摘要
介绍了在石油勘探中用于识别介形类化石及分析岩芯图象的人工神经元网络方法。讨论了以Hopfield网络为基本模型的加深联想记忆神经网络及加速的反向传播模型(Back-Propagation Model),将其应用于介形类化石的识别获得了很好的效果,并成功地实现了岩芯中的空隙分析。
In this paper, artificial neural net methods are introduced for use in two aspects of petroleum prospecting—the recognition of mussel-shrimp fossils and the analysis of core images. The deeper associative memory based on Hopfield net and the accelerating back-propagation model are discussed. The former is applied to recognize mussel-shrimp fossils with very good results, and the latter is used to successfully realize the analysis of gaps in core. The result of the experiment indicates that the artificial neural net method is superior to the traditional model-recognition method and has vast prospects in image analysis.
关键词
介形类化石
石油勘探
人工神经网络
image processing, pattern recognition, neural networks