Pattern recognition in electrical images - digital image analysis with special focus on Gas Hydrate Leg 204

点击翻译按钮获取中文摘要

Electrical images are widely used in oil industry for borehole inspection and reservoir characterisation. They consist of artifically colour-coded micro-resistivity measurements on the borehole wall. In ODP these are commonly used to study the internal structure of the oceanic crust. At present, the reconstruction of lithology from electrical images is purely based on visual inspection and subjective interpretation. Our objectives are to (1) develop methods for an object- and texture-based pattern recognition of electrical images; (2) develop methods for an automatic classification of rocks; (3) apply this methodology to study the occurrence and characteristics of gas hydrates at ODP Leg 204. Avaible methods from seismic data interpretation and biomedical imaging will be adapted for use with resistive images. In particular, we will implement an algorithm for automatic object identification for a study of the internal rock morphology with respect to sediments drilled at ODP Leg 204. This way, resistivity anomalies caused by gas hydrates will localised them in electrical images. Their proportion, size, and shape will be quantified in order to analyse anisotropy and spatial heterogeneity of identified gas hydrates. Finally, we will correlate the 1D information from all available boreholes and interpolate them into a 2D map of the gas hydrates distribution at Leg 204.