Image Recognition–Based Identification of Multifractal Features of Faults

Abstract

Geologists have made several advances in applying multifractal theory in geology; however, some questions such as a large statistical workload and low efficiency remain unanswered. Thus, this study proposes an image recognition–based method for calculating fault multifractality. First, grayscale processing and binarization of the fault distribution map were performed. The image was then gridded, and the grids were numbered. Subsequently, computer image recognition technology was used to count the number of faults in each grid as a replacement for manual counting. Finally, the fractal dimensions of the faults were calculated using a multifractal box-counting algorithm. This method was successfully applied to fracture studies of the Maokou Formation in southeast Sichuan. Compared to the conventional approach, the proposed method demonstrated considerably improved work efficiency and accuracy. The results showed that the faults in the study area exhibited good statistical self-similarity in the scale range, indicating fractal characteristics. The fractal dimensions of faults with different orientations and the planar distribution of the fractal dimension contours indicate tectonic stages and stress magnitude in the study area. The results indicate that the tectonic setting of southeast Sichuan was formed primarily during the Indosinian, Yanshanian, and Himalayan periods. From the Indosinian to the early Yanshanian periods, NE-trending faults with relatively large fractal dimensions developed under NW–SE compressional tectonic stress. From the Late Yanshanian to Early Himalayan, EW-trending faults were formed by relatively weak N–S compressional stress and had the lowest fractal dimensions. The NW-trending faults formed by intense NE–SW compressional tectonic stress in the Late Himalayan region had the highest fractal dimensions. To promote oil and gas migration and ensure that faults do not destroy the caprock, oil and gas reservoirs must be in a relatively mild tectonic environment. Thus, the fractal dimensions of faults in favorable areas should be neither too high nor too low. The relationship between the fractal dimensions of faults and well test results in southeast Sichuan indicates that the region along the wells “ls1–xia14–guan3” (with fractal dimensions of 1.49–1.57) in the study area is a relatively favorable region for oil and gas preservation.