Neural network for robust pattern recognition in real seismic data

Kou-Yuan Huang*

*此作品的通信作者

研究成果: Paper同行評審

摘要

The multilayer perceptron of neural network is trained as the classifier and is applied to the robust recognition of seismic patterns. Three classes of seismic patterns are analyzed in the experiment: bright spot, pinch-out, and horizontal reflection patterns. Seven moments that are invariant to translation, rotation, and scale, are employed for feature generation of each seismic pattern. The training set includes noise-free, low noisy, and misclassified seismic patterns. The testing set includes seismic patterns with various noise levels. The multilayer perceptron is initially trained with the training set of noise-free and low noisy seismic patterns. After convergence of training, the network is applied to the classification of the testing set of noisy seismic patterns. Some misclassified testing seismic patterns with higher noise level are added to the training set for retraining. Repeat the classification and the training through several steps. The retraining can significantly improve the robustness of the network in higher steps. Finally we apply the network at each training step to the real seismic data at Mississippi canyon, the bright spot pattern can be detected. From experiments, the multilayer perceptron is shown to have the capability of robust recognition of seismic patterns and the recognition results are encouraged to the seismic interpretation.

原文English
DOIs
出版狀態Published - 1999
事件1999 Society of Exploration Geophysicists Annual Meeting, SEG 1999 - Houston, United States
持續時間: 31 10月 19995 11月 1999

Conference

Conference1999 Society of Exploration Geophysicists Annual Meeting, SEG 1999
國家/地區United States
城市Houston
期間31/10/995/11/99

指紋

深入研究「Neural network for robust pattern recognition in real seismic data」主題。共同形成了獨特的指紋。

引用此