Remote-sensing image semi-supervision classification method based on customized step-size learning

Abstract

The invention relates to a remote-sensing image semi-supervision classification method based on customized step-size learning. The remote-sensing image semi-supervision classification method comprises the following steps of: pre-processing remote-sensing images, and obtaining marked samples of all kinds of surface features; selecting some or all of unmarked samples, constructing a sparse graph by combining the selected unmarked samples with all of the marked samples, and carrying out initial category calibration on the unmarked samples, so that expanding the number of the marked samples; based on a customized step-size learning algorithm, selecting or rejecting category information with initially-marked samples after the expansion; and selecting a supervision classifier to carry out pixel one-by-one classification on the remote-sensing images. According to the invention, the number of training samples is expanded, and the purpose of reducing falsely-calibrated samples in the training samples is achieved.

Description

technical field [0001] The invention relates to a remote sensing image semi-supervised classification method for self-determined step length learning. Background technique [0002] Due to its high efficiency, economy, and large-scale synchronization, remote sensing data has become one of the important technical means for monitoring, planning, and management of the earth's resources and environment, and has been widely used regionally and even globally. The main principle of remote sensing image classification is to identify the attribute category of ground objects according to the characteristics of different spectrum, texture and spatial geometric characteristics in remote sensing images. At present, there are two main methods of remote sensing image classification, visual interpretation and computer classification. Among them, computer automatic classification has the advantages of high efficiency and objectivity, and is an effective method for large-scale and repetitive ...

AI Technical Summary

Problems solved by technology

However, the acquisition of training samples by visual interpretation requires high professional quality of interpretation experts, and the degree of subjectivity is relatively large, so there may be some errors when selecting a large number of samples; on the other hand, obtaining training samples by field investigation Although the samples can obtain high-precision training samples, the work efficiency is low, and the cost of field surveys is high. Generally, only a small number of training samples are suitable for selection

The above shows that in the computer classification of remote sensing images, the acquisition of training samples will face two problems: (1) or the training sample data has certain errors or noise; (2) or only a small amount of high-precision samples can be obtained

However, semi-supervised learning methods based on self-learning or graph methods have their own limitations. For example, the recursive fitting process based on self-learning classifiers is easy to cause the cumulative propagation of errors; information limitation

Images