Rock classification method, terminal equipment and storage medium

Abstract

The invention relates to a rock classification method, terminal equipment and a storage medium, and the method comprises the steps: S1, collecting different types of texture images to form a first training set, and collecting different types of rock images to form a second training set; s2, constructing a classification model based on the residual block stacking network, adopting a transfer learning method, setting a first training set as a source domain of transfer learning and a second training set as a target domain of transfer learning, training the model under the condition that only parameters of a full connection layer in the classification model are changed, and taking the trained classification model as a rock classification model; and S3, performing type identification on the to-be-identified rock image through the rock classification model to obtain the type of the rock corresponding to the to-be-identified rock image. According to the method, over-fitting or under-fitting of the model is prevented through transfer learning, and the rock classification recognition effect is improved.

Description

technical field [0001] The invention relates to the field of rock classification, in particular to a rock classification method, a terminal device and a storage medium. Background technique [0002] There are many types of large rocks in nature, and more than 3,000 kinds have been discovered by humans so far. The identification of rock lithology is the basis for studying the characteristics of geological reservoirs, calculating reserves and geological modeling, especially in the exploration of mineral resources, lithology identification also plays an important role. The intelligent identification of rocks in the target area can help determine the layout and quantity of different rocks, and can provide specific geological information for the description of regional characteristics. The improvement of rock identification efficiency means that the efficiency of geological exploration work has been improved. Therefore, how to quickly identify rock lithology is an issue that req...

AI Technical Summary

Problems solved by technology

[0003] Traditional rock sample identification methods mainly include gravity and magnetic, well logging, seismic, remote sensing, electromagnetic, geochemical, hand specimen and thin section analysis, etc. Traditional rock sample identification is mainly based on human eye observation, manual operation and empirical classification It is classified by professionals by extracting effective information features from rock images through professional equipment, mainly relying on the experience of the classifier and the sensitivity of the equipment. This method has the disadvantages that it cannot be quantitatively analyzed, the efficiency is low, and it is relatively affected by human subjective factors. Large scale, relatively high professional level and difficulty in popularization

Images