This invention proposes an intelligent classification method based on an amphibious in-situ fracture acquisition device, involving technical fields such as geological surveying, oil and gas exploration, and
machine vision. The invention acquires raw digital images containing geological structural surfaces transmitted back by the amphibious acquisition device; inputs a pre-set deep neural
network model, outputting a two-dimensional binary
mask image of the fractures; performs
skeletonization processing on the images to construct a topological central axis network, extracts topological key nodes based on the connection degree of skeleton pixels, and separates spatially independent fracture samples; for each fracture sample, constructs a multi-dimensional feature
space vector, inputs it into an expert
decision tree, and determines the geological origin and evolutionary stage of each fracture sample according to preset threshold conditions, outputting a visualized
classification result. This invention improves the walking and obstacle-crossing performance and
image acquisition stability in complex amphibious flow environments through a composite
chassis design of tracks and retractable vector propellers, combined with a pressure-resistant, waterproof, streamlined, layered cabin.