Shale large-vision-field image classification method based on machine learning

Abstract

The invention discloses a shale large-view-field image classification method based on machine learning, and the method reduces the calculation pressure of a computer through the cutting of a large-view-field image to generate sub-images, and is higher in processing efficiency. In order to realize accurate extraction of image features of all pores in each sub-image for subsequent classification, typical features and texture features of the pores are considered at the same time, that is, a first matrix representing the typical features of all the pores in each sub-image is determined through a machine learning algorithm; then determining a second matrix representing texture features of all pores in each sub-graph; coupling the first matrix and the second matrix of each sub-image to obtain a feature matrix representing image features of all pores in the large-view-field image; and according to the feature matrix, through a clustering analysis algorithm, all the pores in the large-view-field image and the category of each pore can be accurately and effectively determined, and accurate identification of all the pores of different types is ensured.

Description

Technical field [0001] The present invention relates to the field of image classification, in particular to a shale large field of view image classification method based on machine learning, systems and apparatus. Background [0002] As a supplement to conventional oil and gas reservoirs, shale oil and gas reservoirs have huge reserves and are an important part of oil and gas reservoirs. The results show that the microstructure of shale oil and gas reservoirs shows typical heterogeneity, which is manifested in the multi-scale spread of pore sizes, and the types of pores in them are very complex and diverse, and microcracks, dissolution pores, intergrandular pores, intragrainular pores, organic matter and organic matter pores are developed. [0003] In order to achieve the analysis of the structural characteristics of pores in shale reservoirs, it is necessary to first effectively separate the different types of pores in shale reservoirs. The main method used in the prior art for ...

AI Technical Summary

Problems solved by technology

[0004] However, as mentioned above, since shale oil and gas reservoirs contain many types of pores, the gray values ​​of some different types of pores are very close, which makes it difficult to effectively determine the threshold in this method, which leads to Using this method can easily lead to misidentification of pores of different types but with close gray values; in addition, this method is also difficult to identify some of the smaller nanoscale pores; and the large field of view image has a huge data volume. In the method, the entire large field of view image is directly processed, which makes the requirements for the hard disk storage space, memory space and CPU processing efficiency of the computer very high during the entire processing process, and also limits the application of the method to a certain extent. use

Images