the common Canny edge detector.