Semantic segmentation method and system for space science experiment data

Abstract

The invention discloses a semantic segmentation method and system for space science experiment data, and relates to the field of image processing. The method comprises the following steps: simulatinga space science experiment on the ground, acquiring ground experiment data, performing semantic segmentation annotation on the ground experiment data, and dividing the annotated ground experiment datainto a training set and a verification set; training a pre-adjusted UNet semantic segmentation network according to the training set and the verification set to obtain a semantic segmentation model;and obtaining space science experiment data, and inputting the space science experiment data into the semantic segmentation model to obtain a semantic segmentation result of the space science experiment data. According to the method, the problems of few space science experiment samples and non-uniform distribution are solved, accurate segmentation of space science experiment objects is realized, and the problem that real-time accurate semantic segmentation of small samples and small targets is difficult is solved.

Description

technical field [0001] The invention relates to the field of image processing, in particular to a semantic segmentation method and system for space science experiment data. Background technique [0002] Semantic segmentation provides rich semantic and visual information by segmenting regions of interest, and is widely used. At present, the traditional image segmentation methods mainly include threshold method, pixel clustering method, edge detection method, region generation method, image cutting and other methods. However, the existing semantic segmentation methods based on deep learning are all based on daily scenes or medical image datasets, and the traditional image segmentation methods need to control a lot of thresholds, and the applicable environment is also very limited, and cannot be well adapted to lighting, In situations such as background transformation, the robustness is poor, and its network structure is not suitable for target segmentation in space science ex...

AI Technical Summary

Problems solved by technology

[0005] The technical problem to be solved by the present invention is to provide a semantic segmentation method and system for space science experimental data based on a deep learning semantic segmentation model for the difficult problem of real-time and accurate semantic segmentation of small samples and small targets

Images