Search for three-dimensional images using the contour comparison method in problems of geological reservoir modeling

Methods for comparing three-dimensional images in problems of geological modeling of a reservoir are studied in order to improve their quality. The proposed method combines such advantages as global representation of shape, invariance to transformations, noise resistance, and computational efficiency. An approach based on the use of image moments for analyzing geological data in problems of geological modeling of a reservoir is developed and substantiated. The problem of comparing three-dimensional images is solved using the mathematical apparatus of algebraic invariants. The essence of the proposed approach is to calculate the moments of three-dimensional images for comparing the invariants of the contours of the standard and sample. The result of the comparison is a quantitative metric of the conformity of the compared contour to the desired standard. Developed software tools were built into the overall modeling and analysis pipeline of the Gempy package. The method showed satisfactory results on the test geological model. The recognition accuracy allows using the developed tools as a recommender system. The possibility of using the proposed method to search for a given object in a geological model and limited applicability for verifying a simplified model during iterative calculations are confirmed. The proposed method is compared with the Hausdorff metric, the cross-section comparison method, the SIFT and SURF methods, and the grid interpolation method. It is shown that the proposed method can be expanded to more complex geological formations for working with heterogeneous structures. The developed tools can be integrated with geological modeling systems, database management systems, and analytical platforms.