METHOD FOR MODELING A HETEROGENEOUS UNDERGROUND ENVIRONMENT FROM MULTIPOINT STATISTICS

摘要

- construction process of an image representing the distribution of apropertyrepresentative of a physical block underground area.we constructed a learning image representative of astructurethe specific geometric property. then, from the firstimagelearning, learning, we constructed images representative of thethe distributions ofa number of ancillary properties. on the basis of all the images.in learning, weto determine a probability distribution of the category and property lawthe probability of eachfor a given pixel auxiliary property as a function of the values of pixelsthe neighbors. fromthese images represent the distribution and properties ofancillary physicalin the area, and for each pixel of the image to construct is a lawthe likelihood ofthe ownership category. finally, the value of the property.in 1982,in a random sample from this probability distribution calculated in this way.applications include the exploitation of oil deposits.