Two-stage saline-alkali land monitoring method based on SAR polarization decomposition and convolutional neural network

Abstract

The invention belongs to the field of land use classification, and relates to a two-stage saline-alkali land monitoring method based on SAR polarization decomposition and convolutional neural network, specifically: selecting an operation area, obtaining GPS point data of remote sensing image R2SLC and vector, and closing the point data generation polygons; visually interpret more land use vector polygons, and rasterize the vector polygons; divide sub-areas, calculate 10 features in each sub-area, and combine them to obtain multi-polarization and multi-feature datasets; in each sub-area The region of interest ROI extraction operation is performed to obtain a multi-polarization and multi-feature data set for each pixel; the data over-fitting technique is used to obtain the first-stage equalized data set; 75% of the balanced data is used for training, and 25% is used for training. In the verification; in the second stage, the CNN neural network structure is designed and optimized; the unknown image is identified pixel by pixel with the optimized parameter model, and the land type of saline-alkali land is extracted. The method of the invention can effectively improve the classification accuracy of samples to be classified with unbalanced type distribution.

Description

technical field [0001] The invention relates to the field of saline-alkali monitoring, in particular to a two-stage saline-alkali monitoring method based on SAR polarization decomposition and convolutional neural network. Background technique [0002] Salinization and soil secondary salinization are one of the main types and causes of land degradation, which seriously affects the ecological environment and even threatens food security. Therefore, the rapid and accurate monitoring of saline-alkali land is of great significance to timely obtain the expansion change information of saline soil, understand the ecological environment changes, and then formulate reasonable saline-alkali land management and environmental improvement programs. The rapid development of remote sensing technology makes it possible to monitor large areas of saline-alkali land based on remote sensing technology. [0003] The research on saline-alkali land monitoring based on remote sensing technology eme...

AI Technical Summary

Problems solved by technology

[0007] In order to solve the following problems in the existing technology: there are obstacles in the acquisition of optical remote sensing saline-alkali land distribution data, the accuracy of SAR remote sensing data is not enough, the reference ground object labels are few and unevenly distributed, and the accuracy is low in large-scale monitoring

Images