The research and application of mathematical morphology in seismic events edge detection and machine vision

Abstract

The event represents the formation interface with different lithology, namely sediment interface. It can also represent the geochronologic isochronous stratigraphic interfaces. This is the basis of the seismic interpretation, including the research of sequence stratigraphy, reservoir prediction and characterization. Events picking play a decisive role in the determination of the reflection interface. This paper is a research on the events extraction from pre/post stack reflection seismic data. The mathematical morphology is widely used in denoising, feature extraction, edge detection and other fields. In this paper, we proposed a mathematical morphology algorithm that is suitable for the seismic events picking. Based on cognitive computing, the mathematical morphology was applied to the SSPA section to reduce the noise, where eight structure elements with the size of 5 × 5 and weighing fusion based on the theory of entropy were proposed to improve the accuracy and computational efficiency. The proposed structure elements can approximate the shape of the events which are hyperbolic curve in pre-stack seismic data, VSP data and the crosswell seismic data. The event curves obtained by applying the proposed method to the synthetic layered model and field record is continuous and correspond to the events in the original data. The results indicate the high accuracy and efficiency of the proposed method.