Coal mine underground image semantic segmentation method

Abstract

The invention discloses a coal mine underground image semantic segmentation method, and belongs to the field of computer vision. The method comprises the following steps: firstly, preprocessing an acquired underground scene image to generate a data set; then constructing a feature extraction network taking ResNet-101 as a skeleton, and inputting and using images with different scales in each stage of the network to enhance the extracted features; constructing a fusion attention module to fuse the features of each stage, enhancing global information by using a global attention module, and obtaining a remote dependency relationship; and finally, inputting the obtained features into a classifier to generate a semantic graph, and completing semantic segmentation of the image. According to the method, the calculation amount and complexity are greatly reduced, an attention mechanism is adopted for a complex scene, semantic information of a target area is highlighted, and the image segmentation effect is improved.

Description

technical field [0001] The invention relates to an image semantic segmentation method, especially a coal mine image semantic segmentation method suitable for underground coal mines, and belongs to the field of computer vision. Background technique [0002] It is of great significance to study the structural characteristics and restoration methods of the visual scene of underground tunnels for the square structural analysis of complex underground scenes. Aiming at the complex environment caused by strong direct light, dark light, dust, water mist and smog in the coal mine underground scene, the underground feature analysis method is studied. Traditional image analysis methods mainly include: Lucas-Kanade algorithm, matching method, energy method, phase method, etc. In complex underground scenes, sudden changes in lighting conditions will cause brightness to change at any time, and small movements in narrow scenes may also will cause a large change in position. Traditional m...

AI Technical Summary

Problems solved by technology

However, the research on structured scenes mostly focuses on the indoor scenes of buildings, and there is no related work on the scene of underground tunnels in coal mines.

Images