Deep learning-based inland salt lake artemia strip remote sensing extraction method

Abstract

The invention relates to an inland salt lake artemia strip remote sensing extraction method based on deep learning. The method comprises the following steps: firstly, preprocessing obtained remote sensing data to obtain surface reflectance data, determining a water body range, preliminarily obtaining artemia-water body data, then selecting typical data, generating a sample through cutting and augmentation, establishing an artemia-water body data set, and then constructing and training an artemia extraction deep learning model to obtain an artemia extraction deep learning model; and performing precision and robustness evaluation on the trained model, enriching samples through data simulation, further generalizing the application range of the model, and finally extracting artemia strips by using the generalized model. According to the method, the salt lake artemia strip is identified and extracted by using the deep learning method, the threshold problem does not need to be considered, automatic processing can be basically realized, the artemia strip can be more accurately obtained, and the method has a business popularization prospect.

Description

technical field [0001] The invention belongs to the technical field of deep learning semantic segmentation and remote sensing image processing, and in particular relates to a remote sensing extraction method for inland salt lake Artemia strips based on deep learning. Background technique [0002] Artemia is a small crustacean that lives in high salinity waters and has very important economic and ecological value. Artemia is rich in protein, fat and easy to preserve and circulate, making it not only a live bait widely used in aquaculture at home and abroad, but also an important component of carbon flux and biological chain in salt lakes. According to statistics, the global annual output of Artemia is about 3000-4000 tons (dry mass of finished product), involving RMB 2.5 billion, so Artemia is also known as "soft gold". Artemia resources are widely distributed all over the world, and there are about 350 Artemia production areas. Among them, the amount of artemia resources i...

AI Technical Summary

Problems solved by technology

At the same time, deep learning, as a new type of algorithm capable of interpreting a large amount of data and with strong adaptability, can make the extracted features more robust. It has been successfully applied to image classification, change detection, target recognition, cloud shadow removal, etc. However, few scholars have carried out deep learning identification and extraction of Artemia strips

Images